Document o9eD18vz3M6aOB1R52V6R768

2 .0 . MECHANISMS OF FORMATION OF DIOXIN-LIKE COMPOUNDS DURING COMBUSTION OF ORGANIC MATERIALS More than a decade of combustion research has contributed to a general understanding of the central m olecular m echanisms th a t form CDDs and CDFs em itted from combustion sources. Current understanding of the conditions necessary to form CDDs and CDFs w ere prim arily derived from studying full-scale municipal solid w aste incinerators (MSW Is), augmented w ith observations involving the experimental combustion of synthetic fuels and feeds w ith in the laboratory. However, the form ation mechanisms elucidated from these studies are generally relevant to m ost com bustion system s in w hich organic material is burned w ith chlorine. Intensive studies have examined MSWIs from the perspective of identifying the specific form ation mechanism(s) that occur w ithin the system. This knowledge may lead to m ethods th a t prevent the form ation of CDDs and CDFs and their release into the environm ent. A lthough much has been learned from such studies, how to com pletely prevent CDDs/CDFs from forming during the combustion of certain organic materials in the presence of a source of chlorine and oxygen is still unknow n. The w ide variability of organic materials incinerated and therm ally processed by a wide range of com bustion technologies that have variable temperatures, residence tim es, and oxygen requirements adds to this com plex problem. However, central chemical events th a t participate in form ing CDDs and CDFs can be identified by evaluating emission te st results from MSW Is in com bination w ith laboratory experiments. CDD/CDF emissions from com bustion sources can potentially be explained by three principal mechanisms, w hich should not be regarded as being m utually exclusive. The first is th a t CDDs and CDFs are present as contam inants in the com busted organic m aterial, and pass through the furnace and are em itted unaltered. This mechanism is discussed in Section 2.1. The second is th a t CDD/CDFs ultim ately form from the thermal breakdown and m olecular rearrangement of precursor ring com pounds, w hich are defined as chlorinated arom atic hydrocarbons w ith a structural resemblance to the CDD and CDF molecules. Ringed precursors emanated from the com bustion zone are a result of the incomplete oxidation of the constituents of the feed (i.e., products of incomplete com bustion). The precursor mechanism is discussed in Section 2.2. The third mechanism , similar to the second and described in Section 2 .3 , is th a t CDD/CDFs are synthesized de novo. De novo synthesis describes a pathway of forming CDD/CDFs from DRAFT--DO NOT QUOTE OR CITE 2-1 December 2003 heterogeneous reactions on fly ash involving carbon, oxygen, hydrogen, chorine, and a transition m etal ca talyst. W ith these reactions, interm ediate com pounds having an arom atic ring structure are form ed. Studies in this area suggest th a t aliphatic com pounds, w h ich arise as products of incom plete com bustion, may play a critical role in initially form ing simple ring molecules, w h ich later evolve into com plex arom atic precursors. CDD/CDFs are then form ed from the interm ediate com pounds. In both m echanisms (2) and (3), form ation occurs outside the furnace, in the so-called post-com bustion zone. Particulate bound carbon is suggested as the primary reagent in the de novo syntheses pathway. Section 2 .4 gives an overview of studies th a t investigate the role th a t chlorine plays in form ing CDDs and CDFs. A lthough chlorine is an essential com ponent for the form ation of CDD/CDFs in com bustion system s, the empirical evidence indicates th a t for commercial scale incinerators, chlorine levels in feed are not the dom inant controlling facto r for rates of CDD/CDF stack emissions. There are com plexities related to the com bustion process itself, and types of air pollution control equipm ent th a t tend to mask any direct association. Therefore, the chlorine content of fuel and feeds to a com bustion source is not a good indicator of levels of CDDs and CDFs em itted from the stack of the same source. Section 2.5 discusses the generation and form ation of coplanar PCBs. The presence of coplanar PCBs in stack emissions to com bustors is an area in need of further research. Evidence to date suggests th a t PCB em issions are m ostly attributed to PCB contam ination in w aste feeds, and th a t emissions are related to mechanism (1). However, new ly published research has also indicated th a t it is possible to form PCBs in much the same w ay as described in mechanisms (2) and (3) identified in the form ation of CDD/CDFs w ithin the post-combustion zone. Section 2.7 provides a closing summary of the three principal formation m echanisms and the role of chlorine. From the discussion in this chapter, it should be evident th a t no clear distinction exists between the precursor and de novo synthesis m echanisms for form ing CDDs and CDFs. Both form ation pathw ays depend on the evolution of precursors w ithin combustion gases, the interaction of reactive fly ashes, a generally oxidative environment, the presence of a transition metal catalyst, the presence of gaseous chlorine, and a favorable range of tem perature. Temperature of the DRAFT--DO NOT QUOTE OR CITE 2-2 December 2003 com bustion gases (i.e., flue gases) is perhaps the single m ost im portant factor in form ing dioxin-like com pounds. Tem peratures between 2 0 0 and 4 5 0 Celsius (C) are most conducive to form ing CDD/CDFs, w ith maximum form ation occurring at around 3 5 0 C . If tem perature falls outside this range in tem perature, the am ount of CDD/CDFs form ed is minimized. 2.1. MECHANISM 1: CDD/CDF CONTAMINATION IN FUEL AS A SOURCE OF COMBUSTION STACK EMISSIONS The first mechanism involved in the stack emission of CDDs and CDFs is the incom plete destruction of CDD/CDF contam inants present in the fuel or feeds delivered to the com bustion chamber. Not all of these molecules are destroyed by the com bustion system , thus allow ing trace am ounts to be em itted from the stack. M ost w o rk in this area has involved the study of municipal solid w aste incineration (MSWI), where CDDs and CDFs were analytically measured in the raw refuse fed into the incinerator. As discussed in Volum e 2 to this report, CDD/CDFs are ubiquitous in the environm ent (air, w ater, soil) and in foods and paper. Therefore, CDD/CDFs are clearly present in municipal waste. Tosine et al. (1983) first reported detecting trace am ounts 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 4 0 0 to 600 ppt. W ilken et al. (1992) separated the various solid w aste fractions of MSW collected from m unicipalities in Germany and analyzed them for the presence of CDD/CDFs and other organochlorine com pounds. Total CDD/CDFs were detected in all MSW fractions in the follow ing range of concentrations: paper and cardboard = 3.1 to 4 5 .5 ppb; plastics, w ood, leather, and textiles combined = 9.5 to 109.2 ppb; vegetable m atter = 0.9 to 16.9 ppb; and "fine debris" (defined as particles < 8 mm) = 0 .8 to 8 3 .8 ppb. Ozvacic (1985) measured CDD/CDFs in the raw MSW fed into tw o MSWIs operating in Canada. In one MSW I, CDDs were detected in the refuse; concentration ranged from 10 to 30 ppb, but no CDFs were detected (detection lim it: 1 pg/g). In the MSW fed to the second MSW I, CDDs were detected in a range of 75 to 4 39 ppb, and CDFs were detected only in one of three samples at a tota l concentration of 11 ppb. EPA has reported detecting CDD/CDFs in refuse derived fuel (RDF) burned in a large urban MSWI (Federal Register, 1991a). CDDs were detected in 13 MSW samples taken prior to incineration at DRAFT--DO NOT QUOTE OR CITE 2-3 December 2003 concentrations ranging from 1 to 13 ppb; CDFs ranged from not detected to 0 .6 ppb. In these samples, OCDD predominated; the lower chlorinated congeners were not detected. Clement et al. (1988) performed a mass balance involving an input versus o u tput of CDD/CDFs at tw o operational MSWIs in Canada. The mass balance showed th a t the mass of CDDs and CDFs emitted at the stack point was much greater than the mass of CDD/CDFs in the raw MSW fed into the incinerator, and th a t the profiles of the distributions of CDD/CDF congeners were strikingly different. Primarily, higher chlorinated congeners were detected as contam inants in the w aste; whereas, the total array of tetra octa CDD/CDFs could be detected in the stack gases. CDDs/CDFs present in the w aste feeds may account for some fraction of the CDD/CDFs released from the stack. However, mass balance studies have clearly show n th a t more CDD/CDF can be detected dow nstream of the furnace than w h a t is detected in the feed, indicating th a t CDD/CDFs are being synthesized after the feed has been combusted (Commoner et al., 1984, 1985, 1987; Clement et al., 1988; Hay et al., 1986; Environm ent Canada, 1985). M oreover, it is expected th a t the conditions of thermal stress imposed by high tem peratures reached in typical com bustion w ould destroy and reduce the CDDs and CDFs present as contam inants in the w aste feed to levels th a t are 0.0001 to 10 percent of the initial concentration, depending on the performance of the com bustion source and the level of com bustion efficiency. Stehl et al. (1973) demonstrated that the moderate temperature of 8 00 C enhances the decomposition of CDDs at a rate of about 9 9 .9 5 percent, but th a t low er tem peratures result in a higher survival rate. Theoretical modeling has shown that unimolecular destruction of CDDs/CDFs at 9 9 .9 9 percent can occur at the follow ing tem peratures and retention tim es w ithin the combustion zone: 9 7 7 C w ith a retention time of 1 second; 1,000C at a retention tim e of / second; 1 ,2 2 7 C at a retention tim e of 4 m illiseconds; and 1 ,7 2 7 C at a retention tim e of 5 m icroseconds (Schaub and Tsang, 1983). Thus, CDDs and CDFs w ould have to be in parts per m illion concentration in the feed to the com bustor to be found in the part per billion or part per trillion level in the stack gas emission (Shaub and Tsang, 1983). However, it cannot be ruled out is th a t CDDs/CDFs in the w aste or fuel may contribute (up to some percentage) to the overall concentration leaving the stack. This leaves the only other possible explanation for CDD/CDF emissions from high tem perature com bustion of organic m aterial, form ation outside and dow nstream of the DRAFT--DO NOT QUOTE OR CITE 2-4 December 2003 furnace. These studies point to form ation mechanisms other than simple pass through of non-com busted feed contam ination. These form ation mechanisms are discussed and reviewed in the sections w h ich fo llo w . 2 .2 . MECHANISM 2: FORMATION OF CDD/CDFs FROM PRECURSOR COMPOUNDS The second mechanism involves the form ation of CDDs and CDFs from arom atic precursor com pounds in the presence of a chlorine donor. This mechanism has been elucidated from laboratory experim ents involving the com bustion of know n precursors in quartz ampules under starved-air conditions, and in experim ents th a t investigate the role of com bustion fly ash in prom oting the form ation of CDD/CDFs from precursor com pounds. The general reaction in this form ation pathw ay is an interaction between an arom atic precursor compound and chlorine prom oted by a transition metal catalyst on a reactive fly ash surface (Dickson and Karasek, 1987; Liberti and Brocco, 1982). Examples of w ell studied precursor com pounds include chorobenzenes, chlorophenols, phenol, and benzene (Esposito et al., 1980). Examples of diverse chlorine donor com pounds are polyvinyl chloride (PVC), and gaseous hydrogen chloride (HCl). CDD and CDF form ation results from heterogeneous gas-phase reactions involving chlorinated precursor com pounds and a source of chlorine. Chlorophenol and chlorobenzene com pounds are measured in flue gases from MSWIs (Dickson and Karasek, 1987). Precursors are carried from the furnace to the flue duct as products of incom plete com bustion. These com pounds can adsorb on the surface of com bustion fly ash, or entrain in the gas phase w ith in the flue gases. In the post-com bustion region outside the furnace, heterogeneous reactions ensue to form CDD/CDFs. Laboratory experiments involving the controlled combustion of precursor compounds have caused the breakdown of the precursor reagent and the subsequent appearance of CDD/CDFs as products of the reaction. For example, Jansson et al. (1977) produced CDDs through the pyrolysis of w ood chips treated w ith tri-, tetra-, and pentachlorophenol in a bench-scale furnace operated at 5 0 0 -6 0 0 C . Stehl and Lamparski (1977) com busted grass and paper treated w ith the herbicide 2,4 ,5 trichlorophenoxyacetic acid (2,4,5-T) in a bench-scale furnace at 6 0 0 -8 0 0 C and generated ppm v levels of TCDD. Ahling and Lindskog (1982) reported CDD form ation during the com bustion of tri- and tetrachlorophenol form ulations at tem peratures of 5006 0 0 C . Decreases in oxygen during com bustion generally increased the CDD yield. DRAFT--DO NOT QUOTE OR CITE 2-5 December 2003 Ahling and Lindskog (1982) noted th a t adding copper salts to the tetrachlorophenol form ulation significantly enhanced the yield of CDDs. This may have been an early indication of copper's role in catalyzing the condensation of chlorophenol to dioxin. Com bustion of pentachlorophenol (PCP) resulted in lo w yields of CDDs. H o w eve r, w hen PCP w as burned w ith an insufficient supply of oxygen, investigators noted the form ation of tetra- through octa-chlorinated congeners. Buser (1979) generated CDD/CDFs on the order of 0 .0 0 1 -0 .0 8 percent (by w eight) by heating tri-, tetra-, and pentachlorobenzenes at 6 2 0 C in quartz ampules in the presence of oxygen. It was noted th a t chlorophenols form ed as com bustion byproducts; Buser (1979) speculated th a t these were acting as reaction interm ediates in the form ation of CDD/CDFs. Temperature of the com bustion gases is, perhaps, the m ost dom inant facto r in the form ation of CDDs and CDFs from arom atic precursor com pounds (Fangmark et al., 1994; Vogg et al., 1987, 1992; Oberg et a l.,1 9 8 9 ; W eber and Hagenmaier, 1999). Vogg et al. (1987) found th a t form ation probably occurs outside and dow nstream from the com bustion zone of a furnace to a com bustion source in regions where the tem perature of the combustion offgases has cooled w ith in a range of 2 0 0 to 45 0 C . A fte r carefully removing organic contam inants from MSWI fly ash, Vogg et al. (1987) added known concentrations of isotopically labeled CDD/CDFs to the matrix. The MSWI fly ash was then heated for 2 hours in a laboratory furnace at varying temperatures. The treated fly ash was exposed to increasing tem peratures in 5 0 C increm ents in a tem perature range of 1500 to 5 0 0 C . Table 2-1 summarizes these data. Because the relative concentration of CDD/CDFs increased while exposed to varying temperature, it w as concluded th a t the tem perature of the com bustion gas is crucial to prom oting the form ation of CDD/CDFs on the surface of fly ash. W ithin a temperature range of 2 0 0 to 4 5 0 C , the concentration of CDD/CDFs increases to some maxima; outside this range, the concentration diminishes. The region of cooler gas tem perature is often referred to as the "post-com bustion zone." The heat loss may be inherent to the conduction and transfer through the com bustion gas metal ducting system , or related to adsorbing/exchanging heat to w ater in boiler tubes. This region extends from near the exit of the furnace to the point of release of the combustion gases at stack tip. DRAFT--DO NOT QUOTE OR CITE 2-6 December 2003 Fangmark et al. (1994) found th a t CDD/CDFs exhibit a similar dependence on tem perature and residence tim es between 2 6 0 and 4 3 0 C , w ith m axim um form ation occurring around 3 4 0 C . Using a pilot-scale com bustor, Behrooz and A ltw icke r (1996) found the form ation of CDD/CDFs from the precursor 1,2-dichlorobenzene rapidly occurred w ith in the post-com bustion region in a tem perature range of 3 9 0 to 4 0 0 C , w ith residence tim es of only 4-5 seconds. On the other hand, CDD/CDF form ation from 1,2dichlorophenol seemed to require higher tem peratures; still outside the furnace, but likely in the exit to the furnace where gas tem peratures are > 4 0 0 C . Oberg et al. (1989) investigated the role th a t tem perature plays in the form ation kinetics using a full-scale hazardous w aste incinerator operating in Sweden. Oberg et al. (1989) observed th a t m axim um CDD/CDF form ation transpired in the boiler used to extract heat for co-generation of energy. In this investigation, significant increases in total concentration of I-TEQDF occurred between tem peratures of 2 8 0 to 4 0 0 C , and concentrations declined at tem peratures above 4 0 0 C . W eber and Hagenmaier (1999) showed th a t in gas phase reactions chlorophenols react in the presence of oxygen at above 3 4 0 C to form CDDs and CDFs. Phenoxyradicals were form ed w h ich, in turn, caused the form ation of CDDs. Polychlorinated dihydroxybiphenyls were identified as reaction interm ediates in the gas phase dim erization of chlorophenols, and these interm ediates could form PCDFs. Other conditions postulated to regulate the synthesis of CDDs and CDFs from the arom atic precursor compound are adsorption and interaction w ith the reactive surface of com bustion generated fly ash (particulate m atter) entrained in the com bustion plasma, and the presence of a transition metal catalyst (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). The molecular precursor leaves the gas-phase and condenses to the fly ash particle. This places greater emphasis on heterogeneous surface reactions and less emphasis on homogeneous gas-phase reactions. This condition was first postulated by Shaub and Tsang (1983) using therm al-kinetic models based on heats of form ation, adsorption, and desorption. Shaub and Tsang (1983) modeled CDD production from chlorophenols and concluded th a t gas-phase form ation w ith in an incineration system is likely to be of low probability and im portance given the short (i.e., seconds) residence tim e of the com bustion gases. Konduri and A ltw icke r (1994) proposed th a t rate lim iting DRAFT--DO NOT QUOTE OR CITE 2-7 December 2003 factors were the nature and the concentrations of the precursors, the reactivity and availability of the fly ash surface, and the residence tim e in the post-com bustion zone. Dickson and Karasek (1987) investigated fly ash reactivity w ith 13C6-chlorophenol compounds. Several fly ashes from a variety of com bustion fuels were heated at 3 0 0 C in quartz tubes under conditions know n to catalyze the conversion of chlorophenols to CDD/CDFs (i.e., MSW I, and copper sm elter fly ashes). The MSW ashes included a sample from a poorly-operated mass burn refractory incinerator and a sample from a well-operated fluidized bed com bustor. The MSWI fly ashes proved to be the m ost active catalytic m edium, despite sim ilarities w ith respect to specific surface area and average pore diameters. The ash from the refractory MSWI generated about seven tim es more mass of dioxin-like com pounds than the fluidized-bed MSW incinerator. In the MSW ashes, all CDD/CDF congener groups were formed from labeled chlorophenols; however, only trace am ounts of heptachloro- and octachlorodioxin were form ed w ith the copper smelter/refiner. X-ray photoelectron spectroscopy revealed the presence of chlorine adsorbed to the surface of the MSWI fly ashes, but an absence of chlorine sorbed to the copper sm elter fly ash. CDD congener groups were postulated to form from the labeled pentachlorophenol precursors by: (1) firs t form ing octachlorodioxin by the condensation of tw o pentachlorophenol molecules, and (2) form ing other low er chlorinated dioxins through dechlorination of the more highly chlorinated isomers. These steps seemed to proceed by an increased reactivity of the chemisorbed precursor molecule caused by the removal of one or more hydrogen or chlorine atom s along the ring structure (Dickson and Karasek, 1987), an observation consistent w ith the kinetic model of Shaub and Tsang (1983). In related experim ents, Dickson and Karasek (1987) more specifically reported on form ing CDD/CDFs from condensation reactions of chlorophenols on the surface of MSWI fly ash heated in a bench-scale furnace. Their experim ent was designed to m im ic conditions of MSW incineration, to identify the step-w ise chemical reactions involved in converting a precursor com pound into dioxin, and to determine if MSWI fly ash could prom ote these reactions. MSWI fly ash was obtained from facilities in Canada and Japan. The MSWI fly ash w as rinsed w ith solvent to remove any organic constituents prior to initiating the experiment. T w e n ty grams of fly ash were introduced into a bench-scale oven (consisting of a simple flo w -tu b e com bustion apparatus) and heated at 3 4 0 C overnight to desorb DRAFT--DO NOT QUOTE OR CITE 2-8 December 2003 any remaining organic com pounds from the m atrix. 13C12 -labeled pentachlorophenol (PCP) and tw o trichlorophenol isotopes (13C12- 2,3,5-Trichlorophenol and 3,4,5-Trichlorophenol) were added to the surface of the clean fly ash m atrix and placed into the oven for 1 hour at 3 0 0 C . Pure inert nitrogen gas (flow rate of 10 mL/min) was passed through the flo w tube to maintain constant temperatures. Tetra- through octa- CDDs were formed from the labeled pentachlorophenol experiment; over 100 jg /g of total CDDs were produced. The congener pattern was similar to th a t found in MSWI emissions. The 2,3,5-Trichlorophenol experim ent prim arily produced HxCDDs and very small am ounts of tetra- and octa-CDD. The 3,4,5-Trichlorophenol experim ent mainly produced OCDD and 1,2,3,4 ,6,7,8-H pC D D . Dickson and Karasek (1987) proposed th a t CDDs on the fly ash surface may result from chlorophenol undergoing molecular rearrangement or isomerization as a result of dechlorination, dehydrogenation, and trans-chlorination before condensation occurs. These reactions were proposed as controlling the types and am ounts of CDDs th a t are ultim ately form ed. Born et al. (1993) conducted experim ents on the oxidation of chlorophenols w ith fly ash in a quartz tube reactor heated to about 3 0 0 C . The MSWI fly ash mediated the oxidation of chlorophenols to produce carbon dioxide and carbon m onoxide as major products, and polychlorinated benzenes, m onobenzofurans, and nonhalogenated dibenzo-p-dioxins as trace species. Formation of these trace aromatic species occurred after residence times of only 7 - 8 seconds w hich was consistent w ith the later experimental result of Behrooz and A ltw icke r (1995) w hich showed the potential for rapid form ation from a precursor. Milligan and A ltw icke r (1996) fitte d experimental flow -tube reactor data to classical catalytic reaction models to empirically explain the interaction of 2,3,4,6-tetrachlorophenol (as a model precursor) w ith reactive MSWI fly ash during MSW incineration. The precursor w as found to be highly adsorptive on fly ash, w ith a first-order dependence on gas-phase precursor concentration to CDD form ation. Milligan and A ltw icke r (1996) concluded th a t chlorophenol's dependence on gas-phase concentration to form CDD on fly ash reflects the highly heterogeneous nature of the fly ash surface. Moreover the estim ated 6 x 1 0 18 adsorption sites per gram of fly ash suggest the presence of highly energetic sites, w hich may be im portant in the surfacecatalyzed reactions form ing CDDs. An interesting observation of Milligan and A ltw icke r (1996) was that precursor molecules appeared to compete w ith oxygen molecules for the reactive sites; therefore, chlorophenols are expected to adsorb less readily to the fly ash DRAFT--DO NOT QUOTE OR CITE 2-9 December 2003 surface in the presence of oxygen. Experimental evidence suggests th a t condensation to CDD of chlorophenol com pounds via isomerization and the Smiles rearrangement on reactive M SW I fly ash surfaces is a proven pathw ay for form ing dioxins from a precursor com pound (Addink and Olie, 1995). However, no detailed m echanisms have been presented for CDD/CDF form ation from other precursors, such as chlorobenzenes under conditions simulating incineration. A condition to the synthesis of CDD/CDFs from arom atic precursor com pounds is that the presence of a transition metal catalyst promotes the chemical reaction on the surface of fly ash. Copper chloride is a strong catalyst for prom oting surface reactions on particulate m atter to convert arom atic precursor com pounds to chlorinated dioxins and dibenzofurans (Vogg et al., 1987). Copper chloride promotes ring condensation reactions (e.g., chorophenols) on fly ash to form CDD/CDFs (Addink and Olie, 1995) via the Ullman reaction (Born et al., 1993). In the Ullman reaction, copper catalyzes the form ation of diphenyl ethers by the reaction of halogenated benzenes w ith alkali metal phenolates (Born et al., 1993), w ith copper participating in a nucleophilic arom atic substitution reaction. Thus, Born et al. (1993) proposes a similar mechanism in catalyzing the form ation of dioxin-like com pounds. Using the Ullman reaction as a model, Born et al. (1993) proposed that the copper-catalyzed condensation of tw o ortho-substituted chlorophenol molecules form chlorine-free dibenzo-p-dioxins. Vogg et al. (1987) proposed an oxidation reaction pathw ay, giving rise to the form ation of CDDs and CDFs in the post-furnace regions of the incinerator in the follow ing order: (1) hydrogen chloride gas (HCl) is therm olytically derived as a product of the com bustion of heterogeneous fuels containing abundant chlorinated organic chem icals and chlorides; (2) oxidation of HCl, w ith copper chloride (CuCl2) as a catalyst, yields free gaseous chlorine via the Deacon reaction; (3) phenolic com pounds (present from com bustion of lignin in the w aste or other sources) entrained in the com bustion plasma are substituted on the ring structure by contact w ith the free chlorine; and (4) the chlorinated precursor to dioxin (e.g., chlorophenol) is further oxidized (w ith copper chloride as a catalyst) to yield CDDs and CDFs and chlorine. G ullett et al. (1990a; 1990b; 1991a; 1991b; 1992) studied the form ation m echanisms through extensive com bustion research at EPA, and verified the observations of Vogg et al. (1987). It w as proven th a t CDDs and CDFs could be ultim ately produced from low tem perature reactions (i.e., 3 5 0 C ) between Cl2 and a phenolic precursor, DRAFT--DO NOT QUOTE OR CITE 2-10 December 2003 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 form ing dioxin is the form ation of chlorine from HCl in the presence of oxygen (the Deacon process), as fo llo w s (Vogg et al., 1987; Bruce et al., 1991): A 2HCl + / O2 ---------- > H2O + Cl2 (2) Phenolic com pounds adsorbed on the fly ash surface are chlorinated to form the dioxin precursor, and the dioxin is form ed as a product from the breakdown and molecular rearrangement of the precursor. The reaction is prom oted by copper chloride acting as a catalyst (Vogg et al., 1987; Dickson and Karasek, 1987; G ullett et al., 1992): (a) phenol + Cl2 ------------ > chlorophenol (dioxin precursor) CuCl2 (b) 2-chlorophenol + / O2 ------------ > dioxin + Cl2 On the other hand, Eklund et al. (1986) observed the high tem perature form ation of a large variety of chlorinated to xic com pounds, including CDDs and CDFs, from precursors during a simple experim ent in w hich phenol was oxidized w ith HCl at 5 5 0 C . One m illigram of phenol w as placed in a quartz tube reactor w ith an aqueous solution (10/vL) of HCl and heated at 5 5 0 C for 5 minutes. Trichlorobenzene, dichlorophenol, dichlorobenzofuran, tetrachlorobenzene, trichlorophenol, and tetrachlorophenol were identified as major products form ed. M onochlorobenzene, chlorophenol, dichlorobenzene, tetrachloropropene, pentachloropropene, trichlorobenzofuran, tetrachlorodibenzofuran, trichlorodibenzodioxin, tetrachlorodibenzodioxin, hexachlorodibenzodioxin, hexachlorodibenzofuran, pentachlorobenzene, pentachlorobiphenyl, and pentachlorodihydroxycylohexane were seen as minor products. Trace species formed included: m onochlorodibenzofuran, pentachlorodibenzofuran, pentachlorodibenzodioxin, octachlorodibenzofuran, and octachlorodibenzodioxin. Eklund et al. (1986) hypothesized DRAFT--DO NOT QUOTE OR CITE 2-11 December 2003 th a t chlorinated organic com pounds can be produced from phenols, acids, and any chlorine source in the hot post-com bustion region (e.g., ju st exit to the furnace). The reaction w as seen as very sensitive to HCl concentration. A t HCl < 10-3 moles, no chlorinated com pounds could be detected. Nestrick et al. (1987) reported th a t the therm olytic reaction between benzene (an unsubstituted precursor) and iron (III) chloride on a silicate surface yielded CDD/CDFs at temperatures >150C . The experimental protocol introduced 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 com bustion processes because benzene is the usual com bustion byproduct of organic fuels. Inert nitrogen gas carried the benzene vapor to the furnace area. The exit from the glass tubing to the furnace was plugged w ith glass w o ol, and silica gel was introduced from the entrance end to give a bed depth of 7 cm to w hich the FeCl3 was added to form a FeCl3/silica reagent. The therm olytic reaction took place in a tem perature ranging from 1 50 -4 00C , at a residence tim e of 20 minutes. Although di- through octa-CDD/CDF were form ed by this reaction at all the tem peratures studied, the percent yields were extrem ely small. Table 2-2 summarizes these data. 2.3. MECHANISM 3: THE DE NOVO SYNTHESIS OF CDDS/CDFS DURING COMBUSTION OF ORGANIC MATERIALS The third and last mechanism , de novo synthesis, prom otes CDD/CDF form ation in com bustion processes from the oxidation of carbon particulate catalyzed by a transition metal in the presence of chlorine. As in mechanism 2, synthesis is believed to occur in regions outside of the furnace zone of the com bustion process, where the combustion gases have cooled to a range of temperatures considered favorable to form ation chem istry. A key com ponent to de novo synthesis is the production of interm ediate com pounds (either halogenated or nonhalogenated) th a t are precursors to CDD/CDF form ation. Research in this area has produced CDD/CDFs directly by heating carbonaceous fly ash in the presence of a transition metal catalyst, w ith o u t the apparent generation of reactive interm ediates. Thus, the specific steps involved in the de novo process have not been fu lly and succinctly delineated. However, laboratory experim entation has proven th a t M SW I fly ash, itself, is a reactive substrate, and the DRAFT--DO NOT QUOTE OR CITE 2-12 December 2003 m atrix can actually catalyze the de novo form ation chem istry. Typically, fly ash is composed of an alum ina-silicate construct, w ith 5-10 percent concentrations of silicon, chlorine (as inorganic chlorides), sulfur, and potassium (NATO, 1988). T w e n ty percent of the w e igh t of fly ash particles are carbon, and the particles have specific surface areas in the range of 2-4 m 2 (NATO, 1988). The de novo synthesis essentially is the oxidative breakdow n of m acromolecular carbon structures, and CDD/CDFs are form ed partially from the arom atic carbon-oxygen functional groups embedded in the carbon skeleton (Huang et al., 1999). The distinguishing feature of the de novo synthesis over the precursor synthesis is the oxidation of carbon in particulate at the start of the process to yield precursor com pounds. In mechanism 2, the precursor com pound is the starting molecule to the condensation reactions form ing CDD/CDFs (Dickson et al., 1992). By this distinction, how ever, one could argue th a t mechanism 3 is really an augm entation to mechanism 2, because the production of CDD/CDFs may still require the form ation of a CDD/CDF precursor as an interm ediate species. Nevertheless, a distinction is presented here to describe additional pathw ays suggested for the thermal form ation of these compounds. To delineate the de novo synthesis of CDD/CDFs, Stieglitz et al. (1989a) conducted experiments that involved heating particulate carbon containing adsorbed m ixtures of MgAl silicate in the presence of copper chloride (as a catalyst to the reaction). The authors described heating m ixtures of Mg-Al silicate w ith activated charcoal (4 percent by w eight), chloride as potassium chloride (7 percent by w e igh t), and 1 percent copper chloride (CuCl2) (in w ater) in a quartz flo w tube reactor at 3 0 0 C . The retention tim e w as varied at 15 m inutes, 30 m inutes, and 1, 2, and 4 hours to obtain differences in the am ounts of CDD/CDFs th a t could be form ed. The results are summarized in Table 2-3. In addition to the CDD/CDFs form ed as primary products of the de novo synthesis, the investigators observed precursors form ed at the varying retention tim es during the experiment. In particular, similar yields of tri- though hexa-chlorobenzenes, tri- through heptachloronaphthalenes, and tetra- through hepta-chlorobiphenyls were qua ntified; this was seen as highly suggestive of the role these com pounds may play as interm ediates in the continued form ation of CDD/CDFs. Stieglitz et al. (1989a) made the follow ing observations: The de novo synthesis of CDD/CDFs via the oxidation of carbonaceous particulate matter occurred at a temperature of 30 0 C . Additionally, the experiment yielded DRAFT--DO NOT QUOTE OR CITE 2-13 December 2003 ppb to ppm concentrations of chlorinated benzenes, chlorinated biphenyls, and chlorinated naphthalenes through a similar mechanism. When potassium bromide w as substituted for potassium chloride as a source of halogen for the organic com pounds in the reaction, polybrom inated dibenzo-p-dioxins and dibenzofurans form ed as reaction products. The transition metal compound copper chloride catalyzed the de novo synthesis of CDD/CDFs on the surface of particulate carbon in the presence of oxygen, yielding carbon dioxide and chlorinated/brom inated arom atic com pounds. Particulate carbon, w hich is characteristic of com bustion processes, may act as the source for the direct form ation of CDD/CDFs, as well as other chlorinated organics. More recently, Stieglitz et al. (1991) investigated the role th a t particulate carbon plays in the de novo form ation of CDD/CDFs from fly ash containing appreciable quantities of organic chlorine. Stieglitz et al. (1991) found th a t the fly ash contained 900 jg / g of bound organic chlorine. Only 1 percent of the organic chlorine was extractable. Heating the fly ash at 3 0 0 -4 0 0 C for several hours caused the carbon to oxidize, leading to a reduction in the total organic chlorine in the m atrix and a corresponding increase in the total extractable organic chlorine (TOX) (e.g., 5 percent extractable TOX at 3 0 0 C and 25-30 percent extractable tota l organic chlorine at 4 0 0 C ). From this, Stieglitz et al. (1991) concluded th a t the oxidation and degradation of carbon in the fly ash are the source for the form ation of CDD/CDFs; therefore, they are essential in the de novo synthesis of these compounds. Addink et al. (1991) conducted a series of experim ents to observe the de novo synthesis of CDD/CDFs in a carbon-fly ash system. In this experim ent, 4 grams of carbon-free MSWI fly ash were combined w ith 0.1 gram of activated carbon and placed into a glass tube between tw o glass wool plugs. The glass tube was then placed into a furnace at a specific temperature, ranging from 200 to 400 C . This was repeated for a series of retention tim es and tem peratures. The investigators observed th a t CDD/CDF form ation was optim ized at 3 0 0 C and at the furnace retention tim es of 4-6 hours. Figure 2-1 displays the relationship between retention tim e, tem perature, and CDD/CDF production from the heating of carbon particulate. A ddink et al. (1991) also investigated the relationship between furnace tem perature and CDD/CDF production from the heating of carbonaceous fly ash. Figure 2-2 displays this relationship. In general, the concentration began to increase at 2 5 0 C and crested at 3 5 0 C , w ith a sharp decrease in concentration above 3 5 0 C . The authors also noted a relationship between tem perature DRAFT--DO NOT QUOTE OR CITE 2-14 December 2003 and the CDD/CDF congener profile; at 3 0 0 C to 3 5 0 C , the low er chlorinated tetra- and penta-CDD/CDF congeners increased in concentration, w hile hexa-, hepta-, and octaCDD/CDF congeners either remained the same or decreased in concentration. The congener profile of the original MSWI fly ash (not subject to de novo experim entation) was investigated w ith respect to changes caused by either tem perature or residence tim e in the furnace. No significant changes occurred, leading the authors to propose an interesting hypothesis for further testing: after form ation of CDD/CDFs occurs on the surface of fly ash, the congener profile remains fixed and insensitive to changes in tem perature or residence tim e, indicating some form of equilibrium is reached in the form ation kinetics. G ullett et al. (1994) used a pilot-scale com bustor to study the e ffe ct of varying the com bustion gas com position, tem perature, residence tim e, quench rate, and sorbent (Ca[OH]2) injection on CDD/CDF form ation. The fly ash loading w as simulated by injecting on fly ash collected from a full-scale MSWI. Sampling and analysis indicated CDD/CDF formed on 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 com bustor operating parameters of CDD/CDF form ation is interactive and very com plicated, substantial reduction in CDD/CDF form ation can be realized w ith high tem perature sorbent injection to reduce HCl or Cl2 concentrations, control of excess air (also affects ratio of CDDs to CDFs form ed), and increased quench rate. Several steps may be involved in the copper-catalyzed form ation of CDDs and CDFs, w ith residual carbon on fly ash at 3 0 0 C (Addink and Olie, 1995). Copper initially reacts w ith chlorine to form CuCl2, and then the ligand transfers the halide to a carbon atom of an organic m acromolecule. The chlorinated m acromolecular structure oxidizes into small com pounds. Milligan and A ltw ic k e r (1995) found that increases in the carbon gasification rate caused increases in the am ounts of CDDs and CDFs form ed, and gave further evidence linking the oxidation of carbon to the form ation of CDD/CDFs. Neither the gas-phase CO2 nor CO (products of carbon oxidation) act as precursors to chlorobenzenes or CDD/CDF from reactions w ith carbon particulate (Milligan and A ltw ic k e r, 1995). A ctivated carbon, w ith a high surface area and excellent adsorptive characteristics, also has the highest gasification rate of all residual carbon (Addink and Olie, 1995). Experimental evidence suggest that the conditions for the de novo synthesis of CDDs and CDFs from carbon are: (a) the carbon consists of im perfect and degenerated DRAFT--DO NOT QUOTE OR CITE 2-15 December 2003 layers of graphite; (b) oxygen m ust be present; (c) chlorine m ust be present; (d) the reactions are catalyzed by copper chloride or some other transition m etal; and (e) tem peratures in the range of 2 0 0 C to 3 5 0 C (Huang and Buekens, 1995). The oxidation of carbon in fly ash is apparently inhibited at tem peratures below 2 0 0 C , thus indicating the lower temperature lim it for the thermal inertization of de novo synthesis (Lasagni et al., 2000). Lasagni et al. (2000) determined th a t at a tem perature of 2 5 0 0 C, the primary product of the gasification of carbon in fly ash is CO2 , but in a tem perature range of 2503 2 5 C , organic com pounds are form ed as products of the oxidation of the carbon. Addink and Olie (1995) raised the possibility th a t the molecular backbone of CDDs and CDFs may be present in carbon. If this is the case, the generation of dioxins and furans from the oxidation of carbon w ould not require the form ation of interm ediate arom atic ring structures. M ore w o rk is needed to identify these possibilities. The de novo synthesis of CDD/CDFs also involves the possibility that aromatic precursors could be form ed w ith in the post-com bustion zone as in the follow ing m anner: (1) fuel molecules are broken into smaller molecular species (e.g., C1f C2 molecules) during primary com bustion; and (2) these simple molecules recombine in the post-com bustion zone to form larger molecular arom atic species (i.e., chlorobenzenes and chlorophenols) (A ltw icker et al., 1993). Thus, small molecular products th a t evolve in the hot-zone of the furnace as a consequence of the incom plete fuel or feed material com bustion may be im portant foundation molecules to the subsequent form ation of precursor com pounds in the cooler, post-com bustion region. Eklund et al. (1988) reported form ation of a w ide range of chlorinated organic com pounds, including CDDs, CDFs, and PCBs, from the oxidation of methane w ith HCl at tem peratures of 4 0 0 to 9 5 0 C in a quartz flo w tube reactor. No active catalysts nor reactive fly ashes were added to the com bustion system. From these experimental results, Eklund et al. (1988) hypothesized th a t chlorocarbons, including CDDs and CDFs, are form ed at high tem peratures via a series of reversible reactions starting w ith chlorom ethyl radicals. The chlorom ethyl radicals can be formed from the reaction of m ethyl radicals and hydrogen chloride in a sooting flame. Methane is chlorinated by HCl in the presence of oxygen at high tem peratures, form ing chlorinated methanes, w hich react w ith methyl radicals at higher temperature (e.g., 800 C ) to form arom atic com pounds. In an oxidative atmosphere, chlorinated phenols are form ed, but DRAFT--DO NOT QUOTE OR CITE 2-16 December 2003 alkanes and alkenes are the primary products. The chlorinated phenols then act as precursors for the subsequent form ation of CDD/CDFs. A liphatic com pounds are com mon products of incom plete com bustion, and may be critical to the form ation of simple ring structures in the post-com bustion zone (Weber et al., 1999; Sidhu, 1999; Froese and Hutzinger, 1996a; Froese and Hutzinger, 1996b; Jarmohamed and Mulder, 1994). The arom atic precursor com pounds may be form ed in a potentially rich reaction environm ent of aliphatic com pounds, reactive fly ash particles, HCl, and oxygen. Sidhu (1999) noted th a t com bustion of acetylene on carbon (a com mon com bustion effluent) in the presence of gaseous HCl and copper chloride (as a catalyst) at 3 0 0 C form ed interm ediate precursors, and subsequently, CDDs and CDFs. Propene oxidized at 3 5 0 to 5 5 0 C in contact w ith reactive MSWI fly ash in a flo w tube reactor form s a wide range of chlorinated aromatic compounds, when the resulting combustion gases are mixed w ith hydrogen chloride gas (Jarmohamed and Mulder, 1994). Although the conversion was low (i.e., 1-3 percent), the oxidation of propene on fly ash in the presence of HCl can yield chlorinated benzenes and m onobenzofurans. Incorporating an oxygen atom into the monobenzofuran structure leads then to the form ation of m onodibenzofuran. The HCl contributes chlorine to the arom atic ring through the Deacon reaction, and cyclization on the fly ash surface can yield cyclohexadienyl-substituted benzenes, w h ich , in turn , can be further oxidized into CDFs (Jarmohamed and Mulder, 1994). Froese and Hutzinger (1996a) investigated the heterogeneous com bustion reactions of the nonchlorinated C2 aliphatics. Acetylene, as a model aliphatic com pound, w as allowed to react w ith pre-cleaned MSWI fly ash in a tube flo w reactor at ca. 6 0 0 C . Metal oxides (e.g., SiO2, Fe2O3, and CuO) were added separately as catalysts, instead of the metal chlorides used in other precursor experiments. The reactants were put into contact w ith HCl vapor, introduced at a constant flo w rate. The acetylene flo w w as set at 1.1 mL/min and constantly fell to near 0.9 mL/min over 30 m inutes. Regulated air flo w maintained hom eostatic oxidation conditions. Chorobenzenes and chlorophenols were form ed, w ith isomer patterns generally resembling isomer patterns of chlorobenzene and chlorophenol em issions from MSWIs. CuO w as seen as catalyzing condensation and chlorination reactions under heterogeneous conditions to form the chlorinated CDD/CDF precursor compounds. Additional more volatile compounds formed were short-chain aliphatic products, such as chlorom ethane, dichlorom ethane, and chloro-and DRAFT--DO NOT QUOTE OR CITE 2-17 December 2003 dichloroacetylene. Chlorobenzene congeners were not the major products formed; perchlorinated aliphatic com pounds dom inated as gas-phase reaction products. Froese and Hutzinger (1997) noted th a t perchlorinated aliphatic com pounds (e.g., hexachloropropene, hexachloro-1,3-butadiene, and hexachlorocyclopentadiene) are im portant interm ediates in arom atic ring form ation; they concluded th a t the catalytic reaction of C2 aliphatic compounds at 6 0 0 C dramatically contributes to form ation of chlorinated and nonchlorinated arom atic com pounds during com bustion. Thus, aliphatic compounds can form CDD/CDF precursor compounds. Variable temperature effects were observed in the form ation of CDD/CDF in the same reactions. M axim al OCDD form ation occurred at 4 0 0 C , and the tetra-hepta homologue groups were m axim ally form ed at 6 0 0 C . For CDFs, production of higher chlorinated homologues occurred at 4 0 0 C , and 5 0 0 C favored the form ation of tetrachlorodibenzofurans. Froese and Hutzinger (1996a) noted a 100-fold increase in TCDF form ation at 5 0 0 C , when compared to form ation at 4 0 0 C . An explanation for this is th a t the higher tem perature of 5 0 0 C maximized the form ation CDD/CDF precursor (chlorophenol) from the aliphatic starting compound; whereas, at the low er tem perature of 3 0 0 C , practically no ring structures were observed. Froese and Hutzinger (1996b) have produced polychlorinated benzene and phenol compounds from the high temperature (i.e., 3 0 0 to 600C ) heterogeneous combustion reactions of ethylene and ethane over fly ash in the presence of HCl, oxygen, and a metal catalyst in a com bustion flo w tube. No chlorobenzene congener precursors were formed from ethylene and ethane at 3 0 0 C ; how ever, the form ation rate increased w ith tem perature, until a m axim um production was achieved at 6 0 0 C . No definitive temperature dependence was observed for the form ation of chlorophenols from the aliphatic starting compounds. However, at 5 0 0 C , 2,4,6-trichlorophenol dominated the reaction products; at 3 0 0 C , pentachlorophenol was initially produced. Froese and Hutzinger (1996b) also investigated the effects of elemental catalysts on potentiating the heterogeneous com bustion reactions by measuring the am ount of chlorobenzene and chlorophenol product formed from the reactions of ethylene/HCl over each catalyst at 6 0 0 C . The reaction w ith Si02 did not have a catalytic effect. A l2O3 catalytic action showed high intensity for the dichlorobenzene isomers, and decreasing intensity for the higher chlorinated isomers. Comparison of the amount of dichlorobenzene product formed indicated th a t an equal quantity w as produced w ith either A l2O3 or fly ash; how ever, A l2O3 DRAFT--DO NOT QUOTE OR CITE 2-18 December 2003 form ed four to five tim es more product than the CuO catalyst. For tri- to hexachlorobenzene congeners, MSWI fly ash reactions produced 5 to 10 times more product th a t the metal catalysts. However, the presence of the CuO catalyst in these reactions produced a chlorobenzene congener pattern comparable to the fly ash reactions. W ith regard to chlorophenol production, A l2O3 also produced a unique dichlorophenol pattern, suggesting th a t A l2O3 has a unique catalytic e ffe ct in the high-tem perature reactions of C2 aliphatic compounds. Reactions over CuO produced additional products, including chlorinated m ethyl com pounds, chlorinated C2 aliphatics, and perchlorinated C3 - C5 alkyl com pounds. Froese and Hutzinger (1996b) noted th a t these perchlorinated alkyl groups, form ed by reacting ethylene and ethane over fly ash in the presence of the CuO catalyst, are key interm ediate com pounds to the form ation of first arom atic rings in typical com bustion systems. This emphasizes the im portance of copper's catalytic effects in a com bustion fly ash system. A l2O3 catalyzed reactions produced nonchlorinated naphthalene and akyalbiphenyl com pounds. Furthermore, the organic chlorine in aliphatic com pounds may also act as a direct source of chlorine for the form ation of CDDs, CDFs in a carbon fly ash system (Weber et al., 1999). In an earlier experim ent using a similar flo w -tu b e apparatus, Froese and Hutzinger (1994) form ed chlorinated benzenes and phenols in fly ash catalyzed reactions w ith trichloroethylene at tem peratures of 4 0 0 to 5 0 0 C . In this case, metal oxides (CuO, FeO3, A l2O3) were used as catalysts, but no HCl w as added for oxychlorination of product compounds. Under combustion conditions, a temperature dependent formation of chlorinated aromatics occurred from the trichloroethylene starting compound. Reaction w ith fly ash at 6 0 0 C form ed hexachlorobenzene in concentrations th a t were about 1,000 tim es greater than at 4 0 0 and 5 0 0 C , w ith similar results for chlorophenols. Froese and Hutzinger (1994) hypothesized th a t key arom atic precursors for CDD/CDFs are form ed in the higher tem perature region of a post-com bustion zone (ca. 6 0 0 C ), w hich are then carried to the cooler post-com bustion region (ca. 3 0 0 C ), where the precursors form CDDs and CDFs. DRAFT--DO NOT QUOTE OR CITE 2-19 December 2003 2.4. THE ROLE OF CHLORINE IN THE FORMATION OF CDDS AND CDFS IN COMBUSTION SYSTEMS The form ation of CDDs and CDFs in the post-com bustion region of com bustion systems via either the precursor or de novo synthesis pathways requires the availability of gaseous chlorine (Luijk et al., 1994; Addink et al., 1995). Chlorine concentration in this region is related som ehow to the chlorine content of com bustion fuels and feed materials in incineration/com bustion system s, because there can be no other source. The central question of the role of chlorine in form ing CDDs and CDFs is w hether or not their exists a positive and direct correlation between the am ount of chlorine in feeds and the am ount of CDDs and CDFs form ed and em itted from the stack. If a direct relationship appears, then reductions in the chlorine content of fuels/feeds prior to com bustion should result in a corresponding reduction in the concentrations of CDDs and CDFs form ed after combustion. If the oxychlorination reactions require a number of steps, then the relationship between chlorine in uncombusted fuels and CDD/CDFs form ed after com bustion may not be linear, although still dependent in some nonlinear association. The central question can best be addressed by examining both form ation m echanisms revealed in laboratory scale com bustion experim ents and correlations between Cl inputs w ith CDD/CDF outputs in com mercial scale com bustors. 2 .4 .1 . Review of Laboratory-Scale Studies A w ide body of experimental evidence has elucidated the direct and indirect associations between chlorine in feeds and fuels and the potential form ation of CDDs and CDFs during com bustion. The de novo synthesis of CDDs and CDFs requires tw o basic reactions: the transfer of chlorine to residual carbon particulate w ith subsequent form ation of carbon-chlorine bonds, and the oxidation of this m acrom olecular com plex to yield carbon dioxide and volatile and sem ivolatile organic com pounds as side products (Weber et al., 1999). Transition metal com pounds, such as copper chloride, catalyze these reactions. Hydrogen chloride gas is the major direct source of chlorine available for participating in the form ation of CDD/CDFs, w hich is initially form ed as a com bustion by product from the inorganic and organic chlorine contained in the fuel (Vogg et al., 1987; Bruce et al., 1991; Gullet et al., 1990; Commoner et al., 1987; Addink et al., 1995; Luijk et al., 1994; Dickson et al., 1992; W agner and Green, 1993; Halonen et al., 1994; Rigo DRAFT--DO NOT QUOTE OR CITE 2-20 December 2003 et al., 1995; Rigo, 1998; A ltw icke r et al.,1993). MSW contains approxim ately 0 .4 5 -0 .9 0 percent (by w eight) chlorine (Domalski et al., 1986). If le ft uncontrolled, MSW incinerators are a major stationary com bustion source of HCl air em issions, w hich average between 4 0 0 to 6 00 ppm in the com bustion gas (U.S. EPA, 1987a). In the presence of oxygen, HCl may oxidize to yield free chlorine gas by the Deacon process, and the free chlorine directly chlorinates a CDD/CDF precursor along the aromatic ring structure. Further oxidation of the chlorinated precursor in the presence of a transition metal catalyst (of w hich copper chloride w as found to be the m ost active) yields CDDs and CDFs (A ltw icker et al.,1993 ). Increasing the yield of chlorine in vapor phase from HCl oxidation generally increases the rate of CDD/CDF form ation. Formation kinetics are m ost favored at tem peratures between 2 0 0 C to 4 5 0 C . Chlorine production can be reduced either by lim iting initial HCl concentration or by shortening the residence tim e in the Deacon process temperature (Bruce et al., 1991; G ullett et al., 1990b; Commoner et al., 1987). Bruce et al. (1991) observed a general increase in CDD and CDF form ation, w ith increases in the vapor phase concentration of chlorine. Figure 2-3 shows the apparent dependence of the extent of form ation 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 form ation of CDD/CDFs in the post-com bustion zone. This is in agreement w ith a simple experim ent of Eklund et al. (1986) in w hich unsubstituted phenol was mixed w ith HCl at 5 5 0 C in a quartz tube reactor and form ed a w ide range of to xic chlorinated hydrocarbons, including CDDs and CDFs as reaction products. Eklund et al. (1988) also found a dependence of the amounts of chlorinated phenol product formed from the nonchlorinated starting material w ith the increased am ount of HCl introduced into the reaction. Under the conditions of this experiment, no chlorinated compounds were formed at an HCl concentration of less than 10-3 moles, and m axim um chlorophenol concentration occurred at ca. 10 +8 M. Born et al. (1993) also observed th a t increasing levels of HCl give rise to increasing rates of oxychlorination of precursors, w ith increasing chances for the post-com bustion form ation of CDDs and CDFs. However, recently Addink et al. (1995) observed th a t an HCl atmosphere and/or chlorine produced approxim ately equal quantities of CDD/CDFs during the de novo synthesis from oxidation of particulate carbon. These experimental results suggest that chlorine production via the Deacon process reaction in the de novo synthesis may not be the only chlorination pathw ay, and may DRAFT--DO NOT QUOTE OR CITE 2-21 December 2003 indicate th a t the HCl molecule can be a direct chlorinating agent. In addition, some chlorine is expected to be form ed from the oxidation of metal chlorides (e.g., CuCl2), but Cl2form ation from the Deacon process is greater because of the continuous supply of HCl delivered from the com bustion chamber (Bruce et al., 1991). In this case, a first order dependence of HCl to Cl2is observed. W agner and Green (1993) investigated the correlation of chlorine content in feed to stack em issions of chlorinated organic com pounds in a pilot-scale incinerator, using HCl flue gas measurements as a surrogate for fuel-bound organic chlorine. In addition to M SW as a fuel, variable am ounts of PVC resin were added during 6 of 18 stack te st runs. The resulting data were regressed to determine the coefficie nt of correlation between HCl measurements and total chlorobenzene com pound emission measurements. In nearly all of the different regression analyses perform ed, the relationship between HCl emission and emissions of chlorinated organic com pounds w as positive and w ell-defined. In addition, W agner and Green (1993) found a direct dependence of HCl emission levels to the level of PVC in the w aste, w ith generally increasing am ounts of HCl form ed as increasing am ounts of PVC were added. From these experim ents, W agner and Green (1993) concluded that decreases in the levels of organically bound chlorine in the input to an incinerator led to decreases in chlorinated organic compound stack emissions. Kanters and Louw (1994) investigated a possible relationship of chlorine content in w aste feed to chlorophenol emissions in a bench-scale therm al reactor. MSW I, w ith a higher content of chlorine, caused a higher emission of chlorophenols via the de novo synthesis pathway. Kanters and Louw (1994) lowered the chlorine content of the prototype MSWI by replacing Clcontaining fractions w ith cellulose. They observed appreciable decreases in the am ounts of chlorophenol form ed from com bustion. Kanters and Louw (1994) concluded th a t reductions in the chlorine content of w aste feeds or elim ination of PVC prior to MSWI com bustion should result in a corresponding reduction in chlorophenol and CDD/CDF emissions. In a similar experim ent, W ikstrom et al. (1996) investigated the influence of chlorine in feed m aterials to the form ation of CDDs, CDFs, and benzenes in a laboratoryscale fluidized bed reactor. An artificial fuel (composed of 34 percent paper, 30 percent w heat flour, 14 percent saw dust, 7 percent polyethylene, and 2 percent metals) to w hich varying am ounts of organic chlorine (PVC) and inorganic chlorine (CaCl2. 6H 2O) were DRAFT--DO NOT QUOTE OR CITE 2-22 December 2003 added, w as com busted. Seven fuels were studied, and the chlorine content w as varied from 0 .1 2 to 2 percent. Flue gases were sampled for CDDs, CDFs, and chlorobenzenes. All combustion was performed w ith a high degree of combustion efficiency (e.g., 99.999 percent) to avoid form ing polyvinylidene chloride and naphthalenes as products of incomplete com bustion of pure PVC. W ith the com bustion conditions held constant, only the chlorine content of the fuel was varied. From these experiments, about 1,000-fold higher concentrations of PCB isomers were produced, as compared to CDD/CDF (expressed as concentration of I-TEQDF). M oreover, a correlation w as found between ITEQdf and PCB levels in the flue gases and the chlorine content of the fuel. A 5-fold increase in both I-TEQDF and PCB concentration w as observed in the flue gases from com bustion of fuels containing 0 .5 and 1.7 percent tota l chlorine. M oreover, no differences were observed in the am ount of chlorinated product produced or w hether the source of chlorine in the fuel was organic or inorganic. No correlation was observed between total CDD/CDF and PCB form ation and tota l chlorine in the feed when chlorine levels in feed were 0.5 percent or lower. Highest am ounts of CDD/CDFs and PCBs were formed from the fuel having the highest total chlorine content (1.7 percent). Under the conditions of this experim ent, W ikstrom et al. (1996) observed th a t a chlorine fuel content of 1.0 percent w as a threshold for form ing excess CDDs, CDFs, and PCBs during combustion. The authors noted that Swedish MSW contains about 0.7 percent chlorine, of w hich approxim ately 40 percent are organic chlorine. They concluded th a t Swedish MSW is below the observed threshold value of 1.0 percent chlorine associated w ith a general increase in CDD, CDF, and PCB form ation in the post-com bustion region. W ikstrom et al. (1996) stated th a t their study does not support the thesis th a t elim ination of only PVC from the w aste prior to com bustion w ill cause a significant reduction of CDD/CDF emissions if the com bustion process is well controlled (high com bustion efficiency). A primary byproduct of com busting PVC is the generation of HCl. Paciorek et al. (1974) therm ally degraded pure PVC resin at 4 0 0 C and produced 550 mg/g HCl vapor as a primary therm olysis product, w hich w as observed as being 94 percent of the theoretical am ount based on the percent w e igh t chlorine on the molecule. Ahling et al. (1978) concluded th a t HCl can act as a chlorine donor to ultim ately yield chlorinated aromatic hydrocarbons from the therm olytic degradation of pure PVC, and th a t these yields are a DRAFT--DO NOT QUOTE OR CITE 2-23 December 2003 function of transit tim e, percent oxygen, and tem perature. They observed data from 11 separate experiments, conducted w ith a range of temperatures from 570 to 1,130C . These data indicated that significant quantities of various isomers of dichloro-, trichloro-, tetrachloro-, and hexachlorobenzenes could be produced. Choudhry and Hutzinger (1983) proposed th a t the radical species Cl- and H- generated in the incineration process may attack the chlorinated benzenes thus form ed, and abstract hydrogen atom s to produce ortho-chlorine substituted chlorophenol radicals. These intermediate radical species then react w ith molecular oxygen to yield ortho-substituted chlorophenols. As a final step, the ortho-substituted chlorophenols act as ideal precursors to yield CDD/CDFs w ith heat and oxygen. The chlorine in aliphatic com pounds has been observed as both yielding high am ounts of HCL during com bustion, and also acting as a direct chlorine source for the de novo synthesis of CDDs/CDFs (Weber et al., 1999). Recently Addink and A ltw icke r (1999) have reported on the role of the inorganic chloride ion in the form ation of CDD/CDFs using the labeled com pound, Na37Cl. The inorganic chloride ion form s carbon-chlorine bonds on soot particles during com bustion. The chlorine in the soot can both be directly inserted into a CDD/CDF molecule during form ation, or can exchange w ith the chloride ions in the transitional metal catalyst w hich prom otes CDD/CDF form ation. Thus, the inorganic chlorine ion participates as a chlorine donor to CDD/CDF formation. De Fre and Rymen (1989) reported on form ing CDDs and CDFs from hydrocarbon com bustion in a dom estic gas/oil heating burner in the presence of 15 and 3 00 ppm concentrations of HCl. Over 100 chlorinated organic com pounds were detected in the flue gases whenever HCl w as injected into the system. De Fre and Rymen (1989) observed form ation of CDDs and CDFs in all experim ents where HCl was injected in a hydrocarbon flame. In this case, CDFs were always more abundant th a t CDDs. De Fre and Rymen (1989) concluded th a t the relationship between the HCl concentration and the em itted concentration of CDD/CDF under fixed com bustion conditions appeared to be exponential for a w ide range in tem perature (e.g., 2 4 0 to 9 0 0 C ). 2 .4 .2 . Review of Full Scale Combustion Systems The review of experimental data clearly indicates an association between chlorine content of feed/fuels and the potential synthesis of CDDs and CDFs. Paradoxically, the DRAFT--DO NOT QUOTE OR CITE 2-24 December 2003 review of full-scale operating incineration processes does not yield such unequivocal results indicating th a t com plex kinetic events make strong associations d iffic u lt in fullscale systems. The follow ing is a review of studies of the association between chlorine in feeds and stack releases of CDD/CDFs in full-scale incineration systems. In the stack 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/w aste and the m agnitude of em issions of CDDs and CDFs from the stack of the tested facilities (U.S. EPA, 1987a). In general, com bustion units w ith the highest CDD emission concentrations had greater quantities of chlorine in the fuel; conversely, sites w ith the lowest CDD emission concentrations contained only trace quantities of chlorine in the feed. The typical chlorine content of various com bustion fuels w as reported by Lustenhouwer et al. (1980) as: coal: 1,300 >wg/g; MSW: 2 ,5 0 0 >wg/g; leaded gasoline: 3 0 0 -1 ,6 0 0 >wg/g; and unleaded gasoline: 1-6 >wg/g. Thomas and Spiro (1995) also analyzed the relationship of CDD/CDF emissions from com bustion to the chlorine content of feed materials. Thomas and Spiro (1996) plotted average CDD/CDF emission factors for a variety of com bustion processes (black liquor boilers, unleaded gasoline com bustion, leaded gasoline com bustion, wire incineration, cigarette com bustion, sewage sludge incineration, MSWI, PCP-treated wood com bustion, hazardous w aste incineration, and hospital w aste incineration) against the average chlorine concentration of the combusted material. The plot showed that average CDD/CDF emissions of com bustion source categories tend to increase w ith the average chlorine content of the combusted fuel. The analysis clearly indicated that combustion sources w ith relatively high com bustion efficiency and adequate air pollution controls tended to have tw o order of magnitude lower emissions than poorly operated sources. This suggested a strong dependence on chlorine concentration in fuels and the m agnitude of CDD/CDF emissions in the more poorly controlled com bustion sources. The slope of the log-log plot w as between 1 and 2, indicating th a t the relationship of chlorine content to CDD/CDF emissions was more than proportional. Recently Costner (1998) reported finding a positive correlation between chlorine content of feed material and CDD/CDF em issions at a full-scale hazardous waste incinerator. Costner concluded that emissions at this facility were dependent on chlorine DRAFT--DO NOT QUOTE OR CITE 2-25 December 2003 input at a chlorine concentration as low as 0.031 percent, and th a t there was no evidence of a threshold in the relationship between chlorine in feed and CDD/CDF emissions. Rigo et al. (1995) summarized the results of a study com missioned by the American Society of Mechanical Engineers (ASME, 1995). The study was a statistical evaluation of the relationship of HCl concentration in flue gases to various com bustion systems (i.e., MSWI, hospital waste incineration, hazardous waste incineration, biomass com bustors, laboratory and bench-scale com bustors) to the stack emission of total CDDs and CDFs. HCl was a surrogate for total chlorine feed content in this study. The data analysis w as su fficie n t for 92 facilities in the data base th a t showed both HCl and CDD/CDF emissions. From the 92 facilities, 72 did not show statistically significant relationship between chlorine input and CDD/CDF o u tput in emissions stream s; 2 facilities showed increasing CDD/CDF concentrations w ith increasing chlorine; and 8 facilities showed decreasing CDD/CDF concentrations w ith increasing chlorine. AMSE (1995) concluded that: "The failure to find sim ultaneous increases in m ost cases and finding inverse relationships in a fe w indicates th a t any e ffe ct chlorine has on PCDD/F emissions is smaller than the variability of other causative factors. W hatever e ffe ct chlorine has on PCDD/F em issions in com mercial scale system s is masked by the e ffe ct of APCS (air pollution control systems) tem perature, ash chem istry, com bustion conditions, measurement im precision, and localized flo w stra tifica tio n ." Liberson and Belanger (1995) reported the results of an analysis of the form ation and emission of CDDs and CDFs as a function of total chlorine in com bustion feed materials at a rotary kiln hazardous waste incinerator (HWI). The data were generated from m ultiple test series conducted over a 13-m onth period at the HWI, while operating a carbon injection system specifically designed to control and reduce CDD and CDF stack emissions. The chlorine feed rates ranged from 0 to 3 ,3 0 0 pounds per hour, while the CDD/CDF emission rates ranged between 0 .7 and 39 ng/DSCM. The authors noted that m ultiple series of CDD/CDF control systems were employed on this HWI (e.g., a high temperature secondary combustion chamber, a spray dryer-evaporative quench that DRAFT--DO NOT QUOTE OR CITE 2-26 December 2003 further cools the combustion gases, activated carbon injection to adsorb semi-volatile organics, and a cool-side electrostatic precipitator follow ed by an acid gas scrubber to collect HCl and Cl2). From the analyses of data, Liberson and Belanger (1995) concluded no correlation exists between CDD/CDF emissions and chlorine feed in a modern MW I, using carbon injection for CDD/CDF control. 2.5. POTENTIAL PREVENTION OF CDD/CDF FORMATION IN COMBUSTION SYSTEMS Given w h a t is currently understood about oxychlorination reactions in the synthesis of CDDs and CDFs, researchers have identified certain interventions th a t could be taken to reduce or impede form ation in com bustion systems. Recently, Haghunathan and G ullett (1996) dem onstrated in a pilot-scale incinerator th a t sulfur com pounds can combine w ith the metal catalyst necessary to stim ulate the Deacon reaction of HCl and O2 to yield Cl2, thereby, neutralizing the catalyzing agent and reducing the form ation of CDDs and CDFs. The Deacon reaction, w hich form s free chlorine in the com bustion plasma, is seen as only occurring in the presence of a catalyst. Thus, the SO2 molecule (formed when sulfur in the fuel combines w ith oxygen) can either inhibit the catalytic a ctivity of the fly ash by com bining w ith a metal-based Deacon catalyst in the fly ash, or by depleting the Cl2 form ed. Haghunathan and G ullett (1996) observed th a t the principal action of sulfur for inhibiting the form ation of CDDs and CDFs in com bustion system s is through SO2depletion of Cl2, as follow s: Cl2 + SO2 + H2O 2HCl + SO2 The relevance of this finding is th a t the co-com bustion of municipal solid w aste w ith coal (that contains sulfur) should lead to dram atic reductions in the am ount of CDDs and CDFs form ed and em itted, and may explain w h y, in the United States, coal com bustion at power plants results in over a m agnitude low er CDD/CDF emission rate than MSWIs. Naikwadi and Karasek (1989) investigated the addition of calcium oxide (CaO) and triethylam ine (TEA) to the flue gases of a com bustion system as an inhibitor of the catalytic a ctivity of fly ash. They placed 500 j g C-13-labeled pentachlorophenol (a dioxin precursor) in a com bustion flo w tube and allowed it to react w ith organic-extracted MSWI fly ash at 3 0 0 C under an air stream. Under these condition, CDD/CDFs were form ed at DRAFT--DO NOT QUOTE OR CITE 2-27 December 2003 concentrations ranging from 1,660 ng to 2 ,2 0 0 ng per 100 ug 13C-PCP. The experimental method was then m odified by mixing reactive MSWI fly ash w ith either CaO or TEA. The results showed th a t the am ount of CDD/CDF form ed could be reduced by an order of m agnitude from the reaction of PCP w ith fly ash and the addition of TEA as an inhibitor. W hen CaO was mixed w ith fly ash, the am ount of CDD/CDFs form ed decreased by over 20 times. 2 .6 . THEORY ON THE EMISSION OF POLYCHLORINATED BIPHENYLS The air emission of PCBs from MSW incineration is less w ell studied. Probably the form ation mechanisms th a t apply to CDDs/CDFs w ould also apply to PCBs. Mechanism 1 (pass through) is im plicit in the TSCA rule w hich requires 9 9 .9 9 9 9 percent destruction in hazardous waste incinerators. When this occurs, 0.0001 percent of the initial amount of PCBs fed into the hazardous w aste incinerator may be em itted out the stack. This may indicate th a t some small fraction of the PCBs present in the fuel fed into an incineration process may result in PCB emissions from the stack of the process. PCBs have been measured as contam inants in the raw refuse prior to incineration in an MSWI (Choudhry and Hutzinger, 1983; Federal Register, 1991a). Using this inform ation, it is possible to te st Theory mechanism 1 involved in CDD/CDF emissions: th a t the PCB contam ination present in the fuel is responsible for emissions from the stack. The mass balance of total PCB, beginning w ith m easurement in the raw refuse and ending w ith m easurement at the stack to an RDF MSW incinerator (Federal Register, 1991a), can be used to calculate the destruction rated efficiency (DRE) of incineration of the PCB contam inated MSW. Using results from te st number 11 at the RDF fa cility (Federal Register, 1991a), a com putation of DRE can be made w ith the follow ing equation (Brunner, 1984): W. - W DRE = -- 1-------- ^ x 100% W, Where: W, = mass rate of contam inant fed into the incinerator system W o = mass rate of contam inant exiting the incinerator system DRAFT--DO NOT QUOTE OR CITE 2-28 December 2003 In te st 1 1 ,8 1 1 nanograms of tota l PCBs/gram of refuse (ng/g) were measured in the MSW fed into the incineration system , and 9 .5 2 ng/g of total PCB were measured at the inlet to the pollution control device (i.e., outside the furnace region, but preceding emission control). From these measurements, a DRE of 9 8 .8 percent can be calculated. Therefore, it appears th a t PCB contam ination in the raw MSW , w hich was fed into this particular incinerator, may have accounted for the PCB emissions from the stack of the MSW incinerator. PCBs can be therm olytically converted into CDFs (Choudhry and Hutzinger, 1983; U.S. EPA, 1984). This process occurs at temperatures som ewhat lower than typically measured inside the firebox of an MSWI. Laboratory experim ents conducted by EPA (U.S. EPA, 1984) indicate th a t the optim um conditions for CDF form ation from PCBs are near a tem perature of 6 7 5 C in the presence of 8 percent oxygen and a residence tim e of 0.8 seconds. This resulted in a 3 to 4 percent efficiency of conversion of PCBs into CDFs. Because 1 to 2 percent of the PCBs present in the raw refuse may survive the thermal stress imposed in the com bustion zone to the incinerator (Federal Register, 1991a), then it is reasonable to presume th a t PCBs in the MSW may contribute to the tota l mass of CDF emissions released from the stack of the incinerator. A lthough it appears th a t contam ination of w aste feeds w ith PCBs may be an im portant factor to detecting PCBs in stack emissions from com bustion processes, recent research has indicated the possibility th a t these com pounds may also be form ed in the post-com bustion zone either from de novo synthesis or from precursor compounds. Zheng et al. (1999) observed the form ation of PCBs in the post-com bustion zone from the pyrolysis of chlorobenzenes using a laboratory scale furnace. Zheng and cow orkers (1999) observed th a t PCBs were optim ally form ed from low er chlorinated chlorobenzenes (i.e., 1,3-dichlorobenzene) catalyzed by copper chloride. In this experim ent, m axim um PCB production occurred at a tem perature of 3 5 0 C . W ikstrom et al. (1998) reported secondary form ation of PCB in the post-com bustion zone similar to the de novo synthesis of CDDs and CDFs, albeit, PCBs were form ed in only small am ounts relative to CDD/CDFs. Fangmark and cow orkers (1994) have postulated th a t form ation of PCBs, CDDs, and CDFs in the post-com bustion zone may occur either from a com m on precursor, or by side reactions affected in a similar w ay by tem perature and residence time. On the other hand, Blum enstock et al. (1998) produced results in a pilot-scale furnace th a t were inconsistent DRAFT--DO NOT QUOTE OR CITE 2-29 December 2003 w ith the de novo form ation of CDDs and CDFs in the post-com bustion zone (i.e., PCBs seemed to be optim ally form ed at high tem peratures in oxygen deficient atmospheres). Shin and Chang (1999) have noted a positive correlation between PCB concentrations on MSW incineration fly ash and fly ash concentrations of CDDs and CDFs, suggesting that high PCBs levels in fly ash may be a contribu tory cause to the post-com bustion form ation of CDDs and CDFs (i.e., PCBs are precursors to CDD/CDFs). Nito et al. (1997) noted the form ation of CDFs and CDDs from the pyrolysis of PCBs in a fluidized bed system indicating th a t PCBs in feeds may account for CDFs form ed in municipal solid w aste incineration. More com bustion related research needs to be conducted to firm ly establish w hether or not PCB contam ination in feeds or post-com bustion form ation (or both) may explain the presence of PCBs in com bustion flue gases. 2.7. SUMMARY AND CONCLUSIONS 2 .7 .1 . Mechanisms of Formation of Dioxin-Like Compounds There are three primary mechanisms for CDD/CDF emissions from com bustion sources: Mechanism 1 : This refers to CDD/CDFs contained in the feed w hich pass through the com bustor intact and are subsequently released to the environm ent. For most system s, this is not th o u g h t to be a major contributor to CDD/CDF emissions for tw o reasons. First, for commercial systems w ith good com bustion controls, the temperatures and residence tim es should result in the destruction of m ost CDD/CDFs in the feed. Second, mass balance studies of a number of com bustion systems show that more CDD/CDFs can be detected dow nstream of the furnace than in the feed. Consequently synthesis appears to be a more im portant mechanism than pass through. Mechanism 2 : This is the form ation of CDD/CDFs from the therm al breakdown and molecular rearrangement of arom atic precursors either originating in the feed or form ing as a product of incom plete com bustion. Actual synthesis of CDD/CDF occurs in the post com bustor environment. The CDD/CDFs form when the precursors sorb onto binding sites on the surface of fly ash particles. This reaction has been observed to be catalyzed by the presence of a transition metal sorbed to the particulate. The m ost potent catalyst is copper chloride. Heat in a range of 200 to 4 5 0 C has been identified as a necessary condition for these reactions to occur, w ith either lower or higher temperatures inhibiting DRAFT--DO NOT QUOTE OR CITE 2-30 December 2003 the process. Since these reactions involve heterogenous chem istry, the rate of emissions is less dependent on reactant concentration than conditions th a t prom ote form ation such as tem perature, retention tim e and availability of catalytic surfaces. For this mechanism to be significant, tw o broad conditions are needed. Mechanism 3 : This is the form ation of CDD/CDFs in the post-com bustion environm ent in the absence of arom atic precursors. This de novo synthesis involves the oxidative breakdown of macromolecular carbon structures (e.g., graphite) leading to the form ation of aromatic CDD/CDF precursors. These precursors then undergo the transform ations associated w ith mechanism 2 to form CDD/CDFs. As w ith mechanism 2, since this mechanism involves heterogenous chem istry, the rate of emissions is dominated by the same physical conditions as discussed in mechanism 2. M echanisms 2 and 3 can occur sim ultaneously, share a number of com m on reaction pathw ays, occur in the same physical environm ent and are controlled by many of the same physical conditions. In well designed and operated com bustion system s, the precursor species needed for mechanism 2 are not in abundant supply; consequently de novo synthesis can become the dom inant pathw ay for form ation. In system s w ith incom plete com bustion, it is d iffic u lt to sort out the relative contribution of these tw o mechanisms to total emissions. Both mechanisms, how ever, can be curtailed if steps are taken to minimize the physical conditions needed to support form ation (i.e., tim e, tem perature and reactive surface). The com bustion form ation chem istry of PCBs is less w ell-studied than for CDD/CDFs, but it is reasonable to assume th a t these same three m echanisms would apply. For w aste incineration, PCBs can exist in significantly higher concentrations in the feed than CDD/CDFs. Consequently, mechanism 1 may play a more prom inent role in PCB emissions from some forms of waste combustion. 2 .7 .2 . Role of Chlorine From the various analyses on the role and relationship of chlorine in feeds to CDD/CDF form ation and em issions, the follow ing observations and conclusions are made: 1. A lthough chlorine is an essential com ponent for the form ation of CDD/CDFs in com bustion systems, the empirical evidence indicates that, for commercial scale incinerators, chlorine levels in feed are not the dom inant controlling factor for rates of DRAFT--DO NOT QUOTE OR CITE 2-31 December 2003 CDD/CDF stack emissions. Im portant factors w hich can affect the rate of CDD/CDF form ation include the overall combustion efficiency, post-com bustion flue gas tem peratures and residence tim es, and the availability of surface catalytic sites to support CDD/CDF synthesis. Data from bench, pilot and commercial scale com bustors indicate that CDD/CDF form ation can occur by a number of mechanisms. Some of these data, prim arily from laboratory and pilot scale com bustors, have show n direct correlation between chlorine content in fuels and rates of CDD/CDF form ation. O ther data, prim arily from com mercial scale com bustors, show little relation w ith availability of chlorine and rates of CDD/CDF form ation. The conclusion th a t chlorine in feed is not a strong determinant of CDD/CDF emissions applies to the overall population of commercial scale com bustors. For any individual com mercial scale com bustor, circum stances may exist in w hich changes in chlorine content of feed could a ffe ct CDD/CDF emissions. For uncontrolled com bustion, such as open burning of household w aste, chlorine content of w astes may play a more significant role in affecting levels of CDD/CDF emissions than observed in commercial scale com bustors. 2. Both organic and inorganic chlorine in com bustion fuels yield HCl in the post com bustion region. HCl vapor is the dom inant source of chlorine leading to the form ation of CDD/CDFs. The reaction proceeds via the oxidation of HCl in the presence of an inorganic chloride catalyst (the Deacon reaction). Although the preponderance of scientific evidence suggest th a t this is a dom inant pathw ay for producing chlorinated com pounds in em issions, it is still unclear if HCl can also directly chlorinate arom atics, or m ust first be oxidized to yield free chlorine. 3. Laboratory scale experiments have examined correlations between chlorine content of feeds w ith total CDD/CDF form ation. These experiments have suggested that for feeds containing less than 1% Cl, the rate of CDD/CDF form ation is independent of Cl. For feeds w ith Cl content greater than 1% , a positive correlation is seen. A lthough this relationship is observed at the laboratory scale, it has been show n not to apply to com mercial scale com bustors (see 1 above). It has not been determ ined, how ever, if these relationships are relevant to other types of com bustion such as backyard barrel burning, landfill fires and agricultural burning. DRAFT--DO NOT QUOTE OR CITE 2-32 December 2003 2 .7 .3 . General Conclusion The trace chem istry of combustion appears to involve a wide variety of form ation pathw ays indicating th a t the chem istry of CDD/CDF form ation is more com plicated than the relatively simple constructs described in this review . Despite this com ple xity, the current w e igh t of evidence w ould suggest th a t the role of chlorine and the form ation mechanisms outlined above w ill account for most of the CDD/CDF emissions associated w ith com bustion. DRAFT--DO NOT QUOTE OR CITE 2-33 December 2003 Table 2-1. Concentration of CDD/CDFs on Municipal Incinerator Fly Ash at Varying Temperatures Congener CDD Tetra Penta Hexa Hepta Octa CDF Tetra Penta Hexa Hepta Octa 200 15 40 65 100 90 122 129 61 48 12 CDD/CDF Concentration on Fly Ash (ng/g) Temperature (C) 250 300 350 26 188 110 517 217 1029 208 1103 147 483 220 590 550 430 200 560 1379 1185 367 1256 1010 236 944 680 195 689 428 74 171 72 Source: Adapted from Vogg et al. (1987). 400 50 135 110 60 15 530 687 260 112 12 DRAFT--DO NOT QUOTE OR CITE 2-34 December 2003 Table 2-2. CDD/CDFs Formed from the Therm olytic Reaction of 690 mg Benzene + FeCl3Silica Complex Congener DiCDD TriCDD TCDD PeCDD HxCDD HpCDD OCDD Total CDDs DiCDF TriCDF TCDF PeCDF HxCDF HpCDF OCDF Total CDFs Mass Produced (ng) 4.9 54 130 220 170 98 20 696.9 990 7,800 12,000 20,000 33,000 40,000 74,000 187,790 Number of Moles Produced 0.019 0.019 0.400 0.620 0.440 0.230 0.040 1.940 4.200 29.00 39.00 59.00 88.00 98.00 167 4 8 4 .2 Percent Yielda 4 .3 E-7 4 .3 E-6 9.0 E-6 1.4 E-5 9.9 E-6 5.2 E-6 9.0 E-7 4 .4 E-5 9.5 E-5 6.6 E-4 8.8 E-4 1.3 E-3 2.0 E-3 1.1 E-3 3.8 E-3 1.1 E-2 a Percent yield = (number of moles of CDD or CDF/moles benzene) x 100. Source: Nestrick et al. (1987) DRAFT--DO NOT QUOTE OR CITE 2-35 December 2003 Table 2-3. De Novo Formation of CDDs/CDFs after Heating Mg-Al Silicate, 4% Charcoal, 7% Cl, 1 % CuCl2.2H 2O at 3 0 0 C Congener TCDD PeCDD HxCDD HpCDD OCDD Total CDD TCDF PeCDF HxCDF HpCDF OCDF Total CDF 0.25 2 110 730 1700 800 3342 240 1360 2500 3000 1260 8360 Concentrations of CDD/CDF (ng/g) Reaction Time (hrs) 0.5 1 2 4 120 780 1840 1000 14 250 1600 3500 2000 30 490 2200 4100 2250 3744 7364 9070 280 1670 3350 3600 1450 670 3720 6240 5500 1840 1170 5550 8900 6700 1840 10350 17970 24160 Source: Stieglitz et al. (1989a). 4 100 820 3800 6300 6000 17020 1960 8300 14000 9800 4330 38390 DRAFT--DO NOT QUOTE OR CITE 2-36 December 2003 nanograms/gram CDDs/CDFs Formed at 300 Degrees Centigrade Total CDDs Total CDFs Figure 2-1. The de novo Synthesis of CDD/CDFs from Heating Carbon Particulate at 300 C at Varying Retention Times DRAFT--DO NOT QUOTE OR CITE 2-37 December 2003 nanograms/gram Temperature Effects on CDD/CDF Production p Total CDD e " Total CDF Temperature-degrees centigrade Source: Addink, Drijver, and Olie (1991) Figure 2-2 Temperature Dependence on CDD/CDF Formation DRAFT--DO NOT QUOTE OR CITE 2-38 December 2003 Chlorine Concentration Dependence for the Formation of CDDs 20 n Legend 18 - TCDD 16 - PeCDD 14 - HxCDD 12 - HpCDD 10 - OCDD 8 6- 4- 2 0 79.1 158.3 316.6 633.2 Vapor Phase Chlorine Concentration (mg/cu. meter) Chlorine Concentration Dependance for CDF Formation 2.2 n 79.1 158.3 316.6 633.2 Vapor Phase Chlorine concentration (mg/cu. meter) Figure 2-3. The Association Between Vapor Phase Cl2 and the Formation of CDDs/CDFs DRAFT--DO NOT QUOTE OR CITE 2-39 December 2003