Environ. Sci. Technol. 2003, 37, 1108-1113
Importance of Chlorine Speciation on de Novo Formation of Polychlorinated Dibenzo-p-dioxins and Polychlorinated Dibenzofurans EVALENA WIKSTRO ¨ M , †,§ S H A W N R Y A N , † ABDERRAHMANE TOUATI,‡ M A R N I E T E L F E R , †,§ D E N N I S T A B O R , ‡ A N D B R I A N K . G U L L E T T * ,† National Risk Management Research Laboratory, Air Pollution Prevention and Control Division, Air Pollution Technology Branch, U.S. Environmental Protection Agency, Research Triangle Park, North Carolina 27711, ARCADIS Geraghty & Miller, Inc., P.O. Box 13109, Research Triangle Park, North Carolina 27709, Oak Ridge Institute for Science and Education Postdoctoral Program, Oak Ridge, Tennessee 37831
The role of chlorine speciation on de novo formation of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDDs/Fs) has been studied thoroughly in an entrained flow reactor during simulated waste combustion. The effects of gas-phase chlorine species such as chlorine (Cl2), hydrogen chloride (HCl), and chlorine radicals (Cl•), as well as ash-bound chlorine, on PCDD/F de novo formation were isolated for investigation. The ashbound chlorine alone was observed to be a sufficient chlorine source for PCDD/F formation. The addition of HCl to the system did not influence the yields of the PCDDs/ Fs nor the degree of chlorination due to its poor chlorinating ability. Addition of 200 ppm of Cl2 to the ash-feed system resulted in increased PCDD/F yields, especially for the octaand hepta-chlorinated congeners. Altering the reaction temperature to enable the presence of only Cl2 to the system did not change the yields of PCDD/F compared to those when both Cl2/Cl• were present. However, comparison between ash-bound and gas-phase chlorine, the latter at a concentration typical of a realistic combustion process, revealed ash-bound chlorine to be the more important chlorine source for de novo formation of PCDD/F in a fullscale incinerator.
Introduction Chlorine is an essential ingredient in the formation of the toxic and persistent polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs). Hence, it is often suggested that removal of chlorine from waste streams will result in decreased emissions of these compounds. Several recent studies have investigated whether the total chlorine level in the fuel was important for the formation of polychlorinated aromatics but the results in general provided contradictory conclusions. Some research * Corresponding author phone: (919)-541-1534; fax: (919)-5410554; e-mail:
[email protected]. † National Risk Management Research Laboratory. ‡ ARCADIS Geraghty & Miller, Inc. § Oak Ridge Institute for Science and Education Postdoctoral Program. 1108
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003
groups reported correlations between chlorine input and the emissions of PCDDs/Fs (1-6), while others did not (79). The discrepancies between the studies suggest that parameters other than the total chlorine content, such as gas-phase and solid-phase chlorine speciation, which in turn are affected by combustion efficiency and reaction temperature, may also affect formation of PCDDs/Fs. Various heterogeneous and homogeneous mechanisms have been proposed for PCDD/F formation. The two lowtemperature (250-450 °C) mechanisms are (i) reactions involving fly ash carbon and organic or inorganic chlorine in the fly ash (i.e., de novo synthesis) (10), and (ii) heterogeneous, catalytic reactions with the fly ash surface and gasphase precursors such as chlorinated phenols (ClPhs) and chlorinated benzenes (ClBzs) (11). Thus, deciphering the role of chlorine speciation is further compounded by the lack of understanding of the predomination of their formation mechanism. The aim of this study is to investigate the role and importance of ash-bound chlorine as well as gas-phase chlorine, e.g., hydrogen chloride (HCl), chlorine (Cl2), and chlorine radicals (Cl•), for de novo formation of PCDDs/Fs. The most predominant gas-phase chlorine species in a combustion zone is HCl, which is a weak carbon chlorinating agent (12, 13) due to the strength of the H-Cl bond. As temperature decreases in the postcombustion zone, HCl is predicted by thermo-equilibrium to convert to molecular Cl2, a much stronger oxidant and chlorinating agent (14). However, measurements taken at coal-powered utilities reveal that Cl2 concentrations are much lower than expected (14). It should be acknowledge that on-line measurements of all three chlorine species (HCl, Cl2, and Cl•) is difficult, however, predicting gas-phase speciation on thermodynamic equilibrium alone will also provide considerable inaccuracies because equivalence ratio, reaction temperature, and kinetics also influence the speciation. Recent chemical kinetic studies on the chlorocombustion system (14-16) have developed and verified mechanisms outlining the elementary reaction steps of chlorine species, thus enabling its prediction in the post-flame region. In summary, chlorine that is released from a chlorine fuel is rapidly transformed to HCl via hydrogen abstraction reaction with the hydrocarbon fuel. Fuel-rich environments therefore produce greater amounts of HCl. Under fuel lean conditions, HCl can readily react with the more abundant oxidizing radicals (OH, HO2) to produce Cl•, which is very reactive. Procaccini’s (16) studies revealed that a significant amount of Cl•, higher than 1% of the total chlorine (corresponding to ppm levels of Cl• in the flue gas), can be present in the flue gas in the post combustion zone at temperatures as low as 500 K (227 °C) due to kinetic limitations of Cl• recombination reaction to Cl2. This phenomena was also found to be enhanced with increasing quench rate. The affect of quench rate on chlorine speciation is of great interest in the area of PCDD/F formation because it has also been found to be a very significant parameter for the formation of polychlorinated aromatic compounds such as PCDDs, PCDFs, polychlorinated biphenyls, polychlorinated benzenes, and polychlorinated phenols (17). Investigations by Wikstro¨m and Marklund (18) revealed that the most important reactions for PCDD/F formation in the cooling stage of the incinerator were further chlorination of aromatics and chloroaromatic compounds. Microscale experiments have also confirmed the importance of chlorination reactions for homogeneous formation of chlorinated aromatics (19-21). Fixed-bed studies of heterogeneous formation with metal chlorides (e.g., copper or iron chlorides) revealed that metal10.1021/es026262d CCC: $25.00
2003 American Chemical Society Published on Web 01/31/2003
TABLE 1. Experimental Settings and Conditions Employed in the EFR parameter gas environment total gas flow Cl2 addition HCl addition HR temperature VR temperature residence time HRa residence time VRa fly ash feeding
FIGURE 1. Schematic drawing of the EFR employed in this study, drawing not to scale. bound chlorine present in fly ash can also be a sufficient chlorinating agent for PCDDs/Fs formation via de novo reactions (22-25). Chlorine in fly ash can be present in at least three different forms: namely, as inorganic chlorides such as metal- and/or earth metal- (e.g., copper, iron, sodium, and calcium) chlorides, extractable organic chlorides (e.g., condensed chloroaromatics), and nonextractable organic chloride (OCl) compounds present in the fly ash matrix. The role of the different forms of solid-phase chlorine for de novo formation of PCDDs/Fs was investigated by Addink and Altwicker (26) with isotopically labeled 37Cl. The most important chlorine source in the ash matrix for de novo formation of PCDDs/Fs was found to be the metal chlorides and the preexisting C-Cl bonds present in the OCl in the ash. The water-soluble earth metal chlorides were observed to be an insignificant source. The latter suggests that the transfer of chlorine between a metal, e.g. copper, and carbon is much faster than that between sodium and carbon for example, which supports the theory that CuCl2 is one of the important chlorination/oxidation agents for de novo formation of PCDDs/Fs as postulated by Stieglitz et al. (25). Furthermore, P. Weber et al. (27) showed that the transfer of chlorine from CuCl2 to the macromolecular carbon structure in the ash to form OCl occurs very fast, even in a helium atmosphere, according to the following reaction:
2 CuCl2 + R-H f 2CuCl + R-Cl + HCl Moreover, the amount of OCl formed was found to be directly dependent on the reduction of Cu(II) to Cu(I). Additional investigations on gas-phase and solid-phase chlorination during de novo PCDD/F formation would supplement the above findings and help elucidate some of the important pathways of PCDD/F formation. This in turn aids development of new control techniques and strategies to minimize the formation of PCDDs/Fs from combustion processes.
Experimental Section Experiments were conducted in an entrained-flow reactor (EFR), consisting of two concentric quartz tube reactors connected in series, viz., a horizontal reactor (HR) and a vertical reactor (VR). A schematic drawing of the EFR setup is shown in Figure 1. The HR was operated at a temperature of 1000 °C to simulate the high-temperature post-flame zone in an incinerator, while the lower post-flame zone was studied in the VR using a quenched temperature profile of 650-240 °C. The temperature conditions and the profiles were determined in separate experiments to avoid interference with the PCDDs/Fs measurements. A mass-flow controller, delivering a constant gas flow of 11 L/min (at 25 °C and 1 atm) of either N2 or air, was employed in all experiments,
setting N2 or air constant 11 L/min at STP 0 to 25 mL/min at STP, equal to 0-200 ppm in the gas phase 0 to 100 mL/min at STP, equal to 0-400 ppm in the gas phase 1000 or 500 °C 650-240 °C or 500-240 °C 0.6 s at T ) 1000 °C; 1.2 s at T ) 500 °C 1.0 s 1.0 ( 0.1 g/h
chlorine added
chlorine speciation in the VR
HCl added in N2 HCl added in air Cl2 added into N2 or air at 1000 °C Cl2 added into N2 or air at 500 °C
HCl HCl/Cl2/Cl• Cl2/Cl• Cl2
a The residence times in the HR and VR are calculated from the reaction temperature employed and the reactor volume.
giving rise to a gas-phase residence time of 0.6 and 1.0 s in the HR and VR, respectively. An extracted derivative of a fly ash collected from the electrostatic precipitator of a grate-fired municipal solid waste incinerator, designated as EX (extracted) ash, was used in all experiments as the carbon source for this study. The fly ash was continuously fed into the VR during each experiment at a rate of 1 ( 0.10 g/h, approximately the particle concentration found in the raw gas of real MSW and coal incinerators. In addition to constant feeding of the ash into the gas-phase, a gradual buildup of fly ash deposit on the walls of the VR was also achieved to further simulate the environment of a real combustor. The active fly ash in each experiment is thus a combination of the gas-phase ash (with a residence time of seconds) and a thin layer of wall ash (with residence time of hours). To study the role of gas-phase chlorination agents (HCl, Cl2, and Cl•) for de novo formation of PCDDs/Fs, Cl2 and HCl were added independently into the HR at various temperatures and gas-phase environments to set the desired chlorine speciation in the gas-phase. Isotopically labeled 37Cl2 gas (Cambridge Isotope Laboratories, CAS Number 13981-73-2) was also used in one experiment to decipher the role of ashbound vs gas-phase chlorine on PCDDs/Fs de novo formation. Table 1 lists all the experimental settings and conditions employed in the 32 de novo experiments conducted in this study. The reactor was flushed with compressed air, rinsed with 250 mL of distilled H2O, and baked to 1000 °C in an air environment between each experiment to clean the surface of remaining fly ash and minimize the memory affects between the individual runs. Sampling of PCDDs/Fs before a selection of the experimental runs showed nondetectable PCDDs/Fs levels. A detailed description of the EFR, as well as preparations and properties of the fly ash can be found elsewhere (28). Sampling of the entire gas flow from the EFR and cleanup and analyses of the mono-octa PCDD/F homologues were conducted according to a slightly modified version of USEPA Method 0023a (29). About 50% of the ash added into the reactor during each experiment was further trapped in the sampling train as part of the gas-flow; meanwhile the ash present on the wall of the EFR during each experiment was excluded in the analyses. A high-resolution gas chromatograph (HRGC)/low-resolution mass spectrometer (LRMS) (Agilent) with a 60-m J&W DB Dioxin column and single ion monitoring (SIM) was used for the PCDD/F analyses. The VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1109
FIGURE 2. CHEMKIN/REKINET simulations of the amounts of Cl radicals present in the EFR at six experimental conditions.
FIGURE 3. CHEMKIN/REKINET simulations of the amounts of Cl2 present in the EFR at six experimental conditions. SIM ion windows for the PCDD/F compounds were modified to include all of the ions for the molecule regardless of 35Cl and 37Cl distribution (e.g., for octa chlorinated dibenzofuran m/z 440-469 was scanned), to ensure unambiguous results. To evaluate the results of the experimental work, the theoretical gas-phase concentrations of HCl, Cl2, and Cl• in the EFR were simulated and calculated using REKINET, a program based on CHEMKIN-II (30) to solve homogeneous gas-phase kinetics. The HR was represented as a plug-flow reactor at a constant temperature of 1000 °C. The VR was simulated in REKINET as a homogeneous plug-flow reactor with an exponentially decaying temperature profile of 650 to 240 °C in 1 s. A gas-phase C1-C2 chlorocombustion mechanism originally compiled by Ho et al. (31), and refined and verified by Procaccini (16), was used to simulate the gas-phase chemistry.
Results and Discussion 1. CHEMKIN/REKINET Gas-Phase Calculations. Unless otherwise noted, all CHEMKIN/REKINET simulations employed a constant temperature profile of 1000 °C in the HR and a quenched temperature profile from 650 to 240 °C in the VR. Figures 2 and 3 display the concentration profiles of Cl• and Cl2, respectively, in the HR followed by the VR at six different conditions. As a result of the quenched temperature profile in the VR, Cl• recombined to Cl2. Simulations revealed that the Cl2 dissociates into Cl• when it is exposed to a reaction temperature of 1000 °C in the HR, regardless of whether a N2 or air environment was used. Thus, the main chlorine species entering the VR for these conditions is Cl• (Figures 2 and 3, N2-Cl2/Cl and Air-Cl2/Cl). HCl, on the other hand, does not dissociate into hydrogen radicals and Cl• in an inert (N2) environment (Figures 2 and 3, N2-HCl) due to a stronger bond between H-Cl (103 kcal/mol) than Cl-Cl (58 kcal/ mol). In an air environment, approximately 10 ppm of Cl2 and 1-14 ppm (depending on the quenched temperature profile) of Cl• are formed in the VR due to reactions between HCl and OH and O radicals (Figures 2 and 3, Air-HCl). 1110
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003
Furthermore, CHEMKIN/REKINET simulations showed that, because of the temperature dependence of the Cl2 dissociation into Cl•, no dissociation will occur if the reaction temperature is equal to or less than 500 °C. Simulations with a reaction temperature of 500 °C in the HR and 500 instead of 650 °C in the first reaction zone of the VR (equal to 1/9 of the length of the VR) were conducted to verify that a Cl• free environment in the EFR could be achieved (Figures 2 and 3, N2-500-Cl2). An additional simulation with the default temperature of 1000 °C in the HR and the lower quenched temperature profile (500-240 °C) in the VR in an N2 environment was conducted to further study the impact on the chlorine speciation when only the temperature in the VR was changed. The amount of Cl• in the VR was found to be the same (Figure 2, N2-500-Cl2/Cl) as when the default conditions were employed (N2-Cl2/Cl or Air-Cl2/Cl), i.e., the amount of Cl2/Cl• was not affected by the change in temperature (500 instead of 650 °C) in the first reaction zone of the VR. 2. Results from the EFR Experiments. To study the importance of chlorination agent for the de novo formation of PCDDs/Fs a total of 32 experiments at 10 different conditions were performed with EX ash in both nitrogen and air environments adding either HCl, Cl2, or no chlorine to the gas phase. Table 2 presents the average total of mono to octa PCDDs and PCDFs homologues emitted (( one standard deviation) and the PCDF/PCDD ratio at the different experimental conditions studied. The average degree of chlorination (#Cl) of the PCDDs and PCDFs is also featured to highlight changes in homologue profiles due to changes in experimental conditions. All data presented in this study, expressed in units of picomoles emitted per experiment, are the total PCDD/F emitted during each experiment (wall ash excluded), i.e., a function of formation and destruction reactions occurring during the experimental time. 2.1. Role of Ash-Bound Chlorine. The total chlorine level in the EX ash is 5.7% with 94 µg/g (2.7 µmol/g) present as OCl, i.e., a level significantly higher than the amount of chlorine required to form picomoles of PCDDs/Fs. A total of five experiments (Table 2, expts 1 and 2) were conducted without gas-phase chlorine added, and revealed that the chlorine present in the ash is sufficient to form between 170 and 755 picomoles of PCDDs/Fs. This supports the theory that ash chlorine can be active as a chlorine source for de novo formation of PCDDs/Fs (22-27). The PCDD/F levels detected and the degree of chlorination were significantly higher when experiments were conducted in air compared to those conducted in an inert N2 environment. Despite this, the relative rate of formation of the PCDDs and PCDFs was very similar for the two gas environments, resulting in a PCDF/PCDD ratio of 1.2. Oxygen in the gas-phase has been postulated to have a dual role in the de novo formation of PCDDs/Fs by Stieglitz (25); viz, as an oxidizer of the carbon in the ash and of reduced metals such as Cu(I). Presence of metal chlorides has been shown to enhance the transformation of inorganic chlorine in the ash to OCl and, thus, the activity of the fly ash chlorine for the PCDD/F formation will increase in an air environment. 2.2. Role of HCl. Given that HCl is the most abundant form of chlorine during a real combustion situation, the affect of HCl as a chlorinating agent during de novo synthesis of PCDDs/Fs was investigated. A level of 400 ppm HCl was added to both a N2 and an air environment in conjugation with EX ash. The CHEMKIN/REKINET simulation revealed that negligible dissociation to Cl• would occur in an inert N2 gasphase environment, i.e., the only chlorinating agent presentwould be HCl (Figure 2, N2-HCl). The yields and profiles of the PCDDs/Fs detected when only HCl is present (Table 2, expt 3) were found to be very similar to the results from the experiments performed in only N2 with no additional
TABLE 2. Average Total Picomoles Per Train, Ratio of PCDF/PCDD, and Degree of Chlorination of PCDD and PCDF ( One Standard Deviation Detected in Each of the 10 Experimental Conditionsa expt 1 2 3 4 5 6 7 8 9d 10
N2e-EX air-EX N2e -EX-HCl air-EX-HCl N2e -EX-Cl2 air-EX-Cl2 N2e -EX-Cl2 N2e -EX-Cl2 N2e -EX-Cl2 air-EX-Cl2
temp HR (°C)
temp VR (°C)
Cl2 added (ppm)
Cl speciation gas phase
PCDD picomoles
PCDF picomoles
PCDF/PCDD
#Cl PCDD
#Cl PCDF
1000 1000 1000 1000 1000 1000 500 1000 1000 1000
650-240 650-240 650-240 650-240 650-240 650-240 500-240 500-240 500-240 650-240
0 0 400 400 200c 200 200 200 100 5
none none HCl HCl/Cl2/Cl Cl2/Cl Cl2/Cl Cl2 Cl2/Cl Cl2/Cl Cl2/Cl
20 ( 20 350 ( 120 30 ( 30 540 ( 300 400 ( 200 700 ( 450 1600 950 480 ( 100 100 ( 20
150 ( 150 405 ( 20 140 ( 10 1500 ( 900 3000 ( 1200 2500 ( 450 15000 14300 1700 ( 180 370 ( 100
NDb 1.2 NDb 2.8 7.2 4.7 9.5 15.0 3.6 3.8
NDb 7.2 NDb 6.5 4.5 7.3 7.9 8.0 6.4 7.5
3.6 5.7 2.6 5.2 6.8 7.5 7.9 7.9 7.0 5.5
a Experimental time is 2 hours in all experiments. b ND: No data. When the total level is lower than 20 picomoles/train no degree of chlorination (#Cl) or PCDF/PCDD ratio is calculated because the number is most likely to be affected by detection limits. c The amount of Cl2 added was slightly lower (170 ppm) in some of the experiments due to a problem with the feeding system. d The same experimental condition was employed for the 37Cl2 experiment; the data presented is an average between the experiments conducted with native Cl2 gas before and after the 37Cl2 run. Experimental time 45 min. e Low levels of O2 can be present from leaks in the experimental setup or N2 bottle gas impurity.
FIGURE 4. Amount of PCDDs and PCDFs detected when various amounts between 0 and 200 ppm of Cl2 were added into an air or N2 gas-phase environment in the HR. chlorine added to the gas phase (Table 2, expt 1). This confirms that HCl alone does not behave as an active gasphase chlorine source for de novo formation of PCDDs/Fs. On the other hand, when HCl was added into an air environment, the CHEMKIN/REKINET simulation showed low levels of Cl2 and Cl• in the gas phase (Figures 2 and 3, Air-HCl). The de novo experiments at this condition revealed higher yields of PCDFs, suggestive of Cl2/Cl• activity in the gas phase. The PCDD levels, on the other hand, were almost within the same range as those of experiments conducted in air environment without gas-phase chlorine added, implying that the Cl2/Cl• only affects the formation of PCDFs and not PCDDs. As a result of this, the PCDF/PCDD ratio increased from 1.2 to 2.8 (Table 2, expts 2 and 4). 2.3. Role of Cl2 Concentration in the Gas Phase. Further experiments were conducted to study the importance of different amounts of Cl2 added to the gas phase for de novo formation of PCDDs/Fs. Levels between 0 and 200 ppm of Cl2 were added into both a N2 and an air environment at temperature conditions of 1000 °C in the HR and a quenched temperature profile between 650 and 240 °C in the VR (thus, achieving both Cl2 and Cl• in the flue gas, Figures 2 and 3, N2-Cl2/Cl or Air-Cl2/Cl). A significant increase in the yields, from hundreds to thousands of picomoles of total PCDDs/ Fs was obtained as the Cl2 levels in the gas phase increased (Figure 4). Linear regression analysis of the PCDDs and PCDFs, independently, and their relationship with the amount of Cl2 added results in R2-values of 0.63 and 0.88, respectively. For some of the experiments, a rather large PCDD/F yield variation between the repeats decreased the correlation between the Cl2 level and the PCDD/F yields. A possible explanation could be variations in the location of the ash on the wall in the VR among the experiments and,
thus, variations in the reaction temperature as a quenched profile (650 to 240 °C) was employed in these experiments. Despite the large variation in results, PCDD/F yields appear to be proportional to the amount of Cl2 added into the gas phase. The positive linear relationship found in Figure 4 (R2 ) 0.88) and the significant increase in PCDF levels with the concentration of Cl2 added in the gas phase, indicates that the formation mechanism of the PCDFs is very dependent on the concentration of gas-phase Cl2. Thus, a significant increase in the PCDF/PCDD ratio can be noticed in Table 2 when additional Cl2 was added into the gas phase, an increase even more pronounced between N2 and N2/Cl2 than air versus air/Cl2, due to enhanced PCDDs formation in an air environment. This can be explained by the different reaction environment in the two experimental conditions, a phenomena discussed in detail in a corresponding study (28). With only the fly-ash-bound chlorine present as a chlorine source, a uniform PCDF homologue profile was detected (#Cl PCDF ) 3.6 and 5.7, expts 1 and 2). However, when 200 ppm of gas-phase Cl2 was added, a clear shift in the homologue profile toward the higher chlorinated homologues (#Cl PCDF ) 6.8 and 7.5, expts 5 and 6) was noticed; in addition more than 85% of the PCDFs were detected as hepta- and octa-congeners (75% as octa-CDF). Thus, the presence of additional gas-phase chlorine (Cl2/Cl•) in the flue gas does not only act to increase the de novo formation rate significantly, but it promotes further chlorination of the PCDFs toward hepta- and octa- congeners as well. PCDD formation, on the other hand, appears to be less affected by the presence of gas-phase chlorine, where the yields only increased to a couple of hundred picomoles, and a rather low R2 value (0.63) was calculated between the concentration of Cl2 added and the amount of PCDDs detected. Gas-phase chlorine can theoretically form new carbonchlorine bonds via four possible reactions: (i) by hydrogen replacement in the matrix, (ii) via ligand transfer of chlorine from a metal chloride complex in the ash matrix, (iii) by reactions between Cl• in the gas phase and the ash matrix, or (iv) via secondary chlorination by either Cl2 (electrophilic substitution) or Cl• of aromatics already formed in, and/or released from, the ash matrix. A fifth (v) possible reaction between the gas-phase chlorine and the ash matrix involves the enhancement of carbon degradation properties of the ash by converting the less active metal oxides to more active metal chlorides (32). Additionally, chlorinated aromatic structures have been shown to be more resistant to oxidative degradation in the ash matrix (33), which could promote higher PCDF yields and degrees of chlorination (#Cl) when gas-phase chlorine is present. VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1111
De novo formation of PCDDs/Fs is only a side reaction occurring in the oxidative degradation of the carbonaceous structure in the ash matrix. To study the changes in the flyash composition due to reactions occurring in the VR during the experimental time, total carbon, chlorine, and OCl were analyzed in some of the deposited wall ash after the completed experiments. The total amount of carbon in the EX fly ash was significantly reduced from 1.00% to 0.19% when it was exposed to an N2/Cl2 (200 ppm) environment in the VR for 2 h. Meanwhile, experiments conducted in an environment of only N2 revealed a carbon degradation of 66% (0.44% carbon remaining), which indicates that a low amount of oxygen was present in the EFR even if a so-called O2 free N2 environment was employed. This is likely due to small leaks in the experimental set up or to bottle gas impurity. It has been shown in other studies (23, 34, 35) that very low levels of oxygen in the gas-phase are sufficient to enhance the carbon degradation and the formation of PCDDs/Fs significantly. Nevertheless these data suggest that the change in carbon degradation from 1.00% to 0.19% carbon remaining when 200 ppm Cl2 is added is more significant, which implies that gas-phase Cl2 does indeed enhance carbon degradation (reaction v). 2.4. Role of Cl2 and Cl•. To elucidate the role of the two gas-phase chlorination agents (Cl2 and Cl•) for de novo formation of PCDDs/Fs, an additional experiment was conducted in which the temperature conditions were set below those required for dissociation of Cl2 into Cl• (Figures 2 and 3, N2-500-Cl), i.e., only Cl2 was present as chlorination agent. The amount of PCDDs/Fs detected increased significantly from 3400 (Table 2, expt 5) to 16 600 picomoles, especially the octa-congeners, which can be seen in the increased degree of chlorination (Table 2, expt 7). These results indicate that Cl2, even with or without Cl2/Cl•, is an active chlorinating agent in the de novo synthesis of PCDDs/ Fs. To discern whether the detected higher formation rate originated from changes in chlorine source or from changes in reaction temperature in the VR, additional experiments were performed with the higher reaction temperature of 1000 °C in the HR (to allow dissociation of Cl2 into Cl•) and with the lower quenched temperature profile from 500 to 240 °C in the VR (Figures 2 and 3, N2-500-Cl2/Cl). The results revealed that the formation and profile of PCDDs/Fs when both Cl2/Cl• were present were the same as the experiment producing Cl2 alone in the HR (Table 2, expts 7 and 8). It furthermore implies that the additional formation of Cl• from 200 ppm Cl2 added in the HR had little impact on the PCDD/F de novo formation in the VR. Therefore, only three of the four listed chlorination reactions appear to have significant influence on de novo formation of PCDDs/Fs at these conditions, i.e., reaction (i) via chlorination through hydrogen replacement, or reactions involving metal complexes in the ash (ii) or electrophilic substitution reactions (iv). As can be seen in Table 2, almost all of the PCDDs and PCDFs in these two experiments are present as fully chlorinated molecules (degree of chlorination between 7.9 and 8.0). Considering that the levels of the two octa-congeners were over the calibration range of the analytical method in these samples, the values reported are semiquantitative, but they are still sufficient for comparison between different experiments conducted in this study. Additionally, other very intensive high peaks were detected in expts 7 and 8. The peaks were identified to be fully chlorinated-, naphthalene (8 Cl), monobenzofuran (6 Cl), and monodioxin (6 Cl). The exact levels cannot be discerned because of the lack of sample spikes, but approximate levels of 10 000 picomoles (about the same level as octa-CDF) of hexa-chloro-monobenzofuran and hexa-chloro-monobenzodioxin and in the order of hundreds of picomoles of octa-naphthalene were estimated to be present in the samples. This shows that the chlorination 1112
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 6, 2003
reactions of aromatic compounds are very pronounced when Cl2 (with or without Cl•) is present in the gas-phase and the reaction temperature in the top zone of the VR is changed from 650 to 500 °C. The role of reaction temperature for the de novo formation is discussed in detail in a corresponding study (28). 2.5. Gas-Phase Chlorine versus Solid-Phase (Ash) Chlorine. Two types of chlorine sources, one particle-bound (ash) and one gas-phase (Cl2/Cl•), were present in the N2/Cl2 experiments. To reveal the relative importance of the flyash-bound chlorine and the gas-phase chlorine, one 45-min experiment in a N2 environment with EX ash (native chlorine source) and 100 ppm of 37Cl2 added into the gas phase was conducted. Experiments with exactly the same conditions, with the exception of native chlorine gas instead of labeled 37 Cl, were conducted the day before and the day after the original experiment to be used as references (Table 2, expt 9). The exact levels of the PCDDs/Fs were not quantified in the experiments where labeled 37Cl2 was employed because of the lack of labeled spikes in the sampling, cleanup, and analysis procedures. The data are used only for internal comparison within the homologues of the sample, and thus, are not listed in Table 2. No significant difference in isomer distribution could be noticed between the two chlorine isotopes, however, a shift in the ratio between isotopes (35 and 37) from the normal 75/25 ratio to nearly 15/85 was observed in the 37Cl2 experiment. This indicates that the major chlorine source for de novo formation at these conditions was the gas-phase chlorine and that 1 or 2 chlorines in each PCDDs/Fs molecule (corresponding to 15/85 ratio) originated from the ash matrix. It should be noted that the amount of total chlorine added via the ash was more than 30 times greater than the gas-phase chlorine. This suggests that the 100 ppm of gas-phase chlorine is more reactive than the fly-ash-bound chlorine for formation of PCDDs/Fs or that only a low fraction of the 5.7% of chlorine in the ash is present in an active chlorine form. A recent fixed-bed study (26) revealed that water-soluble chlorides in the ash, such as sodium chloride, are rather poor chlorine sources compared to insoluble chlorides, such as copper chlorides. Water leaching tests of the EX fly ash showed that about 80% of the ash-bound chlorine is water soluble. If the amount of chlorine added via the ash is recalculated as “active” chlorine, i.e., without the water-soluble chlorides, an almost equal amount of chlorine added through the gas and ash phases (2 times higher ash bound chlorine) can be deduced. The importance of metal chlorides for the de novo formation of PCDDs/Fs have been shown in numerous studies before, which leads to the conclusion that the 1 to 2 native chlorines in each PCDDs/Fs molecule most likely originate from reactions with the metal chlorides present in the EX ash. To investigate the individual importance of the previously listed (i to iv) carbon chlorination reactions, an additional sample was taken for 60 min directly after the fly ash feeding and 37Cl2 gas addition was terminated, a native (i.e., with 75:25 isotopic ratio) Cl2 gas was added into the gas phase at a level of 100 ppm. The active chlorine sources in this experiment were, therefore, the additional 100 ppm native gas-phase chlorine and the remaining fly-ash-bound chlorine on the wall (0.2 g of ash) which can theoretically contain both native ash chlorine and 37Cl incorporated into the ash matrix by reactions between the gas phase and the ash during the previous run. After the experiments, the deposited wall ash was analyzed via neutron activation analysis for the distribution in 35 and 37 chlorine isotopes. No significant increase in the amount of 37Cl was found incorporated into the ash; moreover, all of the PCDDs/Fs detected in the gasphase sample taken during this run had normal chlorine isotopic ratios (75/25) that originated from reactions by native chlorine. The absence of increased abundance of the 37Cl in
the wall ash after the additional 60-min experiments could be the result of very fast reactions between the gas-phase chlorine and the active metals in the ash matrix. All potentially incorporated 37Cl in the ash due to reactions occurring in the previous run, if any, were replaced by native gas-phase chlorine during the additional 60-min experiment. As mentioned before, metal chlorides, such as the one present in fly ash, enhance the oxidation/chlorination reaction of a carbon matrix more than the corresponding metal oxides (32), and our data show that some part of the ash chlorine is incorporated into the final PCDDs/Fs molecule. Thus, one likely explanation of the increased de novo formation when Cl2 is added into the gas phase and at the same time incorporation of 1 or 2 chlorine(s) from the matrix could be that the gas-phase chlorine reacts (fast reaction) with the metal oxides and transforms them into an active form of ash chlorine. The newly formed metal chloride will then form new C-Cl bonds in the matrix via ligand transfer reactions, i.e. reaction (ii). Another explanation, albeit less not as likely, could be that carbon chlorination reactions without incorporation of gas-phase chlorine into the ash (reactions i and iv) are more important for de novo formation of PCDDs/Fs than reactions through ligand transfer of the chlorine (ii). 2.6. Implications for Real Combustion Situation. Experiments conducted in this study to investigate the role of gas-phase chlorine on the PCDD/F de novo formation employed concentrations of hundreds of ppm Cl2, resulting in 50 to 5 ppm of Cl• in the gas phase. CHEMKIN/REKINET simulations of chlorine speciation in a typical combustion process revealed that only low ppm levels of Cl2/Cl• are present in the gas phase, with the majority of the chlorine remaining as HCl due to hydrogen reactions (36), which is an inactive chlorination agent for de novo formation of PCDDs/Fs. Experiments with only 5 ppm of Cl2 added into an air environment were thus conducted to study the role of gas-phase chlorine for de novo formation of PCDDs/Fs in a more realistic combustion environment instead of the high ppm levels previously discussed. The effect of higher amounts of Cl2/Cl• in the gas-phase in our experiments can as well be seen in the higher degree of chlorination (#Cl) of the PCDDs and PCDFs homologues detected (Table 2, expts 5-9) compared to a real combustion situation where the average degree of chlorination is closer to 4.5 for both the PCDDs and PCDFs. The difference in degree of chlorination can also be an indication of the absence of other important reactions occurring in a combustion situation, such as hydrogen and precursor reactions, which may yield the lower chlorinated homologues. It should be noted that the amount of chlorine added via the ash is a couple of hundred times higher when only 5 ppm of Cl2 was added into the gas phase. The levels and homologue profiles detected were about the same as when no additional chlorine was added into the gas phase (Table 2, expts 6 and 10). This suggests that the major chlorine source for PCDD/F de novo formation in a real combustion situation is the flyash-bound chlorine, which agrees with the results from the 37Cl experiments, i.e., the major reactive form of chlorine is 2 the metal chlorides in the ash, whose activity is enhanced by the presence of gas-phase chlorine. Nevertheless, the gasphase chlorine can still be an important source for formation of PCDDs/Fs via reaction pathways other than de novo synthesis, such as reactions involving gas-phase products (precursors).
Acknowledgments This research was performed at USEPA and supported in part by an appointment for E.W. and M.T. in the Postgraduate Research Participation Program at the National Risk Management Research Laboratory administered by the Oak Ridge Institute for Science and Education through an integrated
agreement between the U.S. Department of Energy and the U.S. Environmental Protection Agency.
Literature Cited (1) Mattila, H.; Virtanen, T.; Vartiainen, T.; Ruuskanen, J. Chemosphere 1992, 25, 1599-1609. (2) Takeshita, R.; Akimoto, Y. Chemosphere 1989, 19, 345-352. (3) Wikstro¨m, E.; Lo¨fvenius, G.; Rappe, C.; Marklund, S. Environ. Sci. Technol. 1996, 30, 1637-1644. (4) Lemieux, P. M.; Lutes, C. C.; Abbott, J. A.; Aldous, K. M. Environ. Sci. Technol. 2000, 34, 377-384. (5) Hatanaka, T.; Imagawa, T.; Takeuchi, M. Environ. Sci. Technol. 2000, 34, 3920-3924. (6) Yasuhara, A.; Katami, T.; Okuda, T.; Ohno, N.; Shibamoto, T. Environ. Sci. Technol. 2001, 35, 1373-1378. (7) Lenoir, D.; Kaune, A.; Hutzinger, O.; Muetzenich, G.; Horch, K. Chemosphere 1991, 23, 1491-1500. (8) Giugliano, M.; Cernuschi, S.; Ghezzi, U. Chemosphere 1989, 19, 407-411. (9) Wikstro¨m, E.; Marklund, S. Chemosphere 2001, 43, 227-234. (10) Stieglitz, L.; Vogg, H. Chemosphere 1987, 16, 1917-1922. (11) Dickson, L. C.; Karasek, F. W. J. Chromatogr. 1987, 389, 127-137. (12) Gullett, B. K.; Bruce, K. R.; Beach, L. O. Waste Manage. Res. 1990, 8, 203-214. (13) Wikstro¨m, E.; Touati, A.; Telfer, M.; Gullett, B. K. Organohalogen Compd. 2000, 46, 264-267. (14) Senior, C. L.; Sarofim, A. F.; Zeng, T.; Helble, J. J.; Mamani-Paco, R. Fuel Process. Technol. 2000, 63, 197-213. (15) Booty, M. R.; Bozzelli, J. W.; Ho, W. P.; Magee, R. S. Environ. Sci. Technol. 1995, 29, 3059-3063. (16) Procaccini, C. Thesis. The chemistry of chlorine in combustion systems and the gas-phase formation of chlorinated and oxygenated pollutants; Massachusetts Institute of Technology: Cambridge, MA, 1999. (17) Fa¨ngmark, I. E.; van Bavel, B.; Marklund, S.; Stro¨mberg, B.; Berge, N.; Rappe, C. Environ. Sci. Technol. 1993, 27, 1602-1610. (18) Wikstro¨m, E.; Marklund, S. Environ. Sci. Technol. 2000, 34, 604609. (19) Gullett, B. K.; Sarofim, A. F.; Smith, K. A.; Procaccini, C. Trans. IChem. 2000, 78, 47-52. (20) Young, C. M.; Voorhees, K. J. Organohalogen Compd. 1990, 3, 203-206. (21) Sommeling, P. M.; Mulder, P.; Louw, R. Chemosphere 1994, 29, 2015-2018. (22) Bruce, K. R.; Beach, L. O.; Gullett, B. K. Waste Manage. 1991, 11, 97-102. (23) Ryan, S. P.; Altwicker, E. R. Chemosphere 2000, 40, 1009-1014. (24) Addink, R.; Altwicker, E. Environ. Sci. Technol. 1998, 32, 33563359. (25) Stieglitz, L. Environ. Eng. Sci. 1998, 15, 5-18. (26) Addink, R.; Altwicker, E. R. Chemosphere 2001, 44, 1361-1367. (27) Weber, P.; Dinjus, E.; Stieglitz, L. Chemosphere 2001, 42, 579582. (28) Wikstro¨m, E.; Ryan, S.; Touati, A.; Gullett, B. K. Environ. Sci. Technol. 2002, Submitted for publication. (29) Test Method 0023A. In SW-856 (NTIS PB88-239223) Office of Solid Waste and Emergency Response; Office of Solid Waste, USEPA: Washington, DC, 1996. (30) Kee, R. J.; Rupley, F. M.; Miller, M. A.; Coltrin, M. E.; Grcar, J. F.; Meeks, E.; Moffat, H. K.; Lutz, A. E.; Dixon-Lewis, G.; Smooke, M. D.; Warnatz, J.; Evans, G. H.; Larson, R. S.; Mitchell, R. E.; Petzold, L. R.; Reynolds, W. C.; Caracotsios, M.; Stewart, W. E.; Glarborg, P.; Wang, C.; Adigun, O. REKINET program. Reaction Design, Inc.: San Diego, CA, 2000. (31) Ho, W. P.; Yu, Q.; Bozzelli, J. W. Combust. Sci. Technol. 1992, 85, 23-63. (32) Mul, G.; Kapteijn, F.; Moulijn, J. A. Appl. Catal., B: Environmental 1997, 12, 33-47. (33) Hall, C. R.; Holmes, R. J. Carbon 1992, 30, 173. (34) Stieglitz, L.; Eichberger, M.; Schleihauf, J.; Beck, J.; Zwick, G.; Will, R. Chemosphere 1993, 27, 343-350. (35) Ryan, S. Thesis, The formation of polychlorinated dibenzo-pdioxins/dibenzofurans (PCDD/F) via the de novo synthesis in the catalytic extraction process (CEP) and the role of iron chloride; Rensselaer Polytechnic Institute: Troy, NY, 2001. (36) Procaccini, C.; Bozzelli, J. W.; Longwell, J. P.; Smith, K. A.; Sarofim, A. F. Environ. Sci. Technol. 2000, 34, 4565-4570.
Received for review October 22, 2002. Revised manuscript received December 9, 2002. Accepted January 6, 2003. ES026262D VOL. 37, NO. 6, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1113
