Environ. Sci. Technol. 2004, 38, 5196-5200
Formation of Polychlorinated Dibenzo-p-dioxins/Dibenzofurans from Soot of Benzene and o-Dichlorobenzene Combustion R. ADDINK† AND E. R. ALTWICKER* Department of Chemical and Biological Engineering, Ricketts Building, Rensselaer Polytechnic Institute, Troy, New York 12180-3590
Soots were prepared from flame combustion of benzene and o-dichlorobenzene (ODCB), creating one soot without carbon-chlorine bonds (benzene soot) and one with such bonds (ODCB soot). ODCB soot was tested for PCDD/F formation between 277 and 600 °C without additional chlorine, but levels were very low. Copper and Cu2O were added as potential catalysts for ODCB soot oxidation, but levels of PCDD/F observed were even lower than without these additives. Both benzene soot and ODCB soot produced PCDD/F after adding CuCl2 to the reaction mixtures, suggesting that a (volatile) metal chloride was needed in order for PCDD/F formation to take place. Under the various conditions of [Cu2+], time, and temperature tested, ODCB soot was always more reactive than benzene soot in forming PCDD/F. It seemed plausible that, despite the fact that CuCl2 was very effective in creating C-Cl bonds in benzene soot, the C-Cl bonds created in ODCB soot during preparation were of a reactivity so as to make this soot especially prone to PCDD/F formation. High temperature (gas phase) chlorination of soots by HCl or other chlorinating agents, followed by deposition of these soots and condensed metal chlorides on the ducts and walls of the postcombustion zone, could create an effective mechanism for de novo formation of PCDD/F.
Introduction Formation of polychlorinated dibenzo-p-dioxins (PCDD) and dibenzofurans (PCDF) in municipal solid waste combustion was first reported in 1977 (1). One of the key questions in the formation of these toxic compounds is the role of chlorine since this element is obviously required to produce PCDD/ F. Model studies in the laboratory showed that both inorganic and organic chlorine could be incorporated in PCDD/F (2). Tests in pilot combustors suggested equal reactivity of inorganic and organic chlorine with regards to PCDD/F formation (3, 4). Recent work by Wikstro¨m et al. (5) indicated that the phase of the chlorine played an important role, as these authors found that fly ash-bound inorganic chloride was a more effective chlorinating agent than a gas-phase mix of Cl2/Cl radicals; their data also suggested that gasphase chlorine was converted to solid-phase fly ash-bound * Corresponding author phone: (518)276-6927; e-mail: altwie@ rpi.edu. † Present address: New York State Dept. of Health, Empire State Plaza, Wadsworth Center D-400, P.O. Box 509, Albany, NY 122010509. 5196
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 19, 2004
chloride before inclusion in PCDD/F. In commercial combustion systems, no correlation was found between [chlorine] and [PCDD/F] (6). This could be explained by the fact that chlorine is usually present in great excess as compared to the amount of PCDD/F formed. Even a very small amount of chlorine in the feed would be sufficient to give PCDD/F formation. However, in some laboratory studies, the overall PCDD/F formation rate showed a dependence on [HCl], whereas zero-order dependence would be expected if HCl were not limiting (7). It has been postulated that especially CuCl2 is very active in de novo formation of PCDD/F from carbon, both by chlorinating carbon (first step) and by oxidizing the carbon to release these compounds (second step) (8). Other work has also shown that CuCl2 is indeed both catalyst and chloride source at the same time (9). In comparison with other possible metal ions with some chlorination/oxidation ability, such as Ca2+ and Fe3+, Cu2+ tends to produce much more PCDD/F (10), although it has been suggested that the redox combination of copper/iron could in fact be very reactive (11). A molecular growth mechanism has been proposed by Wehrmeier et al. (12) in which acetylene and CuCl2/CuO react to form perchlorinated C2-C8 compounds, whereas it also has been found that acetylene in HCl/air and CuO produced PCDD/ F (13). Both these studies were carried out at temperatures between 150 and 500-600 °C (i.e., typical postcombustion zone temperatures of an incinerator). Acetylene could also produce polycyclic aromatic hydrocarbons (PAHs) and soot at higher temperatures. Lee et al. passed a O2/CO2/CO/HCl synthetic flue gas over tubes containing a soot/copper deposit at ∼320 °C, producing up to 150 µg/m3 PCDD/F (14). PCDD/F formation has also been reported for a sooting methane flame with Cl2 without copper present (15). Combined, these observations showed the importance of flame chemistry and soot formation in postcombustion zone formation of PCDD/F. Soot deposited on ducting walls could perhaps remain reactive for a considerable amount of time. To obtain more insight in the relevance of soot in PCDD/F formation at moderate temperatures and in the role of chlorine, soot from the combustion of benzene and another soot from o-dichlorobenzene (ODCB) combustion was prepared. ODCB soot was expected to contain a significant amount of carbon-chlorine bonds, in analogy with residual carbon on fly ash. Stieglitz reported on a fly ash with 0.01 g of organic chlorine per gram of carbon (i.e., 1 wt % (16)). This way PCDD/F formation from soots with and without preexisting carbon-chlorine bonds could be compared. CuCl2 was added as a catalyst and chloride source and SiO2 as a support. Soots prepared from benzene and ODCB combustion could also be expected to have a reasonably high [carbon]/[hydrogen] ratio and hence high degree of aromaticity (e.g., a soot prepared from dodecane (C12H26) combustion was found to already have a molar [carbon]/ [hydrogen] ratio of 7:1 (17)). Although not a prerequisite for PCDD/F formation (2), the presence of aromatic rings in organic starting material should certainly stimulate formation of these compounds. Results reported here include formation of PCDD/F from ODCB soot without added CuCl2; the role of [Cu2+], t, and T on formation rates with both soots; and evolution of the amount of organic chlorine in the form of C-Cl bonds in both soots.
Experimental Procedures Soots were prepared from combustion of benzene or o-dichlorobenzene in a setup similar to the laboratory thermal 10.1021/es035197k CCC: $27.50
2004 American Chemical Society Published on Web 08/07/2004
oxidizer described in ref 18. The flame zone temperature was between 700 and 740 °C (flame temperature ∼1500 °C), and soots were collected in a downstream particulate trap kept at 300-375 °C. The pressure was ∼ 30 mm Hg, and the equivalence ratio was ∼1.8. Fuel rates were 2.42 mL/h for benzene and 3.55 mL/h for o-dichlorobenzene; gas flows were 30 mL/min N2 and 220 mL/min air. After preparation, the soots were extracted with hexane and toluene for 24 h to remove extractable organics and dried. Another soot prepared from o-dichlorobenzene combustion (19) under identical conditions showed that soot had reactivity with regards to PCDD/F formation significantly different from the ODCB soot discussed here. Reproducibility between different samples of ODCB soot (even if prepared under nominally identical conditions) has not been established. Reaction mixtures consisted of silica, benzene soot, or ODCB soot, CuCl2, and glass beads as a diluent. Total sample weight was 1 g. The components of the samples were mixed physically. The mixtures were placed in a Pyrex glass or quartz tube against a glass frit with a plug of glass wool at the upstream end of the bed. They were heated for 60 min in a furnace to the desired temperature under a stream of N2. Once a mixture had reached the reaction temperature, a mixture of 10% O2 in N2 was passed through the bed. The flow of gases was regulated with electronic mass flow controllers. Any PCDD/F desorbing from the bed were collected in toluene (cooled with ice) throughout the experiment. Sample cleanup and analysis have been described elsewhere (20). Only the T4CDD-OCDD and T4CDF-OCDF were quantified. Data given are for solid phase and gas phase combined. Blanks of all reaction components contained negligible amounts of PCDD/F. PCDD/F formed from the soots during warming up in nitrogen and due to CuCl2 (which contained no PCDD/F prior to runs but gave some formation on its own, even after 24 h of toluene extraction) were subtracted from the data presented here. The average reproducibility of experiments was 25%. Benzene soot and ODCB soot were tested for organic chlorine (measured as evolved HCl via combustion) present prior to any experiment (i.e., after preparation and hexane/ toluene extraction). In addition, chlorination of both soots with CuCl2 at 299 °C was studied, the temperature at which most PCDD/F formation experiments were carried out. These runs were done in the flow system described previously. After each experiment with CuCl2, chloride was removed from the reaction mixture by repeated extraction with diluted HNO3. Samples were combusted at 698 °C for 90 min in 10% O2/N2. The organic chlorine evolved as HCl and was collected in an ice-cooled impinger containing deionized water. The Cl- was titrated with a 0.02 M AgNO3 solution (standardized against KCl). Organic chlorine was taken as a measure of C-Cl bonds.
Results and Discussion Benzene soot was first tested for presence of any chlorine in its structure. Not surprisingly, no carbon-chlorine bonds could be detected (detection limit ) 4 µmol of chlorine/g of soot), as no chlorine source had been present during its combustion. A subsequent experiment with this soot at 299 °C for 60 min in 10% O2/N2 yielded no detectable amount of PCDD/F. Next, ODCB soot was determined to contain 400 µmol of chlorine/g of soot, an amount equivalent to ∼1.4 wt % and a molar [carbon]/[chlorine] ratio of ∼200: 1. A series of experiments with ODCB soot mixed with silica support, but without added copper, was carried out (Figure 1). There was some PCDD/F formation at each temperature; the minimum found at 450 °C may simply have been an outlier. At no temperature was formation comparable to levels observed with CuCl2 added: ODCB soot on its own appeared to have only limited reactivity. In each run, 0.010 g of ODCB
FIGURE 1. Formation of PCDD/F from ODCB soot without CuCl2. Conditions: mixture of 0.50 g of SiO2, 0.010 g of ODCB-soot, and 0.49 g of glass beads; reaction time ) 60 min; T ) 277-600 °C; flow: 10% O2/N2 at 91 mL/min; n ) 1 for all experiments. soot was used, which contained 4000 nmol of chlorine. In the run with the highest yield in Figure 1, at 299 °C, 0.08 nmol () 0.16 nmol/g) of PCDD/F was formed, which incorporated a total of 0.38 nmol of chlorine. Even a 100× higher yield (i.e., 8 nmol of PCCD/F, incorporating ∼40 nmol chlorine) would still have consumed only 1% of the amount of chlorine present in OCDB soot. When only considering PCDD/F formation, a large excess of chlorine was available. However, in carbon or soot oxidation, the main products will mostly be CO2, CO, short-chain aliphatics, and HCl if the carbon/soot contains chlorine. Thus, formation of PCDD/F is typically a side reaction with only 0.01-0.04 wt % of the carbon forming these compounds (21). From that perspective, most of the chlorine in ODCB soot would likely react to form HCl or chlorinated short-chain aliphatics, so that [chlorine] still could have been rate limiting for PCDD/F formation. The [carbon]/[chlorine] ratio of 200:1 implied that it would be unlikely for two neighboring carbon atoms to both contain chlorine, and resulting PCDD/F may have contained only one or two chlorine atoms. Mono-, di-, and tri-chloro-DD/F were not considered in this study. The experiment at 600 °C could perhaps have generated some HCl through decomposition of carbon-chlorine bonds, creating a strong chlorinating agent to form some higher chlorinated PCDD/F, which could then have escaped destruction after evaporation. However, this was not observed, as the amount of PCDD/F formed at this temperature was not significantly different from yields at lower temperatures. It should be pointed out that at 698 °C (90 min, 10% O2/ N2), ODCB soot was completely destroyed after treatment, and all carbon-chlorine bonds quantitatively converted to HCl. This was established while developing the method for measuring the amount of carbon-chlorine bonds in soots. Therefore, it seems unlikely that any PCDD/F would form from ODCB soot at T > 600 °C. In any case, a soot deposit at the beginning of the postcombustion zone would likely not be exposed to T >600-700 °C. Next, the possibility of lowering the temperature for ODCB soot oxidation was considered by adding metal ions. Since the soot already contained chlorine, addition of an unchlorinated metal (salt) could be sufficient to produce PCDD/F. X-ray fluorescence and TEM analyses of ODCB soot showed that it contained a very low amount of iron (∼0.004 wt %) and no copper (
