Environ. Sci. Technol. 2011, 45, 1034–1040
Formation of PCDD/Fs from the Copper Oxide-Mediated Pyrolysis and Oxidation of 1,2Dichlorobenzene SHADRACK NGANAI, SLAWOMIR M. LOMNICKI, AND BARRY DELLINGER* Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803, United States
Received August 27, 2010. Revised manuscript received September 29, 2010. Accepted November 24, 2010.
Formation of polychorinated dibenzo-p-dioxins (PCDDs) has been demonstrated to occur via surface-mediated reactions of chlorinated phenols. However, polychlorinated dibenzofurans (PCDFs) are observed in much lower yields in laboratory studies than in full-scale combustors where PCDFs are in higher concentrations than PCDDs. This has led to the suggestion that at least PCDFs are formed from elemental carbon in the de novo process. However, the potential for PCDF formation from reactions of chlorinated benzenes has been largely overlooked. In this study, we investigated the potential contribution of chlorinated benzenes to formation of PCDD/Fs using 1,2dichlorobenzene as a surrogate for reactions of other chlorinated benzenes and CuO/silica (3 wt % Cu) as a surrogate for fly ash. Results were similar for oxidative and pyrolytic conditions with a slight increase in more chlorinated products under oxidative conditions. Reaction products included chlorobenzene, polychlorinated benzenes, phenol, 2-monochlorophenol (2MCP), dichlorophenols, and trichlorophenols with yields ranging from 0.01 to 2% for the phenols and from 0.01 to 10% for chlorinated benzenes. 4,6-Dichlorodibenzo furan (4,6-DCDF) and dibenzofuran (DF) were observed in maximum yields of 0.2% and 0.5%, respectively, under pyrolytic conditions and 0.1% and 0.3%, respectively, under oxidative conditions. In previous studies of the pyrolysis of 2-MCP under identical conditions, 4,6-DCDF and dibenzo-p-dioxin (DD) were observed with maximum yields of ∼0.2% and ∼0.1%, respectively, along with trace quantities of 1-monochlorodibenzo-p-dioxin (1-MCDD). Under oxidative conditions, 1-MCDD, DD, and 4,6-DCDF were observed with maximum yields of 0.3%, 0.07% and 0.1%, respectively. When combined with the fact that measured concentrations of chlorinated benzenes are 10-100× that of chlorinated phenolsinfull-scalecombustionsystems,thedatasuggestsurfacemediated reactions of chlorinated benzenes can be a significant source of PCDD/F emissions.
Introduction Seventy percent of emissions of polychlorinated dibenzop-dioxins and dibenzofurans (PCDD/Fs) from combustion and thermal processes have been estimated to be a result of transition metal-mediated reactions in the postflame, cool zone (1-3). Two surface-mediated formation pathways have * Corresponding author e-mail:
[email protected]. 1034
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 45, NO. 3, 2011
been proposed: (1) the de novo pathway in which elemental carbon and transition metals in combustion-generated particles react with oxygen and chlorine to liberate PCDD/F from the carbon matrix, and (2) the precursor pathway in which organic species such as chlorophenols react on a transition metal surface to form PCDD/F (4-7). In the de novo pathway, it has been proposed that nonchlorinated dibenzofuran and dibenzo-p-dioxins scaffolds are first formed from the carbon and then chlorinated to form PCDFs and lesser amounts of PCDDs (8-12). Research on the precursor pathway has demonstrated chlorinated phenols are readily converted to PCDDs and much lesser amounts of PCDFs by reaction over a copper oxide surface (13-15). The high yields of PCDFs from the de novo process and the high yields of PCDDs from the precursor pathway have led to the suggestion that PCDFs are formed primarily from elemental carbon (or PAHs) and PCDDs are formed primarily from chlorinated phenols in full-scale combustors (16, 17). In apparent contradiction to the chlorinated phenol hypothesis, correlations between various incineration parameters and PCDD/F emission concentrations indicated chlorinated benzene, rather than chlorinated phenols, concentrations were the best predictors of total PCDD/F emissions (1, 18-22). These data are complicated by the fact that chlorinated phenols can be difficult to analyze due to their propensity to strongly bind with the surface of particulate matter (23). However, there is little convincing experimental evidence of formation of PCDD/Fs from chlorinated benzenes (24, 25). In spite of the lack of data and a plausible mechanism, it is frequently assumed chlorobenzenes form PCDD/Fs (18, 22). Although research has demonstrated copper oxidemediated reactions can convert very simple hydrocarbons such as ethylene and acetylene to chlorinated benzenes and chlorinated PAHs (26, 27), chlorinated phenols reaction products have not been detected. Chlorophenols do not form from chlorobenzenes during gas-phase pyrolysis, as chlorobenzenes predominantly undergo ring rupture with a small fraction converting to other substituted aromatics under oxidative conditions (28, 29). Consequently, it is currently unclear whether chlorinated benzenes can be converted to PCDD/Fs in sufficient yields to make a significant contribution to their emissions from combustion systems. In an attempt to resolve this issue, this manuscript reports the results of an experimental study of the formation of PCDD/Fs and other byproducts from the copper oxide mediated reaction of 1,2-dichlorobenzene (1,2-DCBz). 1,2DCBz was selected because it is nearly isoelectronic with previously studied 2-monochlorophenol (2-MCP), which provides a basis for comparison of product distributions and yields, and because it is among the highest concentration chlorobenzenes detected in the exhaust of incinerators (30, 31). Three % by weight of copper (5% by weight of copper oxide) on silica was used as the catalytic substrate to facilitate comparison to previously published studies (30, 31). Although oxygen concentrations in combustors are typically reported as 8-12%, this is a spatial and temporal average of a wide range of reaction atmospheres resulting from mixing that can result in nearly oxygen-starved to highly oxygen-rich conditions (32, 33). Consequently, pyrolysis in nitrogen and oxidation in 20% O2 in nitrogen were used to bracket the conditions in full-scale combustors. 10.1021/es102948f
2011 American Chemical Society
Published on Web 12/21/2010
Experimental Section Surface-mediated reactions of 1,2-DCBz over a CuO/silica surface were studied using the System of Thermal Diagnostic Studies (STDS), which is described in detail elsewhere (34). Briefly, the system is composed of a thermal reactor located in a high-temperature furnace housed within a gas chromatographic oven, which facilitates precise temperature control as well as sample introduction. A computer-interfaced control module is used to set and monitor all experimental parameters. A high resolution GC-MS is interfaced in-line with the thermal reactor for chemical analysis of the reactor effluent. The method of incipient wetness was used to prepare the catalytic material, which serves as a surrogate for combustiongenerated, copper-rich fly ash. A water solution of copper(II) nitrate (Aldrich 98% pure), of concentration chosen to obtain 5% copper(II) oxide, was used as the active phase precursor. Silica gel powder (Aldrich, grade 923, 100-200 mesh size) was introduced into the solutions of copper(II) nitrate in the amount for incipient wetness to occur. The samples were then ground and sieved to a mesh size of 100-120 corresponding to a particle size of 120-150 µm. The presence of the CuO phase was confirmed by XRD. A 50-mg aliquot of catalytic material was placed between quartz wool plugs in a 0.3-cm i.d. fused silica reactor in the STDS. Fresh samples of catalyst were used for each experimental run. To avoid condensation of the reaction products, all transfer lines were maintained at a constant temperature of 180 °C. Prior to each experiment, fresh catalytic material was oxidized in situ at 450 °C for 1 h at an air flow rate of 5 cc/min to activate the surface of the sample. Air was replaced with the experimental carrier gas for the actual experiments. The 1,2-DCBz (Aldrich 99% pure) was introduced into the flow stream using a digital syringe pump (KD Scientific, model 100) through a vaporizer maintained at 180 °C. Nitrogen or 20% oxygen in nitrogen was used as a carrier gas for the pyrolysis and oxidation experiments, respectively, and the rate of injection was selected to maintain a constant concentration of 50 ppm of 1,2-DCBz for temperatures ranging from 200 to 550 °C. The overall flow rate of the gas stream was maintained at 5 cc/min. The products from the reaction were analyzed using an in-line Agilent 6890 GC-MSD system. For product separation, a 30-m, 0.25-m i.d., 0.25 film thickness column was used (Restek RTS 5MX) with a temperature hold at -60 °C for the reaction period, followed by a temperature-programmed ramp from -60 to 300 at 10 °C/min. Detection and quantification of the products were obtained on an Agilent 5973 mass spectrometer operated in the full-scan mode from 15 to 350 amu for the duration of the GC run. Products were identified based on both mass spectra of the standards (or alternatively NIST mass spectra library) and the retention time. The yields of the products were calculated using the following: yield ) ([product]/[1,2-DCBz]o) × 100, where [product] is the amount of specific product formed (in moles) and [1,2-DCBz]o is the initial amount of 1,2-DCBz (in moles) injected into the reactor. Each experimental data point presented in the manuscript is an average of three experimental runs. One-sigma error bars are given for each data point on the product curves. Curve fitting of the data was performed using polynomial functions with manual adjustment of parameters to obtain the best fit. Quantitative standards were used to calibrate the MS response for all observed products. The MS responses for nonobserved products were estimated based on calibration responses for standards for observed products. Blank runs of the empty reactor and transfer lines were performed to confirm no
FIGURE 1. Yields of chlorobenzenes from the pyrolysis of 1,2DCBz over copper oxide/silica.
FIGURE 2. Yields of chlorophenols from the pyrolysis of 1,2DCBz over CuO/Silica. PCDD/F formation on the walls of the system and lack of subsequent transfer to the next run.
Results Figures 1-3 present the yields of 1,2-DCBz decomposition products as a function of temperature under pyrolytic conditions. Three analogous figures in the Supporting Information display the data under oxidative conditions. There was little difference between the two data sets. In the following discussion, the pyrolysis data will be discussed in detail, and the differences observed in the oxidation studies will be highlighted as appropriate. Sixty-five percent of the 1,2-DCBz was decomposed by 200 °C under pyrolytic conditions and 75% was decomposed under oxidative conditions. 1,2,3-Trichlorobenzene (1,2,3VOL. 45, NO. 3, 2011 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1035
were detected. However, 4,6-dichlorodibenzofuran (4,6DCDF) and dibenzofuran (DF) were observed with maximum yields of 0.2% at 350 °C and 0.5% at 450 °C, respectively. In spite of the increase of the yield of chlorinated phenols, the yields of 4,6-DCDF and DF were similar under oxidative conditions. However, the temperature of the maximum yield was shifted to 200-250 °C for 4,6-DCDF and 500-550 °C for DF.
Discussion
FIGURE 3. Yields of PCDD/Fs from the pyrolysis of 1,2-DCBz over CuO/silica. TriCBz), which was a 0.4% impurity in the 1,2-DCBz, decomposed to 0.1% at 200 °C. The yields of the chlorinated benzenes and chlorinated phenols from the pyrolytic decomposition of 1,2-DCBz are depicted in Figures 1 and 2. Benzene (Bz), monochlorobenzene (MCBz), tetrachlorobenzenes (1,2,3,4 + 1,2,3,5 TeCBzs), phenol, and 2-MCP were detected in yields of 0.004%, 0.5%, 0.005%, 0.1%, and 0.01%, respectively, at 200 °C. The yields of Bz, trichlorobenzenes (1,2,3- + 1,2,4-TriCBzs), tetrachlorobenzenes (TeCBzs), pentachlorobenzene (PeCBz), phenol, and 2-MCP increased with increasing temperature, achieving maxima of 0.02% at 400 °C, 10% at 400 °C, 10% at 400 °C, 1% at 350 °C, 0.6% at 500 °C, and 0.03% at 350 °C, respectively. The yield of MCBz decreased with increasing temperature to below detection limit (0.0001%) at 550 °C. Dichlorophenols (2,4-DiCP + 2,6DiCP) were observed with a yield of 0.01% at 500 °C. Biphenyl and naphthalene (data not shown) achieved maximum yields of 0.03 at 500 °C and 0.1 at 450 °C, respectively. The total yields of chlorinated benzenes under oxidative conditions were similar but distributed slightly differently than under oxidation. In addition 1 which is not in agreement with ratios measured in full scale combustor where the PCDD/PCDF was observed to be ,1 (3, 17, 42-46). This disagreement is a subject of vigorous discussion within the dioxin research community. In our recent manuscript, it was reported that the reaction of 2-MCP over an Fe2O3/silica surface resulted in a PCDD to PCDF ratio of 0.38, which led to the proposal that iron oxide mediated reactions could be a source of some of the discrepancies between previously reported laboratory and full-scale results (47). We also postulated when chlorophenols are the reactant, the surface radical-radical condensation reaction, which forms PCDF, is inhibited by a competitive reaction between the gas-phase chlorophenols and surface bound radicals that instead form dichloro-, hydroxy-, biphenyl ether (cf. Scheme 1) (39). We suggested if chlorobenzenes, rather than chlorophenols were the reactant, the competitive reaction would not occur and higher PCDF yields (and lower PCDD to PCDF ratios) would be observed. The 1,2-DCBz data support this hypothesis, as only PCDFs and no PCDDs were formed. The competing reaction product, dichloro-, hydroxy-, diphenyl ether was also not observed. Moreover, the temperature-integrated yield of PCDFs (DF + DCDF) from the reaction of 1,2-DCBz were approximately equal for the pyrolysis and 1.5× higher for oxidation than the combined PCDD and PCDF yield (DD + MCDD + DCDF) from the reaction of 2-MCP over copper oxide under the same conditions (cf. Figure 4). An important difference between the 2-MCP and 1,2DCBz results was the presence of two temperature maxima at ∼250-350 °C for 4,6-DCDF and 450-550 °C for DF from the pyrolysis and oxidation of 1,2-DCBz. This can be
SCHEME 1. Comparison of Surface Reactions of Gas-Phase Chlorophenols (Upper Pathway) and Chlorobenzenes (Lower Pathway) (Competitive Reaction of Gas-Phase Chlorophenol with a Surface Radical Forms Dichloro-Dihydroxy Diphenyl Ether Rather than PCDFs. When the Reactant Is a Chlorobenzene, the Competitive Reaction Is Slow, and PCDFs Are Formed)
1036
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 45, NO. 3, 2011
FIGURE 4. (a) PCDD/Fs yields from the pyrolysis of 1,2-DCBz (50 ppm) and 2-MCP (55 ppm) over copper oxide/silica. (b) PCDD/Fs yields from the oxidation in 20% O2 in nitrogen of 1,2-DCBz (50 ppm) and 2-MCP (55 ppm) over copper oxide/silica.
SCHEME 2. Two Pathways of Transformation of Chemisorbed Chlorobenzenes and Chlorophenols
SCHEME 3. Formation of Oxygen- and Carbon-Centered Phenoxyl Radicals from Chemisorbed 1,2-DCBz
attributed to the differences in the chemisorption reactions of 2-MCP and 1,2-DCBz (48) (cf. Scheme 2). Both 2-MCP and 1,2-DCBz adsorb to the surface of copper oxide through the reaction with terminal hydroxyl groups of the surface by H2O or HCl elimination. The experimental results indicate (48) 2-MCP reacts almost exclusively by water elimination resulting in the formation of 2-chlorophenoxy radicals on the surface (upper pathway in Scheme 2 where X ) OH and Y ) Cl). This is not the case for 1,2-DCBz (lower pathway in Scheme 2 where X and Y ) Cl), which can adsorb through both pathways with a preference for the formation of a bidentate surface species. Electron transfer from the ligand to the surface results in reduction of Cu2+ to Cu1+ with concomitant formation of an oxygen-centered organic radical. At lower temperatures, the surface chlorophenoxy species can undergo surface condensation reactions to form 4,6DCDF, desorption to form 2-MCP, and chlorination followed by carbon-oxygen bond cleavage to form chlorinated benzenes. The yields of polychlorinated benzenes are the highest (0.3-10%) followed by the yield of 4,6-DCDF (0.22%)
and the yield of 2-MCP (0.03%). This indicates that the C-O bond cleavage is more favorable than Cu-O bond cleavage. The observation of phenol as low as 200 °C suggests the bidentate species has already formed by this temperature. At higher temperatures, one carbon-oxygen bond can cleave to form phenoxyl radicals in both its oxygen-centered and carbon-centered structures (cf. Scheme 3). These transformations are manifested by the appearance of the nonchlorinated products such as dibenzofuran and phenol with maximum formation yields at 450-500 °C versus other products at 350 °C. More precisely, the phenoxyl radicals formed at these temperatures can either desorb and scavenge H · to form phenol or undergo a condensation reaction to form dibenzofuran (cf. Scheme 4) by a mechanism similar to that previously proposed for formation of 4,6-DCDF (39). The high yields of PCDFs and lack of formation of PCDD observed in our laboratory study may have implications concerning PCDD to PCDF ratios observed in the full-scale combustion systems. Previous laboratory studies of chlorophenols have not been able to reproduce the low PCDD to PCDF ratios observed for full-scale combustors and usually overestimated PCDD emissions (36, 49). Since the concentrations of chlorobenzenes in the exhaust of waste incinerators are comparable to or exceed that of chlorophenols by more than an order of magnitude (30, 31, 50, 51), and the PCDD to PCDF ratio for our 1,2-DCBz study is
