Environ. Sci. Technol. 2003, 37, 4148-4156
Assessing Influence of Experimental Parameters on Formation of PCDD/F from Ash Derived from Fires of CCA-Treated Wood N. W. TAME, B. Z. DLUGOGORSKI,* AND E. M. KENNEDY Process Safety and Environment Protection Group, School of Engineering, The University of Newcastle, Callaghan, New South Wales 2308, Australia
Ash residues from fires of radiata pine timber, both untreated and treated with chromated copper arsenate (CCA), were analyzed for the presence of polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/F). Fire conditions were simulated using a cone calorimeter. The sensitivity of the magnitude and profile of PCDD/F in the ash under controlled experimental conditions were examined to gain an insight into the formation of PCDD/F in a system containing CCA. The total amount of PCDD/F increased from 2.0 ng/kg of ash (0.05 ng of TE/kg of ash, using WHO-TEF) for untreated radiata pine to a maximum of 2700 ng/kg of ash (78 ng of TE/kg of ash) for 0.94% CCA. Ash containing CCA showed a distinct preference for formation of PCDFs, particularly the tetrachloro homologue. It is concluded that PCDD/F formation predominantly occurred via de novo synthesis during smoldering of the char rather than during the initial flaming and pyrolysis. Furthermore, the composition of the CCA constituents present in the timber was controlled to assess whether the physical presence of Cu, a known catalyst in PCDD/F production, was sufficient to account for the formation of PCDD/F in fires of timber impregnated with CCA.
Introduction Treatment of wood using water-borne solutions is currently the most popular form of wood preservation in Australia, with an estimated 850 000 m3 produced in 1996 (1). Of this quantity, wood impregnated with chromated copper arsenate (CCA) makes up the greatest percentage, with the remainder treated with alkaline copper quaternary salts. CCA contains Cr(VI), Cu(II), and As(V), with the most common formulation containing oxides of each metal. For specific applications, different retentions of CCA in the final product must be achieved to provide the desired protection of the wood. Of the three metals contained in the treatment solution, chromium is present as a fixing agent, while copper acts as a fungicide, and arsenic provides protection against insects (2). A mechanism for the fixation process of Cr, Cu, and As was proposed by Dalhgren and later modified by Pizzi (3-5) and can be summarized as follows: (i) adsorption of Cr(VI) onto cellulose; (ii) reduction of adsorbed Cr(VI) to Cr(III) and subsequent oxidation of functional groups in the timber; * Corresponding author telephone: +61 2 4921 6176; fax: +61 2 4921 6920; e-mail:
[email protected]. 4148
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003
(iii) complexation of Cu(II) with lignin and cellulose; (iv) formation of various precipitates such as CrAsO4, CuCrO4, and Cr2(OH)4CrO4, which may then complex with wood components or be precipitated onto them, although the presence of combined Cu and Cr precipitates has been questioned recently by Bull (6). Despite the availability of other disposal methods, thermal utilization of waste-treated wood is the most attractive option for recovering energy from this renewable material. For this reason, the partitioning between the gas emissions and the ash of the different metals during pyrolysis and combustion of treated wood has been of interest. Previous research has concentrated on reducing the emission of toxic arsenic compounds during combustion of CCA-treated wood as well as on the leaching characteristics of metal compounds from the resulting ash (7-11). Terkildsen (8) concluded that copper was significantly retained in the bottom ash, while chromium was also predominantly found in the bottom ash, and arsenic was mainly in the fly ash from measurements of the metal behavior in the co-combustion of CCA-treated wood with municipal solid waste (MSW). These conclusions are generally supported by the findings of other researchers (7-11). Helsen and Van den Bulck (10, 11) have found that the bulk of metals could be retained in the ash if the treated wood is pyrolyzed at low temperatures, although at the optimum temperature of 350 °C 25% of the As was still released, while negligible amounts of Cr and Cu were given off. The bulk of previous work on PCDD/F from wood and waste wood has concentrated on the measurement of emissions from natural and anthropogenic sources in order to form accurate pollution inventories. Some investigations (12-16) have concluded that the ash residue of native wood contains very little PCDD/F, ranging from 0.23 (15) to 4.9 (13) ng of TE/kg, although the actual levels would be dependent on factors such as wood type and carbon burnout. The ash remaining after the combustion of a wood treated with a solution containing boron, chromium, and copper was found to have increased levels of PCDD/F as reported by Pohlandt and Marutzky (14). In their study, wood was combusted in a two-stage furnace with furnace ash, boiler ash, and fly ash being collected and analyzed. The furnace ash did not exhibit significant PCDD/F concentrations, but the boiler and fly ash showed increases of several magnitudes, even though the copper was found to be more concentrated in the furnace residue. The complex process leading to formation of PCDD/F is commonly described using two theories: (i) the precursor formation from aromatic compounds such as chlorobenzenes or chlorophenols and (ii) de novo synthesis, where fragmentation of aromatic structures from a carbon matrix (17) or coupling of short-chain aliphatics (18) leads to precursor compounds. Different mechanistic studies have investigated the effect of copper, usually as CuO or CuCl2, as a catalyst in various steps for the formation of PCDD/F. Especially CuCl2, which can be formed from CuO, has been widely acknowledged to facilitate chlorination of precursors on fly ash in municipal solid waste incinerators (MSWI). Taylor et al. (19) and Lenoir et al. (20) proposed that acetylene is initially chlorinated before forming stable radical complexes with copper species, followed by the condensation and subsequent desorption of chlorobenzene. CuO also promotes the combination of aromatic rings of PCDD/F precursors. Gullet and Bruce (21) concluded that chlorophenols condense via an Ullman reaction to form PCDD in the presence of a metalbased catalyst, the catalytic effect being the most pronounced for CuO. 10.1021/es0304143 CCC: $25.00
2003 American Chemical Society Published on Web 08/15/2003
FIGURE 1. Schematic of the cone calorimeter used to simulate small-scale fires. Several research groups have investigated the relationship between the amount of copper in the fuel and the formation of PCDD/F. Increasing the concentration of copper, in the form of CuCl2, was found to increase the total amount of PCDD/F formed upon heating the fly ash from MSW incineration (22) and PVC combustion (23). Since the total amount of PCDD/F formed did not exhibit an eventual zero order dependence with CuCl2 loading expected for a catalyst, it was resolved that CuCl2, added as a model component, acted as a source of chlorine. Chlorination of PCDD/F or precursor molecules can occur either via the Deacon reaction (22) or by direct donation of chlorine from the metal when O2 is not available to facilitate the Deacon reaction (24). The dominance of higher chlorinated PCDD/F on congener profiles from experiments containing CuCl2 supports this conclusion. A conflicting result for the yield of PCDD/F with CuCl2 was reported by Luijk et al. (25), showing a nonmonotonic dependence of the formation of PCDD/F on the loading of CuCl2. In a previous paper, we have reported a significant increase in PCDD/F levels in the bottom ash from fires of CCA-treated wood (26). The aim of the current study is to examine the influence of certain experimental parameters on PCDD/F formation and therefore elicit a more detailed understanding of the synthesis of PCDD/F in this system. To test the role of the metal compounds, treated wood samples of increasing CCA content were combusted. Experiments with radiata pine mixed with CuO powder and soaked in CuSO4‚5H2O, K2Cr2O7, and 0.03% CCA were also conducted to assess the influence of the metal species. The temperature profiles in the wood sample during flaming and smoldering combustion were recorded in an attempt to link the PCDD/F formation with the temperature history. The temperature measurements provided the necessary information to test the hypothesis as to whether dioxins are formed in the solid residue during the flaming combustion.
Materials and Methods Materials. Commercial samples of CCA-treated Pinus radiata (radiata pine) were selected according to their varying metal content, designated as H levels in Australia, while for comparison a piece of untreated, seasoned P. radiata was also obtained. Deionized water was used for metal and chlorine analyses, while GC residue-analysis grade hexane, dichloromethane, acetone, and toluene (obtained from Omnisolv,
EM Science, Germany) were used for the PCDD/F analysis. Other reagents used were supplied by Ajax Finechem, Australia (H2SO4), and Riedel-de Hae¨n, Germany (H2O2). 13C -labeled PCDD/F standards were supplied (in nonane) 12 by Wellington Labs, Canada. Untreated pine was impregnated with CuSO4‚5H2O, K2Cr2O7, and CuO, which were acquired from Chem-supply (Australia), BDH (Australia), and Hopkin and Williams (England), respectively. A 50% (w/w) CCA salt solution (Cu:Cr:As, 24:39:37) for treatment was obtained from Koppers-Arch (Australia). Apparatus. A cone calorimeter (Fire Testing Technology, United Kingdom) illustrated in Figure 1 was employed to simulate the burning of wood under conditions corresponding to those present in small-scale combustion heaters, domestic stoves, and outdoor fires (27). In fire research, the cone calorimeter serves as a standard testing apparatus (28) for evaluating the performance of a specimen in model fires by measurement of time to ignition, heat release rate, mass loss, and production of smoke and gases. Fire conditions are simulated by an incident radiant heat flux produced by a temperature-controlled, copper heating element that is wound into a conical shape. The dimensions of the cone are designed to impose a one-dimensional heat flux upon the horizontal surface of a sample. The temperature of the infrared radiator was adjusted to obtain a desired heat flux (e.g., 50 kW/m2). Samples were loaded in aluminum foillined, stainless steel holders with depth of 40 mm and surface area of 0.0088 m2. Packed samples were placed under the cone element with an external electric spark unit used to ignite the sample. Air was drawn up into the exhaust hood at a set flow rate using an exhaust fan, while analyzers and a load cell connected to a data logger allowed the continuous measurement of O2, CO, and CO2 concentration and mass loss data. The oxygen content of the exhaust stream, in conjunction with the total flow rate and composition of the ambient gases, was used to calculate the rate of heat released by the fuel with the effect of CO production assumed negligible. A sample holder, adapted to fit thermocouples at four depths, facilitated measurement of the sample temperature during firing. A logger independent of the cone information acquisition system recorded the temperature data. Sample Analysis. Metal concentrations in each initial wood sample were quantified according to AS1605 (29). Wood samples dried at 105 °C and ground to 2.4, a shift toward lignin occurs (33). At a CCA VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4153
TABLE 2. PCDD/F Concentrations in Ash from Samples with Different Copper Agents Presenta CuO (no. 6) Cu, % ∑PCDD:∑PCDF total PCDD/F Cl4-Cl8 WHO98-TEQ ∑TCDD 2,3,7,8-TCDD ∑PeCDD 1,2,3,7,8-PeCDD ∑HxCDD 1,2,3,4,7,8-HxCDD 1,2,3,6,7,8-HxCDD 1,2,3,7,8,9-HxCDD ∑HpCDD 1,2,3,4,6,7,8-HpCDD OCDD ∑TCDF 2,3,7,8-TCDF ∑PeCDF 1,2,3,7,8-PeCDF 2,3,4,7,8-PeCDF ∑HxCDF 1,2,3,4,7,8-HxCDF 1,2,3,6,7,8-HxCDF 2,3,4,6,7,8-HxCDF 1,2,3,7,8,9-HxCDF ∑HpCDF 1,2,3,4,6,7,8-HpCDF 1,2,3,4,7,8,9-HpCDF OCDF a
9.00 6.8 56 0.6 0.3 nd 0.8 nd 1.7 0.8 nd nd 4.4 2.1 41 2.3 0.8 1.5 0.7 0.6 0.9 0.2 0.8 0.6 nd 2.4 1.1 nd nd
CuSO4 (no. 7) 1.47 0.046
K2Cr2O7 (no. 8) nd 1.7
Sample Concentration (ng/kg of ash) 3300 19 8.2 0.6 82 0.3 0.4 0.2 33 0.3 0.7 0.3 9.9 2.9 0.3 nd nd nd 0.7 nd 7.8 1.5 1.7 1.0 15 7.1 2400 2.9 13 0.9 690 1.6 6.8 0.1 10 0.1 110 1.0 3.1 nd 3.4 nd 2.3 nd 0.2 nd 11 0.4 2.9 nd nd nd 1.6 1.0
0.30% CCA (surrogate) (no. 9) 0.12 0.081 340 11 2.4 1.2 2.7 0.6 0.5 nd 0.1 0.2 1.9 0.8 18 250 51 51 7.0 6.4 14 2.9 1.8 1.4 0.2 0.7 1.0 0.2 1.2
0.94% CCA (commercial) (no. 3) 0.23 0.050 2700 78 76 14 24 4.8 7.9 0.2 1.5 1.3 5.0 2.1 16 2000 240 490 73 58 69 13 13 6.3 1.3 7.5 3.7 0.4 2.0
nd, target analyte was not detected. ∑PCDD:∑PCDF, ratio of total tetra-octa-PCDD to total tetra-octa-PCDF.
FIGURE 5. Distribution of total peak area among the assigned peaks in the TCDF homologue from a DB-5 column. Data are presented for samples 3 and 7. The 4-digit numbers refer to the positions substituted with Cl in the parent dibenzofuran molecule. The toxic 2,3,7,8 congener is highlighted in bold, and coeluting congeners with three Cl on one ring are indicated with gray text. solution pH of 5, the segregation of Cr was found by Pizzi (34) to be 95% bound to lignin, with the remainder to the holocellulose. To reproduce the magnitude of PCDD/F observed in ash from commercial CCA-treated radiata pine, 4154
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003
adjustment of the initial solution to a pH of 2 might be required. Effect of Temperature History. Figure 6 represents the temperature history of sample 3 (0.94% CCA) as a function
FIGURE 6. Representative ash temperature through the experimental run, with total PCDD/F amounts resulting from samples quenched.
FIGURE 7. Homologue distributions for samples 10-12 and 3. of experimental time for a heat flux of 50 kW/m2. Measurements taken at the bottom of the sample and at a height of 15 mm from the bottom are presented only, as the two thermocouples set higher in the sample became exposed above the sample surface prior to the commencement of the cooling period. Two samples (10 and 11) were quenched with liquid nitrogen immediately following flame extinction and removal of the radiation source. Upon analysis of these remains, the PCDD/F levels in the ash were calculated as 26 and 40 ng/kg of ash for samples 10 and 11, respectively. A further sample quenched at 45 min contained 320 ng/kg of ash, while the average concentration for sample 3 was 2700 ng/kg of ash. The change in homologue distribution among these samples is presented in Figure 7. Pyrolysis of the sample below the surface during flaming combustion is not responsible for the bulk of PCDD/F presence, based upon the analysis of samples 10-12. The steady increasing concentration of total PCDD/F throughout the smoldering and cooling periods suggest that these conditions are more conducive to de novo synthesis. Although a pyrolytic environment would imply the dechlorination of
PCDD/F, samples 10 and 11 contain appreciable fractions of OCDD. This may be attributed to contamination during the sample preparation or is an indication of the amount of OCDD remaining after decomposition prevalent during pyrolysis. Sample 12, quenched at 45 min, displays the same chlorination pattern as samples that underwent complete cooling. It is possible that the presence of a char structure is required before considerable formation is initiated, so increased PCDD/F levels are not observed until the wood structure has been thermally stressed. A recent characterization of the char structure of red pine after thermal alteration up to 350 °C (35) concluded that the final matrix contained an enhanced amount of aryl compounds. An increase of oxygen containing aryl rings, as in furans, was also observed within the char. The smoldering in the experimental runs, indicated by elevated levels of CO in the exhaust stream, involves the gasification of the char allowing such predibenzofuran structures to decouple and participate in de novo synthesis of PCDF. As previously mentioned, the presence of Cu atoms bound to O atoms, present in carboxylic groups of holocellulose and phenolic hydroxyl and ester VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4155
groups of lignin (6, 36), seems to promote this fragmentation. The experimental observations infer that the magnitude and profile of PCDD/F in the ash from fires of CCAimpregnated wood are a function of the amount of copper bound directly to the wood structure in combination with the residence time of the ash at temperatures suitable for the de novo synthesis. The final PCDD/F concentration and distribution are also the product of competing formation and dechlorination/destruction reactions, which are dependent on O2 levels and the amount of CCA within the sample.The fixation of Cu(II) to lignin and holocellulose is facilitated during the reduction of Cr(VI) to Cr(III); therefore, the physical presence of copper alone is not sufficient to explain the PCDD/F formation in the current system.
Acknowledgments The authors thank the staff of the DAU at the Australian Government Analytical Laboratories for their continued support and patience. This study was funded by the Australian Research Council.
Literature Cited (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
Greaves, H. Proc. Timber Des. Conf. 1997, 7 pp. Humphrey, D. Rev. Inorg. Chem. 2002, 22, 1-40. Pizzi, A. J. Polym. Sci. Chem. Ed. 1982, 20, 707-724. Pizzi, A. J. Polym. Sci. Chem. Ed. 1982, 20, 725-738. Pizzi, A. J. Polym. Sci. Chem. Ed. 1982, 20, 739-764. Bull, D. Wood Sci. Technol. 2001, 34, 459-466. Van den Broeck, K.; Helsen, L.; Vandecasteele, C.; Van den Bulck, E. Analyst 1997, 122, 695-700. Terkildsen, L. Solid Waste Manage.: Therm. Treat. Waste-toEnergy Technol., Proc. Int. Spec. Conf. 1996, 441-450. Solo-Gabriele, H.; Townsend, T.; Messick, B.; Calitu, V. J. Hazard. Mater. 2002, 89, 213-232. Helsen, L.; Van den Bulck, E. J. Anal. Appl. Pyrolysis 2000, 53, 51-79. Helsen, L.; Van den Bulck, E. Environ. Sci. Technol. 2000, 34, 2931-2938. Wunderli, S.; Zennegg, M.; Dolezˇal, I.; Noger, D.; Hasler, P. Organohalogen Compd. 1996, 27, 231-236. Wunderli, S.; Zenneg, M.; Dolezˇal, I.; Gujer, E.; Moser, U.; Wolfensberger, M.; Hasler, P.; Noger, D.; Studer, C.; Karlaganis,
4156
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003
(14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27) (28)
(29) (30) (31) (32) (33) (34) (35) (36)
G. Chemosphere 2000, 40, 641-649. Pohlandt, K.; Marutzky, R. Chemosphere 1994, 28, 1311-1314. Oehme, M. Chemosphere 1995, 30, 1527-1539. Someshwar, A. J. Environ. Qual. 1996, 25, 962-972. Milligan, M.; Altwicker, E. Environ. Sci. Technol. 1995, 29, 13531358. Froese, K.; Hutzinger, O. Environ. Sci. Technol. 1997, 31, 542547. Taylor, P.; Wehrmeier, A.; Sidhu, S.; Lenoir, D.; Schramm, K.; Kettrup, A. Chemosphere 2000, 40, 1297-1303. Lenoir, D.; Wehrmeier, A.; Sidhu, S.; Taylor, P. Chemosphere 2001, 43, 107-114. Gullet, B.; Bruce, K. Chemosphere 1992, 25, 1387-1392. Addink, R.; Altwicker, E. Environ. Eng. Sci. 1998, 15, 19-27. Wang, D.; Xu, X.; Zheng, M.; Chung, C. Chemosphere 2002, 48, 857-863. Grabic, R.; Peka´rek, V.; Ullrich, J.; Puncˇocha´ˇr, M.; Fisˇerova´, E.; Karban, J.; Sˇ ebestova´, M. Chemosphere 2002, 49, 691-696. Luijk, R.; Akkerman, D.; Slot, P.; Olie, K.; Kapteijn, F. Environ. Sci. Technol. 1994, 28, 312-321. Tame, N.; Dlugogorski, B.; Kennedy, E. Chemosphere 2003, 50, 1261-1263. Bhargava, A.; Dlugogorski, B.; Kennedy, E. Fire Saf. J. 2002, 37, 659-673. AS/NZS 3837. Method of test for heat and smoke release rates for materials and products using an oxygen consumption calorimeter; 1998. AS/NZS 1605. Methods for sampling and analysing timber preservatives and preservative-treated timber; 2000. McKee, D. Carbon 1981, 16, 1-119. Iino, F.; Tabor, D.; Imagawa, T.; Gullet, B. Organohalogen Compd. 2001, 50, 447-450. Ryan, J.; Conacher, H.; Panopio, L.; Lau, B.; Hardy, J. J. Chromatogr. 1991, 541, 131-183. Weber, R.; Iino, F.; Imagawa, T.; Takeuchi. M.; Sakurai, T.; Sadakata, M. Chemosphere 2001, 44, 1429-1438. Pizzi, A. Int. J. Wood Preserv. 1983, 3, 89-95. Baldock, J.; Smernik, R. Org. Geochem. 2002, 33, 1093-1109. Zhang, J.; Pascal, D. Holzforschung 2000, 54, 119-122.
Received for review March 31, 2003. Revised manuscript received June 19, 2003. Accepted June 20, 2003. ES0304143