Environ. Sci. Technol. 2005, 39, 3321-3329
Chromium Release from Waste Incineration Air-Pollution-Control Residues T. ASTRUP,* C. ROSENBLAD, S. TRAPP, AND T. H. CHRISTENSEN Environment & Resources DTU, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark
Cr release over time was investigated in batch experiments for eleven air-pollution-control residues from eight different municipal solid waste incinerators covering all major flue gas cleaning technologies. Cr released during 168 h of contact with water showed significant variations among the residues studied. Also for the individual residue, large variations were observed depending on the liquidto-solid ratio used in the leaching test and the degree of carbonation. It is argued that Al(0) present in the residues can control Cr leaching by reducing Cr(VI) released from the solid phase by dissolution and that exposure to oxygens either prior to or during the leaching testsdepletes the reduction capacity of Al(0) leading to increased Cr leaching. A dynamic model is shown to describe Cr release from all investigated residues by accounting for Al(0) oxidation with Cr(VI), O2, and water as well as Cr(VI) dissolution. The paper reveals that Al-O2-Cr(VI) interactions must be considered very carefully when interpreting Cr leaching data.
Introduction Solid residues from waste incineration contain high levels of soluble salts and heavy metals and have potentials for substantial leaching. Incineration residues, especially residues from flue gas cleaning, i.e., air-pollution-control (APC) residues, have been shown to maintain alkaline pH-values and to be able to leach high levels of salts and heavy metals for possibly thousands of years (1, 2). Within the recent decade, research has mainly focused on leaching of cationic heavy metals such as Pb, Cd, and Zn and the means to stabilize the APC residues with respect to leaching of these elements (3-5). However, significantly less emphasis has been put on leaching of anionic heavy metals such as Cr (6). While it is possible to stabilize APC residues chemically with respect to Pb, Cd, and Zn, no attempts have yet been successful in controlling Cr leaching. In consequence, Cr is probably one of the most critical elements in the leachate from these residues. Cr in incineration residues is likely to be present mainly as Cr(VI) due to their oxidized and high pH characteristics. Cai et al. (7) studied relations between solid Al and Cr leaching from APC residues. They found that low Cr leaching was correlated with hydrogen generation under moist anaerobic conditions and that the presence of oxygen in leaching tests resulted in increased Cr leaching compared to similar tests at anaerobic conditions. In addition, it was shown that drying * Corresponding author phone: +45 4525 1600; fax: +45 4593 2850; e-mail:
[email protected]. 10.1021/es049346q CCC: $30.25 Published on Web 03/17/2005
2005 American Chemical Society
of ash samples (previously stored moist and anaerobically) also increased Cr leaching in a following test. Hydrogen generation from incineration ashes has been observed on several occasions and is linked to Al(0) in the ashes (8-10). Hydrogen gas as such cannot reduce Cr(VI) in these alkaline systems (7); however, reduction of Cr(VI) by Al(0) has been shown to occur in residue leachate by addition of Al-foil (7, 11). Cr(VI) reduction by Al(0) has also been suggested to occur in MSW fluidized bed combustion ashes (12). The apparent correlation of high Cr release with the presence of oxygen has often been interpreted as a possible reoxidation of already reduced Cr. However, Cr(VI) reduced by Al(0) in residue systems has been shown to be stable over several weeks and not easily reoxidizable by atmospheric oxygen (11). From the previous research (7-11), it appears most likely that Al(0) reactivity in the ashes is a critical factor affecting Cr leaching from incineration residues. The main candidate for oxidation of Cr(III) in natural systems is Mnoxides (13-16). The most important reactions potentially affecting Cr leaching from alkaline incineration residues can be summarized as
Cr(VI)(s) T Cr(VI)(aq)
(1)
Al(0) + CrO42- + 4H2O f Al(OH)4- + Cr(OH)30 + OH(2) Al(0) + 0.75O2 + 1.5H2O + OH- f Al(OH)4-
(3)
Al(0) + OH- + 3H2O f Al(OH)4- + 1.5H2
(4)
Cr(III)(aq) + 1.5Mn(IV)(s) f Cr(VI)(aq) + 1.5Mn(II)(aq) (5) Cr present in the residues as Cr(VI) is released to solution at a rate likely controlled by the dissolution of Cr(VI)containing phases (eq 1). Once dissolved, Cr(VI) can be reduced to Cr(III) by Al(0) (eq 2). When dissolved oxygen is present, Al(0) can be oxidized following eq 3, and oxidation with water can occur in situations without oxygen (eq 4). Gibb’s free energy of reaction for the oxidation of Al suggests that oxidation of Al(0) by O2 (∆Gr0 ) -842 kJ/mol) may be more favorable than oxidation by Cr(VI) (∆Gr0 ) -661 kJ/ mol) and that oxidation of Al(0) by water (∆Gr0 ) -456 kJ/ mol) may be the least favorable. The above reaction for Cr(III) oxidation by Mn(IV) (eq 5) assumes that Mn(IV) oxides are reduced directly to Mn(II); it should however be noted that Mn(III) oxides are likely formed as intermediates in the reaction and that these oxides may also oxidize Cr(III) to Cr(VI) (15, 17). Although indications of possible interactions between Cr, O2, and Al have been suggested, the mechanisms controlling Cr release from incineration residues are not yet fully resolved. In leaching tests on these residues, highly varying and unpredictable results for Cr are often observed (7). These variations may be due to varying effects of interactions between Cr, O2, Al, and maybe Mn depending on the actual residue and the conditions in the leaching test. With respect to chemical stabilization, Cr leaching may in some cases increase as a result of the stabilization process; even in cases where reducing agents are applied (5). To stabilize incineration residues and control future Cr leaching, more detailed information about the release of Cr from the solid phase is needed. In terms of stabilization processes, the Cr release kinetics is highly important since retention times and reactor VOL. 39, NO. 9, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3321
TABLE 1. Overview of Investigated APC Residuesa R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11
type
filter type
sampling point
incinerator
dry dry semidry semidry semidry wet wet wet wet wet wet
FF FF FF FF ESP ESP + FF ESP + FF ESP ESP ESP + FF ESP + FF
silo big bags silo big bags after ESP, before silo silo ESP after ESP, before silo after ESP, before silo ESP ESP
Nordforbrænding (furnace 1-3) Vega Amagerforbrænding Refa Aarhus (furnace 3) Kara Nordforbrænding (furnace 4) Aarhus (furnace 1) Aarhus (furnace 2) Vestforbrænding (furnace 1-4) Vestforbrænding (furnace 5)
a FF and ESP denote Fabric Filters and Electrostatic Precipitators, respectively. All sampling was done in July-August 2002, except R10 which was sampled July 1998. Incinerator Vestforbrænding, furnace 1-4 are equipped with a rotary kiln.
TABLE 2. Chemical Composition of the Investigated APC Residues SiO2 Al2O3 CaO Fe2O3 K2O MgO MnO2 Na2O TiO2 S Zn Pb As Ba Cd Cr Cu Ni
g/kg g/kg g/kg g/kg g/kg g/kg g/kg g/kg g/kg g/kg g/kg g/kg mg/kg mg/kg mg/kg mg/kg mg/kg mg/kg
R1
R2
R3
R4
R5
R6
R7
R8
R9
R10
R11
78 38 430 11 29 11 0.7 34 2.6 27 36 10 280 460 200 200 880 66
59 31 430 6.7 49 11 0.5 41 2.1 27 13 2.1 250 290 180 180 530 49
67 25 500 6.4 20 10 0.6 22 2.1 36 12 2.8 110 370 90 130 430 42
110 39 390 8.9 41 12 0.7 36 3.8 27 17 3 200 420 170 160 560 26
72 37 290 9.6 70 14 0.6 68 5.2 42 32 7.2 300 600 300 320 1100 40
180 80 240 15 71 23 1.2 68 14 38 24 4.5 280 990 280 450 1500 84
200 80 240 18 59 20 1.2 66 15 56 26 4.1 630 1300 340 590 1500 80
140 67 230 20 85 21 1.1 78 12 45 37 6.3 400 970 530 420 1300 72
190 94 260 19 58 26 1.3 56 16 51 23 3.5 250 1100 190 470 1000 99
200 110 270 27 57 24 1.4 54 16 24 19 6.5 160 1200 250 510 1500 100
93 48 190 10 100 17 0.8 92 8.5 55 55 11 420 790 500 590 1600 60
volumes are important at an industrial scale. Variations among residues may also be significant, since waste incinerators are subject to large variations in waste input, technology, and operational conditions with respect to both the incineration process and the flue gas cleaning. The aim of this paper is to provide an improved understanding of Cr release by investigating a range of APC residues covering all major flue gas cleaning technologies used in waste incineration and thereby to offer an overview of variations in Cr release from these residues. The aim is further to suggest possible control mechanisms for Cr and provide a conceptual model explaining the variations in the observed Cr release.
Experimental Section Solid Residues. Eleven APC residues originating from eight different municipal solid waste incinerators in Denmark were investigated. Two residues were produced in dry flue gas cleaning systems, three were produced in semidry systems, and five residues were fly ashes from incinerators with wet flue gas cleaning. The residues were separated from the flue gas either by electrostatic precipitators or by fabric filters (see ref 18 for details). Details regarding the residues (R1R11) are summarized in Table 1, and their chemical composition is listed in Table 2. All residues were sampled from the ash transportation systems and stored in airtight containers below 10 °C until experiment. All glass and plastic equipment was washed in 10% nitric acid before use. Subsamples of the residues were in the laboratory exposed to atmospheric air in a thin layer under moist conditions to let carbonation reactions take 3322
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 9, 2005
place. The changes in pH during carbonation were occasionally tested in small batches at a liquid-to-solid (L/S) ratio of about 2 L/kg, and the carbonation was continued until the residue pH-values were less than 8.5 (about 2-4 weeks of carbonation). Chromium Release Experiments. Release of Cr from the solid residues, carbonated and untreated, was investigated in batch leaching experiments at L/S 5 L/kg and L/S 100 L/kg over 168 h. For L/S 5 L/kg, 380 g of residue and 1900 mL of water were mixed in 2 L glass bottles, and for L/S 100 L/kg, 21 g and 2100 mL of water were used. The bottles were placed in an end-over-end tumbler rotating with about one rotation per minute. Samples of the leaching solution were taken at intervals, starting from the moment that residue and water was mixed. To do this, bottles were removed from the rotation device and opened to the atmosphere for about 1 min. Solution samples were taken with a syringe and filtrated at 0.45 µm, and the supernatant was acidified with 0.1% concentrated HNO3 by volume. In total, less than 10% of the solution volume was removed from each bottle. At each sampling, pH was measured in the bottle. In the solution samples, Cr was measured using graphite furnace atomic absorption spectrometry (GFAAS) ((1 µg Cr/L). Cr(VI) was also measured colorimetrically in selected samples using the diphenylcarbazide method (19). Hydrogen Gas Experiments. All residues, carbonated and untreated, were characterized with respect to hydrogen generation. This was done by adding 30 g of residue and 80 mL of 6.5 N NaOH to a 100 mL glass bottle that was closed with a rubber stop and metal cap. Generated gas was allowed to escape through a needle inserted in the rubber stop. The
gas was collected in initially water filled glass beakers. The volume of produced gas was registered over time as the volume of displaced water ((0.5 mL). The experiments were performed for about 2-3 weeks. The bottles were shaken from time to time in order to increase ash-solution contact. Oxygen Measurements. Residues R2, R3, R8, and R11 were further studied with respect to oxygen depletion over time in batches at L/S 5 L/kg and L/S 100 L/kg. Batches were made by adding 50 g of ash and 250 mL of water and 2.7 g of ash and 270 mL of water to 300 mL glass bottles, respectively. This gave about 30-50 mL of headspace in the bottles. The relative proportions of ash, water, and headspace in these batches were identical to the batches used in the Cr release experiments. Carbonated samples of the four residues were included at L/S 100 L/kg only. The bottles were closed with a rubber stop and metal cap immediately after adding water to the ash. Noninvasive oxygen measurements (Precision Sensing GmbH, Germany) were facilitated by using oxygen-sensitive luminescent sensor foils mounted inside the glass bottles with silicone filler. The oxygen concentration in the bottle is related to the luminescence decay time of the sensor foil and was measured from the outside of the glass bottle by a fiber optical oxygen meter ((0.05 mg O2/L). Solution samples of a few milliliters were taken at intervals with a needle and syringe through the rubber stop, and the samples were filtrated at 0.45 µm and acidified. The experiments were carried out for about 50 h. Cr in the samples was measured using GFAAS.
Results and Discussion Cr Release. Figures 1 and 2 show dissolved Cr as a function of time in the batch release experiments for the untreated and carbonated residues at L/S 5 L/kg and 100 L/kg. Figure 1 depicts Cr release from dry (R1-R2) and semidry APC residues (R3-R5), and Figure 2 depicts Cr release from fly ashes (R6-R11). Although significant differences regarding the released amounts and release patterns for the individual residues can be observed, there are also important similarities to be recognized. Cr release in batches with carbonated residues at L/S 100 L/kg generally showed an initial fast release to the solution during the first 10 h. This was followed by a slow and moderate release during the following 50-100 h. After 100 h, the release was negligible. For most untreated residues at L/S 100 L/kg, the release curves resembled those of the carbonated residues rather well, but in a few cases (e.g. R2-R3, R5, and R11), Cr concentrations increased at a somewhat slower rate although still approaching similar levels. This suggests kinetics effects throughout the 168-h experimental period. For the residues at L/S 5 L/kg, the Cr release patterns were quite different from the patterns observed at L/S 100 L/kg. In case of untreated samples, the initial Cr concentrations had a level similar to that observed at L/S 100; however, Cr concentrations in most cases dropped rapidly 1-2 orders of magnitude within the first 4-8 h of water contact. After this, Cr concentrations increased again at a much slower rate often approaching an apparently stable level toward the end of the experiment. A few residues behaved qualitatively different, either by showing decreasing concentrations throughout the experiment (R1 and R10) or by showing increasing concentrations similar to the L/S 100 L/kg experiments (R7). The releases at L/S 5 L/kg for carbonated samples were in general qualitatively similar to release curves at L/S 100 L/kg, although the actual amounts of Cr leached out usually were smaller. A few residues behaved qualitatively different, either by showing decreasing concentrations within the first 24 h (R1, R5, and R10) or by showing slightly decreasing concentrations throughout the main part of the experiment (R3).
The Cr release during 168 h of contact with water showed significant variations among the 11 residues studied. Also for the individual residue, large variations were observed depending on the carbonation-state of the residues and the L/S ratio of the experiment. However, the observed Cr release may have some common features as discussed later. Hydrogen Gas Production. The residues were characterized with respect to production of H2 under moist, alkaline, and anaerobic conditions in order to estimate maximum reduction capacity of Al in the residues, i.e., the amounts of “reactive” Al. The H2 production is shown as a function of time in Figures 1 and 2 for the untreated and carbonated residues. It should be noted that the conditions (L/S ratio, mixing, and headspace gas composition) in these experiments, for practical reasons, were different from the Cr release experiments. For almost all residues, the H2 production from the untreated residues was significantly higher than from the carbonated. In batches with untreated residues, an initial very fast H2 production was generally observed within the first few hours. After this, a much slower H2 production was seen with a constant rate throughout the remaining part of the experiment (300-360 h). The anaerobic oxidation of Al(0) generating the H2 most likely produces a coating of Al oxides on the surface of Al(0) potentially causing a decrease in the H2 production rate. H2 production from the carbonated residues apparently lacked the initial fast release observed for the untreated residues; however, H2 was still produced at a slow rate similar to the untreated. This indicates that the exposure to air during carbonation inhibits only the initial fast H2 production and not the production later in the experiment. The residues R5 and R11 did not exhibit the initial fast H2 production as observed for the other residues, and the H2 amounts produced were low in both cases. During carbonation, Al(0) is oxidized by O2, and a surface layer of Al-oxide is created on the Al(0) particles. The differences in H2 production between untreated and carbonated residues in these experiments are likely due to this Al oxidation during carbonation. The results also show that although likely passivated at the surface, Al(0) can still produce H2 with a slow rate similar to the untreated residues. This suggests that electron transfer can still occur across the Al-oxide surface layers. Developments in pH. Changes in pH during the experiments are illustrated in Table 3 by values of initial pH and the steady-state pH after 100 h. It is seen that steady-state pH-values for the untreated residues were all comparable (11.1-12.3), only one residue (R11) had a lower pH of about 10. The pH-values of the carbonated samples were somewhat lower for the dry and semidry residues (R1-R5) compared with the fly ashes (R6-R11). Most untreated residues showed no major change in pH during the release experiments, only R7, R8, and R11 initially had significantly lower pH probably due to acidic components on the surface of ash particles. Carbonated fly ash samples generally had initial pH-values about 2-3 units lower than the steady-state pH-values observed after 100 h. The fact that the carbonated residue samples show higher steady-state pH than the initial pH suggests that the residues are not fully carbonated, and only an outer rim on the ash particles has reacted with atmospheric CO2. However, as the steady-state pH in the experiments in all cases but R11 were lower for the carbonated samples compared with the untreated samples, this carbonated rim did play a role in controlling pH during the batch experiments. Variations among Residues. Table 4 illustrates relations between total contents of solid Cr and Al in the residues, released Cr and “reactive Al” after 100 h. Again, 100 h was chosen as a point of comparison, where Cr release and hydrogen production processes generally had reached a VOL. 39, NO. 9, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3323
FIGURE 1. Cr release and hydrogen production as a function of time for dry and semidry APC residues (R1-R5). Solution Cr concentrations are related to the amount of solid residue used in the individual experiments, i.e., mg Cr/kg residue. stable state and only relatively small changes were observed later in the experiments. Values for reactive Al in Table 4 were calculated based on produced amounts of H2 gas at 100 h according to eq 4 as 1 mg of Al(0) can produce 1.36 mL of H2 gas (8). It should be realized that the values of reactive Al may not be directly compared with Cr release data and that the values more likely represent maximum amounts of reactive Al. Overall, the data for reactive Al indicate that the more solid Al present in the residue, the more gas is produced and the more reactive Al is determined. Quite clearly, the fly ashes are seen to contain more Al than the dry and semidry residues, and in many cases the fly ashes also produced more H2. Only a fraction of the solid Al, as specified in Table 4, is likely to 3324
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 9, 2005
be available in the form of Al(0). However, metallic Al in amounts of a few percent has been reported for APC residues (20). Al(0) in incineration ashes likely exists as small droplets (10) condensed from evaporated metal and can potentially originate from aluminum foils used in food packaging. The fact that dry and semidry residues have lower Al contents per kg of ash than fly ashes is likely due to the generation of the ashes: dry and semidry ashes are composed of fly ash mixed with reacted and unreacted lime and thus include extra mass compared with the fly ashes that are removed prior to wet cleaning systems. An estimate of Cr that was reduced in the experiments at L/S 5 L/kg can be obtained by subtracting the released amounts of Cr in batches carried out with untreated samples
FIGURE 2. Cr release and hydrogen production as a function of time for fly ashes from wet flue gas cleaning systems (R6-R11). Solution Cr concentrations are related to the amount of solid residue used in the individual experiments, i.e., mg Cr/kg residue. from the amounts in batches carried out with carbonated samples. This yields 0.2-7.7 mg of Cr/kg or 3-150 µmol of Cr/kg of ash. Comparing this with the values of reactive Al in Table 4 for the untreated residues, it is seen that the amounts of Cr reduced by Al generally constitute a very small fraction of reactive Al(0) in the residues (
