MSWI Fly Ash Particle Analysis by Scanning Electron Microscopy

by Scanning Electron Microscopy-Energy Dispersive X-ray Spectroscopy ... Management of Hazardous By-products of Urban Waste Incineration: Some ...
2 downloads 0 Views 163KB Size
Environ. Sci. Technol. 2004, 38, 6669-6675

MSWI Fly Ash Particle Analysis by Scanning Electron Microscopy-Energy Dispersive X-ray Spectroscopy S. GILARDONI,† P. FERMO,† F. CARIATI,† V . G I A N E L L E , ‡ D . P I T E A , * ,§ E. COLLINA,§ AND M. LASAGNI§ Dipartimento di Chimica Inorganica Metallorganica ed Analitica, Universita` degli Studi di Milano, via Venezian 21, 20133 Milano, Italy, Dipartimento di Milano citta`, ARPA Lombardia, via Juvara 22, 20133 Milano, Italy, and Dipartimento di Scienze dell’Ambiente e del Territorio, Universita` degli Studi di Milano-Bicocca, Piazza della Scienza 1, 20126, Milano, Italy

Municipal solid waste incinerator (MSWI) fly ash was investigated to study metal distribution on the particle surface. A detailed investigation into the distribution of chlorine, copper, iron, and zinc was carried out by electron microscopy coupled with X-ray fluorescence spectroscopy. Compositional and leaching test data were used to identify the correlation of significant variables and to formulate a hypothesis about metals speciation. The presence of copper chloride, iron, and zinc oxides was inferred. The iron and zinc accumulation in the submicron nuclei indicates that these metals came from the condensation of volatile species. As far as concerns copper, morphological data together with the element correlation study suggest that this element accumulates on particles involved in heterogeneous condensation processes. Furthermore, during such processes, particles of small size containing copper are formed.

Introduction Incineration, a major method for the treatment of waste matter, reduces its volume and allows the recovery of energy. However metals in waste matter are not destroyed, they are simply transformed so that they are found in the gas phase, in fly ash, and in bottom ash (1). Metals in waste matter are generally present as inorganic compounds that, being unaffected by combustion processes, enter the flue gases in the solid or liquid phase. The entrained particles generally range in size from 1 to 100 µm. Nevertheless some metals vaporize under the conditions reached in the combustion chamber, and move, as vapor, toward postcombustion devices, where homogeneous and heterogeneous condensation takes place (2). Due to processes that occur in the cold zone of the incinerator (3) municipal solid waste incinerator (MSWI) fly ash acts as a catalyst in the formation, at around 300 °C, of PolyChloroDibenzo-p-Dioxins (PCDDs) and * Corresponding author phone: +390264482823; fax: +390264482890; e-mail: [email protected]. † Universita ` degli Studi di Milano. ‡ ARPA Lombardia. § Universita ` degli Studi di Milano-Bicocca. 10.1021/es0494961 CCC: $27.50 Published on Web 11/12/2004

 2004 American Chemical Society

PolyChloroDibenzoFurans (PCDFs). Some authors (3, 4) suggest that such formation reactions involve chlorine transfer to carbonaceous particulate material with carbonchlorine bond formation, followed by the oxidative degradation of the macromolecular structures to carbon dioxide and the side production of aromatic chlorocompounds. Other authors propose the formation of PCDDs and PCDFs from one-ring chloroaromatic compounds (5-7). The determination of parameters related to chloroaromatic compound formation has been the objective of many experiments (3, 8, 9). Fly ash carbon is the primary source for PCDDs and PCDFs, as was proven by the good linear relationship reported by Stieglitz et al. (3); PCDD yield depends on the nature of the added carbon and is probably a function of surface area (4). Fangmark et al. (10) reported results that support the hypothesis that small fly ash particles, which constitute most of the surface area, are the main contributors to the levels of chlorinated aromatic compounds found in flue gases. In fact, the presence of metals such as copper in fly ash is known to be most important, not only for gasification catalysis but also for the formation of chlorinated aromatic structures (11-13) and PCDD/Fs from chemically similar precursors (14). In addition to copper, also zinc and iron could be involved in the catalytic production of chloroaromatic compounds (5, 8, 15). Hinton et al., through single and multiple correlation analysis conducted on actual ash samples, identified Cl, Cu, Na, K, and Zn as promoters of PCDD formation. Although the exact role of these elements is unclear, copper appears to have the most marked effect (16). Nevertheless, with regard to PCDD/F formation, it is clear that the potential chemical activity of one, single, element is not a sufficient condition to ascribe it an actual role. What does matter is that this element is physically accessible to the reactants involved in forming chloroaromatic compounds. The present paper presents an investigation into fly ash particles using scanning electron microscopy coupled with an energy dispersive fluorescence spectroscopy probe (SEM-EDX). A study was made of the distribution of the elements on the surface of the ash particles with the aim of understanding whether the catalytic role of an element could be enhanced by the simultaneous presence of other elements (e.g. chlorine) or by favorable contact with the reactants (e.g. carbon).

Experimental Section Reagents and Materials. Fly ash samples (native fly ash) were collected from the hoppers of electrostatic precipitators in Italy (17). The MSWI plant consists of a primary and a secondary combustion chamber, a heat exchanger, an electrostatic precipitator, and a wet scrubber system. The incinerator was fed raw municipal solid waste (400 ton/day) with a low heating value of 10 000 kJ/kg. The proximate analysis (% w/w) of the MSW was as follows: moisture 30.0, inert material 22.2; combustible fraction 47.8. The ultimate analysis (% w/w) was as follows: carbon 55.38, hydrogen 7.50, oxygen 34.56, nitrogen 0.47, chlorine 1.67, sulfur 0.42. The fly ash inorganic components were determined in a previous study (18). Methanol HPLC gradient grade (Fluka, Switzerland), hexane analysis grade (Merck, Darmast, Germany), and water obtained by a Milli-Q Water Purification System (Millipore, USA) were used for sample preparation. Dicyclohexyl sulfosuccinate sodium salt (purity higher than 98.0%, Fluka) was used as surfactant for fly ash dispersion in water. VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6669

TABLE 1. Definition of Morphological Parameters Employed in the Characterization of Fly Ash Particlesa variables

definition

variability range

x4A/π

0-∞ 1-∞

circularity

P/2xπA

1-∞

form factor

4πA/P2

0-1

convexity

Pconv/P

0-1

Waddel diameter roundness

x4A/πL2

notes 1 ) circle ∞ ) irregular shape 1 ) circle ∞ ) irregular shape 1 ) circle 0 ) irregular shape

a

A indicates the particle area, L stands for length, P perimeter, and Pconv the convex perimeter.

Polycarbonate filters (22 mm diameter, 0.2 µm pores), supplied by Whatman (Whatman International LTD Maidstone, England), were used as the support for particle deposition. Sample Preparation. Investigations into single particles require careful sample preparation to avoid particle overlapping, so the samples were prepared by dispersing the fly ash in a liquid medium, the chosen solvents being hexane, methanol, and water. About 2 mg of fly ash was dispersed in 5 mL of hexane or methanol. For the samples where water was the solvent, about 2 mg of fly ash was dispersed in 50 mL of water containing 7.5 mg of dicyclohexyl sulfosuccinate sodium salt (added to prevent particle coagulation) that was then magnetically stirred for 5 h. Each prepared suspension was filtered on a 0.2 µm Wathman filter in such a way that the particles on the filter were well separated. The filters were dried, mounted onto stub by a double adhesive carbon tape, and finally coated with a thin layer of carbon to improve conductivity and prevent thermal damage. The filter samples were then analyzed by SEM-EDX. From each solvent treatment, at least three filter samples were obtained and analyzed. From hereon in, the solid samples obtained from the above solvent treatments are respectively called the hexane, methanol, and water insoluble fractions. Analysis. SEM-EDX analyses were performed using a Leica (Cambridge) Stereoscan 420 Scanning Electron Microscope, equipped with a Si (Li) Energy Dispersive X-ray analyzer with a Be window. As a consequence, elements lighter than Na could not be detected. The samples were analyzed automatically using Feature Scan Software (19). Magnifications of 25000×, 5000×, 1000×, and 200× were adopted, respectively, allowing the detection of particles with diameters in the ranges 0.2-1 µm, 1-5 µm, 5-25 µm, and 25-125 µm. The particles were identified by visually comparing the image acquired by backscattered electrons with the image in secondary electrons (20). Backscattered electron signals were also used to localize the particles. First, all the surface grid points and Feret diameters were stored; Feret diameters represent the distance between two parallel tangents taken on opposite sides of the particles. During the present study each particle was characterized by 20 Feret diameters. Particle area was calculated from the number of pixels constituting the particle image, and the perimeter was determined by counting the pixels around the particle edge. In addition, particle morphology was described by five morphological parameters: Waddel diameter, circularity, form factor, roundness, and convexity (for mathematical definitions see Table 1). The Waddel diameter corresponds to the diameter of a circle with the same area as the investigated particle. Circularity is equated with sphericity and is a function of perimeter versus area. Form factor is both an area and perimeter function: when the perimeter 6670

9

TABLE 2. Energetic Windows for X-ray Fluorescence Spectroscopy Study, Relative to the 18 Investigated Elementsa

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

element

energetic window/keV

Na Mg Al Si P PbI S Cl K Ca Ti V Cr

1.009-1.112 1.163-1.303 1.403-1.543 1.663-1.823 1.895-2.038 2.283-2.463 2.243-2.838 2.543-2.703 3.223-3.403 3.603-3.783 4.403-4.623 4.843-5.023 5.313-5.493

element

energetic window/keV

Mn Fe Ni Cu Zn PbII Fo1 Fo2 Fo3 Fo4 Fo5 Fo6

5.763-5.983 6.283-6.483 7.383-7.603 7.903-8.143 8.523-8.723 10.403-10.703 2.963-3.143 4.183-4.363 6.663-6.863 9.113-9.413 10.903-11.123 12.943-13.183

a Windows indicated as Fo were used to determine the background signal (Bremsstrahlung radiation).

increases, the form factor decreases and irregularity increases. Roundness is similar to form factor but uses length instead of perimeter, so the mathematical definitions become equivalent only for round particles. Convexity is the ratio between convex perimeter and actual perimeter where convex perimeter is the shortest length joining all the touch points of Feret diameters. In addition to the morphological parameters, an X-ray spectrum was acquired, for each particle, in the energy range of 0-20 keV, using a beam current of 1 nÅ and an accelerating voltage of 20 kV. For each sample the five morphological parameters (Waddel diameter, circularity, roundness, convexity, and form factor) were investigated, 18 elements (18) were analyzed (Na, Mg, Al, Si, P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, and Pb, see Table 2), and the total counts were recorded. Each element, except for lead, had one single energetic window corresponding to it, the correct counts for these windows were obtained by subtracting Bremsstrahlung radiation deduced using 6 energetic windows (Table 2). Instead the monitoring of lead was done by using two energetic windows to avoid Pb-S interference in the 2.2832.383 kV energetic range. The present study assumed an element to be present on the particle surface when the correct counts for that element were higher than twice the background counts. Statistical Treatment of Data. The correlation among the variables was studied by the Pearson correlation coefficient that is calculated by means of SCAN software (Software for Chemometric ANalysis, Minitab Inc., Pasadena, U.S.A.). The probability of a linear relation was determined by the t-test (21)

t ) |p|

xn - 2

x1 - p2

where p is the Pearson coefficient, and n is the number of particles indicating the simultaneous presence of elements whose correlation has to be studied. The particles were subdivided into different groups on the basis of their morphological parameters, through multivariate statistical analysis preformed by SCAN software using complete linkage clustering; Euclidean distance was employed for cluster definition.

Results and Discussion The aim of this paper was to investigate the presence of metals and their salts on fly ash particles. The results will be

FIGURE 1. Dimensional distribution of fly ash particles dispersed in hexane and water; each dimensional range has a width of 0.5 µm and, for simplicity, is indicated by the upper limits. useful in further studies aimed at defining the role of metals in PCDD/F formation during the incineration of municipal solid waste. It is known that de novo synthesis involves fly ash native carbon which contains trace amounts of metals (22). In previous studies (9) we had observed that the most relevant mass fraction in fly ash samples composed by particles of less than 1.5 µm average diameter consisted of carbon. As a consequence, this paper addresses fly ash particles characterized by small diameters. However, to clarify the behavior of some elements, dimensional distribution and solubility studies were extended to particles of up to 100 µm diameter. Since our interest lay mostly in assessing the presence of inorganic species, hexane, that is unable to solubilize such salts, was used as the nonpolar solvent. Instead methanol, which is characterized by a dielectric constant higher than hexane, partially dissolves alkaline chlorides and was thus employed to specifically remove sodium and potassium chlorides without affecting the transition metal chlorides. Water treatment was used to verify the presence of water soluble salts. Particle Dimensional Distributions and Solubility Investigation. Figure 1 shows the dimensional distributions obtained for particles dispersed in hexane and in water. The examined fractions (hexane and water insoluble fractions, see sample preparation section) show the maximum of particle numbers in the submicron regions. It is known that smaller particles are characterized by a high water soluble chloride content; therefore, it can be expected that the water insoluble fraction, compared to the hexane insoluble one, will contain fewer particles smaller than 1 µm. In our case, the number of observed particles for the water insoluble fraction was similar to that of the hexane insoluble fraction for the 0.2-0.5 µm range and lower for the 0.5-1 µm range. Figure 2 shows, for each dimensional range, the percentage of particles containing chlorine, copper, iron, and zinc, for both the hexane and the water insoluble fractions. No consideration was given to particles of >20 µm diameter as the numbers were too small for a statistical approach. It is evident for the hexane insoluble fraction that chlorine is evenly distributed over all the dimensional ranges; also iron and zinc show a fairly uniform distribution. Copper presents two ranges of highest percentage values: one centered at 1.5 µm, the other between 3 and 15 µm. With regard to the water insoluble fraction, the number of particles characterized by the presence of chlorine was, as expected, dramatically reduced; iron and zinc retained a fairly uniform distribution. The behavior of copper was similar

to that of chlorine, and the presence of a copper soluble species such as copper chloride cannot be excluded. A previous study dealing with fly ash chemical composition (22) reported leaching tests with water and an acid solution, together with total sample digestion. These analyses revealed that almost all the copper was present as insoluble species, it being detected only in the solution obtained after total digestion. However, it is worth noting that these results are consistent with those obtained in the present study where copper and chlorine show similar behavior for the smaller sized particles, these particles being the most abundant in terms of number (see Figure 2) but not significantly contributing to the total mass of fly ash. We therefore deduced that, as far as concerns the investigated dimensional range, chlorine and copper are present mainly as water-soluble compounds, whereas zinc and iron preferentially form water insoluble compounds or are trapped in an insoluble matrix. In fact, IR and SEM-EDX spectra in a previous study (23) led to the detection of Fe2O3 and ZnO2 on fly ash particles of less than 1.5-2 µm diameter. In any case it cannot be excluded that smaller iron and zinc fractions could be present as soluble compounds, as discussed later (see the following section). In an endeavor to understand the results obtained for Fe and Zn let us look at Figure 3. The dimensional distribution of particles carrying Fe and Zn is shown for both the hexane and water insoluble fractions. For all the dimensional ranges a greater number of particles containing Fe and Zn were detected after dispersion in water than after dispersion in hexane. Therefore it can be hypothesized that iron and zinc form the inner core of the particles and were thus detected only after the water leaching process had removed the soluble species that, being on the surface, acted as masking agents. The great number of Fe and Zn enriched particles observed in the submicron region is evidence that their insoluble nuclei are characterized by the presence of these elements. Correlation of Variables. To better understand the potential catalytic role of metals (Cu, Fe, and Zn) and their chloride salts in the formation of PCDD/Fs, EDX measurement data were statistically elaborated to indicate possible elements correlation. Pearson correlation coefficients were calculated for the analyzed elements and used to define the probability that a linear correlation among the elements really exists. This was done by determining the t-test values. The study was applied to both the hexane and the methanol insoluble fractions for Cl, Cu, Fe, and Zn, referred to four dimensional ranges (0-2 µm, 2-5 µm, 5-7 µm, and 7-10 µm). Correlations in the water insoluble fraction were calculated only for Fe and Zn as the number of particles characterized by Cl and Cu after water dispersion was too VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6671

FIGURE 2. Percentages of particles characterized by the presence of chlorine, copper, iron, and zinc for different Waddel diameters, after dispersion in hexane or water. small. The calculated percentage probabilities for actual correlations are reported in Table 3, where negative percentages are indicated by the footnote b. With regard to our study, only those correlations defined by a probability higher than 95% were considered significant. Chlorine. In the hexane insoluble fraction, there is, over all the dimensional ranges, a clear link between chlorine and the sodium and potassium, suggesting the presence of NaCl and KCl, as previously reported (22); instead with Si, Al, Mg, and Ca, typical constituents of a silicate matrix, there is a significant negative correlation. The high concentrations of NaCl and KCl in the fly ash precluded the detection of possible Cl correlations with other metals present in small amounts. Thus it was thought useful to analyze the methanol insoluble fraction where only transition metal chlorides were present. In fact, new correlations were revealed: Cu, Mn, Ni, and Pb are linked to Cl in smaller particles, while correlation with Zn appears in bigger particles. The results obtained from examining the methanol insoluble fraction can be used to formulate some hypotheses about metals speciation. It is known that HCl, present in flue gases at combustion chamber outlets, affects the speciation of heavy metals by forming heavy metal chlorides (24) that 6672

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

generate smaller particles by volatilization-condensation processes. In fact Verhulst et al. (25) reported, about a decade ago, the thermodynamic behavior of metal chlorides under the conditions of incineration furnaces: CuCl2 is stable up to 473 K, CuO is stable in the 373-973 K range, and Cu3Cl3(g) and CuCl (g) are the predominant species at even higher temperatures. Therefore, as temperatures in the postcombustion chamber reach up to 1200 K, copper would be present in fly ash mainly as CuO. In addition to temperature, burning conditions involving oxygen, water, and HCl, all quite critical in determining the kind of compounds formed, could result in other species such as complex oxides. Furthermore, it cannot be excluded that there is the formation of oxychlorides when copper chlorides become exposed to air; such an example is atacamite, CuCl2‚3CuO‚3H2O, formed in air between 373 and 473 K. In the methanol insoluble fraction a copper-chlorine correlation was evidenced and could be justified by the CuCl2 formed at low temperatures through the reaction of CuO with HCl, by the CuCl expected from the reducing combustion conditions, and by the presence of oxychlorides. The presence of manganese as MnCl2 (stable up to 1400 K) can be hypothesized, also as MnO given that its formation

FIGURE 3. Dimensional distribution of particles characterized by the presence of iron (a) and zinc (b) after dispersion in hexane or water. free energy is similar to that of MnCl2 (26). As far as concerns nickel and lead, NiCl2 and PbCl2 volatilize at 1250 and 1230 K, respectively. Results from water dispersion experiments, not reported in the previous section, show only the partial dissolution of Mn and complete dissolution of species containing Pb and Ni; these findings confirm the presence of manganese as chlorides and oxides and that of nickel and lead as chlorides. In addition it has been observed that Ni has good correlation with S in the hexane insoluble fraction, suggesting the presence of NiSO4. The Zn-Cl correlation can be ascribed to ZnCl2 formation, as supported by the thermodynamic data (25); in fact, although the experimental evidence indicates the presence of Zn as, primarily, an insoluble species, small amounts of water soluble compounds could be present; a situation that is confirmed by our earlier study (22). Copper. Copper in the hexane insoluble fraction showed a positive correlation with different metals (Cr, Fe, Mn, Ni, Pb, and V); instead in the methanol insoluble fraction there was a negative correlation with Ca. The same correlations were observed for bigger particles in the hexane insoluble fraction and for smaller particles in the methanol insoluble fraction, a consequence of partial solubilization and changed morphological parameters. Some authors (1) have shown that lead in fly ash comes mainly from the condensation of lead chloride, thus the PbCu correlation leads us to suppose that the presence of copper on particle surfaces can also be ascribed to condensation events. Iron is related to Mn, Ni, and Cu in the hexane insoluble fraction but is negative to Ca in the methanol insoluble fraction. Iron and Zinc. In the hexane insoluble fraction Zn is related only to Na, while in the methanol insoluble fraction it correlates positively with Na, Cl, and negatively with Ca.

For Fe and Zn more interesting information comes from the water insoluble fraction. Fe correlates with Zn and anticorrelates with Si and Al. Zn shows positive correlation with Na and Fe and negative correlation with Ca, Al, and Si. On considering these findings together with those on Fe and Zn particle distribution after the water dispersion process, it can be suggested that Fe and Zn compounds (probably oxides) constitute insoluble nuclei with diameters less than 1 µm and low Si, Al, and Ca concentrations. It is known that particles in the submicron region are generated by the condensation of volatile compounds (2); thus the presence of Fe and Zn oxides in these particles suggests that these metals reach the postcombustion chamber as volatile species, confirming the conclusions of other authors (27). It follows that ZnCl2 and iron volatile compounds, perhaps FeCl2 (26), are generated in the combustion chamber and are later converted to oxides in the postcombustion zone. However, in the presence of small amounts of Cl2, the formation of ZnCl2 and FeCl2 from the oxides is not excluded. Morphological Features. The investigated morphological parameters were Waddel diameter, circularity, roundness, form factor, and convexity (see Table 1). The morphological parameters role was investigated in the hexane insoluble fraction, as in this case the solvent induced no modification of particle shape. Particles were submitted to hierarchical cluster analysis (28) using morphological parameters as the discriminating variables. By choosing a similarity level of 65%, five clusters were identified (Table 4); also centroid cluster coordinates are reported. It can be observed (Table 4) that convexity decreases when the Waddel diameter increases, until particles have a diameter of about 6 µm. Since smaller particles are characterized by a higher surface area, they are mostly involved in condensation events that can be expected to reduce perimeter shape irregularity. Hence the increment in the convexity values is not surprising: high VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6673

TABLE 3. Correlation Probabilities, as Percentages, for Chlorine, Iron, Copper, and Zinc, Calculated for Particles Subdivided into Four Dimensional Ranges: 0-2 µm, 2-5 µm, 5-7 µm, and 7-10 µma diameter

Na

Mg

Al

Si

P

S

0-2 2-5 5-7 7-10

99.9 99.9 99.9 99.9

99.9b 40 99.9b 60b

99.9b 95b 99.9b 90b

99.9b 30b 99.9b 95b

10b

80b 90b 10 20

0-2 2-5 5-7 7-10

50 9

80 40 50

99.9

40b 30b 60b 20b

50b 10b

80b 30b 80b 50b

80 60b 10b

90 80 20b 20b

Chlorine Cl K

Pb

Ca

Ti

V

Cr

Mn

99.9b 99.9b 99.9b 99.9b

70b

40b

20b

70 60 70

Hexane 99.9 99.9 99.9 99.9

60b 30b

40b 10b

10b

10

99.9 99.9 50 10

Methanol 30 10 30b 20b

99.9b 95b 90b 40b

20b 10b 20b

30 40

9.9 95

99.9 95

Fe

Ni

Cu

Zn

10

10

80b 40b 10 40

99.9 95

99.9 10

30 40b

40b 50 10b 10b

70 90 99.9 99.9

Iron 0-2 2-5 5-7 7-10

20b 10b 40b 20b

0-2 2-5 5-7 7-10

80b

0-2 2-5 5-7 7-10

20b

20b

10b

10b

10

30 10b

40 10

60b 40 10b

20b

20 40

70b 50 20b 20b

30b

10b

10b

20 60

50 80

80b 20b 20b 20b

Hexane 30b 30b 40b

70b 10 10b 20b

30b

20

50b 10b

20 30b

Methanol 30b 50 50 10b 10b 10b

90b 80b 60b 40b

20b

60b

90b

60b

30b

90b 70b

98b 80b

80b 60b

50b 30b

40 30

70 60

70 95

70 95

80 95

30b 20

70 40

30

20 30

40b 60

20 40b 40

Water 40b 10b

40b 30b

99b

99.9b

99b 95b

99.9b 98b

10

10

10b

30

99 10

95 90

60

Copper 0-2 2-5 5-7 7-10 0-2 2-5 5-7 7-10

10b 20b 30b

20 40 10 20b

10b

60 10b

10b

20b 10b

60

80

40 50

30

60

80

99

30

60

Hexane 10b 10b 20b

95 70 98 80

10 99.9 10 60

30b

40

40

80

50b 40b

99.9 70

99.9 70

99.9 99.9

99.9

99 10

99.9 10

Methanol 50 98b 40b 30b

10b

99.9

20

80 95

99.9 99.9

20 20

10b 60

99.9 10

10

Zinc 0-2 2-5 5-7 7-10

40 50 95 99

0-2 2-5 5-7 7-10

99.9 99.9 99.9 99.9

0-2 2-5 5-7 7-10

99.9

10 20b 40b 60b

10b 20b 20b 80b

60

10b

10b

70b 40b

10b 20b 60b 60b

20b 95b 70b

30b 10b

30b 20b 40b 10b

70b 50b 70b 80b 50b

20b 20b 50b 30b

Hexane 80b 80b 40b 50b 10 10 40 20b

60b 10 70b 50b

Methanol 60 30b 90 20 99.9 20b 99.9 20b

10b 30b 80b 60b

20b

70b 98b 99.9b 80b

20b 30b 40b

10b

70b

99b

80b

40b

80

60

99

30b 80b

98b 99b

90b 70b

60b 50b

10

10 10

95 90

30b 20b

60

20

20

20

20

20

20 10

20

20 10

20 20

10b 10b

30b 40

10b

Water

a

99.9 99.9

95b 60b 50

70b 70b

99.9b 80b 90b

10b 10b

10b

Unless indicated otherwise, probability is below 10%.

b

Indicates negative percentage values.

convexity values reveal that a particle has undergone condensation processes. Each dimensional range (0.2-2 µm, 2-5 µm, 5-7 µm, and 7-10 µm) was investigated separately, revealing more homogeneous morphological features. In fact, the 0.2-2 µm particles gave interesting results. By choosing a 65% similarity level, the particles were subdivided into eight clusters (Table 6674

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 24, 2004

5); centroid coordinates are also reported. More than 75% of the particles characterized by the presence of copper belong to the same cluster (C1 in Table 5), indicating a common formation mechanism. Compared with the other clusters (Table 5), C1 is characterized by a low value for circularity and high values for convexity and form factor. It therefore follows that particles characterized by copper tend

TABLE 4. Coordinates of Cluster Centroids Determined by Hierarchical Cluster Analysis, Performed on Particles Belonging to the Hexane Insoluble Fractiona variables

C1

C2

C3

C4

C5

Waddel diameter roundness circularity form factor convexity

1.32 0.51 1.57 0.46 0.80

6.16 0.42 1.69 0.39 0.80

9.87 0.40 2.12 0.28 0.68

14.99 0.38 2.60 0.22 0.59

20.79 0.40 2.56 0.20 0.57

a All the reported values are pure numbers except for the Waddel diameters that are expressed in microns.

TABLE 5. Coordinates of Cluster Centroids Determined by Hierarchical Cluster Analysis Performed on Particles, 0.2-2 µm in Diameter, Belonging to the Hexane Insoluble Fractiona variable Waddel diameter roundness circularity form factor convexity

C1

C2

C3

C4

C5

C6

C7

C8

1.30 1.59 0.86 0.34 0.33 0.45 0.54 1.15 0.63 1.21 0.68 0.93

0.47 1.55 0.42 0.81

0.50 1.48 0.46 0.82

0.50 1.67 0.37 0.74

0.35 3.17 0.10 0.47

0.63 1.25 0.65 0.89

0.44 2.38 0.18 0.55

0.40 1.91 0.28 0.71

a All the reported values are pure numbers except for the Waddel diameters that are expressed in microns.

to have a more circular shape and regular profile. This last finding supports our hypothesis concerning copper rich particles, allowing us to deduce that such particles come primarily from condensation processes. The findings for iron and zinc were not so definite. The obtained results will be useful in further studies in order to define the role of metals in PCDD and PCDF formation. In fact the condensation mechanism potentially leads to the best contact between the catalytic species (metals) and the reactants (carbon).

Literature Cited (1) Biswas, P.; Yinn, W.; Lin, C.; Yu, Wu. Formation and emission of metallic aerosols from incinerators. J. Aerosol. Sci. 1992, 23, S273-S276. (2) Barton, R. G.; Clark, W. D.; Seeker, W. R. Fate of metals in waste combustion systems. Combust. Sci. Technol. 1990, 74, 327-342. (3) Schwarz, G.; Stieglitz, L. Formation of organohalogen compounds in fly ash by metal-catalyzed oxidation of residual carbon. Chemosphere 1992, 25, 3, 277-282. (4) Stieglitz, L.; Zwick, G.; Beck, J.; Roth, W.; Vogg, H. On the denovo synthesis of PCDD/PCDF on fly ash of municipal waste incinerators. Chemosphere 1989, 18, 1219-1226. (5) Gullet, B. K.; Bruce, K. R; Beach, L. O.; Drago, A. M Mechanistic steps in the production of PCDD and PCDF during waste combustion. Chemosphere 1992, 25, 1387-1392. (6) Milligan, M. S.; Altwicker, E. R. Chlorophenol Reactions on fly ash. 1. Adsorption/desorption equilibria and conversion to polychlorinated dibenzo-p-dioxins. Environ. Sci. Technol. 1996, 30, 225-229. (7) Hell, K.; Altwicker, E. R.; Stieglitz, L.; Addink, R. Comparison of 2,4,6-trichlorophenol conversion to PCDD/PCDF on a MSWIfly ash and a model fly ash. Organohalogen Compd. 1998, 36, 53-58. (8) Hinton, W. S.; Lane, A. M. Effect of zinc, copper, and sodium on formation of polychlorinated dioxins on MSW incinerator fly ash. Chemosphere 1992, 25, 811-819.

(9) Fermo, P.; Cariati, F.; Santacesaria, S.; Bruni, S.; Lasagni, M.; Tettamanti, M.; Collina, E.; Pitea, D. MSWI fly ash native carbon thermal degradation: A TG-FTIR study. Environ. Sci. Technol. 2000, 34, 4370-4375. (10) Fangmark, I.; Stromberg, B.; Berge, N.; Rappe, C. The influence of fly ash load and particle size on the formation of PCDD, PCDF, PCBz and PCB in a pilot incinerator. Waste Manage. Res. 1995, 13, 259-272. (11) Stieglitz, L.; Zwick, G.; Beck, J.; Bautz, H.; Roth, W. Carbonaceous particles in fly ash - a source for de-novo-synthesis of organochloro compounds. Chemosphere 1989, 19, 283-290. (12) Luijk, R.; Akkerman, D.; Slot, P.; Olie, K.; Kapteijn, F. Mechanism of formation of polychlorinated dibenzo-p-dioxins and dibenzofurans in the catalyzed combustion of carbon. Environ. Sci. Technol. 1994, 28, 312-321. (13) Addink, R.; Altwicker, E. R. Role of copper compounds in the de novo synthesis of polychlorinated dibenzo-p-dioxins/dibenzofurans. Environ. Eng. Sci. 1998, 15, 19-27. (14) Gullet, B. K.; Bruce, K. R.; Beach, L. O. Formation of chlorinated organics during solid waste combustion. Waste Manage. Res. 1990, 8, 203-214. (15) Ryan, S. P.; Altwicker, E. R. The formation of polychlorinated dibenzo-p-dioxins/dibenzofurans from carbon model mixtures containing ferrous chloride. Chemosphere 2000, 40, 1009-1014. (16) Hinton, W. S.; Lane, A. M. Characteristics of municipal solid waste incinerator fly ash promoting the formation of polychlorinated dioxins. Chemosphere 1991, 22, 473-483. (17) Lasagni, M.; Collina, E.; Tettamanti, M.; Pitea, D. Kinetics of MSWI fly ash thermal degradation. 1. Empirical rate equation for native carbon gasification. Environ. Sci. Technol. 2000, 34, 130-136. (18) Fermo, P.; Cariati, F.; Pozzi, A.; Demartin, F.; Tettamanti, M.; Collina, E.; Lasagni, M.; Pitea, D.; Puglisi, O.; Russo, U. The Analytical Characterization and Speciation of a Municipal Solid Waste Incinerator Fly Ash: Methods and Preliminary Results. Fresenius J. Anal. Chem. 1999, 365, 666-673. (19) Oxford Instrument, Rev. 1.2. (20) Reimer, L. Scanning electron microscope; Springer-Verlag: 1985. (21) Desimoni, E. Chimica analitica; equilibri ionici e fondamenti di analisi chimica quantitativa; CLUEB (Bologna), 1996; 544 p. (22) Imagawa, T.; Wai Lee, C. Correlation of polychlorinated naphthalenes with polychlorinated dibenzofurans formed from waste incineration. Chemosphere 2001, 44, 1511-1520. (23) Fermo, P.; Cariati, F.; Pozzi, A.; Tettamanti, M.; Collina, E.; Pitea, D. Analytical Characterization of Municipal Solid Waste Incinerator Fly Ash - Part II. Fresenius J. Anal. Chem. 2000, 366, 267-272. (24) Chiang, K. Y.; Wang, K. S.; Lin, F. L.; Chu, W. T. Chloride effects on the speciation and partitioning of heavy metal during the municipal solid waste incineration process. Sci. Total Environ. 1997, 203, 129-140. (25) Verhulst, D.; Buekens, A.; Spencer, P. J.; Eriksson, G. Thermodynamic Behavior of Metal Chlorides and Sulfates under the Conditions of Incineration Furnaces. Environ. Sci. Technol. 1996, 30, 50-56. (26) Fernandez, M. A.; Martinez, L.; Segarra, M.; Garcia, J. C.; Espiell, F. Behavior of heavy metals in the combustion gases of urban waste incinerators. Environ. Sci. Technol. 1992, 26, 10401047. (27) Thipse, S. S.; Dreizin, E. L. Metal partitioning in products of incineration of municipal solid waste. Chemosphere 2002, 46, 837-849. (28) Massart, D. L.; Kaufman, L. The interpretation of analytical chemical data by the use of cluster analysis; John Wiley & Sons: 1983; 237 p.

Received for review April 2, 2004. Revised manuscript received September 14, 2004. Accepted September 28, 2004. ES0494961

VOL. 38, NO. 24, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

6675