Environ. Sci. Technol. 2005, 39, 8481-8489
Nanostructured Sorbents for Capture of Cadmium Species in Combustion Environments MYONG-HWA LEE,† KUK CHO,† APOORVA P. SHAH,‡ AND P R A T I M B I S W A S * ,† Aerosol and Air Quality Research Laboratory, Environmental Engineering Science Program, Washington University in St. Louis, Campus Box 1180, St. Louis, Missouri 63130, and Triton Systems, Inc., 200 Turnpike Road, Chelmsford, Massachusetts 01824
The pathways of cadmium species to form a submicrometer-sized aerosol in a combustion system exhaust were established. Cadmium oxide was the predominant species formed in the experiments and resulted in particles of a mean size of 26-63 nm with number concentrations in the range of 2-8 × 106 cm-3. Two different nanostructured sorbents, a solid montmorillonite (MMT) and an in situ generated agglomerated silica, were used for capture of the cadmium species. The MMT sorbent was not stable at 1000 °C, and structural changes resulted. MMT did not suppress nucleation of cadmium species and partially captured it by weak physisorption as established by the leachability tests. In contrast, the in situ generated silica nanostructured agglomerates had a high surface area, suppressed nucleation of cadmium species vapors, and chemisorbed them effectively resulting in a firm binding, as compared to the MMT sorbent. There is an optimal temperature-time relationship at which the capture process is expected to be most effective. The leaching efficiency under these conditions was less than 3.2%. The nanostructured silica agglomerate size can be tuned for effective capture in existing particle control devices.
Introduction The heavy metal loading in the atmosphere has been increasing in the last century due to anthropogenic activities (1, 2), and there have been reports outlining their inventory in the atmosphere (3-5). In recent years, single-particle spectrometers have been used to determine concentrations of metallic species in the urban atmospheric aerosol and intraparticle correlations to elucidate potential sources (6). Toxic heavy metals are emitted from coal combustors, incinerators, oil combustors, jet engines, smelters, steel production processes, welding, and several other anthropogenic combustion sources (7). The heavy metal species are typically in the sub-micrometer sizes, and it is well-known that conventional particle control devices have a minima in collection efficiency in these size ranges (8, 9), resulting in an enrichment of these species in the exhaust gases. * Corresponding author phone: (314)935-5482; fax: (314)935-5464; e-mail:
[email protected]. † Washington University in St. Louis. ‡ Triton Systems. 10.1021/es0506713 CCC: $30.25 Published on Web 09/15/2005
2005 American Chemical Society
Cadmium is a toxic metal that is a trace constituent of fossil fuels, electrodes in batteries, sludges, contaminated soils, and municipal wastes (10-12). It is also a common ingredient in munitions that are incinerated (13). When the above-mentioned materials are introduced into the combustor, they result in hazardous emissions of cadmium species. Cadmium concentrations in combustion system emissions have been summarized by Biswas and Wu (7) and are in the range of 24-1500 µg m-3 for municipal solid waste incinerators (14, 15), 493 µg m-3 for municipal wastewater sludge incinerators (16), and 6-25 µg m-3 for coal combustors (17). There are also several deactivation furnaces for demilitarization operations in which heavy metal species are commonly encountered (18). The U. S. Environmental Protection Agency (USEPA) has mandated the use of hazardous waste combustors maximum achievable control technology (HWC-MACT) for such systems. Standards are also being tightened; for example, cadmium species concentrations in emissions of incinerators and chemical demilitarization furnaces (CDFs) are being reduced from 240 to 24 µg m-3. There are a few studies that have reported ambient atmospheric cadmium species concentrations. The cadmium concentration near a wire-drawing factory in south France was as high as 5 µg m-3, resulting in soil contamination and high levels in the human body (19). After the operations in the factory ceased, the value dropped to 5 ng m-3. Another study reported that atmospheric cadmium concentrations were as high as 39 µg m-3 near a battery factory with metallurgical cadmium production in Russia (20). A study conducted in India reported cadmium species concentrations in the range of 0.11-1.5 ng m-3 (21). Ambient cadmium concentrations near demilitarization furnaces, metallurgical furnaces, and smelters could reach high levels requiring special attention and the need to control emissions. The tightening of emission standards and the potential toxicity of the metallic species requires a thorough consideration and development of new control technologies for fine particles and heavy metal species emissions. To do this effectively, there is a need to understand the pathways of metal species transformations in the combustion environment and then design effective control methodologies. There have been several studies on the fate of heavy metal species in combustion environments (22-26). When the volatile heavy metals are introduced into a combustion system, they enter the gas phase at high temperatures, then nucleate and grow in the cooler downstream regions, resulting in the formation of a sub-micrometer aerosol. An understanding of the metallic species formation rates can also be used to design effective control strategies using sorbents (7). There are several methods that have been proposed for control of toxic metal emissions from combustors, and detailed reviews are available in the literature (7, 8). One approach that has been proposed is the injection of sorbents for heavy metal species control in combustion systems. Nanostructured sorbents have been demonstrated to be effective in the prevention of heavy metal species emissions (7). These sorbents can be injected in two forms: one as a solid or as a precursor, resulting in the formation of a very high surface area agglomerate. The objectives of the metal capture process are to prevent the nucleation of the heavy metal species and associate the metal species with the sorbent particles that can be readily captured. Furthermore, it is important that the metal is firmly bound to the sorbent so as to not leach out when disposed. Finally, lowering the ratio of the sorbent to metal species is desirable from the viewpoint of minimizing waste and cost. VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8481
Bulk sorbents have been proposed for the capture of heavy metal species in combustion systems (10, 27, 28). Though they are effective in the capture of the metal species, they are plagued with several physicochemical limitations. Once the outer surface has chemisorbed the metallic species, the inner volume is rendered ineffective, thus requiring a large ratio of bulk sorbent to the trace metal that is being captured. Furthermore, the bulk sorbents have been ineffective in certain environments, for example, when chlorine is present (29). Also, they have not proven to effectively suppress the nucleation of the heavy metal species. An alternate is the use of a novel nanostructured sorbent agglomerate process for the capture of heavy metals in combustion environments (30-32). This technology is potentially more effective at suppressing nucleation and promoting chemisorption of heavy metal species. Studies have been conducted to understand the effectiveness of this process for heavy metals such as lead (23, 31). Many sorbents have been proposed for the capture of cadmium species. Owens et al. (33) used equilibrium calculations to illustrate that silica and silicates are effective in the capture of cadmium species. Naturally occurring alumino-silicate materials, such as kaolinite, are very effective as potential sorbents for removing lead and cadmium from flue gas streams (10, 27, 34, 35). Recently, alumino-silicate pillared montmorillonite (MMT) clays have been developed that not only improve the accessibility of the interlayer sites to reacting metals by increasing the pore diameters but also significantly enhance the surface area of the sorbent particles. Montmorillonite is a naturally occurring layered clay with a typical platelet size of 100-500 nm in length and 1 nm in thickness. Pillaring of MMT has been shown to greatly increase the surface area and produce microporous structures with surface areas of 200-400 m2 g-1 and micropore volumes of 0.1 mL g-1 (36). The characteristics and effectiveness of these materials when injected into a high-temperature combustion environment have not been established. Their effectiveness at capturing heavy metal species and how they compare to in situ generated sorbents have also not been established. This paper describes the pathways of transformation of cadmium species that result in aerosol formation in a laboratory-scale furnace reactor system. The effectiveness of two different types of sorbents, an in situ generated silica sorbent and solid-phase montmorillonite clay sorbent, at capturing cadmium species is established.
Materials and Methods Experimental System. The experimental setup is shown in Figure 1 and consists of a feed system (cadmium and sorbents), alumina tubular furnace (Lindberg/Blue M, model HTF55342C, Tmax ) 1200 °C) with a reactor tube (0.0254 m i.d.), and dilution and measuring system. The cadmium species was introduced as a cadmium acetate (Cd(CH3CO2)2‚ 2H2O, Fisher Scientific) aerosol generated by atomizing (TSI Inc., model 3076) an aqueous solution (concentration of 1 g L-1). The flowrate through the atomizer was maintained at 1.67 × 10-5 m3 s-1 (1 L min-1). Two types of sorbents, a vaporphase precursor (1,1,3,3-tetramethyldisiloxane ((CH3)2SiH)2O, Aldrich; TMDS) and a solid-phase nanostructured MMT sorbent, were used. TMDS was introduced by an impinger (bubbler nozzle with a sintered glass filter) with a nitrogen carrier gas at a flowrate of 6.34 × 10-7 m3 s-1(0.038 L min-1). The MMT sorbent was injected by nebulizing (BGI, Inc., model CN24) a suspension of the clay in water (2.73 wt %) using a flowrate of 5.51 × 10-5 m3 s-1 (3.3 L min-1). The furnace allowed maintenance of a constant temperature and was set to either 700 or 1000 °C. A dilution system was used to obtain a desired temperature profile at the exit of the 8482
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 21, 2005
FIGURE 1. Schematic diagram of the experimental setup used for understanding pathways of cadmium species aerosol formation and capture by nanostructured sorbents (Q1 ) 0.038 L min-1, Q2 ) 1 L min-1, Q3 ) 3.3 L min-1, Q4 varies with conditions (Q1 + Q2 + Q3 + Q4 ) 14 L min-1)). furnace tube. Dilution with particle-free air was used to reduce the temperature and quench aerosol dynamics, so representative aerosol size distributions could be measured (37, 38). An adequate flowrate also was obtained to separate fine (1 µm) particles with a cascade impactor (MARK III, Pollution Control Systems Corp.). Test Plan. The overall objective is to understand the interaction and capture characteristics of cadmium species with nanostructured sorbents, and an experimental plan was designed as outlined in Table 1. As stated in the Introduction, it is essential to understand the mechanistic pathway of the heavy metal species so that effective control strategies can be designed. To that effect, experiments were first done with a cadmium-only feed (set I). The feed concentration of cadmium species in this work is 5 mg m-3 (expressed as the mass of cadmium), close to levels encountered in munitions deactivation furnaces (13). These results also provide a baseline for the sorbent capture experiments. To understand the evolution of the nanostructured sorbent aerosols and their transformations, experiments were conducted with a sorbent-only feed (sets II and III). The fourth and fifth sets of experiments were conducted with a cofeed of cadmium species and the sorbents (sets IV and V). The experiments were carried out at two furnace set-point temperatures to simulate the range of values encountered in industrial incinerators. The temperature gradient at the outlet of the combustor dictates the aerosol growth dynamics, and the measured temperature profile is shown in Figure 2. For the two conditions, the gradients at the outlet were -0.343 and -0.468 °C/m. The gradients with respect to time were -1727.3 and -3085.1 °C/s, respectively, similar to that encountered in full-scale combustors. A series of characterization measurements were conducted as summarized in the last column of Table 1. Measurements. Particles less than 1 µm in aerodynamic size were separated by the cascade impactor and introduced into a real-time size-distribution measuring instrument (SMPS, TSI Inc., St. Paul, MN) to obtain the number concentration in the range of 12-600 nm. The particles were also collected on a glass fiber filter (Whatman, EPM 2000) or 0.5 µm membrane filter (Advantec MFS, Inc., J050A047A) to perform further characterization tests. The crystallinity was determined by X-ray diffraction (XRD, Rigaku, model D-MAX/ A). As outlined in Table 1, this was done for the different tests to establish the role of temperature and the change in speciation of cadmium on the addition of the sorbent. The crystallinity may also affect the dissolution rates and hence affect the leachability. The leachability test used was similar
TABLE 1. Summary of Experiments Performeda total particle generation rate (mg/min) set
test
furnace temperature (°C)
fine (dp < 1 µm)
coarse (dp > 1 µm) 0.06b
other remarksc
1 2 3
room temperature 700 1000
0.10b 0.05 0.05
0.03 0.02
A A, D A, B, D
II (MMT only)
4 5 6
room temperature 700 1000
0.21 0.21 0.03
0.35 0.35 0.37
A, B, C, D, E A, C, D, E A, B, C, D, E
III (SiO2 only)
7 8
700 1000
1.28 1.43
0.14 0.22
A, C A, B, C
IV (MMT + Cd)
9 10
700 1000
0.25 0.22
0.38 0.25
A, D A, B, D
V (SiO2 + Cd)
11 12
700 1000
1.33 1.48
0.14 0.22
A, D A, B, D
I (Cd only)
a The total particle generation rate is calculated by actual collection on an absolute filter and dividing the mass by the sampling time. value. c A, Particle Size Distribution; B, XRD; C, BET; D, leaching test; E, SEM.
b
Estimated
FIGURE 2. Temperature profile vs axial position in the reactor for two furnace set-point temperatures. to the toxicity characteristic leaching procedure (TCLP; EPA method 1311) test and was conducted to determine how strongly the cadmium species was bonded to the sorbent. The dissolution test was performed using 0.04 M acetic acid and Milli-Q water. The filters were placed in 50-mL falcon tubes with each extraction solution. Each sample was fixed by an extraction vessel holder and agitated for 24 h with a rotating speed of 8 rpm (agitator, Barnstead International, model 4152110). After the samples were agitated, 9.76 mL of solution was sampled and filtered with a 0.45-µm membrane filter from each sample. To avoid the deposition of cadmium species in the tube, the final sample for inductively coupled plasma (ICP) analysis was prepared by adding 0.24 mL of concentrated (15.8 M) nitric acid. Cadmium standard solutions (0.05-50 ppm) were also prepared for obtaining a calibration curve. The concentration of cadmium species in the leachate was determined by inductively coupled plasma atomic emission spectrometry (ICP-AES; Varian Liberty). To examine the morphology of the samples, scanning electron microscopy (SEM; Hitachi S-4500) was used to view the particles collected on a grid. The specific surface areas of the sorbents were determined by the Brunauer-Emmett-Teller (BET) method (Micrometrics, Gemini 2375).
Results and Discussion Different sets of experiments were designed as outlined in Table 1 to understand the evolution of the metallic species aerosol and the effectiveness of sorbents at capturing cadmium in combustion environments. Cadmium-Only Tests. The results of the cadmiumspecies-feed-only experiments (set I) are discussed first. The size distributions measured at the outlet of the furnace for two temperatures (700 and 1000°C) are shown in Figure 3. The experiment at 700 °C was also conducted by matching the residence time to that at 1000 °C (16.3 s). On comparison of the resultant size distribution at the exit of the furnace to that at the inlet (feed as cadmium acetate) at both temperatures, there is a reduction in the geometric mean diameter (62.9 and 26.1 nm at 700 and 1000 °C) compared to that at the inlet (87.2 nm). As the cadmium acetate decomposes and converts to cadmium oxide (verified by XRD, Figure 7), there would be an expected reduction in size from 87.2 to 55.0 nm. However, the mean size is smaller at 1000 °C, indicating that vaporization of the cadmium species in the furnace was followed by nucleation at the reactor exit. The high number concentration at the smaller sizes confirms VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8483
FIGURE 3. Particle size distribution of cadmium species compounds in the reactor at room temperature (feed aerosol, test 1), 700 °C (test 2), and 1000 °C (test 3). The 700 °C test is done at two residence times (4, 20.1 s; 2, 16.3 s).
FIGURE 4. Cadmium oxide vapor concentration as a function of temperature and comparison with the critical vapor concentration. that nucleation did occur. This is not clearly evident for the 700 °C experiment. No vapor pressure data is available for cadmium acetate (as it decomposes to the oxide) to conclusively determine that it completely evaporates at 700 °C. However, equilibrium calculations indicate that the cadmium species should be present in the vapor state even at 700 °C (39, 40). The absence of a clear peak in the smaller sizes (as in the 1000 °C test) may indicate that the cadmium acetate is only partially vaporizing and a fraction of the feed particles are directly converting to the oxide (without vaporization and nucleation). This is specific to these laboratory tests, and in most full-scale systems it is anticipated that the Cd species will indeed vaporize and then follow the pathway of growth as in the 1000 °C test. The location (temperature) for the onset of nucleation is important, because it then establishes the time available for the subsequent aerosol growth mechanisms. The critical vapor concentration, C*, at which nucleation is initiated (41) is plotted in Figure 4 as a function of temperature. The critical 8484
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 21, 2005
vapor concentration is determined from the critical saturation ratio, S*, and is given by
C* ) p*/RT ) S*ps/RT
(1)
The nucleation rate (I) is set to 1 particle cm-3 s-1 to calculate S* (7, 41).
[
I) 2
CNavkBT
x2πm1kBT
]
[CNavvm2/3]
[x ] [ σvm2/3 k BT
exp -
16πσ3vm2
3(kBT)3(ln S)2
]
(2)
to yield (for property values of CdO, surface tension, σ ) 0.47 N m-1 (estimated from the pure cadmium value (42)); molecular volume, vm ) 2.6 × 10-29 m3; molecular mass, m1 ) 2.1 × 10-25 kg; Avogadro’s number, Nav ) 6 × 1023 particles
FIGURE 5. Mechanistic description of pathways of cadmium species in the reactor: (a) Cd only, (b) Cd + MMT, (c) Cd + in situ generated SiO2. mol-1, and Boltzmann’s constant, kB ) 1.38 × 10-23 kg m2 s-2 K-1)
S* ) exp
(x
4.5 × 1011 T3(2 ln C + 83.01)
)
(3)
and substituting in eq 1
C* )
[
762.76 17513.9 exp + T T
x
4.5 × 1011 T3(2 ln C + 83.01)
]
(4)
1 6N T avvmCφ 2kB ml πN
x
(
)
2/3
τcoag,fm )
1
(5) Nxπ
where fm is free molecular regime, N is the particle number
(6)
x
2
When the vapor concentration of the cadmium species, C, exceeds C*, nucleation will take place. When C* ) C in eq 4, the temperature at which nucleation is initiated can be determined. For the concentration in the furnace (9.24 mg m-3 as CdO, calculated using the feed rate of Cd) in our system, this happens at 545 K (Figure 4). The corresponding locations in the furnace reactor are indicated in Figure 2, assuming that all of the cadmium species are present as the vapor. The pathways of the cadmium species transformation in the furnace reactor are shown in Figure 5a. To estimate the relative importance of the condensation and coagulation growth mechanisms, characteristic times were computed. The condensation characteristic time, τC, is defined as the time necessary for particles to consume the available vapor and reach the final particle size. The coagulation characteristic time, τcoag, is the time required for the particle number concentration to reduce to half of the initial value. As the particles are in the free molecular regime (Kn > 1), τC is given by (43)
τC,fm )
concentration based on the critical nucleus size, d/p (4σvm/ kBT ln S*, where T is the temperature at the nucleation starting point), and Cφ is the equivalent initial vapor concentration expressed by C0 + (Nπd/p3)/(6Navvm) - Cs, where C0 is the initial concentration of CdO and Cs is the surface concentration (assumed to be 0). τcoag in the free molecular regime is given by
6kBTd/p N Fp
where Fp is the particle density and µ is the gas viscosity. For the conditions in tests 2 and 3, τC is computed to be 4.69 × 10-4 s using eq 5 with property values for CdO and N ) 4.93 × 1012 cm-3, T ) 545 K, and C0 ) 9.24 mg m-3. The corresponding τcoag calculated using eq 6 is 1.57 × 10-3 s for the two tests. In summary, as illustrated in Figure 5a, decomposition, vaporization, and oxidization of the cadmium acetate result in the formation of a vapor-phase species, followed by particle formation by nucleation. The cadmium oxide particles grow quickly (short characteristic times compared to the residence time ≈ 20 s) by condensation and coagulation. The characteristic times for coagulation and condensation are of a similar order of magnitude, and hence both are important growth mechanisms. The chemical structure of the resultant cadmium species (orange color) at 1000 °C was confirmed by XRD analysis to be CdO in Figure 7. Sorbent-Only Tests. The structure of the sorbent and its eventual size are important characteristics that determine its effectiveness for metal capture. Sorbent-only experiments (set II and III) were conducted to see how these characteristics changed with temperature. Tests 4, 5, and 6 (set II, Table 1) were conducted with the MMT sorbent. The mean sizes of the MMT particles were 130, 105, and 96.6 nm at room temperature (test 4), 700 °C (test 5), and 1000 °C (test 6) VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8485
FIGURE 6. Particle size distributions in the cadmium capture experiments: (a) 700 °C; (b) 1000 °C. Also shown are baseline data for Cd-only feed tests and sorbent-feed-only size distributions.
FIGURE 7. Crystallinity of particles collected at 1000 °C for various test conditions. (Figure 6, symbol O), respectively. Due to the structural change of MMT at high temperatures, the particles shrink, and these are confirmed by the SEM pictures (Figure 8). The 8486
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 21, 2005
SEM pictures indicate that not much structural change occurs at 700 °C (Figure 8b) in comparison to room temperature; however there is a distinct change at 1000 °C (Figure 8c). Vassilev et al.44 have reported that MMT can lose its lattice water by dehydroxylation up to 730 °C without destruction of the structure. However, the lattice structure was observed to be destroyed at temperatures greater than 850 °C with a conversion to the amorphous state. The measured BET surface areas were 18 m2 g-1(room temperature), 9 m2 g-1(700 °C), and 13.9 m2 g-1(1000 °C), and not much change was observed. It should be noted that these measured values are a factor of 20 times lower than that reported by the manufacturer, and changes occurred after aerosolization from an aqueous-phase suspension. The MMT particles were also found to be amorphous when analyzed by XRD for all tests (4, 5, and 6). The concentrations (2.6 × 106 and 2.2 × 106 cm-3 at 700 and 1000 °C, respectively, Figure 6) at which the MMT were injected did not result in much change of the size in the aerosol reactor due to minimal particle-particle interactions. Similar experiments were conducted (set III) with the silica sorbent. In this case, a precursor is injected that rapidly reacts at the elevated temperature to form a nanostructured
FIGURE 8. SEM images of MMT particles at different temperatures: (a) room temperature (test 4), (b) 700 °C (test 5), (c) 1000 °C (test 6). agglomerate sorbent aerosol (Figure 5c). The overall objectives are to have a high surface area agglomerate with a sufficiently large mean size that can be captured in an existing particle control device. For the conditions selected in this study (set III, Table 1), the mean size of the silica particles were 214 nm (700 °C, test 7, Table 1) and 169 nm (1000 °C, test 8, Table 1). This was intentionally selected to be in this range (larger and very distinct from the Cd species aerosol sizes, but not very large) so that the same real-time instrument (SMPS system) could be used for the measurement. In a full-scale system, the size of the resultant particles would be larger (so that they are readily captured), and this is achievable by altering the precursor feed conditions. Detailed calculations outlining conditions to obtain micrometer-sized, nanostructured agglomerate sorbents by injecting precursors have been presented elsewhere (45). There is no nucleation barrier for the silica particles, and growth is primarily by collision mechanisms (37). The characteristic times for collisions (τcoag) based on the concentration of silica (90.13
mg m-3) are 4.1 × 10-5 s (700 °C) and 6.6 × 10-5 s (1000 °C), much smaller than the residence time. The sorbent particles are amorphous as analyzed by XRD. The surface areas are 114.6 m2 g-1 at 700 °C and 89.9 m2 g-1 at 1000 °C. Clearly, these are much larger than that of the MMT sorbent and can be engineered to high values by choosing the injection location for the precursor. Cadmium Capture Tests. The effectiveness of two different sorbents, MMT (set IV) and silica precursor (set V), on capture of the cadmium species are discussed. Figures 6a and 6b (symbol 0) show the particle size distribution of MMT with a cofeed of cadmium at 700 and 1000 °C. The size distributions of the cofeed tests (tests 9 and 10) are between the MMT-only (tests 5 and 6) and cadmium-only (tests 2 and 3) experiments for both 700 and 1000 °C. At 1000 °C, the number concentration of particles for the MMT and cadmium cofeed experiment (test 10) between 12 and 60 nm is less than that for the Cd-only experiment (test 3), while it is higher for particles greater than 200 nm. This might be attributed to the fact that some of the cadmium oxide vapors are physically adsorbed on the MMT surfaces and the others nucleate to form cadmium oxide particles. The condensation characteristic times of cadmium species on MMT, τC, based on the concentration of MMT (36.05 mg m-3) are 4.96 × 10-2 and 6.23 × 10-2 s at 700 and 1000 °C, respectively, as calculated using eq 5. The condensation characteristic time (∼10-2 s) is smaller than the time (12 s) it takes to reach the nucleation onset location; thus some of the CdO species vapors are physisorbed by MMT. However, because the surface area of MMT is not too large, the remnant CdO vapor does nucleate, and this is confirmed by the number concentrations for sizes less than 60 nm being higher than that of MMT alone (test 6, Figure 6b). The effectiveness of the MMT sorbent for cadmium capture was evaluated with XRD analysis (Figure 7). For test 10, cadmium oxide peaks are clearly visible, indicating that the MMT sorbent is not effective at preventing the formation of cadmium oxide (i.e., not capturing all of the cadmium species). The orange color of the collected samples in tests 9 and 10 provides a visual indication of the presence of cadmium oxide in the sample. Results of the in situ generated silica sorbent capture experiments (tests 11 and 12) are shown in Figures 6a and 6b (symbol 4). Cadmium oxide particles that were formed by nucleation (with a mean size of 26.1 nm at 1000 °C, test 3) are completely suppressed in the presence of the silica sorbent. The mechanistic pathway is shown in Figure 5c, and the presence of a highly reactive surface results in the scavenging of the cadmium species vapors before they can nucleate on their own. Surprisingly, even at 700 °C, there is no evidence of any remnant cadmium oxide aerosol. It could be that in the presence of the precursor the Cd species does indeed all vaporize and is present in the gas phase, which then subsequently condenses on the high surface area silica sorbent. Or it could be that the higher number concentrations of the larger sorbent particles are coagulating with the smaller-sized cadmium oxide particles. However, sorbentcadmium oxide particle coagulation would not cause the complete disappearance of a preexisting cadmium oxide aerosol in the given time (46); however, a complete disappearance is observed in this work. The condensation characteristic times for cadmium species vapors on the silica sorbent are 2.12 × 10-5 and 2.77 × 10-5 s at 700 and 1000 °C, respectively, much lower than that for the MMT capture tests. This characteristic time is very small, and as the silica particles are formed and present in the high-temperature environment, condensation is the dominant pathway (prior to any possibility of nucleation of the cadmium species, Figure 4) of transfer of cadmium species vapors. The XRD analysis (Figure 7) for the silica sorbent and cadmium feed test (test VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8487
TABLE 2. Analysis of Total Cadmium Content in the Collected Samples by Leaching Tests leached cadmium amounts per sampling gas volume (mg/m3) feed stream
temperature (°C)
Cd only (test 2) MMT + Cd (test 9) SiO2 + Cd (test 11)
700
Cd only (test 3) MMT + Cd (test 10) SiO2 + Cd (test 12)
1000
fine particles (dp < 1 µm) Milli-Q water acetic acid
9
leaching efficiency of cadmium
0.012 0.004 0.013
3.33 3.13 1.88
0.005 0.005 0.023
2.48 2.89 1.73
100% 62.2%
0.013 0.006 0.006
2.60 2.68 0.12
0.008 0.010 0.031
1.76 2.16 0.01
100% 3.2%
12) indicates no cadmium oxide peaks, further confirming the effectiveness of the silica sorbent. Leachability of Cadmium from Produced Particles. The in situ generated silica sorbent captures the cadmium species vapors by a chemisorption mechanism and completely suppresses any cadmium species aerosol formation. In contrast, MMT appears to partially capture the cadmium species by a physisorption mechanism. To confirm the binding characteristics of cadmium to the sorbent, leachability tests were performed, and the results are summarized in Table 2. To obtain a baseline, the dissolution rates of the fine (1µm) fractions of the particles produced in the Cd-only feed experiments (tests 2 and 3) were determined. The dissolution rate in water is rather low (Table 2), consistent with the low solubility of CdO in water (∼20 µg m-3). However, most of the particles are dissolved in the acetic acid solution, and similar distributions of cadmium are present in the fine (slightly higher) and coarse modes. Similar dissolution tests are carried out for the particles collected in experiments with a cofeed of MMT and cadmium (tests 9 and 10). The dissolutions of the cadmium species are very similar to those of the baseline tests, and the leaching efficiency (defined as the ratio of total cadmium dissolved using acetic acid from sorbent tests to that of the Cd-only tests) is close to 100%. For the silica and Cd experiments (tests 11 and 12), the dissolution of cadmium is lower, and leaching efficiencies are 62% and 3% (Table 2). This is consistent with the previous observation and interpretation that the silica sorbent firmly binds the cadmium species, especially in comparison to the MMT solid srobent. The lower leaching efficiency (3.2%) at 1000 °C in comparison to the 700 °C test is due to the greater degree of chemical reaction between the cadmium species and the silica sorbent, leading to firmer binding or generation of a low leachability cadmium species (such as amorphous cadmium silicate). Owens et al. (33) have illustrated the formation of different species at different temperatures by equilibrium calculations. Nanostructured silica sorbents have been shown to be very effective at capturing cadmium species compounds in high-temperature environments. Clearly, pilot-scale tests are necessary to prove the viability of the proposed approach. On the basis of data in Table 1 and the overall size-distribution parameters, the total feed rate of Cd species is approximately 0.07 mg min-1 (as mass of CdO) at 1000 °C. The difference between the silica sorbent and cadmium feed experiment (test 12) and that of the silica-only case (test 8) gives an estimate of the cadmium species captured, which is 0.05 mg min-1. The exact speciation (chemical form of the Cd) is unknown in the sorbent-Cd complex, but by conversion to the mass of oxide, the estimated capture rate is 0.06 mg min-1. Therefore, an estimate of the overall capture efficiency is approximately 86%. On the basis of the more accurate number-based size-distribution measurements, the capture efficiency is estimated to be much higher and close to 99%. If such a process were to be used in a military deactivation 8488
coarse particles (dp > 1 µm) Milli-Q water acetic acid
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 21, 2005
furnace, then the emission rate would be brought down to 14% (conservative estimate) of the current value and would meet the objective of a decrease of the new proposed standard (from 240 to 24 µg m-3). Furthermore, the laboratory-scale study provides valuable insights with regard to the role of temperature (injection location) and other process parameters to scale up such methodologies for pilot-level testing and full-scale application.
Acknowledgments This work was supported in part by a SBIR contract (DACA4203-C-0029) from the U. S. Army Construction Engineering Research Laboratory (CERL) through Triton Systems, Inc. Discussions with Dr. V. Boddu, CERL, Urbana-Champaign, IL, are gratefully acknowledged.
Literature Cited (1) Hong, S.; Park, J. K.; Thompson, L. G.; Boutron, C. F.; Ferrari, C. P.; Francou, B.; Maurice-Bourgoin, L. Changes in the occurrence of heavy metals in the tropical atmosphere during the past 22,000 years as recorded in Bolivian ice core. J. Phys. IV 2003, 107, 633-636. (2) Schuster, P. F.; Krabbenhoft, D. P.; Naftz, D. L.; Cecil, L. D.; Olson, M. L.; Dewild, J. F.; Susong, D. D.; Green, J. R.; Abbott, M. L. Atmospheric mercury deposition during the last 270 years: A glacial ice core record of natural and anthropogenic sources. Environ. Sci. Technol. 2002, 36, 2303-2310. (3) Nriagu, J. O. A global assessment of natural sources of atmospheric trace metals. Nature 1989, 338, 47-49. (4) Kakareka, S.; Gromov, S.; Pacyna, J.; Kukharchyk, T. Estimation of heavy metal emission fluxes on the territory of the NIS. Atmos. Environ. 2004, 38, 7101-7109. (5) Pacyna, J. M.; Graedel, T. E. Atmospheric emissions inventoriess Status and prospects. Annu Rev Energy Environ. 1995, 20, 265300. (6) Tolocka, M. P.; Lake, D. A.; Johnston, M. V.; Wexler, A. S. Number concentrations of fine and ultrafine particles containing metals. Atmos. Environ. 2004, 38, 3263-3273. (7) Biswas, P.; Wu, C. Y. Control of toxic metal emission from combustors using sorbents: A review. J. Air Waste Manage. Assoc. 1998, 48, 113-127. (8) Linak, W. P.; Wendt, J. O. L. Toxic metal emissions from incinerationsMechanisms and control. Prog. Energy Combust. Sci. 1993, 19, 145-185. (9) Zhuang, Y.; Kim, Y. J.; Lee, T. G.; Biswas, P. Experimental and theoretical studies of ultra-fine particle behavior in electrostatic precipitators. J. Electrost. 2000, 48, 245-260. (10) Uberoi, M.; Shadman, F. High-temperature removal of cadmium compounds using solid sorbents. Environ. Sci. Technol. 1991, 25, 1285-1289. (11) Linak, W. P.; Srivastava, R. K.; Wendt, J. O. L. Metal aerosol formation in a laboratory swirl flame incinerator. Combust. Sci. Technol. 1994, 101, 7-27. (12) Hu, C.-W.; Chao, M.-R.; Wu, K.-Y.; Chang-Chien, G.-P.; Lee, W.-J.; Chang, L. W.; Lee, W.-S. Characterization of multiple airborne particulate metals in the surroundings of a municipal waste incinerator in Taiwan. Atmos. Environ. 2003, 37, 28452852. (13) Boddu, V. M.; Hay, K. J. Characterization of lead and cadmium species in emissions from munitions deactivation furnace. In 24th Army Science Conference Proceedings; Department of the Army: Washington, DC, 2004.
(14) Savage, G. M.; Bordson, D. L.; Diaz, L. F. Important issues related to air-pollution at municipal solid-waste facilities. Environ. Prog. 1988, 7, 123-130. (15) Reimann, D. O. Heavy metals in domestic refuse and their distribution in incinerator residues. Waste Manage. Res. 1989, 7, 57-62. (16) Bennett, R. L.; Knapp, K. T. Characterization of particulate emissions from municipal wastewater sludge incinerators. Environ. Sci. Technol. 1982, 16, 831-836. (17) Gullet, B. K.; Raghunathan, K. Reduction of coal-based metal emissions by furnace sorbent injection. Energy Fuels 1994, 8, 1068-1076. (18) Boddu, V. M.; Hay, K. J.; Ghosh, T. K.; Viswanath, D. S. Adsorption of lead vapors on perlite. 23rd Army Science Conference Proceedings; Department of the Army, Washington, DC, 2003. (19) Noack, Y.; Lefloch, M.; Robin, D. Environmental impact of a cadmium atmospheric pollution at Marseille (South France). J. Phys. IV 2003, 107, 961-964. (20) Bustueva, K. A.; Revich, B. A.; Bezpalko, L. E. Cadmium in the environment of three Russian cities and in human hair and urine. Arch. Environ. Health 1994, 49, 284-288. (21) Pandey, P. K.; Patel, K. S.; Subrt, P. Trace elemental composition of atmospheric particulate at Bhilai in central-east India. Sci. Total Environ. 1998, 215, 123-134. (22) Wu, C.-Y.; Biswas, P. Lead species aerosol formation and growth in multicomponent high-temperature environments. Environ. Eng. Sci. 2000, 17, 41-60. (23) Biswas, P.; Zachariah, M. R. In situ immobilization of lead species in combustion environments by injection of gas-phase silica sorbent precursors. Environ. Sci. Technol. 1997, 31, 2455-2463. (24) Sterling, R. O.; Helble, J. J. Reaction of arsenic vapor species with fly ash compounds: Kinetics and speciation of the reaction with calcium silicates. Chemosphere 2003, 51, 1111-1119. (25) Yu, S. Y.; Kennedy, I. M. A Preliminary kinetic model of chromium in a hydrogen/air flame. Combust. Sci. Technol. 2000, 160, 3545. (26) Guo, B.; Kennedy, I. M. The speciation and morphology of chromium oxide nanoparticles in a diffusion flame. Aerosol Sci. Technol. 2004, 38, 424-436. (27) Uberoi, M.; Shadman, F. Sorbents for removal of lead compounds from hot flue gases. AIChE J. 1990, 36, 307-309. (28) Uberoi, M.; Punjak, W. A.; Shadman, F. The kinetics and mechanism of alkali removal from flue-gases by solid sorbents. Prog. Energy Combust. Sci. 1990, 16, 205-211. (29) Scotto, M. V.; Uberoi, M.; Peterson, T. W.; Shadman, F.; Wendt, J. O. L. Metal capture by sorbents in combustion processes. Fuel Process. Technol. 1994, 39, 357-372. (30) Owens, T. M.; Biswas, P. Reactions between vapor phase lead compounds in in situ generated silica particles at various leadsilicon feed ratios: Applications to toxic metal capture in combustors. J. Air Waste Manage. Assoc. 1996, 46, 530-538.
(31) Owens, T. M.; Biswas, P. Vapor phase sorbent precursors for toxic metal emissions control from combustors. Ind. Eng. Chem. Res. 1996, 35, 792-798. (32) Wu, C.-Y.; Lee, T.-G.; Tyree, G.; Arar, E.; Biswas, P. Capture of mercury in combustion systems by in situ generated titania particles with UV irradiation. Environ. Eng. Sci. 1998, 15, 137148. (33) Owens, T. M.; Wu, C.-Y.; Biswas, P. An equilibrium analysis for reaction of metal compounds with sorbents in high temperature systems. Chem. Eng. Commun. 1995, 133, 31-52. (34) Tran, Q. K.; Steenari, B. M.; Iisa, K.; Lindqvist, O. Capture of potassium and cadmium by kaolin in oxidizing and reducing atmospheres. Energy Fuels 2004, 18, 1870-1876. (35) Diaz-Somoano, M.; Martinez-Tarazona, M. R. High-temperature removal of cadmium from a gasification flue gas using solid sorbents. Fuel 2005, 84, 717-721. (36) Gil, A.; Guiu, G.; Grange, P.; Montes, M. Preparation and characterization of microporosity and acidity of silica-alumina pillared clays. J. Phys. Chem. 1995, 99, 301-312. (37) Biswas, P.; Li, X.; Pratsinis, S. E. Optical waveguide preform fabrication. J. Appl. Phys. 1989, 65, 2445-2450. (38) Baron, P. A.; Willeke, K. Aerosol Measurement, 2nd ed.; John Wiley & Sons: 2001, Chapter 31. (39) Wu, C. Y.; Biswas, P. An equilibrium analysis to determine the speciation of metals in an incinerator. Combust. Flame 1993, 93, 31-40. (40) CRC Handbook of Chemistry and Physics; 85th ed.; CRC Press: Boca Raton, FL, 2005. (41) Friedlander, S. K. Smoke, Dust and Haze; Wiley: New York, 1977. (42) International Critical Tables of Numerical Data, Physics, Chemistry and Technology; Washburn, E. W., Ed.; McGraw-Hill: New York, 1933. (43) Wu, C.-Y.; Biswas, P. Particle growth by condensation in a system with limited vapor. Aerosol Sci. Technol. 1998, 28, 1-20. (44) Vassilev, S. V.; Braekman-Danheux, C.; Laurent, P.; Thiemann, T.; Fontana, A. Behaviour, capture and inertization of some trace elements during combustion of refuse-derived char from municipal solid waste. Fuel 1999, 78, 1131-1145. (45) Suriyawong, A.; Smallwood, M.; Biswas, P. A strategic approach for optimizing the use of titania particles for mercury removal in coal-burning utilities. In Proceedings of the 98th AWMA Annual Conference and Exhibition; 2005; paper no. 05-A-1092-AWMA. (46) Lee, S. R.; Wu, C. Y. Size distribution evolution of fine aerosols due to intercoagulation with coarse aerosols. Aerosol Sci. Technol. 2005, 39, 358-370.
Received for review April 7, 2005. Revised manuscript received July 20, 2005. Accepted August 16, 2005. ES0506713
VOL. 39, NO. 21, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8489
