Particle–metal interactions during combustion of pulp and paper biomass in a fluidized bed combustor

Particle–metal interactions during combustion of pulp and paper biomass in a fluidized bed combustor

Combustion and Flame 142 (2005) 249–257 www.elsevier.com/locate/combustflame Particle–metal interactions during combustion of pulp and paper biomass ...

490KB Sizes 1 Downloads 110 Views

Combustion and Flame 142 (2005) 249–257 www.elsevier.com/locate/combustflame

Particle–metal interactions during combustion of pulp and paper biomass in a fluidized bed combustor F. Eldabbagh a , A. Ramesh a , J. Hawari b , W. Hutny c , J.A. Kozinski a,∗ a Energy & Environmental Research Laboratory, McGill University, 3610 University Street, Wong Building, Room 2160,

Montreal, QC, Canada H3A 2B2 b Biotechnology Research Institute, National Research Council of Canada, Ottawa, ON, Canada c Energy Technology Centre—Ottawa, Natural Resources Canada, Ottawa, ON, Canada

Received 7 September 2004; received in revised form 5 March 2005; accepted 21 March 2005 Available online 30 April 2005

Abstract We compare interactions between metals and solid particles during the classic fluidized bed combustion (FBC) and a new low–high–low temperature (LHL) combustion of selected biomass. The biomass was a mixture of bark and pine wood residues typically used by a paper mill as a source of energy. Experiments, conducted on a pilot scale, reveal a clear pattern of surface predominance of light metals (Ca, Na, K) and core predominance of heavy metals (Cd, Cr) within the LHL-generated particles. No such behavior was induced by the FBC. Metal migration is linked to the evolution of inorganic particles. A composite picture of the metal rearrangements in the particles was obtained by a combination of independent analytical techniques including electron probe microanalysis, field emission scanning electron microscopy, inductively coupled plasma spectrometry, and X-ray diffractometry. It is suggested that the combination of (1) the high-temperature region in the LHL and (2) changes in the surface free energy of the particles is the driving force for the metal–particle behavior. Important practical implications of the observed phenomena are proposed, including removal of hazardous submicron particulate and reduction in fouling/slagging during biomass combustion. These findings may contribute to redesigning of currently operating FBC units to generate nonhazardous, nonleachable, reusable particles where heavy metals are immobilized while environmental and technological problems reduced.  2005 The Combustion Institute. Published by Elsevier Inc. All rights reserved. Keywords: Biomass; Metals; Inorganic particles; Fluidized bed combustion

1. Introduction Biomass is a clean and renewable energy source, which stores and converts the solar energy via the photosynthesis process. Not only is it a plentiful fuel,

* Corresponding author. Fax: +1 514 398 4492.

E-mail address: [email protected] (J.A. Kozinski).

but its use also reestablishes the natural carbon cycle, helping mitigate greenhouse gas emissions. This renewable energy source is nearly CO2 neutral [1]. An average heating value of biomass energy crops is comparable to that of a subbituminous coal. Overall, it is possible to achieve 93% reduction in net CO2 emissions per unit heating value by switching from coal to biomass and 84% reduction by switching from natural gas-fired cogeneration to biomass [2]. Due to inherent advantages of the biomass in substitut-

0010-2180/$ – see front matter  2005 The Combustion Institute. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.combustflame.2005.03.013

250

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

ing fossil fuels, and increasing legislative pressures against CO2 emissions (Kyoto Protocol), biomassbased power is genuinely considered. It seems practically impossible to meet Kyoto requirements by replacing fossil fuel combustion with nuclear energy, hydropower, or fuel cells. Simply, there is not enough time. In this context, we believe that there exists a niche for biomass-based power generation. There are several excellent reviews concerning biomass combustion (e.g., [3,4]) and metal behavior (e.g., [5,6]). These works focus on the assessment of emissions from biomass combustion systems and understanding its gasification/pyrolysis. However, biomass combustion has never been thoroughly researched, as most of the resources have been invested in coal, oil, or natural gas combustion. (This trend is changing now due to legislative pressure and price increases.) Although several commercial-scale facilities burn biomass, there is not yet a comprehensive knowledge of mechanisms involved in the evolution of biomass–metal systems during combustion that could allow for optimization of the combustion practices (fundamental interpretation of data from full-scale reactors is difficult). For example, problems concerning metal–ash interactions and removal of hazardous submicron particulates that lead to increased emissions and excessive fouling/slagging are not solved. This article is an attempt to explain particle–particle and particle–metal interactions that may help in finding solutions to these problems. In this article we report on the application of novel multizone temperature approach to biomass combustion. In this method, biomass is initially fed into the low-temperature region (<1000 K), then subjected to the high-temperature treatment (∼1400 K), which is followed by quenching in the second low-temperature zone (1000 K). Deliberate use of the low–high–low (LHL) temperature zones has proven to be successful in encapsulating heavy metals when using a small thermogravimetric furnace [7], but it was not tested on a larger scale. This article compares biomass combustion in the classic fluidized bed combustor (FBC) and the LHL reactor on a pilot scale. Evolution of inorganic particles is linked to metal migration. New data on metal immobilization are presented and mechanisms explained for the first time. In the conclusion, we suggest practical implications of the observed phenomena.

2. Experimental 2.1. Pilot-scale multimode combustion facility Experiments were carried out in a pilot-scale multimode combustion facility (Fig. 1) (described in de-

Fig. 1. Pilot-scale multimode combustion facility capable of operating as the classic FBC or LHL. Locations of selected characteristic regions are illustrated. Bed temperature was maintained using uniformly distributed electrical heaters. Cooling/quenching in the second low-temperature region (∼500 K/s) was achieved via combination of water cooling and vortex quenching.

tail elsewhere [8,9]). The facility can operate in either a fluidized bed or a single burner mode (in addition, bubbling or circulating regimes can be applied). Results discussed in this article come from the fluidized bubbling bed combustion (operating conditions are listed in Table 1 and Fig. 1). The combustion chamber (300 kW) consists of six individual sections reaching 3.90 m in height with a 0.5-m o.d. and 0.15-m i.d. The sections are lined with one 0.15-m-thick castable refractory layer. Each section is cooled or heated independently, and features six ports of different size providing access for in situ observations and measurements of gas composition, velocity, particle properties, and temperature/pressure profiles. The three middle sections are used to create the LHL regions. The high-temperature region (HTR) is created via uniformly distributed natural gas flame (70 kW) and located 1.1–1.4 m above the nozzle plate. The splash zone of the freeboard region is limited to 0.5 m above the bed level. The flue gas exits to a cyclone, which removes large fractions of solids. When the classic FBC combustion was tested, no HTR was created. There-

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257 Table 1 Properties of biomass and experimental conditions (wt%) Ultimate analysis C H N S O

26.9 3.14 0.26 0.05 18.47

Proximate analysis Volatiles Ash Fixed carbon Moisture Heating value

46.89 35.38 1.92 15.81 10,000 kJ/kg

Ash analysis SiO2 Na2 O K2 O Al2 O3 CaO Cr & Cda

38.12 0.30 0.22 28.33 19.10 6000 ppm

Experimental conditions Superficial velocity Bed carbon Residence time O2 Air temperature

0.82 m/s <2.5% 1.8 s 8.4 vol% 823 K

a Final content after doping.

fore, during the LHL combustion, the temperature in the region corresponding to the HTR was ∼470 K higher than the maximum temperature obtained dur-

251

ing the classic FBC combustion mode (Fig. 2). Identical operating conditions were used in both FBC and LHL tests with the exception of the HTR. Biomass used in the experiments consisted of fibers ranging from 20 to 40 µm in width and 100 to 600 µm in length, forming larger (∼15 mm) spherical aggregates. It was a mixture of bark and pine wood residues typically used by a paper mill as a source of energy [9]. Biomass properties are listed in Table 1. It is anticipated that energy costs and legislative measures will stimulate an increase in the usage of this type biomass while increased recycling will enlarge the heavy metal portion of its composition [9]. Thus, the presence of high levels of heavy metals in the biomass is realistic [4,7,9]. It is particularly true in the case of bark and pine wood residues, applied in this experiments, which are widely used in the pulp and paper industry. The amount of heavy metals in this biomass dramatically increases due to increase in the recycling of high-quality, glossy, and plasticized papers. In addition, treated and waste woods may contain high concentrations of heavy metals. Therefore, it makes sense to perform this study to find potential solutions before the industry is overwhelmed by heavy metal-related problems. The biomass was doped with aqueous solutions of metal salts Cd(NO3 )2 ·4H2 O and Cr(NO3 )3 ·9H2 O (6000 ppm each) to quantify metal behavior (Cd and Cr). (The form of the doped species was the same as that present in the biomass [7].) An inert “garnet sand” (SiO2 40%, Fe2 O3 35%, Al2 O3 25%) with an average particle size diameter of 1.1 mm was used as bed material and fluidized to a height of 0.4–0.5 m above the nozzle plate. A predetermined amount of the biomass (50 kg) was con-

Fig. 2. Typical temperature profiles in the FBC and the LHL combustion modes.

252

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

tinuously delivered into the combustion chamber with an auger (14.5 kg/h). To achieve steady-state conditions, the facility was heated for 200 min prior to data acquisition measurements. Data presented here were collected during the stable period (typically 180 min). 2.2. Sampling and analytical techniques Solid particles were collected on a 0.25-µm fiberglass filter using a water-quenched probe. As particle– gas velocities were continuously monitored within the facility, the sampling rates were adjustable, permitting isokinetic sampling [9]. The sample quenching rate was ∼1600 K/s (there was no evidence of secondary reactions in the probe and on the filter [8,9]). Solid samples were collected along the facility at different stages of the LHL and FBC combustion processes. Microstructural and chemical analyses of the samples were based on the techniques previously developed and described [7,10,11]. Approximately 2000 individual particles were analyzed per sample (FBC, LHL). The procedures for the analysis of metal migration and particle surface, based on an electron microprobe analyzer (EPMA) and a field-emission scanning electron microscope (SEM), were further improved to satisfactorily characterize the particle– metal interactions. Collected inorganic particles were quenched, embedded into epoxy, and cross-sectioned for EPMA analyses of microstructural changes and distribution/concentration of metals (a Jeol 8900 SuperProbe equipped with five wavelength-dispersive spectrometers was used). Methods applied for EPMA calibration and correction procedures for overlapping characteristic X rays are described elsewhere [7,11]. The measured concentrations of metals were at least one order of magnitude higher than detection limits. Supermicron particles were quantitatively analyzed, while qualitative trends were determined in submicron particles due to electron beam limitations [12]. (The supermicron fraction constituted more than 70% by weight of all particles.) Particle morphology and size were determined by secondary electron imaging on a Jeol 850 FE-SEM operating at 15 keV. At least 2000 particles were automatically analyzed per each data set to obtain representative information. Inaccuracy of metal analyses was typically below 7%, while precision of particle size/surface measurements, governed by counting statistics, was above 95% [11]. The different phases formed were identified with a Rigaku X-ray diffractometer operating at 50 kV and 150 mA. A 17-kW high-intensity rotating Cu anode was used as a target with point focusing system applying curved crystal monochrometer [7]. Approximately 0.1 g of the fine powder was analyzed per sample between 2θ angles of 10◦ and 130◦ . The bulk concentration of metals in particles and

their leachability (TCLP tests) were determined using a Perkin–Elmer PE-40 inductively coupled plasma spectrometer (ICP) [9]. Details of the surface topography and submicron particulate attachments were characterized with an atomic force microscope (AFM Nanoscope III) at magnifications exceeding one million times in two dimensions. In addition, a computeraided optical stereomicroscope (Olympus Z110) was used to determine ash porosity [9].

3. Results Scanning electron images taken at different locations along the combustion process provide insight into the mechanism of biomass burning and particle evolution in FBC and LHL settings (Fig. 3). Initial stages are similar for both combustion modes. However, significant differences result from the presence of the high-temperature region in the LHL. Initially, the surface of all biomass fibers (Fig. 3A) is simultaneously subjected to the thermal input and undergoes shrinking and descending. As a result, chainlike porous aggregates are formed below 800 K, replacing the initial fibers (Fig. 3B). These aggregates are subjected to internal overpressure caused by devolatilization, triggering the collapse of the porous aggregate structure into smaller, less porous fragments (48% porosity), seen in Fig. 3C (attrition due to particle–particle collisions is also a factor). It was observed that intensive burning between 800 and 1100 K occurred around the holes located at the center of the primary fibers, which provided easy passages for oxygen to enter the aggregate’s interior. Thus, subsequent oxidation reactions cause internal burning and shrinking of the particle chains at around 1100 K and their coalescence, forming spherical shapes. The final fly ash formed in the FBC was dominated by particles with diameters of 5–25 µm and a highly porous (∼42% porosity) surface (Fig. 3D). The final particles formed in the LHL were different (∼14 porosity) (Fig. 3E). At temperatures exceeding 1350 K in the HTR, it was clear that sintering of the inorganic particles occurred. Due to high alkali content most of the ash appeared as a liquid phase in the HTR which, on cooling, transformed into spherical supermicron particles (10–30 µm) with many submicron particles bound to their surface (see Fig. 3E). At this stage, the end of the HTR, the molten ash structures are suddenly quenched in the second lowtemperature region (below 700 K). Surface solidification and structure compacting dominate, preserving changes imposed by the HTR. The surfaces observed in the AFM revealed the presence of individual dendrites 2.0 nm in width. These dendrites which developed as a result of the sudden solidifica-

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

253

Fig. 3. Structural evolution of the biomass during combustion. SEM images illustrate changes in the particle morphology: (A) initial biomass fibers; (B) porous aggregates; (C) disintegrated fragments; (D) individual porous FBC particles; (E) final LHL supermicron particles with attached submicron particles.

tion of the particles in the low-temperature zone after prior melting in the HTR, provide strong and stable bonds between sub- and supermicron particles. Xray analysis of the particles indicated the presence of calcium aluminosilicate (anorthite, CaAl2 Si6 O8 ) as a major phase. The LHL-generated particles also contained sodium aluminosilicate (albite, NaAlSi3 O8 ) and K2 O·Al2 O3 ·4SiO2 . The interior of the particles was densely packed by the aluminosilicate phases. No initial segregation between light and heavy metals was observed in the feed biomass material [13]. Heavy metals (Cd, Pb, Cr, Ni) were rather uniformly distributed over the entire cross section due to doping and thorough mixing, while light metals (K, Na, Ca, Al) were dispersed randomly [9,13]. However, distribution of metals in the FBC- and LHLgenerated final particles was clearly linked to their respective morphological changes. Profiles of metal concentrations are shown in Fig. 4 (heavy metals) and Fig. 5 (light metals). Approximately 2000 particles were analyzed. Quantitative data are illustrated in cross sections of the FBC and LHL particles with identical equivalent diameters of 24 µm (radial profiles are shown). Cr was slightly scattered while Cd was relatively uniformly distributed in the FBC parti-

cles (Fig. 4A). The surface concentration of Cr was a little higher than that in the particle center (4610 ppm vs 4540 ppm, respectively). Cd was located in discrete pockets along the particle’s radius. Its concentration was maximum (3210 ppm) 3.0 µm below the surface. The amount of Cr on the surface was insignificantly higher than that in the center (3080 ppm vs 3060 ppm, respectively). Distributions of Cr and Cd in the LHL-generated particles were entirely different. Both heavy metals were found primarily near the particle center, 9–12 µm from the surface (Fig. 4B). Cd concentrations rose from 170 ppm at the external surface to 3850 ppm in the particle center. Very little Cr (11 ppm) was found in the 3-µm-thick external surface layer. Cr concentration increased from 280 ppm at 6.0 µm from the surface to 5475 ppm in the center. Thus, the final LHL particle had a heavy metaldominated “nucleus.” It is clear that the internal microstructure of particles strongly depended on biomass combustion mode (LHL vs FBC). A mass balance performed for the heavy metals (also including Pb, Ni, Zn, Ba, Sb, and Cu) indicated that 94.3% of the initial amount was retained in the LHL particles (93.2% in the FBC case). Thus, a sig-

254

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

Fig. 4. Radial profiles of heavy metal (Cr, Cd) concentrations in the particles generated during (A) FBC and (B) LHL combustion. The diameter of both FBC and LHL particles is 24 µm.

Fig. 5. Radial profiles of light/alkali metal (Ca, Na + K) concentrations in the particles generated during (A) FBC and (B) LHL combustion. The diameter of both FBC and LHL particles is 24 µm.

nificant portion of these toxic metals was prevented from escaping into the flue gas stream. In addition, little leachability of heavy metals was observed in the LHL-generated particles (0.14 and 0.06 ppm for Cd and Cr, respectively). Immobilization of heavy metals in the LHL was due to their fixation within the aluminosilicate matrix formed in the HTR [7]. The leachability of the FBC-generated particles was 30.7 and 14.3 ppm for Cd and Cr, respectively. This means that LHL combustion significantly improves containment of heavy metals within the fly ash as compared with the conventional FBC technique. It should be mentioned that LHL leachability results fall below the environmental limits while FBC leachability is above

the acceptable levels causing FBC fly ash to be classified as hazardous [14]. Metals that formed the ash matrix, Si and Al, were homogeneously distributed in the particles’ cross sections regardless of the combustion mode (such behavior of Al/Si is typical of inorganic particles due to their dominating amounts). Profiles of light/alkali metals Ca, Na, and K are shown in Figs. 5A (FBC) and 5B (LHL). No distinct pattern was observed in the FBC case (at least not as distinct as that obtained in the LHL case, which is discussed below). The light/alkali metal concentrations varied between 12.3 and 14.5% for Ca and between 0.3 and 0.5% for Na + K. One would expect a relatively constant distribution of alkalis across the particle in the FBC

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

case in which a driving force such as the HTR is lacking. Therefore, dispersion of Ca and Na + K in FBC particles (Fig. 5A) could likely be linked to a transient combustion process observed in porous structures between 800 and 1100 K, which is well within the range of FBC bulk temperatures (localized temperatures would certainly be higher than bulk temperatures). In the LHL case, the concentrations of Na + K and Ca were highest in the near-surface region (Fig. 5B). The cumulative concentration difference for light metals between the particle surface and center was 13.7% (16.3% in the surface region vs 2.6% in the particle core). Thus, contrary to unsystematic scattering in the FBC particles, light metals showed rather regular concentration patterns. (The authors verified that the distribution of semivolatile alkali metals could likely not be attributed solely to vaporization processes [7,10].) These low-melting compounds were mostly located in the external surface and subsurface layers of the LHL-generated particles. X-ray diffraction tests confirmed that these metals existed as either inorganic oxides (Na2 O, K2 O, CaO) or oxide– aluminosilicate complexes (e.g., NaAlSi3 O8 ), which have higher melting points and lower vapor pressures than the initial alkalis.

4. Discussion Why do metals behave in such a peculiar, but practically important, way in the LHL? We believe that the presence of the HTR is the key. (No HTR was present in the FBC mode.) As observed previously [10], the mobility of light metals is about two orders of magnitude higher than the mobility of heavy metals (10−9 m2 /s vs 10−11 m2 /s) in the ash melt. The difference between the migration rate of smaller ions, Na+ and K+ (∼0.6 Å in size), and larger ions, Cd2+ and Cr3+ (∼1.0 Å), forces K+ and Na+ toward the surface, whereas Cr3+ and Cd2+ are concentrated mainly inside the particle melt core [7]. Apart from the difference in metal migration, it is believed that the influence of surface tension or surface free energy on the diffusion of these metals may contribute to their final radial distribution. Both self-diffusion and chemical diffusion are occurring in the LHL particles (precisely, in the aluminosilicate melt). Self-diffusion involves movement of various species and, in general, does not contribute to the establishment of a net flux or a chemical potential gradient. As the limited surface evaporation initiates the concentration/chemical potential gradient between the center and the surface [7,10], chemical diffusion of light and heavy metals inside the particle is stimulated. The origin of this chemical potential

255

gradient is the excess surface free energy, which influences diffusion of the surface-active metals such as Na and K toward the surface. In general, the relationship between the surface tension and surface free energy of the multicomponent system is given by [15]  Γi µsi , σ =fs − (1) i

where σ is surface tension, f s is surface free energy, and Γi and µsi are the surface excess and chemical potential of the ith component, respectively. Equation (1) also implies that for a single-component system, surface tension is equal to surface free energy. The surface tension of the molten particle in the high-temperature region of the LHL is reduced due to diffusion of the surface-active metals to the liquid– solid interface. Na and K are considered very effective surface-active metals [16]. The diffusion of Na and K through the silicate melt proceeds via breakage of the Si–O–Si bonds according to the reaction M2 O + Si– O–Si → Si–OM + MO–Si (where M is Na or K). This chemical diffusion through the particle leads to the excess concentration of Na/K metals on the surface of the LHL particles. The surface excess of the single metal could be calculated as [17] Γi = −Ci /RT (dσ/dCi ) [mol cm−2 ],

(2)

where Ci is the concentration of solute and dσ/dCi is the flux. The surface excess profiles at different temperatures, corresponding to different particle evolution stages, are shown in Fig. 6 (they were calculated using EPMA data). The surface excess for Na and K increases with temperature. The slope, which corresponds to the variation in surface tension with solute concentration, becomes more negative, indicating reduction in the surface free energy of the system. Thus, the combined effect of the high temperature and the increase in concentration of the surface-active metals leads to the reduction in surface free energy. As expected, the surface excess of Cr and Cd decreases with an increase in temperature (Fig. 6). According to surface thermodynamics [18], processes responsible for the reduction in surface free energy occur spontaneously. Thus, to reduce surface free energy (and tension), the system uses the excess energy toward the kinetic motion of the small, surface-active ions such as Na+ and K+ , promoting their diffusion toward the surface. This explains their higher concentrations near the surface than in the particle interior. Because of their surface “inactivity” (see Fig. 6), as well as a diffusion rate two orders of magnitude lower, heavy metals reside mostly in the particle center. This, in turn, enables them to interact efficiently with the Al–Si matrix present in the particle core. Their longer residence time inside the core,

256

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

Fig. 6. Variation of the surface excess of Na + K, Cr, and Cd with the inverse of temperature. The slope corresponds to the variation of surface tension with respect to solute concentration. The surface excess of Na + K increases with temperature, indicating reduction in the particle surface free energy.

compared with alkalis, promotes better interaction. Because the subsequent particle quenching in the second low-temperature region of the LHL is relatively fast (∼500 K/s), radial variations in metal concentrations, established in the HTR region, are preserved. Heavy metals are entrapped in the center of quickly solidifying dense particles, which ensures their immobilization, thus inhibiting their leaching. Another consequence of the surface domination by the alkalis is the observed bonding of submicron particulate to LHL-generated supermicron particles (see Fig. 3E). Because of their adhesive nature, Na and K act as links between super- and submicron particles, promoting their attachment (the nature of these particle–particle bonds is not clear so far). It was observed that during migration, alkalis change “partners,” reacting with other components of the particle multi-oxide system and forming new inorganics. Na and K, which are responsible primarily for fouling and slagging during biomass combustion, form new complexes (NaAlSi3 O8 and K2 O·Al2 O3 ·4SiO2 were identified) revealing melting points higher and vapor pressures lower than those of the initial alkalis. Thus, slagging may be abated because these particles have high melting points. Massive condensation of these metallic complexes in the cold regions of the LHL is also diminished, because of their lower vapor pressures compared with the alkalis. Thus, fouling is considerably reduced in comparison to the classic FBC biomass combustion.

5. Conclusions and practical connotations It was shown that the LHL approach can be successfully applied in biomass combustion on a pilot scale. Moreover, the LHL method offers better handling of ash particles and metals when compared with

classic FBC. It seems that the presence of the HTR in the LHL plays an important role in metal–ash interactions. Quantitative analyses of metal distributions in particles indicated a surface predominance of alkalis/light metals and core predominance of heavy metals. In the FBC case, metals were randomly scattered all over the particle cross section. It is likely that the combination of the HTR and changes in the surface free energy of the particles is the driving force for the observed behavior of metals. Practical implications of the LHL method are severalfold. It allows for generating particles with heavy metals immobilized within the aluminosilicate matrix. Their leachability is two orders of magnitude lower than the leachability of heavy metals from FBC-generated particles. This means that LHL combustion significantly improves heavy metal containment. Consequently, LHL ash particles are classified as nonhazardous and can be safely reused, whereas FBC ash would need to be treated prior to safe disposal. In addition, the presence of Na/K on the particle surface permits bonding of the submicron particles to the surface of the supermicron particles. Thus, the submicron particulates can be removed together with the supermicron solids using the classic air pollution control equipment (it is difficult, if not impossible, to capture submicron particles using the classic air pollution control equipment). Another advantage of LHL over FBC is the possibility of reduced slagging and fouling as the process helps to form metal– aluminosilicate complexes with melting points higher and vapor pressures lower than those of the alkalis, which are primarily responsible for fouling and slagging during FBC biomass combustion. These findings may contribute to redesigning of currently operating FBC units in order to generate nonhazardous, nonleachable, reusable particles where heavy metals are

F. Eldabbagh et al. / Combustion and Flame 142 (2005) 249–257

immobilized while environmental and technological problems are reduced. Acknowledgments We acknowledge the financial support from the Natural Sciences and Engineering Research Council of Canada (NSERC) through Discovery, Strategic, and Equipment grants (NSERC Grants RGPIN170464, STP-246053, EQPEQ-252297). Continuous support provided by Canmet Energy Technology Centre (CETC-O), BIOCAP Canada Foundation, and Biotechnology Research Institute is highly appreciated. We also acknowledge Bowater, KMW Systems, Rio Tinto, and NUI for supporting this research. References [1] R. Hurt, Proc. Combust. Inst. 27 (1998) 2887–2904. [2] K. Scheffer, R. Stülpnagel, in: A.A.M. Sayigh (Ed.), Biomass for Energy, Pergamon, London, 2002, p. 446. [3] A. Williams, M. Pourkashanian, M. Jones, Proc. Combust. Inst. 28 (2000) 2141–2162. [4] T. Abbas, P. Costen, F. Lockwood, Proc. Combust. Inst. 26 (1996) 3041–3058.

257

[5] J.O.L. Wendt, Proc. Combust. Inst. 25 (1994) 277–289. [6] W.P. Linak, J.O.L. Wendt, Prog. Energy Combust. Sci. 19 (1993) 145–167. [7] A. Ramesh, J.A. Kozinski, Combust. Flame 100 (2001) 920–930. [8] F. Eldabbagh, J.A. Kozinski, Rev. Sci. Instrum. 75 (2004) 5308–5314. [9] F. Eldabbagh, Novel Technique and Facility for Thermal Treatment of Solid Residues, Ph.D. thesis (Honor’s), McGill University, Montreal, 2003. [10] J.A. Kozinski, G. Zheng, Proc. Combust. Inst. 27 (1998) 1745–1752. [11] A. Ramesh, J.A. Kozinski, Chem. Eng. Sci. 56 (2001) 1801–1809. [12] A. Ramesh, J.A. Kozinski, Appl. Surf. Sci. 152 (2000) 185–192. [13] R. Saade, J.A. Kozinski, Biomass Bioenergy 18 (2000) 391–404. [14] Regulation No. 625/6-99/014, Hazardous Pollutants, Environment Canada, 1999. [15] A.I. Rusanov, Surf. Sci. 23 (1996) 173–247. [16] T. Dunn, in: M.C. Scarfe (Ed.), Diffusion in Silicate Melts, 1986, pp. 57–92. [17] D. Chattoraj, K. Birdi, in: Adsorption and the Gibbs Surface Excess, Plenum, New York, 1984, pp. 7–44. [18] E.T. Turkdogan, in: Physical Chemistry of High Temperature Technology, Academic Press, New York, 1980, pp. 60–123.