Kinetic Study of Hydrated Lime Reaction with HCl - ACS Publications

Apr 12, 2003 - Environmental Science & Technology .... A fundamental study was carried out to investigate the kinetics of HCl−lime reaction under si...
6 downloads 9 Views 165KB Size
Download PDF
Environ. Sci. Technol. 2003, 37, 2556-2562

Kinetic Study of Hydrated Lime Reaction with HCl RONG YAN,* TERENCE CHIN, DAVID TEE LIANG, KARIN LAURSEN, WAN YEAN ONG, KAIWEN YAO, AND JOO HWA TAY Institute of Environmental Science & Engineering, Nanyang Technological University, Innovation Center, Block 2, Unit 237, 18 Nanyang Drive, Singapore 637723

Hydrochloride (HCl) is an acidic pollutant present in the flue gas of most municipal or hazardous waste incinerators. Hydrated lime (Ca(OH)2) is often used as a dry sorbent for injection in a spray reactor to remove HCl. However, due to the short residence time encountered, this control method has generally been found to have low conversion efficiencies which results in the high lime usage and generates large amount of fly ash as solid wastes. A fundamental study was carried out to investigate the kinetics of HCl-lime reaction under simulated flue gas conditions in order to better understand the process thereby providing a basis for an optimized lime usage and reduced fly ash production. The initial reaction rate and conversion of three limes were studied using a thermogravimetric analyzer by varying the gas flow rate, temperature (170400 °C), and HCl concentrations (600-1200 mg/m3) as well as the associated particle size and surface area of the limes. The initial lime conversions were found to rely mostly on the residence time, while the ultimate lime conversions were strongly influenced by temperature and the reaction products. CaOHCl was found to be the primary product in most cases, while for one specific lime, CaCl2 was the ultimate conversion product after an extended time period. The true utilization of lime in flue gas cleanup is thus higher when CaOHCl is considered as the final product than those based on CaCl2 as the final product, which has been commonly used in previous studies. The initial reaction was controlled by diffusion of HCl in gas phase and the subsequent reaction by gaseous diffusion through the developing product layer. Increasing the HCl concentration raised the initial rate as well as conversion. However, overloading the lime with excessive HCl caused clogging at its surface and a drop in the ultimate conversion. Limes with smaller particle diameters and higher surface areas were found to be more reactive. The effect of gas-phase mass transfer was minimized when an optimum flow rate was chosen, and in the absence of internal diffusion the reaction was found to be first order with respect to HCl concentration.

1. Introduction Hydrogen chloride (HCl) emissions from flue gases are mostly found in municipal and hazardous waste incinerators (MWI * Corresponding author phone: 65-67943244; fax: 65-67921291; e-mail: [email protected]. 2556

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 11, 2003

and HWI), due to the destruction of halogenated organic wastes. HCl emissions can also result from the combustion of certain fossil fuels such as chlorine-rich coals. Published regulations (U.S. EPA) require new and large MWIs undergoing major modifications to remove 95% of HCl, while HWIs must remove 99% of the HCl emissions. New coal-fired power plants and those undergoing major modifications must also report HCl releases if they exceed 10 t/year. Although wet scrubbing is the most common method used, HCl can also be removed from flue gases by dry spraying of Ca-based sorbents (namely Ca(OH)2, CaO, and CaCO3). Hydrated lime (Ca(OH)2) is favorable for scrubbing HCl at relatively low temperatures that allows for higher recovery of energy from the flue gas, lower volume of gas to be treated and enables the capture of some heavy metals as they condense. Depending on the injection temperature, the species reacting with HCl may be Ca(OH)2 or a product of its decompositionsCaOsif injection takes place at a temperature higher than 400 °C. Similarly, when calcium carbonate is injected, the reacting species may be CaCO3 or CaO if the injection takes place at a temperature higher than 650 °C. Studies on various dry Ca-based sorbents for HCl removal have been reported previously (1-16). Fundamental and bench-scale investigations on the kinetics and mechanisms of HCl reaction with Ca-based sorbents have been carried out by thermogravimetric analysis (1-3), using fixed-bed reactors (4, 6, 9-13), or in a cyclone reactor (15). Attempts have also been made to mathematically simulate the process of dry scrubbing of HCl (5, 7-8). Most of the published works assume the overall reaction of Ca-based sorbents with HCl forms CaCl2 (2, 4-11), whereas Jozewicz and Gullett (1) recently identified CaOHCl as the product of HCl reaction with Ca(OH)2 or CaO over the 100-600 °C range in a shorttime differential reactor. Furthermore, Allal et al. (16, 17) studied the mechanism of Ca(OH)2 reaction with HCl and found that the reaction leads to the formation of not only CaCl2 but also CaOHCl. The authors also determined the thermodynamic data of CaOHCl (17). Knowledge of reaction products is necessary to accurately determine lime conversion efficiency following contact with HCl, while sorbent conversion is often an important factor in determining the operation costs of a flue gas cleaning process. Sorbent injection for HCl removal has generally been found to have relatively low conversion, due to the short residence time encountered. This results in high lime usage that in turn generates large amount of fly ash as solid waste. In addition, the kinetics and reaction mechanisms involved in this process have not been fully understood. So far, only limited studies focus on the reaction of hydrated lime with HCl (1, 6, 8, 10, 16). A fundamental study was carried out to investigate the kinetics of HCl-lime (Ca(OH)2) reaction under simulated flue gas conditions in order to better understand the process thereby providing a basis for optimizing lime usage and reducing fly ash production.

2. Experimental Section 2.1. Materials and Apparatus. Three hydrated lime samples (referred as lime 1, 2, and 3) were provided by a local lime supplier. The HCl concentrated gas (2094 ppm, balanced in nitrogen) and high purity nitrogen (99.999%) were supplied by Messer Singapore while purified oxygen (99.99%) was obtained from Soxyl. Several instruments were used to determine the physical properties of the raw and exhausted lime samples. The surface areas and pore volumes of limes were measured by nitrogen adsorption using the Micromer10.1021/es020902v CCC: $25.00

 2003 American Chemical Society Published on Web 04/12/2003

FIGURE 1. Experimental setup diagram of lime reaction kinetic study. itics BET analyzer model ASAP 2010. Powder X-ray diffraction (XRD) data were collected on a Siemens D5005 BraggBrentano diffractometer using CuKR radiation operating at 40 kV and 40 mA, and phase identifications were conducted using the Bruker software Diffrac-plus EVA supported by the Powder Diffraction File (PDF-2). A laser diffraction instrument (Malvern Mastersizer) was used to evaluate particle size distribution (PSD) of the limes. A Thermogravimetric Analyzer (TGA-2050 by TA Instruments) was used as a reactor for investigating the reaction kinetics between lime and HCl. Figure 1 shows the experimental setup. HCl (f) was the reaction gas, while oxygen gas (g) kept at 9% of the total flow was used to simulate incinerator flue gas conditions. Nitrogen served as (1) bypass gas (b) for keeping constant flow rate conditions, (2) makeup gas (c) for adjusting the synthesized gas flow rate, and (3) purge gas (e) when the HCl reaction has been completed. Nitrogen gas (a) was also supplied as an inert gas to continually purge the TGA’s mechanical balance chamber whenever it is switched on. SO2 and CO2 (h) as well as a column of water for the introducing humidity (d) are features for the study currently being conducted. A chamber was connected to ensure a good mixing of the simulated flue gas before it was admitted into the TGA. Digital flowmeters were used to set the desired flow rates, while control valves were also added to each of these gas lines for adjustments. Two bottles of diluted NaOH solutions were used to scrub away excessive reaction gases before they are exhausted. A 4-way valve (i) was used to alternate between the nitrogen bypass gas (b) and the mixed reaction gases (including HCl (f), nitrogen makeup (c), and oxygen (g)). 2.2. Operating Procedures. A thin layer of lime sample (approximately 10 mg) was spread evenly on the platinum pan of the TGA. The pan was then loaded into the furnace of the TGA. The lime sample was heated in a ramp of 10 °C/min to the desired temperature and kept isothermally for around 30 min. Once the constant reaction temperature and a flat baseline of weight was achieved, the mixed reaction gas was introduced into the TGA furnace by switching the 4-way valve. An immediate weight increase was observed (see Figure 2) which resulted from the formation of the limeHCl reaction products, either CaOHCl or CaCl2 (to be identified by XRD in the subsequent text). Both have a higher

FIGURE 2. TGA curve-weight increase and initial reaction rate. molecular weight than Ca(OH)2. The reaction can be stopped at various desired reaction times by switching back the 4-way valve. N2 was immediately purged into the system for another 1 h to ensure a clean system before proceeding to the next test and to prevent HCl from corroding any part of the system. The rate of weight increase (obtained from the TGA curve, see Figure 2) was used to describe the kinetics of the reaction. The determination of weight gain at a specific time was also evaluated for calculating the lime conversion. Generally, rapid increases in weight can be observed for the first hour. Thermal decompositions of lime samples under either nitrogen gas or air were measured first for comparison. This information is also useful for the determination of desired reaction temperatures. The baseline run was carried out under the same experimental conditions as the sample tests, except that the HCl gas was replaced by N2. The average of three baseline tests gave 0.00564 wt %/min in terms of initial reaction rate, which is negligible when compared to that of HCl reaction with lime. Several parameters were investigated in terms of their effects to the reaction of HCl and lime. These included the total gas flow rate, reaction temperature, and HCl concentration under the following conditions: (1) Effect of Gas Flow Rate. Five different flow rates were chosen: 100, 140, 150, 170, and 200 mL/min, while keeping VOL. 37, NO. 11, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2557

other experimental parameters constant (temperature 200 °C, HCl concentration 914 mg/m3, and O2 concentration 9%). (2) Effect of Temperature. Four different temperatures were chosen to simulate the real flue gas temperature conditions: 170, 200, 230, and 300 °C, keeping other experimental parameters constant (total flow rate 150 mL/min, HCl concentration 914 mg/m3, and O2 concentration 9%). (3) Effect of HCl Concentration. Four different HCl concentrations were used: 600, 800, 1000, and 1200 mg/m3, with other experimental parameters kept constant (total flow rate 150 mL/min, temperature 200 °C, and O2 concentration 9%). 2.3. Reaction Rate and Conversion. In general, from the TGA graphs (Figure 2), a slight decrease in weight (about 0.5%-1.0%) was first observed as the temperature increased from ambient to the reaction temperature due to removal of free water from the lime. A flat baseline was achieved after removing all the water, followed by a sharp rise of the curve when the mixed reaction gas (HCl, nitrogen and oxygen) was introduced, indicating the start of reaction between lime and HCl. (1) Calculation of Initial Reaction Rate. The initial reaction rate is given by the slope of the curve-weight change (%) versus time (min). In this study, the initial reaction rate is defined by the slope of the first 10 min from the start of reaction, given in %/min. (2) Calculation of Initial Global Reaction Rate. The initial global reaction rate calculation takes into account the lime particle size and diameter. With the initial reaction rate (mg/ min) and known values of lime surface area, the initial global reaction rate is defined as

RAO )

(dW/dt) Ao(Mproduct - Mreactant)

(1)

where RAO is the initial global reaction rate; dW/dt is the initial reaction rate in mg/min; Ao is the surface area of particles; and Mproduct and Mreactant are the molecular weight of solid reaction product and solid reactant, respectively. (3) Calculation of Lime Conversion at a Specific Time. Lime conversion at a specific time t ) 10, 20, 50, 100, 200, and 400 min were calculated for each set of experiment, based on the reaction: Ca(OH)2 + 2HCl f CaCl2 + H2O, using the formula

lime conversion (%) )

∆ mole of Ca(OH)2 reacted

× initial mole of Ca(OH)2 100% (2)

or lime conversion (%) ) weight increase at spec. time (mg) × 100% (MCaCl2 - MCa(OH)2) × (initial weight)/MCa(OH)2 (3) where MCaCl2 and MCa(OH)2 are the molecular weights of CaCl2 and Ca(OH)2, respectively. The initial weight of dehydrated Ca(OH)2 at the desired time can be obtained from the TGA graph of weight change versus time.

3. Results and Discussion 3.1. Lime Decomposition. The screening for lime decomposition was conducted from ambient temperature to 900 °C under the presence of N2 and air, respectively. Both have the same decomposition pattern occurring from 400 °C to 475 °C. Nevertheless, one sample test conducted with lime 1 at 350 °C showed that decomposition had already begun at that temperature. The rate of decomposition is very slow at 350 °C, taking about 200 min for 20% weight of loss. At 400 °C, it only takes about 10 min for the same percentage weight loss. The decomposition of lime occurring below 400 °C was 2558

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 11, 2003

also reported by Jozewicz and Gullett (1). Using the Differential Scanning Calorimetry (DSC), the author observed that the decomposition of Ca(OH)2 occurs from 300 °C onward. For this study, the temperature was chosen to be close to the real flue gas temperature (∼200 °C). 3.2. Lime Characterization. 3.2.1. XRD. XRD analysis were carried out for three raw lime samples and were confirmed to be predominantly Ca(OH)2 (>90% in weight). The analysis was also carried out for the exhausted lime sample after it had undergone various stages of reaction. For lime 2 sample after about 800 min of reaction, five peaks are observed at 2θ ) 26.7°, 28.2°, 32.3°, 36.2°, and 38.2° which are identified as Ca(OH)Cl (Figure 3a). However, for the lime 3 sample after 800 min of reaction, only one major peak is observed at 2θ ) 29.3°, which was identified as CaCl2 (Figure 3b). Additional scans for lime 3 samples undergoing 70 min and 140 min of reaction, respectively, were carried out, and the results show that for both cases the product formed is Ca(OH)Cl. It is suggested that both lime 2 and lime 3 react with HCl to primarily form Ca(OH)Cl, whereas prolonging the reaction for an extended time (800 min in this study) leads to the formation of CaCl2 but only for lime 3. However, in the case of lime 1 sample after a long reaction time, apart from Ca(OH)Cl detected, XRD analysis also gives one major peak at around 2θ ) 10.7° which represents a new Ca-based crystalline form, but it could not be identified based on the currently available databases and software. These results show that the products of lime-HCl reaction may be different with respect to the reaction stages, lime types, and experimental conditions. The complexity of the lime/HCl reaction products has also been explored previously (1). The varied products from the three lime samples studied are accounted for by their different nature, such as textures (see Table 1) and chemical compositions (∼95% of Ca(OH)2 in lime 3; ∼92% in lime 1 and 2) as well as sources (lime 3 is originally from Germany while the other two are local). Further experimental studies are currently being carried out to understand in depth the chemistry and mechanisms involved in this process. Although the chemical analysis indicates complex products of lime/HCl reactions under different conditions including lime type and reaction extent, a specific reaction path and product were chosen to standardize the calculation in terms of rate and conversion for the ease of comparison, since reaction kinetics is the main focus of this study. This reaction path was also chosen in order to alert those operators who might otherwise attribute poor lime conversions to unrelated process parameters in optimizing their system performances. The following reaction is thus used as the basis for calculation and comparison of lime conversions for all three lime samples, while keeping in mind that Ca(OH)Cl is often the predominant product for most cases and as such, the conversion should be higher:

2HCl (g) + Ca(OH)2 (s) f 2H2O (g) + CaCl2 (s) (4) So far, most previous publications assumed CaCl2 as the product to describe the reaction between Ca-based sorbents and HCl (2, 4-11). By using DSC, TGA, and XRD identifications, Jozewicz and Gullett (1) concluded that the main product of reaction between Ca(OH)2 and HCl at the temperature range from 100 °C to 600 °C is a solid solution of CaCl2‚2H2O, Ca(OH)Cl‚xH2O, and unreacted Ca(OH)2. The author also suggested that it is possible for Ca(OH)Cl to be the primary product, and CaCl2‚2H2O is only formed when this primary product further reacts with HCl. Our experimental findings with respect to lime 3 confirm this point of view. On the other hand, CaCl2‚2H2O may also react with unreacted Ca(OH)2 to form Ca(OH)Cl (1, 17). During the real flue gas cleanup using Ca-based sorbents, Ca(OH)Cl is

FIGURE 3. (a) XRD of lime 2 after 800 min of reaction 3 and (b) XRD of lime 3 after 800 min of reaction.

TABLE 1. Identification of Raw and Exhausted Lime Samples Using PSD and BET lime sample

lime 1

lime 2

lime 3

average particle diameterbefore reaction (µm) average particle diameterafter reaction (µm) BET surface area (m2/g) micropore area (m2/g) external surface area (m2/g) pore volume (cm3/g) average pore diameter (Å)

9.53

6.73

4.50

14.9

8.29

10.06

17.1 1.36 15.7 0.064 150.1

15.0 1.76 13.3 0.062 165.2

38.1 2.46 35.7 0.119 125.3

supposed to be one of the major products of lime-HCl reaction due to the short residence time encountered, and thus the true utilization of sorbents should be higher than those based on the assumption of a CaCl2 product. 3.2.2. Particle Size and Surface Area. Lime 3 has the highest BET surface area of around 40 m2/g and the smallest average particle diameter at 4.50 µm (Table 1). The average particle diameter after reaction refers to those exhausted samples that have undergone almost complete reaction with HCl. It was found that for all three samples, the average particle size increased after reaction. Moreover, the comparison of particle size distribution before and after the reaction shows a significant increase in overall particle diameters after undergoing reaction with HCl. It has been recognized that the volume increase (i.e., the molar volume of the products are larger than those of the reactants) may cause pore plugging that has been assumed as the reason for low lime conversion (5, 7). The time taken to achieve 25% conversion of lime (T25) is compared with the particle sizes of the different lime samples (see Figure 4). The points seem to show a correlation, i.e., the time taken to achieve 25% conversion increases as the average particle size increases. A similar method was used to investigate the effect of surface area on the reaction.

FIGURE 4. Effect of particle size on the lime conversion of HCl at different conditions. A plot of the time taken to achieve 25% conversion versus the BET surface area of the different lime samples shows roughly a reduction in the time with respect to the increase of lime surface areas. The similar trends of particle size and surface area effect to sorbent-HCl reaction were also found in the literature (3, 11, 14), whereas Weinell et al. (6) reported very little surface area effect. 3.3. Effect of Total Flow Rate. The effect of total flow rate on the lime reaction was investigated, and the results for lime 1 are presented in Figure 5 as a representative. Similar patterns were found with lime 2 and 3. The initial reaction rate and conversions at different reaction times increased with flow rate until it reaches the maximum at 150 mL/min. However, the initial reaction rate and conversions started to drop gradually as the flow rate was increased from 150 mL/ min onward. When the flow rate is below 150 mL/min, the reaction is diffusion controlled; the mass transfer in gas is enhanced with the gas flow rate increasing. At flow rates higher than 150 mL/min, the reaction is kinetically limited; VOL. 37, NO. 11, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2559

FIGURE 5. Effect of total flow rate on lime 1 reaction with HCl: (a) initial reaction rate (b) conversion.

FIGURE 6. Conversion (mol %) of lime 1 at different temperatures: (a) conversion ∼ time and (b) conversion at 800 min reaction ∼ temperature. the higher the flow rate, the shorter the residence time of the HCl molecules to be in contact with the lime surface, giving it less time for the reaction. The total flow rate at 150 mL/ min was therefore chosen for all the subsequent experiments. 3.4. Effect of Temperature. The effects of temperature on the reaction were investigated at 170, 200, and 230 °C for all the three lime samples. Additional points at 300, 350, and 400 °C were carried out only for lime 1. The initial reaction rate and the ultimate conversion increased as the temperature was raised for all three samples in the range of 170-230 °C. Similar effects of temperature were found in previous studies (3, 14), whereas Daoudi et al. observed a decrease of limestone conversion at higher temperature (650 °C) (3). Figure 6 gives an example of lime 1 conversion at different temperatures and times. Most of the conversion takes place in the first 50-100 minutes, which is also true for lime 2 and 3 samples. The conversion after that slows down sharply and flattens out (Figure 6(a)), probably due to the formation of the product layer causing the diffusion of the HCl from the gas phase through the product layer becoming the control step (5, 7). Since the grain diffusion resistance is smaller at higher temperatures, more conversion is achieved as the temperature is raised. At the beginning of the reaction with no product layer formed yet, the internal diffusion limit was negligible. Moreover, at the optimized flow rate of 150 mL/ min, the mass transfer effect in the gas phase was minimized as well (2). The rate is likely to be controlled by the chemical reaction at the gas-solid surface, and the initial reaction rate increased with respect to the temperature. Comparing the behavior of the three lime samples in the range of 170-230 °C, lime 1 has the highest initial global reaction rate followed by lime 2 and 3, respectively. However, as the reaction proceeds, the conversion of lime 2 and 3 overtook the conversion of lime 1. Lime 3 has the highest 2560

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 11, 2003

conversion, followed by lime 2 and 1. On the other hand, at 230 °C lime 1 and lime 2 have relatively comparable performances with lime 3 in terms of conversion, indicating that lime 1 has a better performance at a higher temperature (230 °C). As seen in Figure 6(b), the ultimate conversions (conversion after 800 min) for lime 1 strongly increase with temperature, while at higher temperatures (350 and 400 °C) the calculated conversion even exceeded 100%. This is because the calculation is based on the conversion of Ca(OH)2 to CaCl2 (eq 4). However from 350 °C onward, part of the Ca(OH)2 may undergo decomposition to CaO, as mentioned previously. If the latter was used for calculating the conversion instead, the value will be lower and definitively less than 100% due to the different molecular weights of the Ca-species involved (refer to eq 3). The reactions at the different temperature zones involve the following: (1) reaction path I (170 e T e 230 °C, conversion e 50%):

Ca(OH)2 (s) + HCl (g) f H2O (g) + Ca(OH)Cl (s) (2) reaction path II (300 °C e T < 350 °C, 50% e conversion e 100%):

Ca(OH)2 (s) + 2HCl (g) f CaCl2 (s) + 2H2O (g) (3) reaction path III (350 °C e T e 400 °C, 100% e conversion):

Ca(OH)2 (s) f CaO (s) + H2O (g); CaO (s) + 2HCl (g) f CaCl2 (s) + H2O (g) The XRD test for the exhausted lime samples after reaction at 300 and 400 °C confirmed that the predominant formation

reactions may occur over a range of temperatures and forming different products. The present study demonstrates that lime conversion is determined by the kinetic and pathway of HCl/ lime reactions; this fundamental knowledge will provide a better understanding to product speciation and lime utilization under actual practical conditions.

4. Evaluation of Reaction Kinetics

FIGURE 7. Initial reaction rate for lime 1, 2, and 3 at different HCl concentrations. of CaCl2(s) at both path II and path III. The results indicate that the initial conversions rely mostly on the residence time, whereas the ultimate conversions are strongly influenced by temperature and particularly, decided by the actual reaction path. Therefore, knowledge of the reaction products is essential to accurately determine the lime conversion efficiency in the real flue gas cleanup process. 3.5. Effect of HCl Concentration. It can be seen from Figure 7 that the initial global reaction rates for lime 1, lime 2, and lime 3 increased as the HCl concentrations was raised, as expected. However, for lime 1 the initial reaction rate started to drop when the HCl concentration exceeded 1000 mg/Nm3. Over longer time periods, higher HCl concentrations do not necessarily give a greater lime conversion. The highest lime conversion was observed when the HCl concentration is at 800 mg/Nm3 for all the lime samples, whereas the ultimate conversion at 1200 mg/m3 of HCl is lower than those achieved at 600, 800, and 1000 mg/m3, respectively. Among the three studied samples, lime 3 has demonstrated the highest conversion followed by lime 2 and lime 1, with respect to the varying HCl concentrations from 600 to 1200 mg/m3. This reduction in lime conversion at higher HCl concentrations could be due to the clogging of the pores on the lime surface. When the HCl concentration is high, the rate of reaction on the lime surface is more rapid, clogging the external particle surface and isolating the interior from reaction 4. Simons also observed the similar phenomenon of the clogging at the surface of the pores for his study of SO2 reaction with calcined limestones (18). In conclusion, the reaction of lime with HCl involves several dynamic phenomena: bulk gas diffusion to lime particle surface, chemical reaction of lime and HCl, formation of product layer, and gaseous diffusion through the developing product layer. As expected, the initial lime conversion is determined by residence time, whereas the ultimate conversion is strongly influenced by temperature and the reaction path involved. Knowledge of the reaction product is thus essential to accurately determine the real lime utilization. However, the reaction products vary (i.e. Ca(OH)Cl, CaCl2, or other Ca-based species) with respect to the lime used and the reaction time encountered. It is supposed that the true utilization of Ca-sorbents in flue gas cleanup is higher than those based on assumption of a CaCl2 product. Given that lime powders are sprayed into the HClcontaining hot flue gas streams under practical conditions, the lime residence times are therefore expected to be significantly shorter than the present experiments. The gasphase mass transfer effects and formation of product layer are also expected to be less critical due to better gas/solid contact and shorter reaction time. On the other hand, under dry injection conditions, individual lime particles are unlikely to achieve thermal equilibrium, which means lime/HCl

The following theoretical considerations based on the “unreacted core model” as given in Daoudi et al.’s work (2) are adopted to calculate the kinetic parameters related to the studied reaction. The following equations are used to describe the chemical (intrinsic) reaction rate (for the reaction shown by eq 4) and mass transfer rate from the bulk gas to the surface of the solid, after minimizing the internal mass transfer resistance by analyzing the reaction based solely on initial reaction rate

RA ) ksCBOCAsn

(5)

RM ) KMA(CAO - CAs)

(6)

where RA is the molar rate of reaction of HCl (A) per unit surface area of lime solid B, ks is the chemical rate constant per unit surface area, CBO is the initial molar concentration of lime defined by the loading and concentration of Ca(OH)2 in the sample, CAs is the molar concentration of HCl at the surface of the lime, and n is the reaction order. RM is the molar rate at which HCl reaches the surface of lime per unit surface area, KMA is the mass transfer coefficient in the gas film, and CAO is the bulk flow molar concentration of HCl. Assuming that the rate of reaction is first order, the chemical reaction rate and the mass transfer rate can be equated using the following equation

RMAM ) RAAO

(7)

where AM is the characteristic area for external mass transfer (sample holder geometry) and AO is the total surface area of the lime, obtained by multiplying the initial weight of lime with its BET surface area. Substituting eqs 5 and 6 into 7 and taking n as 1, the following global rate equation can be obtained

RA ) K′CAO

(8)

AO 1 1 ) + K′ ksCBO AMKMA

(9)

where

K′ is then the apparent first-order rate constant for the global rate equation. The last term in eq 9 can be reduced to approach zero either by decreasing the total surface area of the lime (initial weight of the sample) or increasing the mass transfer coefficient. Gas-phase mass transfer resistance will be kept at a minimum if the ratio of initial area to gas flow rate can be reduced to its smallest value by increasing the total flow rate. The 1/K′ values and the ratios of the corresponding total surface area over total flow rate were considered for lime 1 and lime 2. They were calculated by the global reaction rate equation (eq 8); initial global reaction rate (RAO) data obtained by eq 1 and the HCl concentration used (914 mg/m3) were substituted into it to calculate the K′ values. The results are plotted in Figure 8. As the ratio of the total surface area over total flow rate reduces, the effect of mass transfer in the reaction becomes insignificant. Theoretically, the mass transfer effect approaches zero when the lines in Figure 8 intercept the y-axis. At this point, the chemical reaction rate constant, ks, is directly VOL. 37, NO. 11, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2561

FIGURE 8. Plot of 1/K′ versus the total surface area over total flow rate ratio.

FIGURE 9. Logarithmic expressions of initial global reaction rate versus HCl concentrations. proportional to K′. For the case of lime 1 and lime 2, the values of the intercept are 0.1046 and 0.7277 min/cm, respectively. Taking the reciprocals, their K′ values (or ksCBO) are then 9.56 and 1.37 cm/min, respectively. This shows that the chemical reaction rate constant, ks, of lime 1 is higher than that of lime 2 since the initial molar concentration of both limes (CBO) used were almost the same. This agrees with the results seen earlier where lime 1 consistently has a higher initial global reaction rate than lime 2 when the temperature and HCl concentration were varied over a wide range. The flow rate of 150 mL/min was kept constant for all the different conditions that were tested in this study since this is the optimal flow rate where the mass transfer effect is either minimal or insignificant. The reaction rate will therefore be determined predominantly by the chemical reaction rate constant. For small values of total surface area over total flow rate ratios, where the gas mass transfer effect is minimal, the global reaction rate (eq 1) can be used to approximate the chemical reaction rate (eq 5). The order of the reaction can be verified by substituting initial rate data obtained under this condition into the chemical rate equation. If the reaction of lime and HCl is first order, a logarithmic plot of eq 5 will produce a line of gradient 1. Taking data from the initial reaction of HCl with lime 1, 2, and 3 carried out at a total flow rate of 150 mL/min, Figure 9 is plotted to express the logarithmic relationship between the initial global rate and HCl concentration for lime 1, 2, and 3. A trend line equation is obtained for each lime sample. As can be seen from the figure, the slopes of lime 2 and 3, being 0.99 and 0.92 respectively, are quite close to 1. Their reactions can therefore be regarded as first order with respect to HCl concentration. As for lime 1, the last data point at 2562

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 11, 2003

[HCl] ) 1200 mg/m3 was left out from the graph since it is quite out of trend. As mentioned earlier, it is postulated that the lime 1 surface could be clogged at high HCl concentrations, leading to the reduced rate. A trend line can still be obtained from the existing three points and the slope is 0.93, indicating the reaction of lime 1 with HCl is first order with respect to lower HCl concentration (e1000 mg/m3). The antilog of the y-intercept of each equation in Figure 9 gives the empirical K′ value (1.91 cm/min for lime 1 and 1.23 cm/min for lime 2); this can be compared to the theoretical values obtained earlier (9.56 and 1.37 cm/min, respectively for lime 1 and 2). The theoretical K′ value for lime 3 is not available. The K′ values obtained from the two different methods are comparable, particularly for lime 2. The difference in the K′ values for lime 1 could be due to the drop in reaction rate at higher HCl concentrations. Most previous studies (2, 8-11) that investigated the HCl reaction with Ca(OH)2 and CaO under different conditions concluded that the chemical reaction rate is first order with respect to HCl concentration. Our results support this conclusion. However, exceptions were found in Chisholm’ work (4) who suggested that the reaction is not first order at high HCl loading due to clogged pores, which was observed here as well in the case of lime 1, and Carminati (14) who reported a complex reaction order. In general, the kinetics of the lime-HCl reaction is influenced significantly by the studied parameters such as temperature, HCl concentration, and particle size of limes. For conditions of optimal flow rate where mass transfer effect is minimal, and in the absence of internal diffusion the chemical reaction seems to be first order with respect to HCl concentration. In this work, the synthesized flue gas is relatively simple as a mixture of HCl, O2, and N2. To closer simulated flue gases to the actual conditions, the investigation of the effect of moisture, SO2 and CO2, including identifying the predominant reaction product under real flue gas conditions is currently being carried out.

Literature Cited (1) Jozewicz, W.; Gullett, B. K. Ind. Eng. Chem. Res. 1995, 34, 607612. (2) Daoudi, M.; Walters, J. K. Chem. Eng. J. 1991, 47, 1-9. (3) Daoudi, M.; Walters J. K. Chem. Eng. J. 1991, 47, 11-16. (4) Chisholm, P. N.; Rochelle, G. T. Ind. Eng. Chem. Res. 1999, 38(10), 4068-4080. (5) Duo, W.; Kirkby, N. F.; Sevill, J. P. K.; Clift, R. Chem. Eng. Sci. 1995, 50(13), 2017-2027. (6) Weinell, C. E.; Jensen, P. I.; Dam-Johansen, K.; Livbjerg, H. Ind. Eng. Chem. Res. 1992, 31, 164-171. (7) Duo, W.; Sevill, J. P. K.; Kirkby, N. F.; Clift, R. Chem. Eng. Sci. 1994, 49, 4429-4442. (8) Fonseca, A. M.; O Ä rfa˜o J. J.; Salcedo, R. L. Ind. Eng. Chem. Res. 1998, 37, 4570-4576. (9) Li, M.; Shaw, H.; Yang, C. Ind. Eng. Chem. Res. 2000, 39, 18981902. (10) Karlsson, H. T.; Klingspor, J.; Bjerle, I. J. Air Pollut. Control Assoc. 1981, 31, 1177-1180. (11) Gullett, B. K.; Jozewicz, W.; Stefanski, L. A. Ind. Eng. Chem. Res. 1992, 31, 2437-2446. (12) Jozewicz, W.; Chang J. C. S. Environ. Prog. 1990, 9(3), 137-142. (13) Mura, G.; Lallai, A. Chem. Eng. Sci. 1992, 47(9-11), 2407-2411. (14) Carminati, A.; Lancia, A.; Pellegrini, D.; Volpiccelli, G. Proc. 7th World Clean Air Congr. 1986, 426. (15) Fonseca, A. M.; O Ä rfa˜o, J. J.; Salcedo, R. L. Ind. Eng. Chem. Res. 2001, 40, 304-313. (16) Allal, K. M.; Dolignier, J. C.; Martin, G. Revue l’Institut Franc¸ ais Pe´trole 1998, 53(6), 871-880. (17) Allal, K. M.; Dolignier, J. C.; Martin, G. Revue l’Institut Franc¸ ais Pe´trole 1997, 52(3), 361-368. (18) Simons, G. A. Resour. Conserv. Recycl. 1992, 7, 161.

Received for review August 26, 2002. Revised manuscript received March 19, 2003. Accepted March 19, 2003. ES020902V