Characterization of a Process for the In-Furnace Reduction of NO x

of calcium magnesium acetate with a suite of five other carboxylic salts (calcium ... acetate, calcium propionate, calcium acetate, calcium benzoate, ...
0 downloads 0 Views 366KB Size
Download PDF
4484

Ind. Eng. Chem. Res. 2005, 44, 4484-4494

Characterization of a Process for the In-Furnace Reduction of NOx, SO2, and HCl by Carboxylic Salts of Calcium W. Nimmo,* A. A. Patsias, W. J. Hall, and P. T. Williams Energy and Resources Research Institute, Houldsworth Building, The University of Leeds, Leeds LS2 9JT, United Kingdom

Calcium magnesium acetate has been assessed as an agent for the reduction of NOx, SO2, and HCl, at the pilot scale, in a down-fired combustor operating at 80 kWth. In addition to this, the chemical and physical processes that occur during heating have been investigated. Benchmarking of calcium magnesium acetate with a suite of five other carboxylic salts (calcium magnesium acetate, calcium propionate, calcium acetate, calcium benzoate, magnesium acetate, and calcium formate) has been performed. NOx reduction involves the volatile organic content of the carboxylic salt being released at temperatures of >1000 °C, where the reaction of CHi radicals with NO under fuel-rich conditions can result in some of the NO forming N2 in a “reburning” process. Thermogravimetry-Fourier transform infrared (TG-FTIR) studies identified the nature of the decomposition products from the low- and high-temperature decompositions. In addition, the rate of weight losses were studied to investigate the influence of the organic decomposition on NOx reduction by reburning. In-furnace reductions of SO2 and HCl are aided by the highly porous, particulate residue, which results from the in situ drying, pyrolysis, and calcination processes. Simultaneous reduction of all three pollutants was obtained, and a synergy between SO2 and HCl capture was identified. A mechanism for this inter-relationship has been proposed. Sorbent particle characterization has been performed by collecting the calcined powder from a spray pyrolysis reactor and compared with those produced from a suite of pure carboxylic salts. Physical properties (including porosity, surface area, and decomposition behavior) have been discussed, relative to reductions in NOx and acid gas emissions. 1. Introduction The injection of calcium magnesium acetate (CMA) has shown potential for the in-furnace reduction of NOx and acid gases (SO2 and HCl) emissions from combustion systems. Small-scale drop-tube experiments have shown1-3 that NOx reduction by CMA is dependent on combustion stoichiometric conditions and temperature. Pilot-scale tests4,5 with CMA injection in a “reburning” configuration showed that, for an equivalent heat input, CMA performed to a similar level of NOx reduction as coal for reburning experiments in the same furnace. Reburning is a form of fuel staging where the emission of NOx is reduced in a secondary fuel combustion zone under fuel-rich conditions. The organic acetate content of the CMA has been shown, in these recent studies, to behave as a conventional fuel. Other studies6-14 have demonstrated the ability of carboxylic salts of calcium to remove SO2 and H2S from combustion or gasification product gases at greater efficiency than limestone or dolomite sorbents. Other alternative fuels, such as biomass, have been demonstrated to perform well as reburn fuels for NOx control.15 The control of the emission of HCl is also a problem for operators of coal combustion plant and incinerators16 burning certain types of plastics. In laboratory-scale coal combustion tests,17 HCl has been shown to increase NOx formation under fuel-rich conditions and may have ramifications under reburning conditions. Therefore, the influence of HCl removal could also have secondary benefits for NOx reduction. In incineration plants, the principal source * To whom correspondence should be addressed. Tel: 0113 343 2513. E-mail: [email protected].

of HCl is chlorinated plastics and is generally formed by decomposition at temperatures of 1100 °C. Temperature profiles are shown in Figure 2 for tests with all the salts at Ca/S or Mg/S ratios of 0.5. Flue gas analysis was performed by drawing sample gas through appropriate sample conditioning lines to on-line gas analysis systems for O2 (Servomex, paramagnetic), NOx (Rotork, Chemiluminescence) and SO2, CO2, and CO (ADC, NDIR). SO2 sampling was performed through heated sample lines (180 °C), coalescing

filters, and driers, to avoid SO2 losses. Samples for HCl analyses were obtained by drawing gas through a heated line (180 °C) into bubbler-type absorbers for offline analysis by ion chromatography (Dionex model DX100). Some interference from HCl on the SO2 analyzer (ADC, NDIR) reading was identified and was determined to be approximately linear, with respect to HCl concentration, and corrections to the SO2 data were applied where appropriate. For example, an HCL concentration of 1500 ppm was observed to give an increased SO2 reading of ∼40 ppm. Thermocouples are installed throughout the length of the furnace to provide temperature profiling during experimental runs. Data from the analyzers and thermocouples were collected by a data-logging system and stored on a personal computer (PC) for post-run processing and analysis. Each sorbent used was dissolved in deionized water, to achieve similar flow rates of ∼130 mL/min at an equivalent of a Ca/S ratio of 2.0. CMA was supplied by Cryotech, USA; calcium propionate (CP, 95%) was supplied by Aldrich Scientific; calcium acetate (CA, 93%) and magnesium acetate (MA, 98%) were provided by Alfa Aesar; calcium benzoate (CB, supersaturated solution) was supplied by Rose Chemicals; and calcium formate (CF, 99%) was provided by Fisher. Initial preparation of the CB solution involved drying the supplied solution in an oven at 110 °C, so that accurate solution concentrations could be composed. The solution concentrations (w/w) were as follows: 25% CMA, 15% CP, 10% CA, 2.6% CB, 15% MA, and 10% CF. The primary zone air-to-fuel equivalence ratio was set to λ1 ) 1.05 (1% O2). Baseline conditions were established with pure water injection, so that initial NOx and SO2 emission levels could be obtained. The reburn zone stoichiometry (λ2) was varied between 1.03 and 0.86. The overall stoichiometry (λ3) was 1.15. The sorbent was then injected and NOx and SO2 reductions were noted at Ca/S levels of 0.5, 1.0, 2.0, and 3.0. When stable readings were obtained, the reacted powder samples

4486

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005

Figure 3. Schematic diagram of the calcination reactor.

were then taken at each Ca/S ratio for imaging and elemental mapping (by energy-dispersive X-ray diffraction (EDX)), using a LEO field-emission gun (FEG) and Camscan 4 scanning electron microscopy (SEM) microscopes, operating in EDX mode. 2.2. Calcination Reactor. Calcined powders were obtained by wet-spraying the solutions of the carboxylic salts through an electrically heated, horizontal furnace (Figure 3). This furnace (2.4 m in length, 70 mm inner diameter (ID)) has been used recently to successfully produce fine ceramic powders by the spray pyrolysis of organic precursor solutions of lead acetate, zirconium acetate, and titanium isopropoxide20,21 and consists of six heated sections that were independently controllable to give a total possible heat input of 7 kW and a calcination temperature of 850 °C. The heating rate of particles in the furnace was ∼400 °C/s and the residence time at 850 °C was 0.5 s. The total time in the reactor was in the region of 2.5 s. Particles were collected after quenching by drawing the gases through ceramic honeycomb filters. A single hole, twin-fluid atomizer was used to generate a spray of solution that was swept through the reactor by a clean air flow of 35 L/min. A water-cooled collar around the injector prevented any conduction of heat from the furnace to the salt solution and any possibility of nozzle blockage due to recrystallization. The nozzle has been previously characterized for droplet size distribution.22 2.3. Thermogravimetric Analysis. Thermogravimetric analysis (TGA) of the carboxylic salt sampless CMA, CP, CB, CA, MA, and CFswas performed using a Stanton and Redcroft TGA system with a Nicolet 560 FT-IR spectrometer. The instruments were linked by a Nicolet interface and a heated line. The samples (3040 mg) were heated to 950 °C (in N2) at a rate of 25 °C/min with an FTIR scan time of 60 s, and the identification of the evolved gases was aided by Nicolet and Aldrich libraries of matching spectra. For part of the carboxylic salt decomposition study, a Shimadzu model TGA-50H instrument was used. The samples (10 mg) were heated, under N2, to 105 °C (15 min hold) and then heated to 950 °C (15 min hold); air then was introduced to combust any residual carbon. 3. Results The results presented in this paper are split into three main sections: (1) NOx reduction data from the 80-kW furnace by carboxylic salts and TGA and TG-FTIR analysis of decomposition behavior, because the volatile organic content of the sorbent is active during NOx reburning. (2) SO2 reduction data from the 80-kW furnace with an analysis of the sorbent powder from calcination tests performed in a horizontal furnace. (3) Simultaneous SO2 and HCl capture results from the 80-kW furnace with evidence for a significant effect of SO2 on HCl capture, supported by SEM/EDX.

Figure 4. Effect of feed rate on NO reduction for (2) CMA, (1) CP, (f) CB, ([) CA, (b) MA, and (9) CF, under reburning conditions of λ1 ) 1.05, λ2 ) 1.03-0.86, and λ3 ) 1.15 (3% O2). Table 1. Relationship between Sorbent Feed Rate, as Expressed as Ca/S or Mg/S, to Reburn Zone Stoichiometry (λ2)a Sorbent Feed Rate, as Ca/S or Mg/S λ2

CB

CMA

CP

CA

MA

CF

0.99 (fuel-lean) 1.0 (stoichiometric) 1.02 (fuel-rich)

0.53 0.42 0.2

0.7 0.56 0.35

1.14 0.91 0.45

1.77 1.4 0.68

1.68 1.4 0.89

4.0 3.15 1.57

a Legend for table is as follows: CB ) calcium benzoate, CMA ) calcium magnesium acetate, CP ) calcium propionate, CA ) calcium acetate, MA ) magnesium acetate, and CF ) calcium formate.

3.1. NOx Reduction (Reburning) by Carboxylic Salts. The effect of carboxylic salt feed rate on NOx reduction is shown in Figure 4 for CMA, CP, CB, CA, MA, and CF. The range of feed rates represented by Ca/S or Ca/Mg ratios were set for fixed initial SO2 concentrations of 1000 ppmv in all cases, and the variation in NO reduction efficiency is due directly to the different levels of combustibles input and the subsequent effect on reburn zone stoichiometry (λ2). The primary zone stoichiometry was fixed at λ1 ) 1.05, and the reburn zone stoichiometry was varied in the λ2 range of 1.03-0.86. The overall stoichiometry (λ3) was 1.15. Table 1 shows the relationship of the Ca/S or Mg/S ratio to the reburn zone stoichiometry (λ2) and indicates that, for a given λ2 value, the extreme cases are those for CB and CF, the Ca/S ratio of CF is eight times greater than that of CB, for each λ2. However, the rate of CB input was limited by its poor solubility and the required solution feed rate to achieve the required Ca/S ratio; consequently, data were only obtained up to a level of Ca/S ) 0.5. It may be expected that, for a given Ca/S ratio, the carboxylic salt with the largest organic fraction would show greater NOx reducing behavior. At the lower feed rate of Ca/S ) 0.5, the predicted order (see Table 1) of CB > CMA > CP > CA ) MA > CF, for NO reduction on this basis compares favorably with the observed NO reductions for experiments in the 80-kW furnace (see Figure 4). However, at higher feed rates (Ca/S ) 2.5), CA outperforms MA by a small but significant degree and CP has now improved almost to the level of CMA. These improvements are related to the changing stoichiometry in the reburn zone by increasing the feed

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005 4487

Figure 5. Decomposition profiles from TGA of CMA, CP, CB, CA, MA, and CF. (See figure for legend of line symbols.)

rates. At low feed rates (Ca/S ) 1), the reburn zone is already operating fuel-rich for CB and CMA. The reburn zone for CP becomes fuel-rich at Ca/S in the region of 1 and coincides with marked improvement in performance from that point forward (see Figure 4). For the extreme case of CF, which shows only slight reductions, the reburn zone is always fuel-lean for the range of feed rates presented, up to Ca/S ) 2.5. The organic fraction of the carboxylic salt is behaving, in many respects, similar to a conventional fuel; however, the performance of each is likely to be dependent, to a large extent, on the ability of the salt to generate CHi radicals, which may be related to the individual structure and decomposition behavior. So that the release of organic fractions from the salts may be compared under more-controlled conditions, a TGA study was performed where an attempt was made to correlate the rates of decomposition with NOx reduction data obtained from the reburn furnace. Approximately 10 mg of each carboxylic salt was heated under N2 to 950 °C. Air was then admitted after 15 min at this temperature, so that any residual combustibles in the powders could be detected by further weight loss that was due to char burnout. Typical weight loss versus temperature data are presented in Figure 5. For the majority of the sorbents tested, three main steps associated with the thermal decomposition were observed: (1) The first occurs at temperatures in the range of 100-200 °C and is associated with the loss of water, either as absorbed moisture or incorporated as a hydrate in the crystal structure. (2) The second occurs in the range of 300-600 °C and relates to the decomposition of organic material. (3) The third occurs in the range of 700-825 °C and is associated with calcination and the loss of CO2.

The samples were tested for losses in air at 950 °C; however, only CB showed significant weight loss, probably because of carbon combustion. The temperatures at the peak rate of weight losses for all the salts are shown in Table 2, along with the compounds identified by FTIR for steps 2 and 3 above. The decomposition products that were released under an inert N2 atmosphere can be linked to their parent carboxylic salt structure. The acetate saltssCMA, MA, and CAsall include 2-propanone (acetone) as a product, at 399, 376, and 444 °C, respectively. CA and CMA include 2butanone, but not MA, which exhibits, essentially, a single-step decomposition23 and includes carbon dioxide (CO2), in addition to acetone. A second, but very small, weight loss was observed at 670 °C; however, no firm identification of any product could be made. CP produced 3-pentanone and 3-methyl-2-butanone at a temperature of 491 °C. CF produced several small-chained molecules at 501 °Cscarbon monoxide (CO), formaldehyde, methanol (CH3OH), and CO2sas would be expected. On the other hand, CB produced benzene and benzophenone as the primary decomposition products at 557 °C. MA exhibited the lowest decomposition temperature (376 °C), whereas the influence of the more strongly bound aromatic group of CB is reflected in the higher decomposition temperature of 557 °C. The principal solid products from the primary decompositions are carbonates of calcium and magnesium. A kinetic analysis of the low-temperature decompositions of the salts in the TGA has been performed by extracting Arrhenius-type parameters from the rates of weight loss and temperature, according to the following method. Data for the calculation are taken from the rising section of the differential thermogravimetry (DTG) plot of the low-temperature decompositions. The rate of mass loss is expressed in terms of the rate constant and the mass:

-

( )

-Ea n dm ) kmn ) A exp m dt RT

(1)

The fraction lost (R) from the TGA data, is given as

R)

mi - mt mi - mf

(2)

and the rate equation of the weight changes described by the conversion factor is given by

( )

-Ea dR ) A exp (1 - R)n dt RT

(3)

dR ) k(1 - R)n dt

(4)

The Arrhenius rate constant (k) is conventionally expressed as

( )

k ) A exp

-Ea RT

(5)

or

ln k ) ln A -

Ea RT

(6)

By the application of the Friedman method,24 the reaction order (n) was determined to be 1 for the

4488

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005

Table 2. Decomposition Products and Temperatures from Thermogravimetric Analysis (TGA) of Calcium Carboxylic Salts Peak 2 magnesium acetate, MA calcium acetate, CA calcium magnesium acetate, CMA calcium propionate, CP calcium formate, CF calcium benzoate, CB

Peak 3

compound(s)

temp (°C)

compound(s)

temp (°C)

2-propanone (acetone), carbon dioxide 2-propanone, 2-butanone 2-propanone, 2-butanone, carbon dioxide 3-pentanone, 3-methyl-2-butanone carbon monoxide, formaldehyde, methanol, carbon dioxide carbon dioxide, benzene, benzophenone

376 444 399/440 390/491 501

negligible loss carbon dioxide, carbon monoxide carbon monoxide, carbon dioxide carbon dioxide carbon dioxide, carbon monoxide

670 779 783 754 802

557

carbon dioxide, carbon monoxide

745

Table 3. Organics Decomposition Data (Peak 1) from Thermogravimetric Analysis of Carboxylic Salts

a

salt

initial weight (mg)a

activation energy, Ea (kJ/mol)

pre-exponential factor, A (s-1)

Arrhenius plot correlation coefficient (n ) 1)

CMA CMA CMA CA MA CP (1st) CP (2nd) CF CB

5 10 15 10 10 10 10 10 10

167.3 153.4 162.9 321 197.6 147.6 247 370.1 358.2

1.22 × 1012 6.66 × 1010 3.91 × 1011 7.56 × 1022 1.31 × 1015 2.01 × 109 8.7 × 1015 1.3 × 1024 6.35 × 1021

0.995 0.991 0.973 0.992 0.991 0.969 0.982 0.989 0.994

With a standard deviation of (0.3 mg.

Figure 6. Arrhenius plot of organic decompositions from TG analysis of weight loss data for CMA, CA, MA, CF, CB, and CP.

decompositions of all the salts and the rate of decomposition was dependent only on the rate constant and temperature. The activation energy can be obtained from a plot of ln k vs 1/T, where the slope is equal to -Ea/R. These data are shown in Table 3 for all the carboxylic salts, along with the correlation coefficients for Arrhenius fits at n ) 1. The Arrhenius plots of the organics decompositions are shown in Figure 6 for all the salts. Repeat runs of the CMA decomposition with initial weights of 5, 10, and 15 mg give similar activation energies (167.3, 153.4, and 162.9 kJ/mol, respectively). Thus, the gradients of the Arrhenius plots are similar and separated by slight differences in the pre-exponential factor (A) for the three CMA samples tested. There seems to be a second peak in the CMA differential of the thermogram (dTG), which is indicated by a shoulder on the downside of the curve (see Figure 5B). The double peak will be a composite of the weight loss from the MA and the CA, which occur at slightly different temperatures (see Table 2). The second peak at 440 °C seems to coincide with the decomposition of CA and the first peak may be the MA or possibly the double salt form of CMA.

The temperature of the first CMA decomposition at 399 °C is slightly higher than that which is expected for pure MA, at ∼376 °C, but is too low for CA and may be indicative of the decomposition of a CMA double salt. However, the second peak is likely to be CA, because of the coincidence of decomposition temperatures. In addition, the activation energy for the main CMA peak (153.4 kJ/mol) is more similar to that of MA (198 kJ/ mol) than that of CA (321 kJ/mol). Therefore, the rate constant plotted for CMA decomposition (Figure 6) is likely to be a combination of MA and the CMA double salt. The rate of the second decomposition has not been determined. CP shows a two-stage decomposition with activation energies of 147.6 and 247 kJ/mol. The first stage is minor and consists of a weight loss of ∼10% of the organic content; however, it may be important, because of the relatively low activation energy. The second stage is the more relevant for reburn consideration, because of the volume of material involved. The salts with the highest activation energiessCF (370 kJ/mol) and CB (358 kJ/mol)sshow the highest decomposition temperatures, but for different reasons. The formate consists of one carboxylic group with no aliphatic chain attached, where bonding is weaker. The benzoate, on the other hand contains an aromatic ring structure, which requires greater energy to dissociate than an aliphatic bonding such as that in an acetate or propionate. The higher temperature decompositions occurred at 783, 754, 745, 779, 670, and 802 °C for CMA, CP, CB CA, MA (negligible loss), and CF, respectively, and represents the loss of mainly CO2 and CO. This represents the calcination step where a proportion of the carbonates are converted to oxides to a degree determined by the final temperature, gas composition, and residence time, particularly during an injection experiment in the 80-kW furnace where the temperatures experienced by the powder may be as high as 1150 °C, with residence times on the order of seconds. 3.2. SO2 Reduction by Carboxylic Salts. Results for SO2 reduction in the 80-kW furnace are shown in Figure 7 for all six carboxylic salts studied. Initial levels of SO2 were set at 1000 ppm (3% O2). These results were

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005 4489

Figure 7. Effect of sorbent feed rate on SO2 reduction for (2) CMA, (1) CP, (f) CB, ([) CA, (b) MA, and (9) CF, under reburning conditions of λ1 ) 1.05, λ2 ) 1.03-0.86, and λ3 ) 1.15 (3% O2).

Figure 8. Sorbent Ca and Mg utilizations for SO2 capture under reburning conditions (λ1 ) 1.05, λ2 ) 1.03-0.86, and λ3 ) 1.15 (3% O2)): (2) CMA, (1) CP, (f) CB, ([) CA, (b) MA, and (9) CF.

obtained under the reburning conditions that were prevailing for the NO reduction tests described previously. In all cases, the reductions in SO2 show an upward trend as the Ca/S ratio increases, because of the increases in the available material for absorption. However, significant differences can be observed between the performances of the sorbents studied. MA was the poorest performing sorbent, achieving 1.1% reduction at an Mg/S ratio of 1.0, followed by CA (19%), CF (23%), CMA in propane (32%), and CP (39%) at Ca/S ) 1.0. CMA and CP performed better than the other salts, showing the highest SO2 reduction throughout the range of sorbent feed rates. Because of the low solubility of CB, it was impractical to inject the large quantities of solution required to obtain Ca/S ratios of >0.5. Reductions for CB were comparable to the other calcium salts at the lower level of feed rate. Calcium utilizations are shown in Figure 8 and reflect the reductions that have been observed. Utilizations are generally higher at excess sulfur conditions, that is, at Ca/S < 1. As expected, the magnesium utilization is 2. This is indicative of the increasing availability of surface sites for chlorine absorption. An example of the EDX spectrum of samples of spent, CMA sorbent powder is shown in Figure 11 for Ca/S ) 1.85. Sample A has no added SO2, and sample B has an initial SO2 concentration of 780 ppm. The main feature to note is the much-reduced intensity of the Cl peak in spectrum B with sulfur absorption. Mercury pore size analysis of the powders produced in the horizontal furnace at 850 °C is shown in Table 5 and Figure 12. Comparison of the pore volume data measured by N2 physisorption and mercury intrusion porosimetry do not match, in most cases. This is to be expected, however, because, in most cases of comparison between the two methods, only moderate concurrence is achieved.25 However, other workers in the field have presented data from mercury pore size analysis on the materials; therefore, we have included our analysis data

Figure 11. EDX spectrum of samples of spent, CMA sorbent powder for Ca/S ) 1.85: (A) sample A, 0 ppm SO2; and (B) sample B, 780 ppm SO2.

for comparison. The range of pore diameters measured by the two techniques is different, and, therefore, there are bound to be differences between the cumulative pore areas and cumulative pore volumes measured by the two different techniques. The number of points measured by N2 physisorption was very small, compared to the number of points measured by mercury intrusion porosimetry, and, therefore, the N2 physisorption measurements may not be as accurate as the mercury porosimetry measurements. The range of pore diameters measured for this work was 0.00312-0.05688 µm for N2 physisorption and 0.0036-20.4 µm for mercury intrusion porosimetry. To compare the data directly, total pore volumes were measured over the same range of pore diameters (0.0036-0.05688) and, on average, the mercury data were 40% greater than the N2 data. There is the possibility that some of this discrepancy may be due to the physical effects of mercury on the fragile

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005 4491 Table 6. Correlation of NOx Reduction with Decomposition Kinetics Obtained from the Thermogravimetric Analysis (TGA) of Organic Decompositions for CP, CMA, MA, CA, and CB

CP CMA MA CA CB

Ea, from TGA (kJ/mol)

NO reduction in the 80-kW furnace (%)

Ca/S or Mg/S

reburn fuel fraction in the 80-kW furnace (%)

148 153 198 321 358

58 52 55 45 49

1.25 2 3 3 0.8

8 8 8 8 8

Hydrogen cyanide (HCN) that is produced in reaction 7 is converted to N2 in the flame via the following reactions:27

Figure 12. Mercury pore size analysis data for CMA, CP, CA, and CF. Table 5. Surface Area, Porosity, and Calcination Data of Samples from a Calcination Reactor and an 80-kWth Furnace BET surface area (m2/g) N2 isosorption total pore area (m2/g) total pore volume (mL/g) average pore diameter (µm) mercury porosimetry total pore area (m2/g) total pore volume (mL/g) average pore diameter (µm) porosity (%) percentage calcination (%)

CMA

CA

CF

CP

31.8

31.0

23.6

22.7

CMAa 10.1

48.2 47.7 31.4 32.4 0.24 0.21 0.14 0.16 0.019 0.018 0.017 0.020 50.04 2.84 0.23 82.6 85.2

48.19 2.32 0.19 79.3 80

55.61 2.15 0.15 73.7 71.3

55.47 1.69 0.12 74 84.1 81.8

a Sample taken from an 80-kW furnace under combustion conditions (propane at 6% O2).

particles that are being tested, resulting in structural damage. The pore merging, which would result from this effect, could give falsely high measurements of pore volume and diameters. Porosity was typically in the range of 0.7-0.8 for the carboxylic salts studied. 4. Discussion and Analysis 4.1. NOx Reduction. Although CMA has the ability to reduce NOx and acid gases (SO2 and HCl) simultaneously, the reaction processes are very different. The former is a gas-phase process whereby hydrocarbon radicals react with NOx to produce CHN or NHi intermediates that may oxidize back to NOx or form N2, depending on the availability of O2. These basic processes are collectively known as “NOx reburning”. On the other hand, the removal of SO2 and HCl are heterogeneous processes whereby the solid particulate residues from CMA decomposition and calcinations principally a mixture of calcium and magnesium oxides and carbonatesschemically react with the gaseous species at the solid surfaces. For each carboxylic salt, the general principle of reburning applies, which is to destroy NOx already formed in the primary combustion zone (see Figure 1) via the following reactions:26

CH + NO f HCN + O

(7)

NO + NH2 f N2 + H2O

(8)

O + HCN T NCO + H

(9)

NCO + H T NH + CO

(10)

NH + H T N + H2

(11)

N + NO T N2 + O

(12)

The N2 forming reaction sequence described previously is encouraged under the fuel-rich conditions of the reburning zone. Under more fuel-lean conditions (i.e., with greater availability of OH and O radicals), HCN and NH may be oxidized,

O + HCN T NO + CH

(13)

OH + NH + 0.5O2 T NO + H2O

(14)

thereby reducing the efficiency of the reburning process. In addition, under these conditions, the pool of active CHi radicals is lower in the reburn zone, because of the greater consumption of hydrocarbons by combustion. Although the Ca/S or Mg/S ratios are fixed for each salt, the amount of organic available for the reburning process varies, depending on the structure. The feed rate of the carboxylic salt can also be presented in terms of the fuel equivalent of reburn fuel fraction (Rff). NOx reductions at 8% Rff (λ2 ) 0.97) are shown in Table 6, along with the Ca/S or Mg/S ratios and the activation energies derived from the TGA study. There is a correlation between the effectiveness of the carboxylic salt and the activation energy, which may explain the poor performance of CA, compared with the other acetates and the propionate. CF is not included in the table, because the low organics content dictates that only Rff levels up to ∼3% were practical at Ca/S ) 2.5. The data for CA were determined by extrapolation to Ca/S ) 3 of the available data, up to Ca/S ) 2.5. (A similar relationship is observed for CB.) The trend is toward slightly lower NO reductions at higher activation energies (see Table 6), which is due to the rapid heating that occurs in the furnace at temperatures of >1000 °C. However, there is a finite residence time in the reburn zone of ∼1 s; the rate at which organic decomposition can proceed will determine the effective residence time for reburn reaction and NO reduction. Slower decompositions will leave less time for reaction before entering the burn-out zone. 4.2. Sorbent Particle Formation and SO2/HCl Reduction. The superior performance of CMA over limestone or dolomite as a sorbent for acid gases is due, in part, to the propensity of the material to undergo structural change during heating, whereby a pore

4492

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005

Table 7. Particle Size Calculation Data Estimated Particle Sizes after In-Furnace Spray Calcination

particle size

typical initial droplet diameter in size range (µm)

calculated particle size (µm)

maximum droplet diameter minimum droplet diameter

50 1

38 0.8

Physical Parameters initial droplet concentration saturation droplet concentration density of salt density of oxide oxide yield of salt porosity of particle

1.6 mol/L 2 mol/L 1.2 g/cm3 3.3 g/cm3 0.3 0.8

structure develops with the formation of large surface holes, which can be several micrometers in diameter. The open pore structure is associated with the drying and decomposition processes that occur when they are heated in the furnace. The process starts with the atomization of the CMA solution to give a mist of droplets with a maximum size of ∼50 µm, which enter the hot gases of the furnace at temperatures of >1100 °C. The solution is comprised of a mixture of mainly CA and MA in a ratio of 3:7. Rapid droplet drying occurs and, as the salt concentration increases, the salts will begin to precipitate, according to the prevailing saturation concentration (Cs) and initial concentration (C0) values for each component. It is likely that significant dissociation within the CMA solution will occur during droplet heating as it enters the hot gases of the 80-kWth furnace, and that, on recrystallization, separation of magnesium and calcium may occur.28 The Cs value for MA is approximately half that for CA;29 therefore, the onset of MA precipitation will precede the onset of CA precipitation. This may account for some variability in Ca/Mg ratios observed by spot SEM/EDX analysis of the reacted sorbent particles where magnesium-rich and calcium-rich areas were identified. However, the droplet heating rates are rapid in the furnace, and significant coprecipitation of calcium and magnesium acetate salts are likely. The process of carboxylic salt decomposition will begin immediately after drying with the evolution of ketones (see Table 2), followed by carbonate formation and then decomposition when the temperature reaches ∼750 °C, to form nascent calcium oxide. Under oxidizing furnace conditions, the organic vapors will ignite as they evolve and form a flame envelope around the particles. The particle size after droplet heating can be estimated by applying the following expression,30 which is normally applied to dense particles:

Dp ) D0

()( C0 Cs

1/3

FsaltW

)

Foxide(1 - P)

1/3

(15)

The equation allows for the formation of large internal pores in the fractional porosity term P. Porosity data from samples of carboxylic salts produced in the horizontal reactor are shown in Table 5 and show that the porosity levels range from 0.7 to ∼0.8 (CMA) and can be used as a guide for porosity input to the model. Application of eq 15 to the spray calcination of CMA from solution gave a reasonable prediction of the range of particle size that is observed via electron microscopy. Typical results are shown in Table 7, which are calcu-

Figure 13. Scanning electron microscopy (SEM) images of reacted sorbent particles after sulfation/reburn: (A) CF, (B) CP, (C) CB, (D) CMA (inset shows a broken particle), (E) MA, and (F) CA.

lated from the estimated maximum and minimum droplet sizes generated from the type of twin-fluid atomizer used in the study. A 20% reduction in size is predicted, with a final porosity of 0.8. If the particles were completely solid, then a reduction of 60% in particle size would be expected, which represents the theoretical maximum for the given initial parameters of concentration and droplet size that have been given in Table 7. SEM images that show the final structure of the sorbent particles can be observed in Figure 13 for the carboxylic salts studied. Most of the sorbents exhibited characteristically open structures with large pores or blowholes, several micrometers in width, extending from the interior of the particle to the external surface. The inset in Figure 13D shows the internal structure of a reacted CMA particle. Typical internal voids are apparent with thin porous walls, with evidence for blowholes to the exterior of the particle. It is this open access structure that gives CMA a superior ability to absorb SO2 and HCl over sorbents such as limestone and dolomite. However, the sorbent from MA formed elongated particles that are devoid of a defined spherical form, with a frondlike structure. In addition, the analysis of powders from the calcination reactor (850 °C) showed surface areas on the order of 30 m2/g for CMA and CA calcination, with lower values for CF and CP (see Table 5). Under the more-severe conditions of the down-fired furnace, the surface area of calcined CMA is 800 °C). On the other hand, the reactions of Ca with SO2 and HCl have equilibrium constants of >1000 across the range of temperatures studied31 and are dominant. However, the role of magnesium in the acid gas removal process by CMA may be more subtle,11 in that the formation of CaSO4‚ 3MgSO4 rather than MgSO4 may be relevant at temperatures of 1100 °C, resulting in a high degree of calcinations. A >80% conversion to the oxide was observed for CMA (see Table 5). To explain the synergistic effect of SO2 on HCl capture (see Figure 9), it is necessary to examine the surface processes that are occurring on the calcined CMA particle. Most sulfation models include a description of the sorbent pore systems as either random or following a defined structure.1 The random model describes a network of interconnecting pores, where there is the theoretical possibility of gas mixing throughout the system, whereas a structured model such as the poretree model describes how larger pores divide or branch into smaller pores. In either model, the volume increase through sulfation and chloridation can cause severe limitation of the degree of reaction and sorbent capacity, because of pore blockage. The cenosphere structure of the calcined CMA particle reduces the influence of pore blockage on capacity, by effectively truncating the length of the pores in the walls of the particle, whether they be external or internal, as the outer surface of a void. The process of gas diffusion to the interior is enhanced by the deeply penetrating blowholes, which are visible on the surface (see Figure 13). The pore and particle structure of sorbents derived from limestones and dolomites does not have these advantages. HCl and SO2 compete for active sites on the CaO particle (see Figure 9) and, as such, must penetrate further into the particles, because the sites near the outer surface and inner shell surface are consumed first. The deeper the pore penetration, the greater the likelihood of pore blockage by truncation, which will have a tendency to inhibit the capture of both chlorine and sulfur. The presence of SO2 drastically reduces the chlorine content of the sorbent (see Figure 10) over the range of feed rates tested, whereas the sulfur content of the sorbent remains relatively constant. This indicates an inhibiting effect on Cl capture due to sulfation. This effect can be observed clearly in the trends of the Ca utilizations for HCl and SO2 (see Table 4). The trend for SO2 in the co-reduction tests showed a declining Ca utilization for S capture, from 34.2% to 23.4%, for increasing Ca/S ratios. The corresponding Cl capture data show an increasing trend in Ca utilization for Cl over the same

range of Ca input. This indicates that, as more calcium becomes available, because of increasing sorbent feed rates, the inhibiting effect of SO2 on HCl becomes less. Molecular volume differences between CaCl2 and CaSO4 on the surface of pores will affect the rate at which the pores become blocked, and, although the Cl- ion (radius of 1.85 Å) is slightly larger than the O- ion, the sulfate group is a much larger molecule, because of the number of atoms. Pore blockage by truncation will be greater by sulfation than by chloridation. 5. Conclusions Simultaneous NOx, SO2, and HCl removal in excess of 60% have been achieved at Ca/S ) 2 by the combination of in-furnace calcium magnesium acetate (CMA) injection and reburning. The volatile organic material from CMA decomposition generates a “NOx reburning” zone (λ2 ) 1.03-0.86) in the furnace when the primary zone stoichiometry is optimized at λ1 ) 1.05. A kinetic investigation via thermogravimetric analysis (TGA) has indicated a link between the rates of organic fraction decomposition on heating and the degree of NOx reduction obtained in the 80-kW furnace. Through the extraction of Arrhenius parameters from the rate plots, it was determined that sorbents with relatively higher activation energies show lower NOx reductions than those exhibiting lower activation energies. This has direct consequences on reburning reactions, in that it is not only the amount of hydrocarbon material present in the reburn zone that is important but also the rate at which it is released. The solid calcium-based residue, which contains mainly CaO with CaCO3, formed after devolatilization and calcination, is then available for reaction with the acid gases, commencing within the reburn zone and continuing into the burn-out zone of the furnace. There is a clear effect of the presence of SO2, which seems to suppress the absorption of HCl by as much as 35%, at Ca/Cl ) 1.5. In other words, the presence of SO2 seemed to reduce the capture efficiency of HCl, because of competition for reactive sites and pore blockage. This effect can be seen clearly in the trends of the Ca utilizations for HCl and SO2. The trend for SO2 in the co-reduction tests showed a declining Ca utilization for S capture from 34.2% to 23.4% for increasing Ca/S ratios. The corresponding Cl capture data shows an increasing trend in Ca utilization for Cl over the same range of Ca input. This indicates that, as more calcium becomes available, because of increasing sorbent feed rates, the inhibiting effect of SO2 on HCl becomes less. Nomenclature λ ) (actual air flow/fuel flow)/(stoichiometric air flow/fuel flow) λ1 ) primary combustion zone stoichiometry λ2 ) reburn zone stoichiometry λ3 ) overall stoichiometry after adding burn-out air mi ) initial mass of material in TGA mf ) final mass of material mt ) mass of material remaining at time t R ) gas constant (kJ mol-1 K-1) Ea ) activation energy (kJ/mol) n ) reaction order for decomposition k ) rate constant for one step decomposition A ) pre-exponential factor Dp ) final particle diameter (µm) Do ) starting droplet diameter (µm)

4494

Ind. Eng. Chem. Res., Vol. 44, No. 12, 2005

Co ) starting droplet concentration Cs ) saturation droplet concentration Foxide ) density of oxide phases Fsalt ) density of salt phases W ) oxide yield of salt(s) P ) fractional porosity of the particle CMA ) calcium magnesium acetate CP ) calcium propionate CB ) calcium benzoate CA ) calcium acetate MA ) magnesium acetate CF ) calcium formate Ca/Cl ) molar ratio of calcium to chlorine Ca/S ) molar ratio of calcium to sulfur Mg/S ) molar ratio of magnesium to sulfur Rff ) reburn fuel fraction

Acknowledgment The authors wish to acknowledge the support of the UK Engineering and Physical Research Council (EPSRC) for a research grant (No. GR/R46120), Dr. A. Cunliffe for help with the TGA, and Mr. E. Woodhouse for technical assistance. Literature Cited (1) Steciak, J. W.; Levendis, Y. A.; Wise, D. L.; Simons, G. A. Dual SO2-NOx Concentration Reductions by Calcium Salts of Carboxylic Acids. J. Environ. Eng. 1995, 121 (8), 596-604. (2) Steciak, J.; Zhu, W.; Levendis, Y. A.; Wise, D. L. The Effectiveness of Calcium (Magnesium) Acetate and Calcium Benzoate as NOx Reducing Agents in Coal Combustion. Combust. Sci. Technol. 1994, 102, 193-211. (3) Steciak, J.; Levendis, Y. A.; Wise, D. L. Effectiveness of Calcium-Magnesium Acetate as a Dual SO2-NOx Emission Control Agent. AICHE J. 1995, 41 (3), 712-722. (4) Nimmo, W.; Patsias, A. A.; Hampartsoumian, E.; Gibbs, B. M.; Williams, P. T. Simultaneous Reduction of NOx and SO2 Emissions from Coal Combustion by Calcium Magnesium Acetate. Fuel 2004, 83 (2), 149-155. (5) Nimmo, W.; Patsias, A. A.; Hampartsoumian, E.; Gibbs, B. M.; Fairweather, M.; Williams, P. T. Calcium Magnesium Acetate and Urea Advanced Reburning for NO Control with Simultaneous SO2 Reduction. Fuel 2004, 83 (9), 1143-1150. (6) Garcia-Labiano, F.; de Diego, L. F.; Adanez, J. Effectiveness of Natural, Commercial, and Modified Calcium-Based Sorbents as H2S Removal Agents at High Temperatures. Environ. Sci. Technol. 1999, 33 (2), 288-293. (7) Adanez, J.; Garcia-Labiano, F.; de Diego, L. F.; Fierro, V. Utilization of Calcium Acetate and Calcium Magnesium Acetate for H2S Removal in Coal Gas Cleaning At High Temperatures. Energy Fuels 1999, 13 (2), 440-448. (8) Nimmo, W.; Agnew, J.; Hampartsoumian, E.; Jones, J. M. H2S Removal Using Spray-pyrolysed Calcium Acetate. Ind. Eng. Chem. Res. 1999, 38, 2954-2962. (9) Sohn, H. Y.; Han, D. H. Calcined Calcium Magnesium Acetate as a Superior SO2 Sorbent: I. Thermal Decomposition. AIChE J. 2002, 48 (12), 2971-2977. (10) Sohn, H. Y.; Han, D. H. Ca-Mg Acetate as Dry SO2 Sorbent: II. Sulfation of CaO in Calcination Product. AIChE J. 2002, 48 (12), 2978-2984. (11) Sohn, H. Y.; Han, D. H. Ca-Mg Acetate as Dry SO2 Sorbent: III. Sulfation of MgO + CaO. AIChE J. 2002, 48 (12), 2985-2991. (12) Levendis, Y. A.; Zhu, W.; Wise, D. L.; Simons, G. A. Effectiveness of Calcium Magnesium Acetate as an SOx Sorbent in Coal Combustion. AICHE J. 1993, 39 (5), 761-773. (13) Levendis, Y. A.; Steciak, J.; Wise, D. L. Reduction of Combustion Generated SO2-NOx by Fine Mists of CMA. In Proceedings of the 19th Coal Utilization and Fuel Systems International Conference, Clearwater: FL, March 22-26, 1994.

(14) Ninomiya, Y.; Zhang, L.; Nagashima, T.; Koketsu, J.; Sato, A. Combustion and De-SOx Behaviour of High-sulfur Coals Added with Calcium Acetate Produced from Biomass Pyroligneous Acid. Fuel 2004, 83 (16), 2123-2131. (15) Maly, M. P.; Zamansky, V. M.; Ho, L.; Payne, R. Alternative Fuel Reburning. Fuel 1999, 78, 327-334. (16) Williams, P. T. Emissions From Solid Waste Management Activities, Report from Environmental and Health Impact of Solid Waste Management. In Issues in Environmental Science and Technology; Hester, R. E., Harrison, R. M., Eds.; The Royal Society of Chemistry: Cambridge, U.K., 2002; p 18. (17) Wei, X.; Han, X.; Uwe, S.; Maier, J.; Worner, H.; Hean, K. R. G. The Effect of HCl and SO2 on NOx Formation in Coal Flames. Energy Fuels 2003, 17, 1392-1398. (18) Shemwell, B.; Atal, A.; Levendis, Y. A.; Simons, G. A. A Laboratory Investigation on Combined In-Furnace Sorbent Injection and Hot Flue-Gas Filtration to Simultaneously Capture SO2, NOx, HCl, and Particulate Emissions. Environ. Sci. Technol. 2000, 34 (22), 4855-4866. (19) Shemwell, B.; Levendis, Y. A.; Simons, G. A. Laboratory Study on the High-Temperature Capture of HCl Gas by Dryinjection of Calcium-based Sorbents. Chemosphere 2001, 42, 785796. (20) Nimmo, W.; Ali, N. J.; Brydson, R.; Calvert, C.; Hampartsoumian, E.; Hind, D.; Milne, S. J. The Formation of PZT powders by Spray Pyrolysis. J. Am. Ceram. Soc. 2003, 86 (9), 1474-1480. (21) Nimmo, W.; Ali, N. J.; Brydson, R.; Calvert, C.; Milne, S. J. Particle Formation During Spray Pyrolysis of Lead Zirconate Titanate. J. Am. Ceram. Soc. 2005, 88 (4), 839. (22) Nimmo, W.; Hind, D.; Ali, N. J.; Hampartsoumian, E.; Milne, S. J. The Production of Ultra-fine Zirconium Oxide Powders by Spray Pyrolysis. J. Mater. Sci. 2002, 37, 3381-3387. (23) Myerson, A. L. The Reduction of Nitric Oxide in Simulated Combustion Effluents by Hydrocarbon-Oxygen Mixtures. In Fifteenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1975; p 1085. (24) Friedman, H. L. Kinetics of thermal Degradation of Charforming Plastics from Thermo-gravimetry-Application to a Phenolic Plastics. J. Polym. Sci., Part C: Polym. Symp. 1965, 6, 183195. (25) Webb, P. A.; Orr, C. Analytical Methods in Fine Particle Technology; Micromeritics Instrument Corporation: Norcross, GA, 1997. (26) Wendt, J. O. L.; Sternling, C. V.; Matovich, M. A. Reduction of Sulphur Trioxide and Nitrogen Oxide by Secondary Fuel Injection. In Fourteenth Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1973; p 881. (27) Thorne, L. R.; Branch, M. C.; Chandler, D. W.; Kee, R. J.; Miller, J. A. Hydrocarbon/Nitric Oxide Interactions in LowPressure Flames. In 21st Symposium (International) on Combustion; The Combustion Institute: Pittsburgh, PA, 1986; pp 965967. (28) Mu, J.; Perlmutter, D. D. Thermal Decomposition of carbonates, carboxylates, oxalates, acetates, formates and hydroxides. Thermochim. Acta 1981, 49, 207-218. (29) Dionysiou, D. D.; Tsianou, M.; Botsaris, G. D. Investigation of the conditions of the Production of Calcium Magnesium Acetate (CMA) Road Deicer in an Extractive Process. Cryst. Res. Technol. 2000, 35 (9), 1035-1049. (30) Zhang, S. C.; Messing, G. L.; Borden, M. Synthesis of Solid, Spherical Zirconia Particles by Spray Pyrolysis. J. Am. Ceram. Soc. 1990, 73 (1), 61-67. (31) Courtemanche, B.; Levendis, Y. A. Control of HCl Emissions from the Combustion of PVC by In-Furnace Injection of Calcium-based Sorbents. Environ. Eng. Sci. 1998, 15 (2), 123135.

Received for review February 15, 2005 Revised manuscript received April 8, 2005 Accepted April 13, 2005 IE0501780