Environ. Sci. Technol. 2005, 39, 5436-5441
Removal of Acetaldehyde Vapor with Impregnated Activated Carbons: Effects of Steric Structure on Impregnant and Acidity T O S H I A K I H A Y A S H I , * ,† MIKIO KUMITA,‡ AND YOSHIO OTANI‡ Research Center, Toyobo Company, Ltd., 1-1, Katata 2-chome, Otsu 520-0292, Japan, and Department of Chemical Engineering, Graduate School of Natural Science and Technology, Kanazawa University, Kakuma-machi, Kanazawa 920-1192, Japan
sulfate (3, 4). As few studies have been conducted on the removal of acetaldehyde by impregnated activated carbons, we do not have enough data to compare the adsorption performance of activated carbons impregnated with various chemicals in selecting proper adsorbents for acetaldehyde. In this work, breakthrough curves of a fixed bed of activated carbons impregnated with various amines were measured to obtain the adsorption capacities for acetaldehyde, and the adsorption mechanisms of impregnated activated carbons were studied.
Experimental Section Preparation of Impregnated Activated Carbon. The properties of activated carbons used as a support of amines are shown in Table 1. Sample A is a coconut-shell activated
TABLE 1. Physical Properties of Activated Carbons The acetaldehyde adsorption capacities of activated carbons impregnated with various amines were experimentally studied by using fixed beds. It was found that the adsorption capacity of impregnated activated carbons is influenced by the steric structure of impregnants as well as their acidity. For activated carbons impregnated with aromatic amines, ortho and meta substituents on the benzene ring hindered the condensation reaction with acetaldehyde. The activated carbon impregnated with aminobenzenesulfonic acids differed from that impregnated with the other amines in the acetaldehyde adsorption mechanism in that a DoebnerMiller reaction was involved. Also, aminobenzenesulfonic acids were not only the reactant but also the acid catalyst in the removal of acetaldehyde. Since paminobenzenesulfonic acid reacts with acetaldehyde without steric hindrance in the Doebner-Miller reaction, it is the most suitable impregnant for the chemisorption of acetaldehyde.
Introduction Acetaldehyde is contained in tobacco smoke, automobile exhaust gas, etc. Since acetaldehyde is odorous at a low concentration of 0.09 mg/m3 and stimulates skin, eyes, nose, and respiratory tract, an effective removal method for acetaldehyde is needed. Activated carbons are commonly used for the removal of gaseous contaminants in air because they adsorb various gases by physisorption. However, the adsorption capacity of activated carbons for acetaldehyde is very small. The use of chemisorption is one of the possible solutions for improving adsorption capacity of activated carbon. The acetaldehyde adsorption capacity of porous materials can be enhanced by the impregnation of chemicals. Because aldehyde groups react with amino groups, amines are possible chemicals for chemisorption of acetaldehyde. There have been a number of studies on the chemisorption of acetaldehyde, e.g., silica gel, activated alumina, activated clay and activated carbon impregnated with hydrazinium aluminum sulfate (1), sepiolite impregnated with 2-aminobenzoic acid (2), and sepiolite impregnated with hydrazinium aluminum * Corresponding author phone: +81-77-571-0074; fax: +81-77571-0077; e-mail:
[email protected]. † Toyobo Co., Ltd. ‡ Kanazawa University. 5436
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
particle size (mm) pH BET surface area (m2/g) total pore volume (cm3/g)
sample A
sample B
sample C
0.55-2.38 7.2 1120 0.537
0.21-0.60 5.8 1350 0.664
0.60-2.00 10.0 1940 0.928
carbon (Nacalai Tesque, Inc.); sample B is a coal-based activated carbon (Taihei Chemical Industry Co., Ltd.), and sample C is a large-surface-area activated carbon (Pica SA). Nitrogen adsorption and desorption data of impregnated activated carbon at 77 K were measured using an accelerated surface area and porosimetry (ASAP2010, Micromeritics Instrument Corp.) to obtain BET surface areas, total pore volumes, and pore size distributions. The measurements of pH values of bare activated carbons conformed to the Japanese Industrial Standard (JIS K 1474: Test methods for activated carbon). A flask containing 3 g of bare activated carbon and 100 mL of water was heated to boil for 5 min. Then the flask was cooled to room temperature, and 100 mL of water was added. The pH value of the water in the flask was measured with a pH meter. Impregnation of amines on the activated carbons was carried out by the following procedure. A given amount of activated carbon was weighed and soaked in an aqueous solution or suspension of amines and stirred at room temperature. Then the activated carbon was taken out from the solution or suspension and dried in an oven at 353 K for 3 h. The amount of amine impregnated on the activated carbons was determined by the difference in mass before and after the impregnation. Thus, the determined impregnated mass was in fairly good agreement with the reduction in impregnant concentration in the solution or suspension. The impregnated mass was varied by changing the amine concentration of the solution or soaking time. Measurement of Breakthrough Curves. Figure 1 shows the experimental setup for the measurement of breakthrough curves of packed beds of impregnated activated carbons. The whole setup was placed in a chamber, the temperature and humidity of which were maintained at 298 K and 50% relative humidity, respectively. The impregnated activated carbon particles were packed in a glass column, and the air with a known concentration of acetaldehyde was fed to the fixed bed at a constant volumetric flow rate. The bed height was 4 cm for 3 g of bare activated carbon (sample A). The volumetric flow rate corresponded to the superficial velocity of 50 cm/s. The effluent concentration of acetaldehyde was measured at a regular interval by a gas chromatograph with a flame ionization detector (G2800-F, Yanaco Analytical 10.1021/es048514b CCC: $30.25
2005 American Chemical Society Published on Web 06/09/2005
FIGURE 1. Experimental setup.
FIGURE 3. Breakthrough curves of activated carbon impregnated with p-aminobenzenesulfonic acid at various inlet concentrations.
FIGURE 2. Breakthrough curves of a bare activated carbon bed. FIGURE 4. Relationship between the instantaneous removal efficiency and the average adsorbed mass on bare activated carbon.
TABLE 2. Experimental Conditions sample A packed mass of impregnated activated carbon (g) bed diameter (mm) bed height (cm) flow rate (cm3/s) space velocity (h-1) acetaldehyde concentration (mg/m3) temperature (K) relative humidity (%)
sample B
sample C
3
1
3
14.5
14.5
14.5
3.2-4.0 1.1-1.5 3.8-5.7 83.3 83.3 83.3 45000-57000 120000-160000 31000-48000 10.7-367
12.4-125
34.2-182
298 50
298 50
298 50
Instrument Corp.) equipped with an auto gas sampler (GSL309A, Yanaco Analytical Instrument Corp.). The inlet acetaldehyde concentration was measured several times during the measurement to ensure the constant inlet concentration. The experimental conditions are given in Table 2.
FIGURE 5. Relationship between the instantaneous removal efficiency and the average adsorbed mass on activated carbon impregnated with p-aminobenzenesulfonic acid at various inlet concentrations. efficiency of acetaldehyde (E) is given by the following equation using instantaneous inlet and outlet concentrations Ci and Co, respectively:
Results and Discussion Figures 2 and 3 show the breakthrough curves of bare activated carbon (sample A) and that impregnated with p-aminobenzenesulfonic acid, respectively, at various inlet acetaldehyde concentrations. The breakthrough curve of bare activated carbon is a weak function of the inlet acetaldehyde concentration, but the breakthrough curve of p-aminobenzenesulfonic acid-impregnated activated carbon exhibits a steeper rise as the inlet acetaldehyde concentration increases. Good reproducibility was confirmed in these breakthrough curves for the adsorbents prepared by different batches as well as different samples from the same batch. Figures 4 and 5 show the relationship between instantaneous removal efficiencies of acetaldehyde (E) and the average adsorbed mass of acetaldehyde (qav), which are obtained from Figures 2 and 3. The instantaneous removal
E)1-
Co Ci
(1)
The average adsorbed mass of acetaldehyde (qav) for the time period of t is obtained by
qav )
Ci Q
∫ E dt t
0
(2)
Mm
where Q is the volumetric flow rate of air, M the molecular weight of acetaldehyde, and m the packed mass of impregnated activated carbon. As seen in Figures 4 and 5, the E-qav relationship of bare activated carbon depends on the inlet acetaldehyde concentration, but that of p-aminobenzenesulfonic acid-impregnated activated carbon is independent VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5437
FIGURE 6. Apparent adsorption isotherm of acetaldehyde for bare activated carbon. of inlet acetaldehyde concentration. The E-qav relationship is not a function of inlet adsorbate concentration when an adsorbate is removed by a first-order chemical reaction (see the Appendix). The adsorbed masses of acetaldehyde until the removal efficiency reaches 0.1, which is hereafter termed the adsorption capacity, qc, are obtained from the breakthrough curves. For the impregnated activated carbon, the value of qc was determined by extrapolating the experimental data by a straight line as shown in Figure 5 (the E-qav relationship can be reduced to a linear function; see the Appendix). Figure 6 shows the adsorption capacity of bare activated carbon as a function of inlet concentration. The dependency of adsorption capacity on the inlet concentration is linear; i.e., a Henry-type adsorption isotherm is obtained for the bare activated carbon. This may explain the weak dependence of breakthrough curves on the influent concentration in Figure 2 because both adsorption capacity and adsorption rate increase in proportion to the influent concentration. However, Figure 5 shows that the adsorption capacity of impregnated activated carbon is not a function of the inlet concentration. Furthermore, we conducted the measurements of breakthrough curves of other amine-impregnated activated carbons and found that they also do not show any dependence on the inlet concentration. These results indicate that the removal mechanism of amine-impregnated activated carbons is not physisorption but chemisorption. The measurements of adsorption capacities for acetaldehyde were carried out using the activated carbon of sample A impregnated with various amines. Figure 7 shows the relationship between the adsorption capacity of acetaldehyde (qc) and the amount of various amines impregnated on the activated carbon (A). In the figure, the solid line represents the stoichiometric reaction between the impregnated amine and acetaldehyde, i.e., equimolar reaction. The activated carbons impregnated with p-aminophenol, p-nitroaniline, and p-phenylenediamine do not adsorb acetaldehyde, indicating that these amines do not react with acetaldehyde on the activated carbon. On the other hand, the adsorption capacities of activated carbons impregnated with 2-aminoethanol, morpholine, aniline, m-anisidine, p-anisidine, pphenetidine, m-toluidine, and p-toluidine increase linearly with the amount of impregnation, and fall on the stoichiometrical line. The activated carbons impregnated with these amines remove acetaldehyde by the following condensation reaction:
R-NH2 + CH3CHO f R-NdCHCH3 + H2O
(3)
Figure 8 shows the relationship between qc and A of aminobenzoic acids (o-aminobenzoic acid, m-aminobenzoic acid, and p-aminobenzoic acid) for sample A. In the figure, 5438
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
FIGURE 7. Relationship between adsorption capacity and impregnated moles of various amines.
FIGURE 8. Relationship between adsorption capacity and impregnated moles of aminobenzoic acids. the solid line represents the equimolar reaction. The agreement between qc and the stoichiometric one for m- and p-aminobenzoic acid-impregnated activated carbon indicates that the activated carbons impregnated with aminobenzoic acids also remove acetaldehyde by the condensation reaction. For o-aminobenzoic acid-impregnated activated carbon, values of qc are lower than that obtained from the stoichiometry; the reactivity of o-aminobenzoic acid is lower than that of m- and p-aminobenzoic acid. The prevention or retardation of the chemical reaction because of neighboring groups on the same molecule is known as a steric hindrance (5). For example, ortho-substituted aromatic acids are more difficult to esterify than are the meta and para substitutions (6). Thus, the low reactivity of o-aminobenzoic acid is probably due to the steric hindrance of the ortho substituent (carboxyl group) present in the benzene ring. Figure 9 shows the relationship between qc and the amount of impregnated mass of aminobenzenesulfonic acids (o-aminobenzenesulfonic acid, m-aminobenzenesulfonic acid, and p-aminobenzenesulfonic acid) on the activated carbon of sample A. In the figure, the adsorption capacities of activated carbon impregnated with m-aminobenzenesulfonic acid and p-aminobenzenesulfonic acid are higher than the stoichiometric line while those impregnated with o-aminobenzenesulfonic acid nearly fall on the stoichiometric line of the equimolar reaction. The enhancement in the
FIGURE 9. Relationship between adsorption capacity and impregnated moles of aminobenzenesulfonic acids.
FIGURE 10. Adsorption capacities of activated carbons impregnated with p-aminobenzenesulfonic acid (samples A-C).
adsorption capacity can be attributed to the acidity of aminobenzenesulfonic acids. Synthesis of 2,3-disubstituted quinolines by cyclocondensation of aromatic amines with aldehydes is known as the Doebner-Miller reaction (7). This reaction takes place only under acidic conditions, where a molecule of aromatic amines reacts with two molecules of aldehydes. The dashed line in Figure 9 represents the stoichiometric relation of the Doebner-Miller reaction, which is in agreement with the adsorption capacity of p-aminobenzenesulfonic acid-impregnated activated carbon. Since the reactivity of aminobenzenesulfonic acids with acetaldehyde decreases in a para compound, meta compound, ortho compound order, the steric configuration of impregnated aminobenzensulfonic acids led to the difference in adsorption capacity. The proposed reaction scheme between aminobenzenesulfonic acids impregnated on the activated carbon and acetaldehyde is as follows:
perhaps because the sulfonic acid group is larger than the carboxyl group. Chemical species formed on the surface of p-aminobenzenesulfonic acid-impregnated activated carbon (sample A) were identified by the following procedure to determine the reaction scheme between aminobenzenesulfonic acids and acetaldehyde. The components on the activated carbon were extracted into various solvents (chloroform, acetone, methanol, water, and 0.6 N hydrochloric acid) under ultrasonic irradiation at room temperature. Then the solvents were evaporated in nitrogen at room temperature. The residues were dissolved into the deuterated solvents (deuteriochloroform, deuterioacetone, deuteriomethanol, and heavy water) for analysis by H NMR. The H NMR spectra of the residues from the impregnated activated carbon challenged with acetaldehyde showed 2-methylquinoline-6-sulfonic acid formed as the major product via the Doebner-Miller reaction. These experimental results showed that the activated carbon impregnated with aminobenzenesulfonic acids differs from that impregnated with the other amines in its acetaldehyde adsorption mechanism; the former removes acetaldehyde by the Doebner-Miller reaction, while the latter does so by the condensation reaction of acetaldehyde. Furthermore, since p-aminobenzenesulfonic acid reacts with acetaldehyde without steric hindrance in the Doebner-Miller reaction, it is said that p-aminobenzenesulfonic acid is the most suitable impregnant for the chemisorption of acetaldehyde. Further measurements of the adsorption capacities for acetaldehyde were carried out using various types of activated carbons impregnated with p-aminobenzenesulfonic acid. Figure 10 shows the relationship between qc and the amount of impregnated mass of p-aminobenzenesulfonic acids (A) for various activated carbons (samples A-C). In the figure, the dashed line indicates the stoichiometric relation of the Doebner-Miller reaction. For sample A, the adsorption capacity is the same as the stoichiometric one up to an A of 0.9 mmol/g but decreases with a further increase in A. The reason for the maximum adsorption capacity at a given impregnated mass can be explained as follows. When the impregnated mass is less than 0.9 mmol/g, p-aminobenzenesulfonic acid covers the surface of activated carbon uniformly as a monolayer so that all p-aminobenzenesulfonic acid is accessible for the reaction with acetaldehyde. On the other hand, when the impregnated mass is greater than 0.9 mmol/g, p-aminobenzenesulfonic acid forms a multilayered structure on the surface of activated carbon. The layered
In the first step, two molecules of acetaldehyde condense to form crotonaldehyde by acid-catalyzed aldol condensation on the surface of activated carbon, where aminobenzenesulfonic acids serve as the acid catalyst in this reaction. Then crotonaldehyde reacts with aminobenzenesulfonic acids to form 2-methylquinoline derivatives by the second reaction. Therefore, aminobenzenesulfonic acids impregnated on activated carbons are not only the reactant but also the acid catalyst in the removal of acetaldehyde. Because the second reaction is hindered for the o- and m-aminobenzenesulfonic acids compared to p-aminobenzenesulfonic acid, the reactivities of o- and m-aminobenzenesulfonic acids are lower than that of p-aminobenzenesulfonic acid. The steric hindrance may occur even for m-aminobenzenesulfonic acid
VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5439
FIGURE 11. Pore size distributions of activated carbon impregnated with p-aminobenzenesulfonic acid (sample A).
FIGURE 12. Pore size distributions of activated carbon impregnated with p-aminobenzenesulfonic acid (sample B). structure grows with an increase in the impregnated mass and eventually fills up the pores so that p-aminobenzenesulfonic acid in pores does not take part in the reaction with acetaldehyde. The pore size distributions of p-aminobenzenesulfonic acid-impregnated activated carbons (sample A) are shown in Figure 11. In the figure, D is the pore diameter and V is the pore volume. The figure shows that there is a large reduction in the pore volume with a diameter of less than ∼2 nm when the impregnated mass is 1.085 mmol/g. This is caused by the obstruction of pores by the impregnated p-aminobenzenesulfonic acid. Figure 12 shows the pore size distribution of p-aminobenzenesulfonic acid-impregnated activated carbons of sample B. The pore volume decreases with an increase in the impregnated mass, but the reduction in pore volume is not so significant, implying that little obstruction of pores occurs for sample B when A < 1.633 mmol/g. Therefore, for sample B, the adsorption capacity is the same as the stoichiometric one and no decrease in qc with A is observed in Figure 10. This is because the surface area of sample B is larger than that of sample A. The decrease in qc with A for sample B would be observed at a higher value of A. What follows from the previous discussion is that an activated carbon with a larger surface area would be suitable 5440
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
FIGURE 13. Adsorption capacities of activated carbons impregnated with p-aminobenzenesulfonic acid potassium salt (sample A). for the impregnation of p-aminobenzenesulfonic acid. Consequently, measurements of adsorption capacity were carried out using a large-surface-area activated carbon of sample C impregnated with p-aminobenzenesulfonic acid, which is shown with triangles in Figure 10. As seen in Figure 10, the dependence of qc on A is peculiar for sample C; i.e., the adsorption capacity remains at zero when A < 0.5 mmol/g and increases with an increase in A. This result indicates that some fraction of impregnated p-aminobenzenesulfonic acid on sample C does not take part in the reaction with acetaldehyde. This may be attributed to the manufacturing process of large-surface-area activated carbons which are usually prepared by employing the intercalation of potassium hydroxide into carbon layers (8, 9); therefore, they probably contain residual potassium. As shown in Table 1, the high pH value of sample C indicates that potassium remains on the surface of sample C. It is conceivable that the residual potassium in the activated carbon reacts with p-aminobenzenesulfonic acid to form p-aminobenzenesulfonic acid potassium salt by a neutralization reaction. Once the salt is formed, the salt would not react with acetaldehyde as shown in Figure 13, where the adsorption capacity of activated carbon impregnated with the potassium salt is plotted against the impregnated mass of the salt. Consequently, residual potassium in sample C serves as an inhibitor in the reaction between p-aminobenzenesulfonic acid and acetaldehyde.
Appendix For the activated carbon-packed bed, the material balance equation is written as
γ ∂q ∂C )∂x u ∂t
(A.1)
where C is the acetaldehyde concentration at depth x in the bed and time t, q is the adsorbed mass of acetaldehyde, γ is the bulk density of a packed bed, and u is the superficial gas velocity. If we can assume that the adsorption of acetaldehyde on the activated carbon is governed by the chemical reaction between physically adsorbed acetaldehyde and the unreacted impregnant on the activated carbon and that the physical adsorption of acetaldehyde follows a Henry-type isotherm, the rate of adsorption is given by eq A.2.
∂q ) k(qs - q)C ∂t
(A.2)
where qs is a saturated adsorbed mass of acetaldehyde and k is the rate constant. The initial condition and the boundary condition are given by eqs A.3 and A.4, respectively.
q ) 0 for 0 e x e L at t ) 0
(A.3)
C ) Ci at x ) 0
(A.4)
These equations are analytically solved to give C and q as
C)
q)
exp(A2t)
Ci
(A.5)
qs
(A.6)
exp(A1x) + exp(A2t) - 1 exp(A2t) - 1 exp(A1x) + exp(A2t) - 1
qav )
qs [ln(1 - E) + A1L] A1
(A.12)
Equation A.12 is rearranged to yield eq A.13.
(q [kγL u
E ) 1 - exp
av
]
- qs )
(A.13)
Equation A.13 proves that the E-qav relationship does not depend on Ci. Incidentally, when the exponential function in eq A.13 is expanded around zero for small argument, we have
E≈-
kγL kγL (q - qs) for ,1 u av u
(A.14)
Equation A.14 indicates that E decreases linearly with qav when kγL/u is small.
where
kγqs u
(A.7)
A2 ) kCi
(A.8)
A1 )
If x ) L (where L is the bed height) in eq A.5, the instantaneous removal efficiency of adsorbate (E) is given by
E)
exp(A1L) - 1
(A.9)
exp(A1L) + exp(A2t) - 1
The average adsorbed mass of adsorbate (qav) is given by
∫ q dx L
qav )
0
(A.10)
L
Substituting eq A.6 into eq A.10, we obtain
qav )
[
qs exp(A2t) + A1L ln A1 exp(A1L) + exp(A2t) - 1
Combining eq A.11 with eq A.9
]
(A.11)
Literature Cited (1) Noda, T.; Suzuki, M. Study on development of adsorvent of acetaldehyde. Kagaku Kogaku Ronbunshu 1998, 24, 646-652. (2) Sugiura, M.; Fukumoto, K. Removal of acetaldehyde by sepiolite2-aminobenzoic acid complex from ambient air. Clay Sci. 1992, 8, 257-271. (3) Noda, T. Adsorvent of acetaldehyde made of sepiolite. Adsorpt. News 1999, 13, 16-19. (4) Noda, T.; Tomura, S.; Sakoda, A. Acetaldehyde adsorption characteristics of adsorvent made of sepiolite impregnated with amino compound. Nendo Kagaku 2001, 41, 75-82. (5) Newman, M. S. Steric Effects in Organic Chemistry; John Willey & Sons: New York, 1956. (6) Hartman, R. J.; Borders, A. M. Effect of polar groups upon esterification velocities of substituted benzoic acids. J. Am. Chem. Soc. 1937, 59, 2107-2112. (7) Doebner, O.; Miller, W. Ueber Chinaldinbasen. Ber. Dtsch. Chem. Ges. 1883, 16, 2464-2472. (8) Marsh, H.; Crawford, D.; O’Grady, T. M.; Wennerberg, A. Carbons of high surface area. A study by adsorption and high-resolution electron microscopy. Carbon 1982, 20, 419-426. (9) Marsh, H.; Yan, D. S. Formation of active carbons from cokes using potassium hydroxide. Carbon 1984, 22, 603-611.
Received for review September 21, 2004. Revised manuscript received April 11, 2005. Accepted May 10, 2005. ES048514B
VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5441