Adsorption of Benzene, Toluene, and Xylenes on Monolithic Carbon ...

Langmuir 2007, 23, 10095-10101

10095

Adsorption of Benzene, Toluene, and Xylenes on Monolithic Carbon Aerogels from Dry Air Flows D. Faireń-Jimeńez, F. Carrasco-Marıń, and C. Moreno-Castilla* Departamento de Quı´mica Inorga´ nica, Facultad de Ciencias, UniVersidad de Granada, 18071 Granada, Spain ReceiVed May 18, 2007. In Final Form: July 2, 2007 Monolithic carbon aerogels were obtained by carbonization of organic aerogels prepared by polymerization of resorcinol and formaldehyde under different conditions. Some carbon aerogels obtained were further CO2-activated. Samples were characterized by gas adsorption, scanning electron microscopy, high-resolution transmission electron microscopy, and mechanical tests. Benzene, toluene and xylenes were adsorbed from dry air by using carbon bed columns, obtaining the breakthrough curves. There was no correlation between the amount adsorbed at the breakthrough point and the volume of micropores narrower than 0.7 nm. Conversely, a good linear relationship between the amount adsorbed at the breakthrough point and the total micropore volume up to a mean micropore width of around 1.05 nm was found. In addition, the height of the mass transfer zone decreased with the mean width of the total micropores up to a value of around 1.05-1.10 nm. One of the best adsorbents obtained showed the lowest height of the mass transfer zone and one of the highest amounts adsorbed at the breakthrough point, either per mass or volume unit. However, it had a lower elastic modulus and compressive strength than other monolithic carbon aerogels, although its compressive strength (3 MPa) was still high enough to use it in carbon bed columns. The sample with the best mechanical properties was a poorer adsorbent. Regeneration of the exhausted adsorbents allowed the recovery of the hydrocarbons adsorbed without any appreciable loss of adsorption capacity of the carbon bed.

Introduction Volatile organic compounds are air pollutants with an adverse health impact, and their emission limits have therefore been progressively restricted.1 These compounds play a key role in the formation of tropospheric ozone and other oxidants in the atmosphere, which can yield photochemical smog.2 Benzene, toluene, and xylenes (known as BTX) are among these pollutants, and they are related to intense traffic in urban areas and the use of organic solvents and industrial paints. Among the different technologies available for controlling gaseous emissions, adsorption on porous media is an interesting option because it offers the possibility of recovering valuable organic compounds. However, the microporous texture of the adsorbents must be carefully optimized to ensure both an adequate adsorption capacity and an easy regeneration process to enable recovery of the pollutant and reutilization of the adsorbents without loss of their adsorptive properties. When the BTX removal is carried out from gas flows, it is necessary to use adsorbent bed columns that allow a good contact time between the adsorbent and the pollutant and a low-pressure drop through the bed. This can be reached by using monolithic adsorbents such as monolithic carbon aerogels (MCAs). These are novel porous carbon materials3,4 whose network structure is formed by interconnected nano-sized primary particles. Micropores are developed within primary particles, and mesopores are produced in the interparticle structure. It is therefore possible to control the concentration of micropores and mesopores independently, which is one of the advantages of carbon aerogels as porous carbon materials. In addition, they can be obtained with high purity and in the form of monoliths, beads, powders, * Corresponding author. Fax: +34-958-248526. E-mail: [email protected]. (1) European Parliament fact sheet on air pollution. http://www.europarl. europa.eu/facts/4_9_6_en.htm (accessed Dec 2006). (2) Atkinson, R. Atmos. EnViron. 2000, 34, 2063-101. (3) Pekala, R. W. U.S. Patent No. 4,873,218, 1989 (4) Pekala, R. W. J. Mater. Sci. 1989, 24, 3221-27.

or thin films. These unique properties make them promising materials for application in adsorption and catalysis.5,6 On the other hand, when adsorbents are going to be used packed inside a column, the knowledge of their mechanical properties is as important as their micropore texture, since they must resist the weight of the adsorbent packed in the bed and the stress produced by vibrations or bed movements. The aim of this work is to study the removal of BTX from dry air (because humidity can interfere with hydrocarbon adsorption) using MCA columns. MCAs were obtained following different recipes to vary their porosity, in order to do a systematic study to determine what characteristics of the microporosity, volume, and mean width govern the BTX removal. This work was complemented by studying the mechanical properties in compression of the adsorbents used. Experimental Section Monolithic organic aerogels were prepared by a sol-gel polymerization reaction of resorcinol (R) and formaldehyde (F) in water (W) as described elsewhere.7,8 Molar ratios were R/F ) 0.5 and R/W ) 0.08 and 0.13. Different polymerization catalysts (C), i.e., sodium carbonate, potassium carbonate, para-toluenesulfonic acid (PTSA) and oxalic acid (OA), were added to the RF mixture. The R/C molar ratio was 800 except in the case of PTSA, when it was 8000 because of the very rapid polymerization reaction and immediate solidification of the mixture. Mixtures were stirred to obtain a homogeneous solution that was cast into glass molds (45 cm length × 0.5 cm i.d.). The glass molds were sealed, and the mixture was cured. After the curing cycle, the gel rods were cut into 5 mm pellets and placed in acetone to exchange water and were then supercritically dried with carbon dioxide to form the corresponding monolithic organic aerogels. Pyrolysis was carried out in N2 flow at 100 cm3/ (5) Pierre, A. C.; Pajonk, G. M. Chem. ReV. 2002, 102, 4243-66. (6) Moreno-Castilla, C.; Maldonado-Ho´dar, F. J. Carbon 2005, 43, 455-65. (7) Faireń-Jimeńez, D.; Carrasco-Marıń, F.; Moreno-Castilla, C. Carbon 2006, 44, 2301-07. (8) Faireń-Jimeńez, D.; Carrasco-Marıń, F.; Djurado, D.; Bley, F.; EhrburgerDolle, F.; Moreno-Castilla, C. J. Phys. Chem. B 2006, 110, 8681-88.

10.1021/la701458h CCC: $37.00 © 2007 American Chemical Society Published on Web 08/28/2007

10096 Langmuir, Vol. 23, No. 20, 2007

Faire´ n-Jime´ nez et al.

Table 1. Carbon Aerogel Recipes (Formaldehyde, 0.224 mol; Resorcinol, 0.112 mol)

sample A E I J K

catalyst precursor (mol) 1.4 × 10-4 1.4 × 10-5 1.4 × 10-4 1.4 × 10-4 1.4 × 10-4

Na2CO3 PTSA OA Na2CO3 K2CO3

other solvents (mL)

solution pH

methanol (2.3) methanol (2.3)

6.4 1.6 2.6 6.4 6.4

water (mL) 26.7 26.7 26.7 24.5 13.1

[( ) ] 2

(1)

(2)

This equation is valid for E0 values between 42 and 20 kJ/mol, which correspond to pore widths between 0.35 and 1.3 nm. The Dubinin equation (eq 3) was used for smaller E0 values:11 L0 (nm) ) 24/E0 (kJ/mol)

kinetic diametera (nm)

benzene toluene o-xylene m-xylene

4.43 × 1.86 × 10-2 6.54 × 10-2 8.09 × 10-2

0.36 0.59 0.73 0.70

10-3

Taken from refs 12 and 13.

Table 3. BET Surface Area, Microporosity Characteristics, and Mechanical Properties of MCAs

where W is the amount absorbed at relative pressure p/p0; W0 is the limiting value filling the micropores; A is the differential molar work given by A ) RT ln P0/P; β is the affinity coefficient, taken to be 0.33 and 0.35 for N2 and CO2, respectively; and E0 is the characteristic adsorption energy. N2 and CO2 molar volumes were taken to be 34.65 and 43.01 cm3/mol, respectively.9 Once the E0 value was known, the mean micropore width, L0, was obtained by applying the Stoeckli equation (eq 2):10 L0 (nm) ) 10.8/(E0 - 11.4 kJ/mol)

P/P0

a

min, heating to 900 °C at a heating rate of 1.5 °C/min and a soaking time of 5 h. The MCAs so obtained are designated, according to their recipes, by different letters in this paper, as indicated in Table 1. Portions of carbon aerogels E, J, and K were activated by heating to 800 °C in CO2 flow to obtain different degrees of activation. These activated carbon aerogels are indicated by adding the percentage weight loss to the letter that designates them. Thus, sample E8.5 is sample E activated to 8.5% weight loss. MCAs were characterized by N2 and CO2 adsorption at -196 and 0 °C, respectively. Adsorption isotherms were measured in conventional volumetric equipment made of Pyrex glass and free of mercury and grease, which reached a dynamic vacuum of more than 10-6 mbar at the sample location. Equilibrium pressure was measured with a Baratron transducer from MKS. Prior to the adsorption measurements, samples were outgassed at 110 °C overnight under high vacuum. N2 adsorption isotherms were analyzed by the Brunauer-Emmett-Teller (BET) equation. In addition, the Dubinin-Radushkevich equation (eq 1) was applied to both N2 and CO2 adsorption isotherms at -196 and 0 °C, respectively: A W ) W0 exp βE0

Table 2. BTX Characteristics hydrocarbon

(3)

Monolithic density, d, was obtained with mercury at atmospheric pressure and room temperature. Scanning electron microscopy (SEM) experiments were carried out with a Zeiss DSM950 (30 kV) microscope. High-resolution transmission electron microscopy (HRTEM) observations were made with a Phillips CM-20 electron microscope. The mechanical properties of the samples were measured under uniaxial compression using Shimazdu equipment, model AGS-J 10kN. Tests were performed at a strain rate of 1 mm/min. All measurements were made at room temperature (around 20 °C) and 40% relative humidity. Specimens were cylinders carefully machined to have perfectly parallel bases and a length/diameter ratio of 1.5. (9) Cazorla-Amoro´s, D.; Alcan˜iz-Monge, J.; De la Casa-Lillo, M. A.; LinaresSolano, A. Langmuir 1998, 14, 4589-96. (10) Stoeckli, F. In Porosity in Carbons: Characterization and Applications; Patrick, J., Ed.; Arnold: London, 1995; p 67. (11) Dubinin, M. M. Carbon 1985, 23, 373-380.

sample

SBET m2/g

W0(N2) cm3/g

W0(CO2) cm3/g

L0(N2) nm

L0(CO2) nm

d g/cm3

E MPa

RF MPa

A E E8.5 E22 I J J14 J35 K K11 K18

934 751 845 1296 740 814 1048 1622 907 1033 1199

0.39 0.28 0.33 0.53 0.31 0.32 0.43 0.67 0.36 0.42 0.47

0.33 0.28 0.30 0.33 0.32 0.28 0.36 0.45 0.27 0.31 0.33

0.95 0.69 0.77 1.55 0.65 0.88 1.03 1.76 1.04 1.07 1.05

0.62 0.61 0.62 0.67 0.61 0.60 0.63 0.68 0.54 0.63 0.69

0.48 1.04 0.89 0.87 0.86 0.35 0.33 0.26 0.72 0.65 0.61

290 1108 448 164 856 214 52 30 489 278 127

7 24 13 4 39 9 3 2 19 13 3

No precautions were taken to prevent moisture adsorption by the carbon aerogels. The compression stress-strain curves obtained were used to determine the elastic modulus, E, and the compressive strength, RF. The elastic modulus corresponds to the slope of the linear region of the curve at low strains. The compressive strength is the maximum stress supported by the specimen during the test, i.e., the stress at which macroscopic failure occurs. The adsorption of benzene, toluene, o-xylene, and m-xylene was carried out at 25 °C by using a dry air flow containing 741 ppmv of the particular hydrocarbon to be studied. This concentration was obtained by bubbling the dry air flow through the hydrocarbon contained in a cold trap at the appropriate temperature to yield the above concentration in the air flow. MCA pellets were placed inside glass columns with an inner diameter of 7.5 mm. The mass of the adsorbent used was 1 g, except in the case of samples from series J, which, due to their low density, were 0.765 g. Some characteristics of the BTX, such as their relative pressure and kinetic diameter, are compiled in Table 2. o-Xylene and m-xylene were selected among the three xylene isomers to carry out this work because they have a greater kinetic diameter than p-xylene (0.65 nm) and toluene. The hydrocarbon-laden air flowed upward through the carbon bed. All flows were controlled by mass flow controllers at a total flow of 60 cm3/min. Inlet and outlet concentrations were measured by on-line gas chromatography using a Varian gas chromatograph, model Chrompack CP-3800, with a flame ionization detector and a Benton 34/DNDP SCOT column. Breakthrough curves for the different carbon beds and hydrocarbons were obtained from these data by plotting the relative concentration, C/C0, against time. Prior to the determination of the breakthrough curves, the adsorbent carbon bed was cleaned by heating at 250 °C for 5 h under He flow. After this treatment, the column was cooled to 25 °C, and the He flow was switched to the hydrocarbon-laden air flow. When the adsorbent beds were exhausted with toluene and m-xylene, they were regenerated to be reused. For this purpose, the hydrocarbonladen air flow was switched to a He flow, and the temperature was increased to 250 °C. This treatment was given for 24 h. After that, the column was cooled to 25 °C, and the He flow was switched again to the hydrocarbon-laden air flow to start a new adsorption cycle.

Results and Discussion Characteristics of the Adsorbents. Surface characteristics of the MCAs obtained from the adsorption isotherms described above are compiled in Table 3. Results obtained with some of these samples have been given and discussed in detail elsewhere.7,8

Adsorption of BTX on MCAs

Langmuir, Vol. 23, No. 20, 2007 10097

Figure 1. SEM microphotographs of samples J and E.

Figure 2. HRTEM microphotographs of samples J and E.

The BET surface area of the MCAs ranged from 740 to 934 m2/g. This value increases with the activation degree, reaching a value as high as 1622 m2/g for sample J35. The micropore volume obtained from CO2 adsorption at 0 °C yields the narrow micropore volume (below about 0.7 nm width), whereas the total micropore volume is obtained from N2 adsorption at -196 °C if there are no pore constrictions at their entrance.9 The value of W0(N2) is similar to or higher than W0(CO2) in all samples, which indicates that there are no pore constrictions. The similarity of both micropore volumes indicates a homogeneous micropore size distribution, whereas this distribution is more heterogeneous, predominating micropores wider than 0.7 nm, when W0(N2) is higher than W0(CO2). The differences between both pore volumes increase with the degree of activation, in parallel with an increase in the average micropore width. This variation indicates that narrow micropores are widened during activation, resulting in a more heterogeneous microporosity. Activation of sample K does not affect the L0(N2) value. However, it increases the other parameters. MCAs with the lowest and highest densities (J and E, respectively (see Table 3)) were observed by SEM, and selected microphotographs are given in Figure 1. This figure shows the increase in the aggregation of the primary particles when the density increases. HRTEM microphotographs of these samples are given in Figure 2. The morphology of sample J shows that the aggregation of the primary particles leaves big holes or

macropores in its structure that are not present in the case of sample E. This leads to a lower density in sample J than in sample E. Therefore, results from SEM and HRTEM clearly indicate that the increase in the density of the MCAs obtained is a consequence of the increase in the aggregation or interconnectivity of the primary particles that constitute the carbon aerogel network. The density of samples prepared under acidic conditions (E and I) was much higher than that of samples prepared using alkaline carbonates as polymerization catalysts. The solution pHs of the initial mixture in samples E and I were 1.6 and 2.6, respectively, which were lower than those used in the preparation of the other samples (Table 1). The polycondensation of resorcinol and formaldehyde involves two main reactions: addition and condensation. The former is catalyzed by bases, and the latter is catalyzed by acids.4 Hydroxymethyl derivatives of resorcinol are negatively charged, whereas formaldehyde is positively charged in acidic medium. The electrostatic attraction between these two species may be responsible for the increase in the aggregation of clusters.14 Compression stress-strain curves of carbon aerogels have a linear region corresponding to elastic deformation, which extends (12) Sing, K. S. W.; Williams, R. T. Part. Part. Syst. Charact. 2004, 21, 71-9. (13) Corma, A.; Corell, C.; Pe´rez-Pariente, J.; Guil, J. M.; Guil-Lo´pez, R.; Nicolopoulos, S.; Gonzalez Calbet, J.; Vallet-Regi, M. Zeolites 1996, 16, 7-14. (14) Bekyarova, E.; Kaneko, K. AdV. Mater. 2000, 12, 1625-28.

10098 Langmuir, Vol. 23, No. 20, 2007


up to the fracture of the material. The shape of these curves is typical of brittle materials. The different parameters obtained from these curves are displayed in Table 3. Highly porous materials such as aerogels show a scaling relationship of elastic modulus with density that is independent of the constituents of the aerogel and the gelation time,15-22 according to eq 4:

E ) Cdn

(4)

where C is a pre-exponential factor, and n is the scaling exponent, which usually ranges from 2 to 4. The open-cell aerogel model23 predicts a value of n ) 2. However, many materials exhibit much higher exponents.20 The relationship between elastic modulus and density in the MCAs prepared shows a scaling exponent and pre-exponential factor of 1.6 ( 0.2 and 10(3.0 ( 0.1), respectively. The scaling exponent obtained was lower than that reported by Pekala et al.17 for their carbon aerogels (2.7 ( 0.2), although their density range differed from that of the present samples (0.35-1.04 g/cm3). The scaling exponent obtained in this study is close to the value of n ) 2 expected for open-cell aerogels. The compressive strength of the MCAs also showed a scaling relationship with their density. The scaling exponent and the pre-exponential factor are 1.4 ( 0.5 and 10(1.4 ( 0.1), respectively The scaling exponent value is close to that found for the elastic modulus and is lower than previously reported values16 for organic and carbon aerogels (2.4 ( 0.3). Results obtained show that the MCA prepared by polymerization of resorcinol and formaldehyde under acidic conditions using PTSA has the best mechanical properties. This is due to the fact that this sample has the highest density, resulting from the greatest aggregation of the clusters formed during the polymerization reaction, as shown by SEM and HRTEM. CO2-activation of samples E, J, and K markedly increased their surface area (Table 3), thereby decreasing their density. There was a major decrease in E and RF values with activation, because of the loss in density. The elastic modulus of activated carbon aerogels also showed a scaling relationship with their density. The scaling exponent and pre-exponential factor are 1.8 ( 0.2 and 10(2.6 ( 0.1), respectively, which shows that carbon aerogels are stiffer than activated carbon aerogels at equivalent density. (Graphs that show the relationship between mechanical properties and density are given in Supporting Information.) BTX Adsorption. Breakthrough curves for toluene and m-xylene on MCAs from series K, as an example, are shown in the Supporting Information. Characteristics of the carbon beds were calculated from these curves following a method described elsewhere.24,25 These characteristics are XB, XB(vol), φ, and HMTZ, (15) Fricke, J. J. Non-Cryst. Solids 1988, 101, 169-73. (16) Pekala, R. W.; Alviso, C. T.; LeMay, J. D. J. Non-Cryst. Solids 1990, 125, 67-75. (17) Gross, J.; Scherer, G. W.; Alviso, C. T.; Pekala, R. W. J. Non-Cryst. Solids 1997, 211, 132-42. (18) Saliger, R.; Bock, V.; Petricevic, R.; Tillotson, T.; Geis, S.; Fricke, J. J. Non-Cryst. Solids 1997, 211, 144-50. (19) Woignier, T.; Reynes, J.; Hafidi, A. A.; Beurroies, I.; Phalippou, J. J. Non-Cryst. Solids 1998, 241, 45-52. (20) Ma, H. S.; Roberts, A. P.; Prevost, J. H.; Jullien, R.; Scherer, G. W. J. Non-Cryst. Solids 2000, 277, 127-41. (21) Ma, H. S.; Prevost, J. H.; Jullien, R.; Scherer, G. W. J. Non-Cryst. Solids 2001, 285, 216-21. (22) Fischer, F.; Rigacci, A.; Pirard, R.; Berthon-Fabry, S.; Achard, P. Polymer 2006, 47, 7636-45. (23) Gibson, L. J.; Ashby, M. F. Cellular Solids, Structure and Properties; Pergamon: New York, 1988. (24) Zogorski, J. S.; Faust, S. D. In Carbon Adsorption Handbook; Cheremisinoff, P. N., Ellerbusch, F., Eds.; Ann Arbor Science: Ann Arbor, MI, 1978; p 753.

which are compiled in Tables A-D in the Supporting Information. XB is the amount of the hydrocarbon adsorbed at the breakthrough of the column, which was arbitrarily chosen at a relative concentration of 0.02. XB was also given per adsorbent volume unit, XB(vol), taking into account the particle density of the MCAs. This value is of great importance when columns or reactors with fixed volumes are to be used. The fractional capacity, φ, is a measure of the efficiency of the carbon located within the mass transfer zone. Values of φ vary between 0 and 1 and were obtained from eq 5:

∫VV

φ)

(C0 - C)dV

1.00

0.02

(5)

(V1.00 - V0.02)C0

where V1.00 and V0.02 are the volumes of gases treated up to a relative concentration of 1.00 and 0.02, respectively. Rapid, sharp breakthrough curves usually show high φ values, and gradual, flat breakthrough curves show low φ values. The results obtained show that φ values are high and increase with the degree of activation of the MCA, which indicates a sharpening of the breakthrough curve in the same sense. HMTZ is the height of the mass transfer zone and indicates the rate of hydrocarbon removal by the adsorbent, with a lower height reflecting a faster adsorption rate. It was calculated from eq 6:

HMTZ ) h

(

V1.00 - V0.02

V0.02 + (V1.00 - V0.02)φ

)

(6)

where h is the bed depth. The HMTZ value should not change with the bed depth. Although in eq (6) HMTZ is directly related with h the other parameters of the equation also change with h. So, variations are balanced. BTX adsorption kinetics on a commercial activated carbon was studied by Kawasaki et al.26 They found that the adsorption kinetic constant increased in the order benzene < toluene < xylenes and that there was no difference among the three xylene isomers. This constant depended on the difference between the melting and boiling points of the hydrocarbons and their molecular size. Lillo-Ro´denas et al.27 studied the adsorption of benzene and toluene on carbon bed columns by using a hydrocarbonladen He flow containing 200 ppmv. They found that toluene adsorption was higher than that of benzene because of the different relative pressure of both hydrocarbons. According to these results, it is expected that BTX adsorption, under the experimental conditions used in this work, will increase when the relative pressure of the hydrocarbon increases. Thus, results obtained show that XB increases in the order benzene < toluene < xylenes (see Tables A-D in the Supporting Information). The highest XB values obtained were 307 mg/g for benzene, 365 mg/g for toluene, 380 mg/g for o-xylene, and 367 mg/g for m-xylene, which correspond to sample J35. However, its XB(vol) values are lower than those found with other MCAs because of the low density of J35. The above maximum values for J35 are higher than others found in the literature. Thus, Chiang et al.,28 using an initial concentration of benzene equal to 768 ppmv and different activated carbons, reached a maximum adsorption (25) Ubago-Pe´rez, R.; Carrasco-Marıń, F.; Faireń-Jimeńez, D.; Moreno-Castilla, C. Microporous Mesoporous Mater. 2006, 92, 64-70. (26) Kawasaki, N.; Kinoshita, H.; Oue, T.; Nakamura, T.; Tanada, S.; J. Colloid Interface Sci. 2004, 275, 40-3. (27) Lillo-Ro´denas, M. A.; Cazorla-Amoro´s, D.; Linares-Solano, A. Carbon 2005, 43, 1758-67. (28) Chiang, H. L.; Huang, C. P.; Chiang, P. C.; You, J. H. Carbon 1999, 37, 1919-28.


Langmuir, Vol. 23, No. 20, 2007 10099

Figure 3. Relationship between the amount adsorbed (as liquid) at the breakthrough point and the micropore volume accessible to CO2 at 0 °C. Sample A, 9; sample I, ×; series E, [; series J, b; and series K, 2.

capacity of 200 mg/g. When the initial concentration was 650 ppmv, the maximum adsorption capacities at saturation for benzene and toluene were 230 and 290 mg/g, respectively.29 However, the XB values for J35 were somewhat smaller than values reported before: 385 mg/g for toluene using activated carbon pellets30 and an initial concentration of 618 ppmv, and 340 mg/g for benzene and 500 mg/g for toluene using a granular activated carbon31 and an initial concentration of 600 ppmv. Although, in all these cases, the adsorption data reported corresponded to the column bed saturation and not to its breakthrough point. BTX adsorption on MCAs will essentially be governed by the microporosity of the adsorbent due to the kinetic diameter of the hydrocarbons (Table 2) being within the size range of the micropores. The relationship between XB and W0(CO2) is depicted in Figure 3. In this case, XB is given as the liquid volume of the hydrocarbon adsorbed per unit of adsorbent mass. So, the bisecting line of the plots indicate the amount of the hydrocarbon needed to fill up the micropores accessible to CO2 at 0 °C, that is to say, micropores below about 0.7 nm width. The results found indicate that there is no correlation between XB and W0(CO2) and that the number of adsorbents that fall on or above the bisecting line increases in the order benzene < toluene < xylenes. This is because, in that order, both the relative pressure and the kinetic diameter of the hydrocarbons increase. The relationship between XB and W0(N2) is plotted in Figure 4, which shows that no adsorbent completely used its total micropore volume for BTX removal and that now there is a good linear relationship between both parameters up to sample K18. Only samples E22 and J35 deviate from this trend. These two samples have a higher W0(N2) value than K18, but their L0(N2) value is also higher than that in the other MCAs. Thus, the above linear relationship between the amount of BTX adsorbed at the (29) Chiang, Y. C.; Chiang, P. C.; Chang, E. E. J. EnViron. Eng. 2001, 127, 54-62. (30) Benkhedda, J.; Jaubert, J. N.; Barth, D.; Perrin, L. J. Chem. Eng. Data 2000, 45, 650-52. (31) Shin, H. C.; Park, J. W.; Park, K.; Song, H. C. EnViron. Pollut. 2002, 119, 227-36.

breakthrough point and the mean width of the total micropores is extended only up to a mean width of around 1.05 nm (the value for K18). This means that micropores wider than this size would be less important for BTX removal from air in the experimental conditions used in this work. In other words, to increase BTX removal, it would be necessary to increase the total micropore volume with a mean micropore width of around 1.05 nm. This maximum limit of the micropore width is 2.9, 1.8, and 1.5 times higher than the kinetic diameters of benzene, toluene, and xylenes, respectively. Values of XB for sample K18 are slightly smaller than those found for J35. However, the former has much higher XB(vol) values than the latter (see Tables A-D in the Supporting Information). This makes sample K18 a better adsorbent than J35 for BTX removal in a fixed volume column. In addition, sample K18 has better mechanical properties than J35 (see Table 3). Results found in this work are different from those reported by Lillo-Ro´denas et al.,27 who found that the adsorption of benzene on activated carbon columns was linearly correlated with the volume of the narrowest microporosity, whereas that of toluene was linearly correlated with both the narrowest and the total micropore volume of the adsorbents. This could be due to the fact that they used a more diluted hydrocarbon concentration in the flow (200 ppmv) than that used in the present work (741 ppmv). On the other hand, the BTX adsorption rate will be faster for samples with lower HMTZ values. This parameter decreases with an increase in the degree of activation of the MCAs (see Tables A-D in the Supporting Information). In addition, HMTZ can be correlated with the mean micropore width L0(N2) as depicted in Figure 5, which shows that HMTZ linearly decreases when L0(N2) increases up to around 1.05-1.10 nm. So, this is the optimum value for a fast BTX adsorption, because HMTZ either remains constant or increases for wider micropores. Sample K18 is one of the best adsorbents for removing BTX from dry air under the experimental conditions used, because it shows the lowest HMTZ value and one of the higher XB and XB(vol) values among the MCAs studied. This is because it has a

10100 Langmuir, Vol. 23, No. 20, 2007


Figure 4. Relationship between the amount adsorbed (as liquid) at the breakthrough point and the micropore volume accessible to N2 at -196 °C. Sample A, 9; sample I, ×; series E, [; series J, b; and series K, 2.

Figure 5. Relationship between the height of the mass transfer zone and the mean micropore width determined from N2 adsorption at -196 °C. Sample A, 9; sample I, ×; series E, [; series J, b and series K, 2.

high total micropore volume with an optimum mean width of around 1.05 nm. From the point of view of the mechanical properties, sample K18 has a lower elastic modulus and RF values than other MCAs, yet its compressive strength seems to be high enough to use this adsorbent in column carbon beds. Conversely, the MCA with the best mechanical properties (sample E) is a poorer adsorbent for BTX than sample K18, with much lower XB and XB(vol) and higher HMTZ values. It is interesting to point out that activation of sample E, to yield E22, drastically reduced the mechanical properties, but the elastic modulus and RF values are still higher than those of sample K18 (Table 3). This sample, E22, shows the highest XB(vol) for toluene, o-xylene, and m-xylene, although its HMTZ values are higher than those for K18.

Once the carbon beds were exhausted with toluene and m-xylene, they were regenerated as indicated above. After regeneration, the breakthrough curves obtained practically matched those obtained in the first adsorption run (see graphs C and D in the Supporting Information). These results are of great interest because BTX adsorbed can be recovered without any appreciable loss of adsorption capacity of the carbon bed.

Conclusions MCAs with BET surface areas in the range between 740 and 934 m2/g were obtained by using different recipes. This parameter increased with the activation degree, reaching a value as high as 1622 m2/g. Microporosity obtained from CO2 and N2 adsorption showed that there were no micropore constrictions.


The elastic modulus and compressive strength of the MCAs showed a power-law scaling with their density, which is expected for open-cell aerogels. Similarly, the activated MCAs also showed a scaling relationship between the elastic modulus and the second power of the density, although carbon aerogels were much stiffer than activated carbon aerogels at equivalent density. Likewise, there was a major decrease in compressive strength with higher activation degree at equivalent density. Under the experimental conditions used, BTX adsorption on MCA will essentially be governed by the microporosity of the adsorbent. The results found indicate that there was no correlation between the amount adsorbed at the breakthrough point and the volume of micropores narrower than 0.7 nm. The number of adsorbents that used micropores wider than 0.7 nm increased with the relative pressure and the kinetic diameter of the hydrocarbons. Conversely, a good linear relationship was found between the amount adsorbed at the breakthrough point and the total micropore volume up to a mean micropore width of around 1.05 nm. So, micropores wider than this size are less important for the removal of BTX from air. In addition, the height of the mass transfer zone linearly decreased with the mean width of the total micropores up to a value of around 1.05-1.10 nm, which means that the adsorption rate increased in the same sense.

Langmuir, Vol. 23, No. 20, 2007 10101

One of the best adsorbents obtained was sample K18 because it showed the lowest height of the mass transfer zone and one of the highest amounts adsorbed at the breakthrough point, either per mass or volume unit. From the point of view of its mechanical properties, it had a lower elastic modulus and compressive strength than other MCAs, yet its compressive strength seems to be high enough to use this adsorbent in carbon bed columns. Conversely, the MCA with the best mechanical properties (sample E) is a poorer adsorbent for BTX than sample K18. Finally, regeneration of the exhausted adsorbents allowed recovery of the BTX adsorbed without any appreciable loss of adsorption capacity of the carbon bed. Acknowledgment. This research was supported by MEC and FEDER Project CTQ2004-03991 and Junta de Andalucıá Project RNM 547. D.F.J. acknowledges the pre-doctoral fellowship from MEC. Supporting Information Available: Graphs that show the relationship between mechanical properties and density and BTX breakthrough curves. Tables with the characteristics of carbon bed columns for the removal of BTX. This material is available free of charge via the Internet at http://pubs.acs.org LA701458H