- Email: [email protected]
Performance evaluation of activated carbon with different pore sizes and functional groups for VOC adsorption by molecular simulation
Chemosphere 227 (2019) 9e16
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Performance evaluation of activated carbon with different pore sizes and functional groups for VOC adsorption by molecular simulation Yaxiong An, Qiang Fu, Donghui Zhang*, Yayan Wang, Zhongli Tang The Research Center of Chemical Engineering, School of Chemical Engineering and Technology, Tianjin University, Tianjin, 300072, China
h i g h l i g h t s Adsorption of VOC investigated by GCMC. Effects of different pore sizes and functional groups of activated carbon. The pressure of capillary condensation is highly depend on the pore size. Functional groups affect intermolecular interactions and thus affect threshold pressure.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 January 2019 Received in revised form 28 March 2019 Accepted 2 April 2019 Available online 5 April 2019
Volatile organic compounds (VOCs) are growing pollutants now that cause air pollution and threaten human health. In this paper, the Grand Canonical Monte Carlo was used to simulate the adsorption performance of activated carbon on VOCs (benzene, toluene, acetone and methanol). After simulating different pore sizes (0.902 nm,1.997 nm,3 nm and 4 nm) adsorption performances of activated carbon, activated carbon with a pore size of 1.997 nm was selected to further study the influence of functional groups (carboxyl, amino, hydroxyl and hydrogen), and the capillary condensation was explained by the Kelvin equation. Furthermore, effects of functional groups under saturated vapor pressure (P0) of VOCs that range from 0 to 0.1 P0 were explained by the accessible volume and intermolecular interaction potential, respectively. Under pressure range of 0e0.1 P0, at the beginning of adsorption of acetone and methanol, carboxyl and amino groups could reduce the threshold pressure while hydroxyl and hydrogen have the opposite effect. For benzene and toluene, all functional groups have little effect on the threshold pressure, and they reduce the adsorption capacity instead. It could be concluded that the activated carbon could achieve the best adsorption effect on acetone and methanol, on the contrary, the addition of functional groups on benzene and toluene will weaken their adsorption performance. © 2019 Elsevier Ltd. All rights reserved.
Handling Editor: Dr. R Ebinghaus Keywords: Volatile organic compounds (VOCs) adsorption Activated carbon Grand canonical Monte Carlo (GCMC) Pore size Functional group
1. Introduction Volatile organic compounds (VOCs), which consist of many complex and hazardous chemicals, are the major organic compounds involved in atmospheric photochemical reactions and they are also the main precursors for the formation of ozone (O3) and fine particulate matter (Derwent et al., 2003). Now all over the world, air pollution problem has not only posed a serious threat to living environment and human health, but also restricted the development of the sustainable economy. Therefore, it is the time that advanced technologies should be applied to reduce VOC
* Corresponding author. E-mail address: [email protected] (D. Zhang). https://doi.org/10.1016/j.chemosphere.2019.04.011 0045-6535/© 2019 Elsevier Ltd. All rights reserved.
emissions and prevent the occurrence of smog at the source in order to improve the air quality effectively and enhance people's happiness. Until now, a variety of VOCs treatment technologies have emerged, such as incineration (Goralski et al., 1998), condensation (Hamad and Fayed, 2004), biodegradation, absorption (Dumont et al., 2011), catalytic oxidation (Azalim et al., 2013) and adsorption (Chiang et al., 2001; Song et al., 2005; Blommaerts et al., 2018). Among these methods, the adsorption technology is a highly effective and economical method, which can achieve the recovery of VOCs and recycle adsorbents. In the adsorption process, the adsorbent's performance plays an important role in overall capital and operating costs, and therefore the choice of suitable adsorbent is crucial for gas separation and recovery. Because of its large specific surface area, rich pore structure and excellent adsorption
10
Y. An et al. / Chemosphere 227 (2019) 9e16
capacity, activated carbon is widely used in gas purification, especially in the treatment and recovery of VOCs(Chiang et al., 2001; Chuang et al., 2003; Cosnier et al., 2005; Zhang et al., 2017). Sasaki T (Sasaki et al., 2008), Szcze˛ sniak B (Szcze˛ sniak et al., 2018) and Do DD (Ahmadpour and Do, 1995) et al. investigated the effect of pore structure of activated carbon on the adsorption performance of VOCs through experimental methods. However, it is difficult to get accurate adsorption data by experiments at normal temperature, because most VOCs, which have a high boiling point, are liquid at normal temperature and pressure. Additionally, the pore structure of activated carbon is complicated, although ordered mesoporous carbon can be obtained (Ryoo et al., 1999; Jun et al., 2000), the variables such as pore size and pore volume are difficult to control and adjust. On the surface of activated carbon, there are different levels of nitrogen-oxygen functional groups and different functional groups in activated carbon also have a significant effect on adsorption performance of VOCs(Jansen and Van Bekkum, 1995; ~ o et al., 2015), which mainly include Tamon and Okazaki, 1996; Patin eight oxygen-containing functional groups and two kinds of nitrogen-containing functional groups. It is also extremely difficult to examine the effects of specific functional groups by experimental methods. Molecular simulation can solve all these problems well and can be used to investigate the adsorption data of VOCs with arbitrary functional groups and activated carbon with arbitrary pore at any pressure and temperature. Nowadays, the molecular simulation is gradually applied in various fields. It can not only solve the above problems by establishing an idealized model, but also get the adsorption data that is difficult to obtain by experiments, and further explain the adsorption phenomenon at the molecular level. Among the simulation calculations reported so far, the most suitable method for studying the gas adsorption behavior of activated carbon is the Grand Canonical Monte Carlo (GCMC) simulation (Li et al., 2013). In some special cases, molecular dynamics is also used to study practical problems. CAO D (Cao et al., 2001) established Steele 10-43 slit sheet model for Monte Carlo simulation of layered microporous material adsorption storage of natural gas. The slit sheet model of activated carbon has been established in our previous work to explore the effect of the pore size of nanoporous carbon materials on the separation performance of CHF3-CHClF2 mixtures and the result has been verified by experiments (Fu et al., 2018). These studies have verified that slit sheet model used in gas adsorption is accurate and reliable. In the study of VOCs adsorption, Do DD (Liu et al., 2012) established carbon nanotube model and slit sheet model to investigate the effects of pore structure on methanol adsorption, but the influence of functional groups on activated carbon was not considered in his study. Liang X (Liang et al., 2018) studied the effects of the oxygen-containing functional groups of activated carbon on the adsorption of acetone at low pressure, but it neglected the effect of the addition of functional groups on the accessible volume. At present, there is no systematic study on the effects of activated carbon pore size and different kinds of functional groups on the adsorption performance of many kinds of VOCs with different structure and properties. In this paper, four representative VOCs (benzene, toluene, acetone and methanol) were selected to study their adsorption performance on activated carbon by establishing a slit sheet activated carbon model. GCMC simulation method was used to investigate the effects of pore size of activated carbon and four functional groups (carboxyl, amino, hydroxyl and hydrogen) on the adsorption performance of different VOCs. The change of capillary condensation pressure with pore size and the effects of functional groups on the accessible volume of activated carbon were discussed in detail. Furthermore, in order to select suitable pore size and functional
group for adsorption of specific VOC, the effects of intermolecular interaction potential on adsorption capacity of VOCs with different properties at 0-0.1P0 partial pressure were also analyzed in a systemic way. 2. Theory 2.1. Molecular model The activated carbon model used in GCMC simulation was simplified into a slit pore model formed by parallel graphite sheets. In recent years, single-layer graphite sheets or three-layer graphite sheets models have been widely used in the simulation of activated carbon according to a number of previous studies (Shevade et al., 2000; Georgakis et al., 2007; Lithoxoos et al., 2012), and what was adopted in this work is a three-layer graphite sheet model. Fig. 1 is a schematic diagram of slit pore model of activated carbon. As is shown in Fig. 1, two graphene layers had an interplanar spacing (d) of 0.34 nm (Fu et al., 2018), which is the van der Waals diameter of carbon atoms. For easier calculation, we chose an activated carbon unit made up of three-layer graphite sheets, where all the periodic boundary is applied in the x-, y- and z-directions, respectively. The specific length of a single-layer graphene in the x-direction is 2.952 nm, which in the y-direction is 2.982 nm. Various distances (D ¼ 1.242, 2.337, 3.34 and 4.34 nm) of graphitic sheet were used to investigate the effect of the activated carbon pore size. The slit pore width (D) is the distance between the centers of carbon atoms in innermost graphite sheets, and the effective pore size (Deff) could be calculated by:
Deff ¼ D 0:34nm
(1)
In order to study the effect of functional groups on the adsorption properties of adsorbed VOCs, four functional groups including hydrogen (He), hydroxyl (OHe), amine (NH2e) and carboxyl (COOHe) were considered. We set the functional group density to 1.73 mmol g1 so as to compare the effects of different functional groups well. The activated carbon models are described in Fig. 1. VOCs (benzene, toluene, acetone, and methanol) are built through the Dmol3 module and optimized for energy minimization. In the optimization project, the basis set selected the Double value orbit and orbital polarization function (DNP), and the
Fig. 1. Slit pore model of activated carbonColor code: gray, C; whit, H; red, O; blue, N. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Y. An et al. / Chemosphere 227 (2019) 9e16
calculation method selected the generalized gradient approximation (GGA) with Perdew-Burke-Ernzerhof (PBE) function (Perdew et al., 1996; Wang et al., 2012) which has high precision in describing long-range forces such as van der Waals interactions and it is widely used to describe the interaction between gas and graphite (Lim et al., 2011). The optimized models are shown in Fig S1.
3 1=2 0 2 εi εj r i r 0j εij ¼ 6 6 r 0i þ r 0j
11
(4)
3. Result 2.2. Grand canonical Monte Carlo simulation A GCMC simulation method, one of the most basic tools, was used to emulate specific adsorption behavior of VOCs on activated carbon. Condensed-phase Optimized Molecular Potentials for Atomistic Simulation Studies (COMPASS) force field (Sun, 1998; Sun et al., 1998; Tao et al., 2014), the first ab initio field, was selected in this simulation to calculate the interaction potential between gasand solid-phase molecules. In order to get more accurate prediction about molecular properties and condensation systems, some empirical corrections were widely used in molecules and condensed phase systems, such as critical state parameters and van der Waals parameters. Besides, the temperature adopted for all simulations in this work was 293.15 K and the fixed pressure task was chosen to calculate the equilibration in the range of the saturated vapor pressure (P0) of the adsorbates. In the Monte Carlo simulation process, 107 Monte Carlo steps were used for simulation. In order to reduce the error caused by the initial configuration and improve the accuracy of the simulation, the entire simulation process was divided into two equal parts, the equilibration steps and the production steps. The Equilibration steps were used for balancing and were discarded, and then the adsorption capacity were calculated on production steps. In the simulation process, the adsorbates were regarded as rigid spheres in the GCMC method and one of creating, transferring, rotating and deleting was selected in a Monte Carlo step, as long as the energy reached a lowest value in the system, then the step will be accepted. Finally, the adsorption capacities with different pressure were obtained through a series of simulations at a specific temperature of 293.15 K, and it is the adsorption isotherm that is at 293.15 K. 2.3. Isosteric heat Intermolecular interaction energy is an important evaluation criterion for adsorption properties, isosteric heat E is displayed by the following equation:
2 !9 !6 3 Xqi qj X r 0ij r 0ij 5 E¼ þ εij 42 3 rij rij rij ij
(2)
ij
The first section is the Coulomb equation, which calculates the electrostatic interaction force between the intermolecular i and j charge groups. The second section is the 9-6 Lennard-Jones (LJ) potential model that represents the intermolecular non-bonding effect and the van der Waals interaction force. q, rij represents the charge of the adsorption site and the distance between the two adsorption sites, respectively. ε is the potential well depth and r0 is the hard sphere radius of the atom. Moreover, the 9-6 LJ parameters of different atoms are determined by using the following equations (Waldman and Hagler, 1993):
0 6 11=6 6 r 0i þ r 0j B C r 0ij ¼ @ A 2
(3)
GCMC method was used to estimate the capacity of VOCs adsorbed by the activated carbon with different pore size. After that, the effects of four functional groups (eCOOH, -OH, -NH2 and eH) on the adsorption properties of benzene, toluene, acetone and methanol were investigated. 3.1. Adsorptions in activated carbon with different pore sizes Fig. 2 shows the adsorption isotherms of the VOCs in activated carbon with various pore sizes (0.902, 1.977, 3.000 and 4.000 nm) at the temperature of 293.15 K and the pressure of the saturated vapor pressure (P0) of the adsorbate. These adsorption data were all modeled independently through GCMC simulations and the total uptake represents the total atom number of gas adsorbed per gram (including gas and condense phase). Fig. 2aed shows the adsorption isotherms of benzene, toluene, acetone and methanol in activated carbon respectively. Obviously, the pore size has a significant effect on the adsorption isotherms of VOCs, and the effects of different VOCs are diverse from each other. For benzene and toluene, the adsorption isotherms in the activated carbon are quite similar. The total capacity was the highest in the graphitic silt pores with effective pore size of 4.000 nm and the total capacity of benzene and toluene follows the sequence of 4.000 > 3.000>1.997 > 0.902 nm under the saturated vapor pressure (P0) of benzene and toluene respectively, which is consistent with the experimental results of Wang X (Wang et al., 2018). However, for acetone and methanol, there is a different law for the changes of total uptake at the pressure of P0 and the order of total uptake becomes 3.000 > 1.997>4.000 > 0.902 nm and 1.997 > 3.000>4.000 > 0.902 nm respectively. It is obvious that there is an optimal pore size for the total uptake of VOCs under the saturated vapor pressure. When the activated carbon pore size is 0.902 nm, the adsorption isotherms of VOCs in Fig. 2 all exhibit type-I Langmuir adsorption behavior, which is a typical characteristic of microporous materials. However, as the pore size increases, the capillary condensation gradually appears and the type-I Langmuir adsorption behavior gradually transforms into type-IV Langmuir adsorption behavior. The adsorption isotherm and molecular size shown in Figs. 2 and S1 indicate that the capillary condensation occurs when the pore size is about four times the molecular diameter (Rafati et al., 2010). Easy to know, with excessive pore size or too small saturated vapor pressure, the capillary condensation is prohibited. 3.2. Adsorptions in activated carbon with functional groups The IDLH (Immediately Dangerous to Life or Health concentration) of benzene, toluene, acetone and methanol are 500, 500, 2500 and 6000 ppm from National Institute of Occupational Safety and Health (Patnaik). Therefore, it is important for adsorbents to adsorb VOCs effectively at lower partial pressure ranging from 0 to 0.1 P0. According to the adsorption isotherm in Fig. 2, activated carbon with pore size of 1.997 nm has better adsorption performance on VOCs at low partial pressure (less than 0.1 P0). Hence, the 1.997 nm activated carbon slit model was chosen in this study and was further used to investigate the effects of functional groups (eCOOH,
12
Y. An et al. / Chemosphere 227 (2019) 9e16
Fig. 2. Adsorption isotherms of VOCs (~ad: benzene, toluene, acetone and methanol) in activated carbon with different pore sizes: Deff ¼ 0.902, 1.977, 3.000 and 4.000 nm.
-OH, -NH2 and eH) on the adsorption of VOCs. 3.2.1. Adsorption of benzene Fig. 3a shows the adsorption isotherm of benzene and the data at low pressure expressed by logarithm, and the type-I Langmuir adsorption behavior is exhibited by all the adsorption isotherm. When the pressure is less than 0.1 P0, the adsorption data is difficult to obtain by experiment, but we can get it by simulation and represent it by logarithm to discover the subtle differences. In order to describe the effect of the functional group on the adsorption of VOCs under low pressure better, a threshold pressure is introduced, which represents the pressure at which the adsorbates begin to be adsorbed and the adsorbates begin to enter the adsorbent. Obviously, the functional group has little effect on the threshold pressure in the adsorption of benzene. All activated carbons with functional groups have reduced adsorption capacity of VOCs in the order: NONE > -H > -OH > -NH2>-COOH. 3.2.2. Adsorption of toluene As shown in Fig. 3b, the effect of functional groups on the adsorption performance of toluene is very similar to that of benzene. All the adsorption isotherms exhibit type-I Langmuir adsorption behavior and the functional groups only reduce the adsorption capacity but have minimal effect on the threshold
pressure. The adsorption capacity of toluene reduces in the order of NONE > -H > -OH > -NH2>-COOH for the pressure between 0-1 P0. 3.2.3. Adsorption of acetone In Fig. 3c, the adsorption isotherms of acetone are presented and the respective threshold pressure in the range of 0e0.01 P0, which modified by different functional groups, are depicted independently in a small graph. It is apparent that the addition of functional groups reduced the adsorption capacity but changed the threshold pressure. As can be seen clearly from the figure, carboxyl and amino decrease the threshold pressure to 1e7 P0 and 1e6 P0 respectively, whereas the hydrogen and hydroxyl increase the threshold pressure. Another important finding is that the total acetone uptake at a high pressure follows the sequence of NONE > -H > -OH z -NH2>COOH, but at a low pressure (0e0.001 P0) the sequence changes into eCOOH > -NH2>NONE > OH > -H. Attention should be paid that there are two intersections between the carboxyl, amino and NONE near 0.001 P0. In summary, the threshold pressure is increased when adding carboxyl, amino and decreased when adding hydrogen, hydroxyl. 3.2.4. Adsorption of methanol For methanol, the functional group has the strongest influence on its threshold pressure for the VOCs selected in this paper. When
Y. An et al. / Chemosphere 227 (2019) 9e16
13
~d: benzene, toluene, acetone and methanol) in activated carbon with different functional groups (eCOOH, -OH, -NH2 and eH). Fig. 3. Adsorption isotherms of VOCs (a
the pressure is less than 0.01 P0, only the activated carbon with carboxyl and amino can uptake methanol and the threshold pressures are 1e6 and 0.001 P0, respectively. At high pressure (0.1e1 P0), the total uptake of methanol in NONE is obviously higher than the uptake in activated carbon with functional groups. At the low pressure (0.01e0.1 P0), the methanol uptake follows the sequence -NH2>-COOH > NONE > -OH > -H. 4. Discussion To provide unique insights into the intrinsic nature of the pore size and surface functional groups of activated carbon for the adsorption of VOCs, the effects of (1) capillary condensation, (2) accessible volume and (3) adsorbate-adsorbent interaction energy are discussed in this section. 4.1. Capillary condensation From section 3, it is obvious that the effect of pore size on the adsorption of VOCs is significant, which mainly reflects in the phenomenon of capillary condensation. For all VOCs used in this work, the pressure (P/P0) of the capillary condensation increases as the pore size increases. The Chan-Kelvin equation (Cohan, 1938) is widely used to describe capillary condensation phenomena which is also applies to the slit pore model. The mechanism of condensation in slit pore model has been expounded by Zeng Y (Zeng et al., 2014), whose studies shows that the adsorption layer on the surface is undulating, and when the adsorption layer reaches a certain thickness, the convex portions are joined together to form a
cylindrical surface. However, when the pores are nanometer-sized, the Chan-Kelvin equation will no longer be applicable because the solid-liquid potential cannot be ignored at the nanometer size. The Derjaguin and Broekhoffde Boer (DBdB) models (Derjaguin, 1992) have a solid-liquid potential (ESL) added to the Cohan-Kelvin equation, which better describes the capillary condensation in nanopore adsorption. The DBdB equation can be written as
RT ln
P P0
syM
¼
r
þ ESL
(5)
where R is the gas constant; T is the temperature; P is the condensation pressure; P0 is the saturation vapor pressure; s is the surface tension; vM is the liquid molar volume; r is the radius of the condensed phase hemispherical surface and ESL is the solid-liquid potential. In Table S1 (Richards and Carver, 1921; Enders et al., 2007), we list the surface parameters of VOCs in DBdB equation. Furthermore, we choose benzene as a representative for further analysis and discussion of capillary condensation of VOCs. From the DBdB equation we can easily conclude that as the pore size increases, the pressure P/P0 at which capillary condensation occurs increases and the rate of increase is proportional to the value of svM. However, the pore size at which capillary condensation occurs is smaller than the actual pore diameter Deff. The r in the DBdB equation is defined as (D2eff t ) where t is an adsorbates “statistical” film thickness. Fig S2 shows the adsorption film of benzene at a pore size of 3 nm before capillary condensation. The adsorption of the molecules on the surface of the adsorbent before the capillary condensation occurs is equivalent to reducing the pore
14
Y. An et al. / Chemosphere 227 (2019) 9e16
size to create a condition for capillary condensation. Therefore, as the number of adsorbed layers increases and the pressure rises, once the capillary condensation condition is reached, the adsorption capacity rapidly increases. As explained above, when describing capillary condensation, the ESL is out of consideration in Chan-Kelvin equation, which means it does not consider the role of the adsorbent. To explain the effect of ESL, we plot PTln(P0/P) versus the inverse of the radius of curvature at the point just after condensation in Fig. 4. It can be obviously seen from Fig. 4 that the simulated value is slightly higher than the value fitted by Chan-Kelvin equation. The difference between these two values is ESL is positively correlated with the inverse of the radius of curvature. This positively correlated variation trend is consistent with the 9-6 LJ potential. In a word, the larger the pore size, the larger the distance between the condensed phase and the adsorbent, and the smaller the ESL. 4.2. Accessible volume As shown in Section 3.2, under the saturated vapor pressure of VOCs, the adsorption capacity all follows the sequence NONE > H > -OH > -NH2>-COOH. Section 4.1 explained the capillary condensation of VOCs. Then, the hypothesis that the adsorption capacity of VOCs under saturation vapor pressure is mainly limited by the accessible volume was proposed (Wang, 2007; Drisko et al., 2009). Liu Y and Wilcox J (Liu and Wilcox, 2012) calculated the accessible volume of the graphite slit-pore by the adsorption capacity of helium. Helium is a non-adsorbed or weakly adsorbed gas that is often used to measure pore volume in experiments or simulations. The excess adsorption (Ne) amount of helium is represented by the following equation:
Ne ¼ Nt V$rbulk
(6)
Where Nt is the number of helium molecular loaded in the pore of unit cell (molecular/unit cell), V is the accessible volume of the unit cell we use, rbulk is the bulk density of helium at a specific temperature and pressure. Since helium is hardly adsorbed on activated carbon, Ne ¼ 0, equation (1) can be expressed as: Nt ¼ V rbulk .At low pressure, helium can be regarded as an ideal gas. So Nt ¼ VM RT P where M is the molar mass of helium. Molecular simulation was performed to obtain a fitting curve of the adsorption
Fig. 4. Chan-Kelvin equation and simulated values of capillary condensation during adsorption.
capacity with pressure. Fig S3 presents the specific adsorption capacity of activated carbon with no functional group and that with different functional groups. Moreover, the accessible volume of each activated carbon was able to be estimated by the slope of each curve showed in Fig S3. The estimation results of accessible volume of different activated carbon are shown in Fig. 5. After adding different functional groups, it is obvious that the accessible volume is significantly reduced. For example, the addition of carboxyl groups reduces the accessible volume of activated carbon by 13.36%. From the data in Fig. 5, it is apparent that the accessible volume of activated carbon with different functional groups follows the sequence NONE > -H > -OH > -NH2>-COOH, which is consistent with the adsorption capacity under saturated vapor pressure. 4.3. Intermolecular interaction energy Intermolecular interaction energy indicates the affinity between the adsorbate and the adsorbent (Shin-ichi et al., 2005; Liang et al., 2018), and it is an important criterion for evaluation of adsorption performance. In order to investigate the effect of functional groups on the adsorption of VOCs, the intermolecular interaction energy curves were researched. As mentioned earlier in Section 3, the addition of functional groups of activated carbon reduced the adsorption capacity of VOCs under saturated vapor pressure and also affected the threshold pressure of different VOCs. This phenomenon can be well explained by Fig. 6, which shows an intermolecular interaction energy of adsorption for VOCs in activated carbon with different functional groups. From the figure, it can be seen that for benzene and toluene, the intermolecular interaction energies were both decreased with the addition of functional groups; however, for acetone, methanol, carboxyl and amino, the intermolecular interaction energy enhanced, especially at low pressure. But hydrogen and hydroxyl groups played an opposite effect on it. In addition, it is noteworthy that the intermolecular interaction energy drops sharply in the range of 0.01e0.1 P/P0 for all VOCs, especially for benzene and toluene. A possible explanation for this tendency might be the capillary condensation, which makes the heat of vaporization play an important role in the intermolecular interaction energy. Under high pressure, the difference of intermolecular interaction energy between different VOCs is significantly reduced. At this point, the difference in adsorption capacity of different VOCs was no longer primarily affected by intermolecular interaction energy, but was mainly restricted by the accessible volume, which has been
Fig. 5. Comparison of accessible volume of activated carbon with different functional groups.
Y. An et al. / Chemosphere 227 (2019) 9e16
15
Fig. 6. Intermolecular interaction energy of adsorption for VOCs (~ad: benzene, toluene, acetone and methanol) in activated carbon with different functional groups.
elaborated in detail in Section 4.2. In summary, under the pressure range of 0e0.1 P0, the threshold pressure is mainly affected by the intermolecular interaction energy. For the four VOCs mentioned above, their threshold pressures on activated carbon with different functional groups show a similar variation trend, which is same as the trend of intermolecular interaction energy as is shown in Fig. 6. That is, the greater the intermolecular interaction energy, the greater the affinity between the adsorbate and the adsorbent, and the lower the threshold pressure.
5. Conclusion In this work, effects of activated carbon pore size and functional groups on adsorption capacity of VOCs were studied by GCMC simulation, and the pore size, accessible volume and intermolecular interaction energy were discussed in detail. The main findings and conclusions were introduced as follows: 1) Adsorption capacity of benzene and toluene under saturated vapor pressure is proportional to the pore size in this paper, however, when the pore size increased to a certain value, the adsorption capacity of acetone and methanol decreased.
2) Within the pressure range of 0.1 P/P0, the activated carbon with 0.902 nme1.997 nm pore size has a stronger adsorption capacity for VOCs than that of a larger pore size. As the pressure increases, the capillary condensation occurs so that the amount of adsorption increases rapidly. According to the Chan-Kelvin equation, the pressure at which capillary condensation occurs increases as the pore size increases. 3) The addition of functional groups reduces the accessible volume of activated carbon that follow the sequence NONE > -H > OH > -NH2>-COOH. Due to the capillary condensation under high pressure, its saturated adsorption capacity sequence is consistent with the accessible volume. 4) The addition of functional groups will change the threshold pressure of VOCs adsorption. For benzene and toluene, the addition of functional groups will slightly increase the threshold pressure, but the influence surface is not significant. Moreover, for acetone and methanol, carboxyl and amino groups can lower the threshold pressure to make adsorption easier to release, while hydroxyl and hydrogen have the opposite effect. The main work of this paper is to evaluate the performance of activated carbon with different pore sizes and functional groups for VOC adsorption. In summary, for low concentration VOCs, the
16
Y. An et al. / Chemosphere 227 (2019) 9e16
partial pressure is often less than 0.1 P0 and an activated carbon with a pore size in the range of 0.902e1.997 nm, and for benzene and toluene, carboxyl groups can be selected to achieve an optimum depuration effect. However, the adsorption performance of acetone and methanol will be weakened with the addition of functional groups. Acknowledgements This work is supported by the State Key Laboratory of Chemical Engineering of Tianjin University (No. SKL-ChE-16B05). Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.chemosphere.2019.04.011. References Ahmadpour, A., Do, D.D., 1995. Characterization of modified activated carbons: equilibria and dynamics studies. Carbon 33, 1393e1398. Azalim, S., Brahmi, R., Agunaou, M., Beaurain, A., Giraudon, J.M., Lamonier, J.F., 2013. Washcoating of cordierite honeycomb with CeeZreMn mixed oxides for VOC catalytic oxidation. Chem. Eng. J. 223, 536e546. Blommaerts, N., Dingenen, F., Middelkoop, V., Savelkouls, J., Goemans, M., Tytgat, T., Verbruggen, S.W., Lenaerts, S., 2018. Ultrafast screening of commercial sorbent materials for VOC adsorption using real-time FTIR spectroscopy. Sep. Purif. Technol. 207, 284e290. Cao, D., Wang, W., Jiao, Q., 2001. Monte Carlo simulation of natural gas adsorption storage in pillared layered material. Acta Chim. Sin. 59, 297e300. Chiang, P., Chang, E.E., Chiang, Y., 2001. Effects of surface characteristics of activated carbons on VOC adsorption. J. Environ. Eng. 127, 54e62. Chuang, C.L., Chiang, P.C., Chang, E.E., 2003. Modeling VOCs adsorption onto activated carbon. Chemosphere 53, 17e27. Cohan, L.H., 1938. Sorption hysteresis and the vapor pressure of concave surfaces. J. Am. Chem. Soc. 60, 433e435. gin, D., Mare ^che , J.F., Barre s, O., 2005. HydroCosnier, F., Celzard, A., Furdin, G., Be phobisation of active carbon surface and effect on the adsorption of water. Carbon 43, 2554e2563. Derjaguin, B., 1992. A theory of capillary condensation in the pores of sorbents and of other capillary phenomena taking into account the disjoining action of polymolecular liquid films. Prog. Surf. Sci. 40, 46e61. Derwent, R.G., Jenkin, M.E., Saunders, S.M., Pilling, M.J., Simmonds, P.G., Passant, N.R., Dollard, G.J., Dumitrean, P., Kent, A., 2003. Photochemical ozone formation in north west Europe and its control. Atmos. Environ. 37, 1983e1991. Drisko, G.L., Cao, L., Kimling, M.C., Harrisson, S., Luca, V., Caruso, R.A., 2009. Pore size and volume effects on the incorporation of polymer into macro- and mesoporous zirconium titanium oxide membranes. ACS Appl. Mater. Interfaces 1, 2893e2901. s, Y., Dumont, E., Darracq, G., Couvert, A., Couriol, C., Amrane, A., Thomas, D., Andre Le Cloirec, P., 2011. VOC absorption in a countercurrent packed-bed column using water/silicone oil mixtures: influence of silicone oil volume fraction. Chem. Eng. J. 168, 241e248. Enders, S., Kahl, H., Winkelmann, J., 2007. Surface tension of the ternary system water þ acetone þ toluene. J. Chem. Eng. Data 52, 1072e1079. Fu, Q., Tanaka, H., Miyahara, M.T., Qin, Y., Shen, Y., Zhang, D., 2018. CHF3eCHClF2 binary competitive adsorption equilibria in graphitic slit pores: Monte Carlo simulations and breakthrough curve experiments. Ind. Eng. Chem. Res. 57, 6440e6450. Georgakis, M., Stavropoulos, G., Sakellaropoulos, G.P., 2007. Molecular dynamics study of hydrogen adsorption in carbonaceous microporous materials and the effect of oxygen functional groups. Int. J. Hydrogen Energy 32, 1999e2004. Goralski, C.T.J., Schmidt, L.D.U.O., Brown, W.L.R.E., 1998. Catalytic Incineration of VOC Containing Air Streams at Very Short Contact Times. United States. Hamad, A., Fayed, M.E., 2004. Simulation-aided optimization of volatile organic compounds recovery using condensation. Chem. Eng. Res. Des. 82, 895e906. Jansen, R.J.J., Van Bekkum, H., 1995. XPS of nitrogen-containing functional groups on activated carbon. Carbon 33, 1021e1027. Jun, S., Joo, S.H., Ryoo, R., Kruk, M., Jaroniec, M., Liu, Z., Ohsuna, T., Terasaki, O., 2000.
Synthesis of new, nanoporous carbon with hexagonally ordered mesostructure. J. Am. Chem. Soc. 122, 10712e10713. Li, Y., Niu, J., Shen, Z., Feng, C., 2013. Size effect of single-walled carbon nanotube on adsorption of perfluorooctanesulfonate. Chemosphere 91, 784e790. Liang, X., Chi, J., Yang, Z., 2018. The influence of the functional group on activated carbon for acetone adsorption property by molecular simulation study. Microporous Mesoporous Mater. 262, 77e88. Lim, D., Negreira, A.S., Wilcox, J., 2011. DFT studies on the interaction of defective graphene-supported Fe and Al nanoparticles. J. Phys. Chem. C 115, 8961e8970. Lithoxoos, G.P., Peristeras, L.D., Boulougouris, G.C., Economou, I.G., 2012. Monte Carlo simulation of carbon monoxide, carbon dioxide and methane adsorption on activated carbon. Mol. Phys. 110, 1153e1160. Liu, Y., Wilcox, J., 2012. Effects of surface heterogeneity on the adsorption of CO2 in microporous carbons. Environ. Sci. Technol. 46, 1940e1947. Liu, Z., Horikawa, T., Do, D.D., Nicholson, D., 2012. Packing effects on argon and methanol adsorption inside graphitic cylindrical and slit pores: a GCMC simulation study. J. Colloid Interface Sci. 368, 474e487. ~ o, Y., Díaz, E., Ordo n ~ ez, S., Gallegos-Suarez, E., Guerrero-Ruiz, A., RodríguezPatin Ramos, I., 2015. Adsorption of emerging pollutants on functionalized multiwall carbon nanotubes. Chemosphere 136, 174e180. Patnaik, P., A Comprehensive Guide to the Hazardous Properties of Chemical Substances. Perdew, J.P., Burke, K., Ernzerhof, M., 1996. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865e3868. Rafati, A.A., Hashemianzadeh, S.M., Nojini, Z.B., Naghshineh, N., 2010. Canonical Monte Carlo simulation of adsorption of O2 and N2 mixture on single walled carbon nanotube at different temperatures and pressures. J. Comput. Chem. 31, 1443. Richards, T.W., Carver, E.K., 1921. A critical study of the capillary rise method of determining surface tension, with data for water, benzene, toluene, chloroform, carbon tetrachloride, ether and dimethyl aniline. [second paper.]1. J. Am. Chem. Soc. 43, 827e847. Ryoo, R., Joo, S.H., Jun, S., 1999. Synthesis of highly ordered carbon molecular sieves via template-mediated structural transformation. J. Phys. Chem. B 103, 7743e7746. Sasaki, T., Matsumoto, A., Yamashita, Y., 2008. The effect of the pore size and volume of activated carbon on adsorption efficiency of vapor phase compounds in cigarette smoke. Colloid. Surf. Physicochem. Eng. Asp. 325, 166e172. Shevade, A.V., Jiang, S., Gubbins, K.E., 2000. Molecular simulation study of wateremethanol mixtures in activated carbon pores. J. Chem. Phys. 113, 6933e6942. Shin-ichi, F., Toshihiro, N., Naoki, A., Tomoshige, N., MMasayoshi, N., 2005. A molecular simulation study on adsorption of acetone/water in mesoporous silicas modified by pore surface silylation. J. Chem. Eng. Jpn. 38, 999e1007. Song, W., Tondeur, D., Luo, L., Li, J., 2005. VOC adsorption in circulating gas fluidized bed. Adsorption 11, 853e858. Sun, H., 1998. COMPASS: an ab initio force-field optimized for condensed-phase Applicationss-overview with details on alkane and benzene compounds. J. Phys. Chem. B 102, 7338e7364. Sun, H., Ren, P., Fried, J.R., 1998. The COMPASS force field: parameterization and validation for phosphazenes. Comput. Theor. Polym. Sci. 8, 229e246. Szcze˛ sniak, B., Choma, J., Jaroniec, M., 2018. Effect of graphene oxide on the adsorption properties of ordered mesoporous carbons toward H2, C6H6, CH4 and CO2. Microporous Mesoporous Mater. 261, 105e110. Tamon, H., Okazaki, M., 1996. Influence of acidic surface oxides of activated carbon on gas adsorption characteristics. Carbon 34, 741e746. Tao, Y., Xue, Q., Liu, Z., Shan, M., Ling, C., Wu, T., Li, X., 2014. Tunable hydrogen separation in porous graphene membrane: first-principle and molecular dynamic simulation. ACS Appl. Mater. Interfaces 6, 8048e8058. Waldman, M., Hagler, A.T., 1993. New combining rules for rare gas van der waals parameters. J. Comput. Chem. 14, 1077e1084. Wang, B., Nisar, J., Ahuja, R., 2012. Molecular simulation for gas adsorption at NiO (100) surface. ACS Appl. Mater. Interfaces 4, 5691e5697. Wang, S., 2007. Comparative molecular simulation study of methane adsorption in MetalOrganic frameworks. Energy Fuel 21, 953e956. Wang, X., Ma, C., Xiao, J., Xia, Q., Wu, J., Li, Z., 2018. Benzene/toluene/water vapor adsorption and selectivity of novel C-PDA adsorbents with high uptakes of benzene and toluene. Chem. Eng. J. 335, 970e978. Zeng, Y., Phadungbut, P., Do, D.D., Nicholson, D., 2014. Anatomy of adsorption in open-end and closed-end slit mesopores: adsorption, desorption, and equilibrium branches of hysteresis loop. J. Phys. Chem. C 118, 25496e25504. Zhang, X., Gao, B., Creamer, A.E., Cao, C., Li, Y., 2017. Adsorption of VOCs onto engineered carbon materials: a review. J. Hazard Mater. 338, 102e123.
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Performance evaluation of activated carbon with different pore sizes and functional groups for VOC adsorption by molecular simulation Yaxiong An, Qiang Fu, Donghui Zhang*, Yayan Wang, Zhongli Tang The Research Center of Chemical Engineering, School of Chemical Engineering and Technology, Tianjin University, Tianjin, 300072, China
h i g h l i g h t s Adsorption of VOC investigated by GCMC. Effects of different pore sizes and functional groups of activated carbon. The pressure of capillary condensation is highly depend on the pore size. Functional groups affect intermolecular interactions and thus affect threshold pressure.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 3 January 2019 Received in revised form 28 March 2019 Accepted 2 April 2019 Available online 5 April 2019
Volatile organic compounds (VOCs) are growing pollutants now that cause air pollution and threaten human health. In this paper, the Grand Canonical Monte Carlo was used to simulate the adsorption performance of activated carbon on VOCs (benzene, toluene, acetone and methanol). After simulating different pore sizes (0.902 nm,1.997 nm,3 nm and 4 nm) adsorption performances of activated carbon, activated carbon with a pore size of 1.997 nm was selected to further study the influence of functional groups (carboxyl, amino, hydroxyl and hydrogen), and the capillary condensation was explained by the Kelvin equation. Furthermore, effects of functional groups under saturated vapor pressure (P0) of VOCs that range from 0 to 0.1 P0 were explained by the accessible volume and intermolecular interaction potential, respectively. Under pressure range of 0e0.1 P0, at the beginning of adsorption of acetone and methanol, carboxyl and amino groups could reduce the threshold pressure while hydroxyl and hydrogen have the opposite effect. For benzene and toluene, all functional groups have little effect on the threshold pressure, and they reduce the adsorption capacity instead. It could be concluded that the activated carbon could achieve the best adsorption effect on acetone and methanol, on the contrary, the addition of functional groups on benzene and toluene will weaken their adsorption performance. © 2019 Elsevier Ltd. All rights reserved.
Handling Editor: Dr. R Ebinghaus Keywords: Volatile organic compounds (VOCs) adsorption Activated carbon Grand canonical Monte Carlo (GCMC) Pore size Functional group
1. Introduction Volatile organic compounds (VOCs), which consist of many complex and hazardous chemicals, are the major organic compounds involved in atmospheric photochemical reactions and they are also the main precursors for the formation of ozone (O3) and fine particulate matter (Derwent et al., 2003). Now all over the world, air pollution problem has not only posed a serious threat to living environment and human health, but also restricted the development of the sustainable economy. Therefore, it is the time that advanced technologies should be applied to reduce VOC
* Corresponding author. E-mail address: [email protected] (D. Zhang). https://doi.org/10.1016/j.chemosphere.2019.04.011 0045-6535/© 2019 Elsevier Ltd. All rights reserved.
emissions and prevent the occurrence of smog at the source in order to improve the air quality effectively and enhance people's happiness. Until now, a variety of VOCs treatment technologies have emerged, such as incineration (Goralski et al., 1998), condensation (Hamad and Fayed, 2004), biodegradation, absorption (Dumont et al., 2011), catalytic oxidation (Azalim et al., 2013) and adsorption (Chiang et al., 2001; Song et al., 2005; Blommaerts et al., 2018). Among these methods, the adsorption technology is a highly effective and economical method, which can achieve the recovery of VOCs and recycle adsorbents. In the adsorption process, the adsorbent's performance plays an important role in overall capital and operating costs, and therefore the choice of suitable adsorbent is crucial for gas separation and recovery. Because of its large specific surface area, rich pore structure and excellent adsorption
10
Y. An et al. / Chemosphere 227 (2019) 9e16
capacity, activated carbon is widely used in gas purification, especially in the treatment and recovery of VOCs(Chiang et al., 2001; Chuang et al., 2003; Cosnier et al., 2005; Zhang et al., 2017). Sasaki T (Sasaki et al., 2008), Szcze˛ sniak B (Szcze˛ sniak et al., 2018) and Do DD (Ahmadpour and Do, 1995) et al. investigated the effect of pore structure of activated carbon on the adsorption performance of VOCs through experimental methods. However, it is difficult to get accurate adsorption data by experiments at normal temperature, because most VOCs, which have a high boiling point, are liquid at normal temperature and pressure. Additionally, the pore structure of activated carbon is complicated, although ordered mesoporous carbon can be obtained (Ryoo et al., 1999; Jun et al., 2000), the variables such as pore size and pore volume are difficult to control and adjust. On the surface of activated carbon, there are different levels of nitrogen-oxygen functional groups and different functional groups in activated carbon also have a significant effect on adsorption performance of VOCs(Jansen and Van Bekkum, 1995; ~ o et al., 2015), which mainly include Tamon and Okazaki, 1996; Patin eight oxygen-containing functional groups and two kinds of nitrogen-containing functional groups. It is also extremely difficult to examine the effects of specific functional groups by experimental methods. Molecular simulation can solve all these problems well and can be used to investigate the adsorption data of VOCs with arbitrary functional groups and activated carbon with arbitrary pore at any pressure and temperature. Nowadays, the molecular simulation is gradually applied in various fields. It can not only solve the above problems by establishing an idealized model, but also get the adsorption data that is difficult to obtain by experiments, and further explain the adsorption phenomenon at the molecular level. Among the simulation calculations reported so far, the most suitable method for studying the gas adsorption behavior of activated carbon is the Grand Canonical Monte Carlo (GCMC) simulation (Li et al., 2013). In some special cases, molecular dynamics is also used to study practical problems. CAO D (Cao et al., 2001) established Steele 10-43 slit sheet model for Monte Carlo simulation of layered microporous material adsorption storage of natural gas. The slit sheet model of activated carbon has been established in our previous work to explore the effect of the pore size of nanoporous carbon materials on the separation performance of CHF3-CHClF2 mixtures and the result has been verified by experiments (Fu et al., 2018). These studies have verified that slit sheet model used in gas adsorption is accurate and reliable. In the study of VOCs adsorption, Do DD (Liu et al., 2012) established carbon nanotube model and slit sheet model to investigate the effects of pore structure on methanol adsorption, but the influence of functional groups on activated carbon was not considered in his study. Liang X (Liang et al., 2018) studied the effects of the oxygen-containing functional groups of activated carbon on the adsorption of acetone at low pressure, but it neglected the effect of the addition of functional groups on the accessible volume. At present, there is no systematic study on the effects of activated carbon pore size and different kinds of functional groups on the adsorption performance of many kinds of VOCs with different structure and properties. In this paper, four representative VOCs (benzene, toluene, acetone and methanol) were selected to study their adsorption performance on activated carbon by establishing a slit sheet activated carbon model. GCMC simulation method was used to investigate the effects of pore size of activated carbon and four functional groups (carboxyl, amino, hydroxyl and hydrogen) on the adsorption performance of different VOCs. The change of capillary condensation pressure with pore size and the effects of functional groups on the accessible volume of activated carbon were discussed in detail. Furthermore, in order to select suitable pore size and functional
group for adsorption of specific VOC, the effects of intermolecular interaction potential on adsorption capacity of VOCs with different properties at 0-0.1P0 partial pressure were also analyzed in a systemic way. 2. Theory 2.1. Molecular model The activated carbon model used in GCMC simulation was simplified into a slit pore model formed by parallel graphite sheets. In recent years, single-layer graphite sheets or three-layer graphite sheets models have been widely used in the simulation of activated carbon according to a number of previous studies (Shevade et al., 2000; Georgakis et al., 2007; Lithoxoos et al., 2012), and what was adopted in this work is a three-layer graphite sheet model. Fig. 1 is a schematic diagram of slit pore model of activated carbon. As is shown in Fig. 1, two graphene layers had an interplanar spacing (d) of 0.34 nm (Fu et al., 2018), which is the van der Waals diameter of carbon atoms. For easier calculation, we chose an activated carbon unit made up of three-layer graphite sheets, where all the periodic boundary is applied in the x-, y- and z-directions, respectively. The specific length of a single-layer graphene in the x-direction is 2.952 nm, which in the y-direction is 2.982 nm. Various distances (D ¼ 1.242, 2.337, 3.34 and 4.34 nm) of graphitic sheet were used to investigate the effect of the activated carbon pore size. The slit pore width (D) is the distance between the centers of carbon atoms in innermost graphite sheets, and the effective pore size (Deff) could be calculated by:
Deff ¼ D 0:34nm
(1)
In order to study the effect of functional groups on the adsorption properties of adsorbed VOCs, four functional groups including hydrogen (He), hydroxyl (OHe), amine (NH2e) and carboxyl (COOHe) were considered. We set the functional group density to 1.73 mmol g1 so as to compare the effects of different functional groups well. The activated carbon models are described in Fig. 1. VOCs (benzene, toluene, acetone, and methanol) are built through the Dmol3 module and optimized for energy minimization. In the optimization project, the basis set selected the Double value orbit and orbital polarization function (DNP), and the
Fig. 1. Slit pore model of activated carbonColor code: gray, C; whit, H; red, O; blue, N. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Y. An et al. / Chemosphere 227 (2019) 9e16
calculation method selected the generalized gradient approximation (GGA) with Perdew-Burke-Ernzerhof (PBE) function (Perdew et al., 1996; Wang et al., 2012) which has high precision in describing long-range forces such as van der Waals interactions and it is widely used to describe the interaction between gas and graphite (Lim et al., 2011). The optimized models are shown in Fig S1.
3 1=2 0 2 εi εj r i r 0j εij ¼ 6 6 r 0i þ r 0j
11
(4)
3. Result 2.2. Grand canonical Monte Carlo simulation A GCMC simulation method, one of the most basic tools, was used to emulate specific adsorption behavior of VOCs on activated carbon. Condensed-phase Optimized Molecular Potentials for Atomistic Simulation Studies (COMPASS) force field (Sun, 1998; Sun et al., 1998; Tao et al., 2014), the first ab initio field, was selected in this simulation to calculate the interaction potential between gasand solid-phase molecules. In order to get more accurate prediction about molecular properties and condensation systems, some empirical corrections were widely used in molecules and condensed phase systems, such as critical state parameters and van der Waals parameters. Besides, the temperature adopted for all simulations in this work was 293.15 K and the fixed pressure task was chosen to calculate the equilibration in the range of the saturated vapor pressure (P0) of the adsorbates. In the Monte Carlo simulation process, 107 Monte Carlo steps were used for simulation. In order to reduce the error caused by the initial configuration and improve the accuracy of the simulation, the entire simulation process was divided into two equal parts, the equilibration steps and the production steps. The Equilibration steps were used for balancing and were discarded, and then the adsorption capacity were calculated on production steps. In the simulation process, the adsorbates were regarded as rigid spheres in the GCMC method and one of creating, transferring, rotating and deleting was selected in a Monte Carlo step, as long as the energy reached a lowest value in the system, then the step will be accepted. Finally, the adsorption capacities with different pressure were obtained through a series of simulations at a specific temperature of 293.15 K, and it is the adsorption isotherm that is at 293.15 K. 2.3. Isosteric heat Intermolecular interaction energy is an important evaluation criterion for adsorption properties, isosteric heat E is displayed by the following equation:
2 !9 !6 3 Xqi qj X r 0ij r 0ij 5 E¼ þ εij 42 3 rij rij rij ij
(2)
ij
The first section is the Coulomb equation, which calculates the electrostatic interaction force between the intermolecular i and j charge groups. The second section is the 9-6 Lennard-Jones (LJ) potential model that represents the intermolecular non-bonding effect and the van der Waals interaction force. q, rij represents the charge of the adsorption site and the distance between the two adsorption sites, respectively. ε is the potential well depth and r0 is the hard sphere radius of the atom. Moreover, the 9-6 LJ parameters of different atoms are determined by using the following equations (Waldman and Hagler, 1993):
0 6 11=6 6 r 0i þ r 0j B C r 0ij ¼ @ A 2
(3)
GCMC method was used to estimate the capacity of VOCs adsorbed by the activated carbon with different pore size. After that, the effects of four functional groups (eCOOH, -OH, -NH2 and eH) on the adsorption properties of benzene, toluene, acetone and methanol were investigated. 3.1. Adsorptions in activated carbon with different pore sizes Fig. 2 shows the adsorption isotherms of the VOCs in activated carbon with various pore sizes (0.902, 1.977, 3.000 and 4.000 nm) at the temperature of 293.15 K and the pressure of the saturated vapor pressure (P0) of the adsorbate. These adsorption data were all modeled independently through GCMC simulations and the total uptake represents the total atom number of gas adsorbed per gram (including gas and condense phase). Fig. 2aed shows the adsorption isotherms of benzene, toluene, acetone and methanol in activated carbon respectively. Obviously, the pore size has a significant effect on the adsorption isotherms of VOCs, and the effects of different VOCs are diverse from each other. For benzene and toluene, the adsorption isotherms in the activated carbon are quite similar. The total capacity was the highest in the graphitic silt pores with effective pore size of 4.000 nm and the total capacity of benzene and toluene follows the sequence of 4.000 > 3.000>1.997 > 0.902 nm under the saturated vapor pressure (P0) of benzene and toluene respectively, which is consistent with the experimental results of Wang X (Wang et al., 2018). However, for acetone and methanol, there is a different law for the changes of total uptake at the pressure of P0 and the order of total uptake becomes 3.000 > 1.997>4.000 > 0.902 nm and 1.997 > 3.000>4.000 > 0.902 nm respectively. It is obvious that there is an optimal pore size for the total uptake of VOCs under the saturated vapor pressure. When the activated carbon pore size is 0.902 nm, the adsorption isotherms of VOCs in Fig. 2 all exhibit type-I Langmuir adsorption behavior, which is a typical characteristic of microporous materials. However, as the pore size increases, the capillary condensation gradually appears and the type-I Langmuir adsorption behavior gradually transforms into type-IV Langmuir adsorption behavior. The adsorption isotherm and molecular size shown in Figs. 2 and S1 indicate that the capillary condensation occurs when the pore size is about four times the molecular diameter (Rafati et al., 2010). Easy to know, with excessive pore size or too small saturated vapor pressure, the capillary condensation is prohibited. 3.2. Adsorptions in activated carbon with functional groups The IDLH (Immediately Dangerous to Life or Health concentration) of benzene, toluene, acetone and methanol are 500, 500, 2500 and 6000 ppm from National Institute of Occupational Safety and Health (Patnaik). Therefore, it is important for adsorbents to adsorb VOCs effectively at lower partial pressure ranging from 0 to 0.1 P0. According to the adsorption isotherm in Fig. 2, activated carbon with pore size of 1.997 nm has better adsorption performance on VOCs at low partial pressure (less than 0.1 P0). Hence, the 1.997 nm activated carbon slit model was chosen in this study and was further used to investigate the effects of functional groups (eCOOH,
12
Y. An et al. / Chemosphere 227 (2019) 9e16
Fig. 2. Adsorption isotherms of VOCs (~ad: benzene, toluene, acetone and methanol) in activated carbon with different pore sizes: Deff ¼ 0.902, 1.977, 3.000 and 4.000 nm.
-OH, -NH2 and eH) on the adsorption of VOCs. 3.2.1. Adsorption of benzene Fig. 3a shows the adsorption isotherm of benzene and the data at low pressure expressed by logarithm, and the type-I Langmuir adsorption behavior is exhibited by all the adsorption isotherm. When the pressure is less than 0.1 P0, the adsorption data is difficult to obtain by experiment, but we can get it by simulation and represent it by logarithm to discover the subtle differences. In order to describe the effect of the functional group on the adsorption of VOCs under low pressure better, a threshold pressure is introduced, which represents the pressure at which the adsorbates begin to be adsorbed and the adsorbates begin to enter the adsorbent. Obviously, the functional group has little effect on the threshold pressure in the adsorption of benzene. All activated carbons with functional groups have reduced adsorption capacity of VOCs in the order: NONE > -H > -OH > -NH2>-COOH. 3.2.2. Adsorption of toluene As shown in Fig. 3b, the effect of functional groups on the adsorption performance of toluene is very similar to that of benzene. All the adsorption isotherms exhibit type-I Langmuir adsorption behavior and the functional groups only reduce the adsorption capacity but have minimal effect on the threshold
pressure. The adsorption capacity of toluene reduces in the order of NONE > -H > -OH > -NH2>-COOH for the pressure between 0-1 P0. 3.2.3. Adsorption of acetone In Fig. 3c, the adsorption isotherms of acetone are presented and the respective threshold pressure in the range of 0e0.01 P0, which modified by different functional groups, are depicted independently in a small graph. It is apparent that the addition of functional groups reduced the adsorption capacity but changed the threshold pressure. As can be seen clearly from the figure, carboxyl and amino decrease the threshold pressure to 1e7 P0 and 1e6 P0 respectively, whereas the hydrogen and hydroxyl increase the threshold pressure. Another important finding is that the total acetone uptake at a high pressure follows the sequence of NONE > -H > -OH z -NH2>COOH, but at a low pressure (0e0.001 P0) the sequence changes into eCOOH > -NH2>NONE > OH > -H. Attention should be paid that there are two intersections between the carboxyl, amino and NONE near 0.001 P0. In summary, the threshold pressure is increased when adding carboxyl, amino and decreased when adding hydrogen, hydroxyl. 3.2.4. Adsorption of methanol For methanol, the functional group has the strongest influence on its threshold pressure for the VOCs selected in this paper. When
Y. An et al. / Chemosphere 227 (2019) 9e16
13
~d: benzene, toluene, acetone and methanol) in activated carbon with different functional groups (eCOOH, -OH, -NH2 and eH). Fig. 3. Adsorption isotherms of VOCs (a
the pressure is less than 0.01 P0, only the activated carbon with carboxyl and amino can uptake methanol and the threshold pressures are 1e6 and 0.001 P0, respectively. At high pressure (0.1e1 P0), the total uptake of methanol in NONE is obviously higher than the uptake in activated carbon with functional groups. At the low pressure (0.01e0.1 P0), the methanol uptake follows the sequence -NH2>-COOH > NONE > -OH > -H. 4. Discussion To provide unique insights into the intrinsic nature of the pore size and surface functional groups of activated carbon for the adsorption of VOCs, the effects of (1) capillary condensation, (2) accessible volume and (3) adsorbate-adsorbent interaction energy are discussed in this section. 4.1. Capillary condensation From section 3, it is obvious that the effect of pore size on the adsorption of VOCs is significant, which mainly reflects in the phenomenon of capillary condensation. For all VOCs used in this work, the pressure (P/P0) of the capillary condensation increases as the pore size increases. The Chan-Kelvin equation (Cohan, 1938) is widely used to describe capillary condensation phenomena which is also applies to the slit pore model. The mechanism of condensation in slit pore model has been expounded by Zeng Y (Zeng et al., 2014), whose studies shows that the adsorption layer on the surface is undulating, and when the adsorption layer reaches a certain thickness, the convex portions are joined together to form a
cylindrical surface. However, when the pores are nanometer-sized, the Chan-Kelvin equation will no longer be applicable because the solid-liquid potential cannot be ignored at the nanometer size. The Derjaguin and Broekhoffde Boer (DBdB) models (Derjaguin, 1992) have a solid-liquid potential (ESL) added to the Cohan-Kelvin equation, which better describes the capillary condensation in nanopore adsorption. The DBdB equation can be written as
RT ln
P P0
syM
¼
r
þ ESL
(5)
where R is the gas constant; T is the temperature; P is the condensation pressure; P0 is the saturation vapor pressure; s is the surface tension; vM is the liquid molar volume; r is the radius of the condensed phase hemispherical surface and ESL is the solid-liquid potential. In Table S1 (Richards and Carver, 1921; Enders et al., 2007), we list the surface parameters of VOCs in DBdB equation. Furthermore, we choose benzene as a representative for further analysis and discussion of capillary condensation of VOCs. From the DBdB equation we can easily conclude that as the pore size increases, the pressure P/P0 at which capillary condensation occurs increases and the rate of increase is proportional to the value of svM. However, the pore size at which capillary condensation occurs is smaller than the actual pore diameter Deff. The r in the DBdB equation is defined as (D2eff t ) where t is an adsorbates “statistical” film thickness. Fig S2 shows the adsorption film of benzene at a pore size of 3 nm before capillary condensation. The adsorption of the molecules on the surface of the adsorbent before the capillary condensation occurs is equivalent to reducing the pore
14
Y. An et al. / Chemosphere 227 (2019) 9e16
size to create a condition for capillary condensation. Therefore, as the number of adsorbed layers increases and the pressure rises, once the capillary condensation condition is reached, the adsorption capacity rapidly increases. As explained above, when describing capillary condensation, the ESL is out of consideration in Chan-Kelvin equation, which means it does not consider the role of the adsorbent. To explain the effect of ESL, we plot PTln(P0/P) versus the inverse of the radius of curvature at the point just after condensation in Fig. 4. It can be obviously seen from Fig. 4 that the simulated value is slightly higher than the value fitted by Chan-Kelvin equation. The difference between these two values is ESL is positively correlated with the inverse of the radius of curvature. This positively correlated variation trend is consistent with the 9-6 LJ potential. In a word, the larger the pore size, the larger the distance between the condensed phase and the adsorbent, and the smaller the ESL. 4.2. Accessible volume As shown in Section 3.2, under the saturated vapor pressure of VOCs, the adsorption capacity all follows the sequence NONE > H > -OH > -NH2>-COOH. Section 4.1 explained the capillary condensation of VOCs. Then, the hypothesis that the adsorption capacity of VOCs under saturation vapor pressure is mainly limited by the accessible volume was proposed (Wang, 2007; Drisko et al., 2009). Liu Y and Wilcox J (Liu and Wilcox, 2012) calculated the accessible volume of the graphite slit-pore by the adsorption capacity of helium. Helium is a non-adsorbed or weakly adsorbed gas that is often used to measure pore volume in experiments or simulations. The excess adsorption (Ne) amount of helium is represented by the following equation:
Ne ¼ Nt V$rbulk
(6)
Where Nt is the number of helium molecular loaded in the pore of unit cell (molecular/unit cell), V is the accessible volume of the unit cell we use, rbulk is the bulk density of helium at a specific temperature and pressure. Since helium is hardly adsorbed on activated carbon, Ne ¼ 0, equation (1) can be expressed as: Nt ¼ V rbulk .At low pressure, helium can be regarded as an ideal gas. So Nt ¼ VM RT P where M is the molar mass of helium. Molecular simulation was performed to obtain a fitting curve of the adsorption
Fig. 4. Chan-Kelvin equation and simulated values of capillary condensation during adsorption.
capacity with pressure. Fig S3 presents the specific adsorption capacity of activated carbon with no functional group and that with different functional groups. Moreover, the accessible volume of each activated carbon was able to be estimated by the slope of each curve showed in Fig S3. The estimation results of accessible volume of different activated carbon are shown in Fig. 5. After adding different functional groups, it is obvious that the accessible volume is significantly reduced. For example, the addition of carboxyl groups reduces the accessible volume of activated carbon by 13.36%. From the data in Fig. 5, it is apparent that the accessible volume of activated carbon with different functional groups follows the sequence NONE > -H > -OH > -NH2>-COOH, which is consistent with the adsorption capacity under saturated vapor pressure. 4.3. Intermolecular interaction energy Intermolecular interaction energy indicates the affinity between the adsorbate and the adsorbent (Shin-ichi et al., 2005; Liang et al., 2018), and it is an important criterion for evaluation of adsorption performance. In order to investigate the effect of functional groups on the adsorption of VOCs, the intermolecular interaction energy curves were researched. As mentioned earlier in Section 3, the addition of functional groups of activated carbon reduced the adsorption capacity of VOCs under saturated vapor pressure and also affected the threshold pressure of different VOCs. This phenomenon can be well explained by Fig. 6, which shows an intermolecular interaction energy of adsorption for VOCs in activated carbon with different functional groups. From the figure, it can be seen that for benzene and toluene, the intermolecular interaction energies were both decreased with the addition of functional groups; however, for acetone, methanol, carboxyl and amino, the intermolecular interaction energy enhanced, especially at low pressure. But hydrogen and hydroxyl groups played an opposite effect on it. In addition, it is noteworthy that the intermolecular interaction energy drops sharply in the range of 0.01e0.1 P/P0 for all VOCs, especially for benzene and toluene. A possible explanation for this tendency might be the capillary condensation, which makes the heat of vaporization play an important role in the intermolecular interaction energy. Under high pressure, the difference of intermolecular interaction energy between different VOCs is significantly reduced. At this point, the difference in adsorption capacity of different VOCs was no longer primarily affected by intermolecular interaction energy, but was mainly restricted by the accessible volume, which has been
Fig. 5. Comparison of accessible volume of activated carbon with different functional groups.
Y. An et al. / Chemosphere 227 (2019) 9e16
15
Fig. 6. Intermolecular interaction energy of adsorption for VOCs (~ad: benzene, toluene, acetone and methanol) in activated carbon with different functional groups.
elaborated in detail in Section 4.2. In summary, under the pressure range of 0e0.1 P0, the threshold pressure is mainly affected by the intermolecular interaction energy. For the four VOCs mentioned above, their threshold pressures on activated carbon with different functional groups show a similar variation trend, which is same as the trend of intermolecular interaction energy as is shown in Fig. 6. That is, the greater the intermolecular interaction energy, the greater the affinity between the adsorbate and the adsorbent, and the lower the threshold pressure.
5. Conclusion In this work, effects of activated carbon pore size and functional groups on adsorption capacity of VOCs were studied by GCMC simulation, and the pore size, accessible volume and intermolecular interaction energy were discussed in detail. The main findings and conclusions were introduced as follows: 1) Adsorption capacity of benzene and toluene under saturated vapor pressure is proportional to the pore size in this paper, however, when the pore size increased to a certain value, the adsorption capacity of acetone and methanol decreased.
2) Within the pressure range of 0.1 P/P0, the activated carbon with 0.902 nme1.997 nm pore size has a stronger adsorption capacity for VOCs than that of a larger pore size. As the pressure increases, the capillary condensation occurs so that the amount of adsorption increases rapidly. According to the Chan-Kelvin equation, the pressure at which capillary condensation occurs increases as the pore size increases. 3) The addition of functional groups reduces the accessible volume of activated carbon that follow the sequence NONE > -H > OH > -NH2>-COOH. Due to the capillary condensation under high pressure, its saturated adsorption capacity sequence is consistent with the accessible volume. 4) The addition of functional groups will change the threshold pressure of VOCs adsorption. For benzene and toluene, the addition of functional groups will slightly increase the threshold pressure, but the influence surface is not significant. Moreover, for acetone and methanol, carboxyl and amino groups can lower the threshold pressure to make adsorption easier to release, while hydroxyl and hydrogen have the opposite effect. The main work of this paper is to evaluate the performance of activated carbon with different pore sizes and functional groups for VOC adsorption. In summary, for low concentration VOCs, the
16
Y. An et al. / Chemosphere 227 (2019) 9e16
partial pressure is often less than 0.1 P0 and an activated carbon with a pore size in the range of 0.902e1.997 nm, and for benzene and toluene, carboxyl groups can be selected to achieve an optimum depuration effect. However, the adsorption performance of acetone and methanol will be weakened with the addition of functional groups. Acknowledgements This work is supported by the State Key Laboratory of Chemical Engineering of Tianjin University (No. SKL-ChE-16B05). Appendix A. Supplementary data Supplementary data related to this article can be found at https://doi.org/10.1016/j.chemosphere.2019.04.011. References Ahmadpour, A., Do, D.D., 1995. Characterization of modified activated carbons: equilibria and dynamics studies. Carbon 33, 1393e1398. Azalim, S., Brahmi, R., Agunaou, M., Beaurain, A., Giraudon, J.M., Lamonier, J.F., 2013. Washcoating of cordierite honeycomb with CeeZreMn mixed oxides for VOC catalytic oxidation. Chem. Eng. J. 223, 536e546. Blommaerts, N., Dingenen, F., Middelkoop, V., Savelkouls, J., Goemans, M., Tytgat, T., Verbruggen, S.W., Lenaerts, S., 2018. Ultrafast screening of commercial sorbent materials for VOC adsorption using real-time FTIR spectroscopy. Sep. Purif. Technol. 207, 284e290. Cao, D., Wang, W., Jiao, Q., 2001. Monte Carlo simulation of natural gas adsorption storage in pillared layered material. Acta Chim. Sin. 59, 297e300. Chiang, P., Chang, E.E., Chiang, Y., 2001. Effects of surface characteristics of activated carbons on VOC adsorption. J. Environ. Eng. 127, 54e62. Chuang, C.L., Chiang, P.C., Chang, E.E., 2003. Modeling VOCs adsorption onto activated carbon. Chemosphere 53, 17e27. Cohan, L.H., 1938. Sorption hysteresis and the vapor pressure of concave surfaces. J. Am. Chem. Soc. 60, 433e435. gin, D., Mare ^che , J.F., Barre s, O., 2005. HydroCosnier, F., Celzard, A., Furdin, G., Be phobisation of active carbon surface and effect on the adsorption of water. Carbon 43, 2554e2563. Derjaguin, B., 1992. A theory of capillary condensation in the pores of sorbents and of other capillary phenomena taking into account the disjoining action of polymolecular liquid films. Prog. Surf. Sci. 40, 46e61. Derwent, R.G., Jenkin, M.E., Saunders, S.M., Pilling, M.J., Simmonds, P.G., Passant, N.R., Dollard, G.J., Dumitrean, P., Kent, A., 2003. Photochemical ozone formation in north west Europe and its control. Atmos. Environ. 37, 1983e1991. Drisko, G.L., Cao, L., Kimling, M.C., Harrisson, S., Luca, V., Caruso, R.A., 2009. Pore size and volume effects on the incorporation of polymer into macro- and mesoporous zirconium titanium oxide membranes. ACS Appl. Mater. Interfaces 1, 2893e2901. s, Y., Dumont, E., Darracq, G., Couvert, A., Couriol, C., Amrane, A., Thomas, D., Andre Le Cloirec, P., 2011. VOC absorption in a countercurrent packed-bed column using water/silicone oil mixtures: influence of silicone oil volume fraction. Chem. Eng. J. 168, 241e248. Enders, S., Kahl, H., Winkelmann, J., 2007. Surface tension of the ternary system water þ acetone þ toluene. J. Chem. Eng. Data 52, 1072e1079. Fu, Q., Tanaka, H., Miyahara, M.T., Qin, Y., Shen, Y., Zhang, D., 2018. CHF3eCHClF2 binary competitive adsorption equilibria in graphitic slit pores: Monte Carlo simulations and breakthrough curve experiments. Ind. Eng. Chem. Res. 57, 6440e6450. Georgakis, M., Stavropoulos, G., Sakellaropoulos, G.P., 2007. Molecular dynamics study of hydrogen adsorption in carbonaceous microporous materials and the effect of oxygen functional groups. Int. J. Hydrogen Energy 32, 1999e2004. Goralski, C.T.J., Schmidt, L.D.U.O., Brown, W.L.R.E., 1998. Catalytic Incineration of VOC Containing Air Streams at Very Short Contact Times. United States. Hamad, A., Fayed, M.E., 2004. Simulation-aided optimization of volatile organic compounds recovery using condensation. Chem. Eng. Res. Des. 82, 895e906. Jansen, R.J.J., Van Bekkum, H., 1995. XPS of nitrogen-containing functional groups on activated carbon. Carbon 33, 1021e1027. Jun, S., Joo, S.H., Ryoo, R., Kruk, M., Jaroniec, M., Liu, Z., Ohsuna, T., Terasaki, O., 2000.
Synthesis of new, nanoporous carbon with hexagonally ordered mesostructure. J. Am. Chem. Soc. 122, 10712e10713. Li, Y., Niu, J., Shen, Z., Feng, C., 2013. Size effect of single-walled carbon nanotube on adsorption of perfluorooctanesulfonate. Chemosphere 91, 784e790. Liang, X., Chi, J., Yang, Z., 2018. The influence of the functional group on activated carbon for acetone adsorption property by molecular simulation study. Microporous Mesoporous Mater. 262, 77e88. Lim, D., Negreira, A.S., Wilcox, J., 2011. DFT studies on the interaction of defective graphene-supported Fe and Al nanoparticles. J. Phys. Chem. C 115, 8961e8970. Lithoxoos, G.P., Peristeras, L.D., Boulougouris, G.C., Economou, I.G., 2012. Monte Carlo simulation of carbon monoxide, carbon dioxide and methane adsorption on activated carbon. Mol. Phys. 110, 1153e1160. Liu, Y., Wilcox, J., 2012. Effects of surface heterogeneity on the adsorption of CO2 in microporous carbons. Environ. Sci. Technol. 46, 1940e1947. Liu, Z., Horikawa, T., Do, D.D., Nicholson, D., 2012. Packing effects on argon and methanol adsorption inside graphitic cylindrical and slit pores: a GCMC simulation study. J. Colloid Interface Sci. 368, 474e487. ~ o, Y., Díaz, E., Ordo n ~ ez, S., Gallegos-Suarez, E., Guerrero-Ruiz, A., RodríguezPatin Ramos, I., 2015. Adsorption of emerging pollutants on functionalized multiwall carbon nanotubes. Chemosphere 136, 174e180. Patnaik, P., A Comprehensive Guide to the Hazardous Properties of Chemical Substances. Perdew, J.P., Burke, K., Ernzerhof, M., 1996. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865e3868. Rafati, A.A., Hashemianzadeh, S.M., Nojini, Z.B., Naghshineh, N., 2010. Canonical Monte Carlo simulation of adsorption of O2 and N2 mixture on single walled carbon nanotube at different temperatures and pressures. J. Comput. Chem. 31, 1443. Richards, T.W., Carver, E.K., 1921. A critical study of the capillary rise method of determining surface tension, with data for water, benzene, toluene, chloroform, carbon tetrachloride, ether and dimethyl aniline. [second paper.]1. J. Am. Chem. Soc. 43, 827e847. Ryoo, R., Joo, S.H., Jun, S., 1999. Synthesis of highly ordered carbon molecular sieves via template-mediated structural transformation. J. Phys. Chem. B 103, 7743e7746. Sasaki, T., Matsumoto, A., Yamashita, Y., 2008. The effect of the pore size and volume of activated carbon on adsorption efficiency of vapor phase compounds in cigarette smoke. Colloid. Surf. Physicochem. Eng. Asp. 325, 166e172. Shevade, A.V., Jiang, S., Gubbins, K.E., 2000. Molecular simulation study of wateremethanol mixtures in activated carbon pores. J. Chem. Phys. 113, 6933e6942. Shin-ichi, F., Toshihiro, N., Naoki, A., Tomoshige, N., MMasayoshi, N., 2005. A molecular simulation study on adsorption of acetone/water in mesoporous silicas modified by pore surface silylation. J. Chem. Eng. Jpn. 38, 999e1007. Song, W., Tondeur, D., Luo, L., Li, J., 2005. VOC adsorption in circulating gas fluidized bed. Adsorption 11, 853e858. Sun, H., 1998. COMPASS: an ab initio force-field optimized for condensed-phase Applicationss-overview with details on alkane and benzene compounds. J. Phys. Chem. B 102, 7338e7364. Sun, H., Ren, P., Fried, J.R., 1998. The COMPASS force field: parameterization and validation for phosphazenes. Comput. Theor. Polym. Sci. 8, 229e246. Szcze˛ sniak, B., Choma, J., Jaroniec, M., 2018. Effect of graphene oxide on the adsorption properties of ordered mesoporous carbons toward H2, C6H6, CH4 and CO2. Microporous Mesoporous Mater. 261, 105e110. Tamon, H., Okazaki, M., 1996. Influence of acidic surface oxides of activated carbon on gas adsorption characteristics. Carbon 34, 741e746. Tao, Y., Xue, Q., Liu, Z., Shan, M., Ling, C., Wu, T., Li, X., 2014. Tunable hydrogen separation in porous graphene membrane: first-principle and molecular dynamic simulation. ACS Appl. Mater. Interfaces 6, 8048e8058. Waldman, M., Hagler, A.T., 1993. New combining rules for rare gas van der waals parameters. J. Comput. Chem. 14, 1077e1084. Wang, B., Nisar, J., Ahuja, R., 2012. Molecular simulation for gas adsorption at NiO (100) surface. ACS Appl. Mater. Interfaces 4, 5691e5697. Wang, S., 2007. Comparative molecular simulation study of methane adsorption in MetalOrganic frameworks. Energy Fuel 21, 953e956. Wang, X., Ma, C., Xiao, J., Xia, Q., Wu, J., Li, Z., 2018. Benzene/toluene/water vapor adsorption and selectivity of novel C-PDA adsorbents with high uptakes of benzene and toluene. Chem. Eng. J. 335, 970e978. Zeng, Y., Phadungbut, P., Do, D.D., Nicholson, D., 2014. Anatomy of adsorption in open-end and closed-end slit mesopores: adsorption, desorption, and equilibrium branches of hysteresis loop. J. Phys. Chem. C 118, 25496e25504. Zhang, X., Gao, B., Creamer, A.E., Cao, C., Li, Y., 2017. Adsorption of VOCs onto engineered carbon materials: a review. J. Hazard Mater. 338, 102e123.