Effect of Immobilized Amines on the Sorption Properties of Solid

Article pubs.acs.org/Langmuir

Effect of Immobilized Amines on the Sorption Properties of Solid Materials: Impregnation versus Grafting Santiago Builes*,† and Lourdes F. Vega†,‡ †

MATGAS Research Center, Carburos Metálicos, Air Products Group, CSIC, UAB, Campus de la UAB, 08193 Bellaterra, Spain Carburos Metálicos, Air Products Group, C/Aragón 300, 08009 Barcelona, Spain

‡

S Supporting Information *

ABSTRACT: The underlying mechanism of the adsorption process in functionalized materials is not yet fully understood. This incomplete understanding limits the possibility of designing optimal adsorbent materials for different applications. Hence, the availability of complementary methods to advance this field is of great interest. We present here results concerning the adsorption of CO2 in amine-functionalized silica materials by Monte Carlo simulations, providing new insights into the capture process. Two different mechanisms of functionalization are compared: impregnation (a physical mixture of the amine and the support) and grafting (a chemical bond is formed between the amine and the support). We evaluate in this work a model of MCM-41 for N2 and CO2 adsorption with varying degrees of density of the functionalized chains. The results indicate that the mobility of the impregnated chains allows the creation of a network of microcavities, which enhance the low-pressure adsorption capabilities of these materials. Molecular simulations allow us to study in detail the conformational changes in the functionalized chains during the adsorption process. Materials functionalized densely by grafting undergo a change in the preferential orientation of the chains, which allows the adsorption of additional molecules close to the surface of the support. The adsorption of gas molecules close to the pore surface is usually the most energetically favorable configuration; however, for densely grafted materials the adsorption close to the surface occurs only at pressures large enough to provide energy to displace the functionalized chains.

■

INTRODUCTION One of the key scientific challenges in recent years has been the search for an effective technology for separating and capturing CO2 from a flue gas stream. The current benchmark technology for industrial separation of CO2 from flue gas is absorption with aqueous amines. However, this process has had a number of disadvantages, such as high regeneration costs, equipment corrosion, and oxidative degradation of the amines.1,2 Therefore, a suitable alternative option to avoiding the energetic penalty and the loss of amines due to degradation consists of immobilizing amines on the surface of a solid. This kind of hybrid material merges the inherent sorptive behavior of porous solids with the large CO2 selectivity of amines, favoring physisorption and reducing the energy cost of regeneration compared to that of the conventional amine process. The presence of amino groups in the adsorbent is interesting for separation purposes because of the possibility of creating responsive charged surfaces with selective CO2 binding properties. Periodic mesoporous silica (PMS) materials are usually employed as solid supports because of their high surface area and pore volume and their ordered structure. Traditionally, zeolites, porous siliceous minerals, and superabsorbent hydrogel polymers have been widely used as sorbents for the separation of fluids. However, industrial applications of © 2012 American Chemical Society

these materials are often limited by the low sorption capacity and selectivity for nonpolar or hydrophobic matter. Currently, functionalized materials are being intensively investigated for new uses in organic matter separation, with applications ranging from the sorption of oil and hydrocarbon contaminants to CO2 capture.3−6 PMS functionalized with amines have been gaining interest as important candidate materials for the separation and capture of CO2. There are two main ways to immobilize amines in a silica material: (i) physically impregnate and (ii) covalently bond an organic chain containing amines.7 Wet impregnation is the simplest method of supporting amines on solid materials. In this method, the amines are physically mixed with the support using a volatile solvent to diffuse the amines into the pore, and the excess solution is removed by the evaporation of the diluted solvent. Conversely, the grafting method consists of covalently tethering amine-containing groups on the silica surface.8 One of the most common ways of grafting chains with amino moieties is to use the reaction of silanol groups in the silica with aminosilanes. The silanol group on the silica surface reacts with Received: September 25, 2012 Revised: November 4, 2012 Published: November 9, 2012 199

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

attainable experimentally. It is possible to neglect certain molecular interactions of the system, centering the study on specific features of the physical system, provided that (i) the main physical features of the system are considered and (ii) the effect of the assumptions on the final results is taken into account. Even though there are numerous experimental studies on the adsorption of CO2 by amine-functionalized silica, the study of molecular simulations in this field is still scarce. There are only a handful of works dealing with complex amine chains functionalizing a silica surface,19,20 and none of them attempts to compare the different grafting techniques. Hence, the purpose of this work is to advance in this area, offering insights into the differences between the impregnation and grafting of organic chains for the adsorption of gases. In the present work, we focus on studying the different interactions of immobilized chains during the adsorption of CO2 using a model of MCM-41 with aminopropyl chains. For this, we have considered two different kinds of models: (i) physisorbed chains on the silica surface and (ii) aminosilanes convalently bonded to the support. Herein, we assume that the same type of organic chain is functionalized to an organic system. We use APTES chains for the study of both types of functionalized materials: grafted and impregnated. Although this type of molecule would react with the solid surface to produce grafted chains, it is possible to neglect the formation of the covalent bond in the simulations and consider the chains as being physically impregnated. This assumption will ease the comparison, allowing us to focus on the type of immobilization rather than on the differences between the functionalized chains. Moreover, the relatively short APTES chains have a large probability of being leached during desorption, in the case of impregnation, because of the mesoporous nature of PMS. However, for comparison purposes, during the simulations, it is assumed that the chains are simply impregnated and cannot be leached out of the pore volume. In the simulations, the physical and chemical interactions of the chains with CO2 are reproduced by explicitly considering the carbamates in the molecular simulations.21 Moreover, the adsorption of purely physisorbed N2 is also analyzed in order to gain a better understanding of the different interactions among the functionalized chains, the support, and the gas molecules. We calculate the CO2 and N2 adsorption isotherms for different degrees of surface functionalization using molecular simulations. This study aims to increase the knowledge of the possibilities of the functionalized materials and the interactions of the functionalized chain with the pore surface. This knowledge will allow for the future design of complex molecular systems.

the silanes in the chains, forming a covalent bond between the surface and the amine chains. The main technical advantages of impregnation over grafting are convenient preparation, reduced corrosiveness, and large amine capacity.9 A larger amount of amine can be loaded onto the support by wet impregnation rather than grafting. In the grafting approach, the quantity of amine that can be retained by the support is limited by the density of the accessible surface silanol groups. However, the main advantage of grafting over impregnation is that the former cannot leach amines unless a temperature strong enough to break the bonds is employed.7 Although there are numerous works that describe the use of amine-impregnated materials and amine-tethered solids for the separation and capture of CO2, to the best of our knowledge there are no reports focusing on the comparison of both methods. The available experimental works usually focus on one amine immobilization technique, attempting to obtain high loading and high stability of the amines. The first amineimpregnated silica used to capture CO2 was reported by Xu et al.10 The authors studied the impregnation of MCM-41 with a low-molecular-weight poly(ethyleneimine) (PEI). The use of low-molecular-weight PEI is associated with the ease of impregnation of the silica pore space; longer polymer chains would likely block pore mouths and prevent complete pore penetration.7 Moreover, different research groups have evaluated MCM-41 impregnated with PEI for CO2 capture.11−13 It has been observed that the CO2 adsorption capacity increases synergistically and a much higher uptake is obtained for the hybrid material than for pure MCM-41 and pure PEI. Although the most commonly used amine for impregnation for CO2 adsorption is PEI, the amines used in research studies range from simple monoamines to large aminopolymers.7 For instance, Yue et al. used SBA-15 impregnated with tetraethylenepentamine (TEPA), improving the dispersion and accessibility of the material by not removing the template after synthesizing the mesoporous silica.14 In the case of covalently tethered amines, research has focused on understanding the interactions among CO2, the functionalized chains, and the silica surface. 3-Aminopropyltriethoxysilane (APTES) has been commonly used as the coupling agent for the modification of silica surfaces. Leal et al.15 studied the adsorption of CO2 on silica gel grafted with APTES. They obtained a functionalization of up to 1.27 mmol amine/g and captured 0.6 mmol/g of CO2 at 23 °C and 1 bar. Knowles et al.16 observed the reversible adsorption of CO2 on silica gel 40 grafted with APTES, although in the presence of water the desorption of CO2 decreased. Knöfel et al.17,18 used in situ FTIR spectroscopy and microcalorimetry to study the reactivity between carbon dioxide and amines functionalizing SBA-16. They proposed that at low loadings chemisorption was the leading mechanism whereas at higher pressures physisorption was more predominant. Previous works have focused on one immobilization technique: either grafting or impregnation. However, no comparison between the two techniques has been attempted, mainly because of the difficulty in finding the proper experimental conditions under which to compare them properly. This difficulty can be overcome by the use of molecular simulations. The complex reactions and chemistry of the surface with the functionalized groups can be simplified by the use of simulations. One of the main advantages of simulations is that they allow the establishment of certain ideal controlled conditions not

■

COMPUTATIONAL METHODS

MCM-41 Model. PMS materials are excellent candidates for scientific studies because they can be reproducibly synthesized and they are easy to characterize because of their ordered structure. The MCM-41 model was generated following the work of Pellenq et al.20 to build atomistic models of MCM-41. The method consists of carving out a hexagonal array of cylindrical pores from a block of amorphous silica, obtained from the Materials Studio database.22 The silica atoms outside the volume of the carved cylinder are kept, and all of the oxygen atoms bonded to these silica atoms are included in the silica model. These nonbridging oxygen atoms are connected to hydrogen atoms to form surface hydroxyl groups. The final structure generated by this procedure provides a sufficiently realistic model of 200

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

amorphous silica, particularly of its surface. The full details of the generation of the model can be found elsewhere.20,23 Functionalized Silica. For the grafting procedure, it was assumed that silane functionalization was solely driven by surface reactions. The silane chains covalently bond to the silanols on the support. This hybrid material was modeled by introducing silane groups into the MCM-41 model, connecting them to a number of silanols. The silane chains were built on the silica segment by segment from the oxygen of the preexisting silanol, according to the procedure explained elsewhere.21,24 Conversely, the physically impregnated molecules were modeled by neglecting the surface reaction that would occur between surface silanols and silane groups. This assumption allows us to compare the same type of chains directly with different types of functionalization mechanisms. Wet impregnation is simulated by filling the pore volume with solvent molecules by GCMC simulations and adjusting the fugacity in order to simulate the loading process and reproduce the concentration of the solvent inside the pore. This procedure is similar to that in previous works for the impregnation of the pore space of solid adsorbents with a solvent.20,24 Adsorption Simulations. CO2 adsorption isotherms were studied in the functionalized materials, both impregnated and grafted. The adsorption isotherms were computed by using the grand canonical Monte Carlo (GCMC) method. For each value of pressure, 1.6 × 107 trials were used for equilibration and 1.0 × 107 trials were used for data collection. The silica atoms in the hybrid materials were kept rigid while the organic chains were allowed to move. The grafted molecules possessed a fixed point, and the rest of the molecule was allowed to undergo regrowth movements. The impregnated molecules, in addition to regrowth, were allowed full canonical moves consisting of translation and rotation. The intermolecular interactions were calculated through pairwiseadditive 12−6 LJ potentials for the repulsive and dispersive terms and Coulombic potentials for the first-order electrostatic contribution.25 The interaction parameters between LJ sites were computed according to the Lorentz−Berthelot combination rules. The interactions of the N2 molecules were modeled using the TraPPE potential,26 and the CO2 molecules were modeled with the force field of Harris and Yung.27 N2 molecules were taken to be rigid, with an N−N bond length of 0.110 nm. CO2 was modeled as a rigid molecule with three interaction sites, a C−O bond length of 0.116 nm, and an O−C−O angle of 180°. The parameters for the intermolecular interactions of the adsorbate molecules are listed in Table 1. The

The reaction of CO2 and the amines was considered to occur at low relative pressures, forming carbamates and a protonated base. A brief description of the method is provided here, although the full details of the method can be found elsewhere.21 A fixed number of molecules not affected by the chemical potential (i.e., with no insertion/removal attempts) are included in the simulations in order to account for the occupancy and the interactions of the chemisorbed CO2 on the adsorption isotherms. Primary amines in the functionalized chains are considered to react with CO2 to form a carbamate and a protonated base (i.e., two different chains are required to capture a single CO2 molecule). Instead of including only amine chains during the functionalization of the models, a predefined number of chains including the carbamates and the protonated amines are also included in the simulation. This inclusion of the carbamates at zero coverage considers that at very low pressures CO2 reacts with the amine chains to form carbamate species. Even though this assumption implies that increasing the pressure has no further effect on the reaction, it is possible to account for this effect by including a varying number of chemisorbed molecules in the system depending on the pressure. This reaction scheme was previously used to account for chemisorbed CO2 with amine-functionalized silica. The inclusion of chemisorbed CO2 in the simulated material allowed us to predict accurately the experimental sorption isotherm of CO2 on APTES grafted on MCM-41.21 Not all of the amine groups immobilized in a solid materials can react with CO2 because the agglomeration of amine molecules in the support limits their mobility, decreasing the conversion of the reaction. The efficiency of the reaction was assumed, and we employed the commonly obtained value of 60% for both reactions, which was determined from low-pressure experimental data.17−19,21

■

RESULTS AND DISCUSSION A series of functionalized materials were generated by molecular simulations of grafted and impregnated APTES on MCM-41. The different configurations employed in the simulations of this work are depicted in Figure 1. For the

Table 1. Parameters for the Intermolecular Interactions of the Adsorbate Molecules and the Silica Support site

σ (nm)

Si Obridging Ononbridging H

0.0 0.2708 0.3000 0.0

C O

0.2785 0.3064

N COM

0.3310 0.0

ε/kB (K) Silica 0.0 228.4 228.4 0.0 CO2 30.00 83.00 N2 36.00 0.0

q (e)

ref

1.283 −0.629 −0.533 0.206

29, 29, 29, 29,

0.66450 −0.33225

27 27

−0.482 0.964

26 26

30 30 30 30

Figure 1. Representation of grafting (G) and impregnation (I) mechanisms in which the hydrolyzed APTES molecule (top right) can impregnate or graft the MCM-41 material (top left). Color key: O (red), Si (yellow), H (white), N (blue), and C (gray).

simulations of G (the grafted material, see Figure 1), the silanol groups on the surface were considered to bond in a monodentate fashion with the siloxanes. Moreover, to simplify the calculations, it was considered that there was no horizontal polymerization among neighboring siloxane chains. Even if polymerization is possible in the experimental material, the main phenomenon in the system is the grafting to the surface. Hence, those assumptions should not have a large influence on

parameters for the siloxane part of the organic chain are taken from the MM2 force field for silane compounds.28 The parameters for the rest of the CHn organic groups are taken from the TraPPE force field. The charges on the surface chains were adjusted to maintain electrical neutrality in the simulation cell. The parameters for the intramolecular energy of the functionalized silica with APTES are listed in Tables S1− S4 (presented in the Supporting Information). 201

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

the final adsorption results. I corresponds to the impregnated material (Figure 1). The silanols on the surface of I were not considered to react with the siloxane chains; only physical forces acted between the functionalized chains and the silica surface. Although this assumption is not realistic, it allows for a direct comparison between the two functionalization mechanisms. Hence, this assumption enables us to control all of the parameters of the simulation and focus on the functionalization mechanism and its effect on the adsorption behavior of the material. An MCM-41 model with a surface area of 983.4 m2/g was used as the raw support for G and I. The agreement between the simulated adsorption properties of N2 and CO2 and experimental data for this PMS model has been demonstrated previously.21 We compared two different degrees of surface functionalization, 0.4 and 2.0 mmol/g APTES, for mechanisms G and I, namely, G04 and G20 for G and I04 and I20 for I. The materials were characterized by nitrogen adsorption isotherms at 77 K, as shown in Figure 2. For all of the samples except for

space. Similar behavior was previously observed by Ho et al.20 on MCM-41 They performed simulations of Lennard-Jones spheres impregnating MCM-41 and computed the adsorption of CO2. They found that the functionalized molecules created micropores inside the mesoporous MCM-41. The complex interplay of apparent surface area, pore volume, and intermolecular interactions affects the overall gas uptake. The surface area and pore volume of the studied samples are presented in Table 2. The surface area and the pore volume are Table 2. Textural Properties of the MCM-41 Support and the Functionalized Materials sample

S_BET (m2/g)

Pvol (cm3/g)

r (nm)

RAW G04 G20 I04 I20

959.5 925.1 706.8 990.1 693.8

0.70 0.64 0.40 0.63 0.38

1.45 1.38 1.14 1.27 1.08

determined using the method proposed by Düren et al.,32 and the equivalent radius was estimated from the pore volume assuming a cylindrical pore. The main adsorption characteristics are similarly modified in spite of the immobilization technique. The volume occupied by one chain in G is constant with respect to the number of functionalized chains. However, the volume occupied by a chain varies with the degree of functionalization; when the loading increases, the chains get closer together and occupy a lower fraction of volume per chain. This indicates that at lower loadings the chains can distribute better and then they start to agglomerate more compactly when the amine loading increases. Moreover, it has to be considered that the decrease in the volume of G might be larger if horizontal polymerization were considered in the simulations. The decrease in the pore volume is very similar for both techniques, although it is consistently lower for I because of the less-compact packing of the nongrafted molecules. The functionalized chains in I at high pressure are not as closed packed as the grafted molecules because of the absence of chemical bonds with the surface. For instance, for G20 the decrease in the pore volume was 43%, whereas for I20 it was 46%. Moreover, the reduction of the surface area is much larger for I than for G. This is caused by the different packing of the impregnated molecules depending on the degree of functionalization. At high loadings, the number of cavities available for adsorption diminish, decreasing the accessible surface. One of the main advantages of using molecular simulations is that they allow the quantification of this effect. This change of behavior can be observed better by the use of snapshots of the corresponding points highlighted in Figure 2 (shown in Figure 3). For I20 at low pressure (Figure 3I), the impregnated chains are distributed around the pore space, creating a dense network that attracts the fluid molecules more strongly. Conversely, in G20 the chains are grafted to the surface. At low pressure (Figure 3III), the molecules are attracted mainly by the surface of the mesopore; however, once the pressure increases (Figure 3IV) the N2 density increases. For I20 at high pressure (Figure 3II), the N2 density is lower than that in G20 because of the presence of the functionalized chains around the entire pore volume.

Figure 2. Adsorption isotherms of nitrogen at 77 K on raw MCM-41 (dashed line), amine-impregnated MCM-41 (circles), and amine grafted MCM-41 (squares). The amine densities were 0.4 mmol/g (top) and 2.0 mmol/g (bottom).

I20, the nitrogen adsorption isotherm is a typical type IV isotherm, with a vertical steep between 20 and 30 kPa that is characteristic of mesopores with a pore diameter smaller than 3.5 nm.31 G20 has a vertical steep at a lower pressure than for G04 and I04. This is due to the larger number of chains around the pore walls, which decrease the apparent pore diameter of the sample. I20 shows a distinct step in the N2 isotherm at low relative pressure (P/P0 < 0.1) characteristic of gas adsorption in micropores. The isotherm exhibits type I behavior because of the created microporous network caused by the agglomeration of the impregnated chains. The microporosity is generated by the proximity of the chains creating interstitial voids (Figure 2b). Hence, the material experiences a transition from mesoporous to microporous as a result of the degree and type of functionalization. In the impregnated materials, at high amine density the chains start to agglomerate, creating small cavities in the pore 202

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

Figure 3. Snapshots of the nitrogen adsorption isotherm in highly loaded materials I20 and G20 at low and intermediate pressures. The numbers in the images correspond to those in Figure 2b. Color key: N in N2 (green); N in NH2 (blue); H (white); C (gray); Si (yellow); O (red).

A different adsorption behavior occurs for CO2. Although N2 does not react with the organic chains, CO2 does react with the amines to form carbamate species. Thus, for the calculations of the CO2 adsorption isotherms it is necessary to account not only for the physisorbed but also for the chemisorbed CO2. It is possible to include the effects of chemisorptions on GCMC simulations by explicitly considering the products of the reaction under the initial conditions of the simulation.21 It is commonly believed that the resulting CO2 adsorption capacity of the composite is proportional to the number of amine groups grafted onto the porous adsorbent.33 However, for grafted materials it is important to note that this reaction results in substantially less than quantitative conversion (∼60% yield) when a densely loaded amine-functionalized silica is used.34 Moreover, although it is expected that the mobility of the molecules contributes to higher amine efficiencies for I, in practice the amine molecules agglomerate, resulting in relatively low efficiencies of the amines in the sorption of CO2.35 Thus, we have assumed an equal efficiency of the CO2 reaction with the amines for the calculation of CO2 adsorption isotherms of G and I. A reaction efficiency of 60% for both kinds of functionalized materials was assumed. This value was determined from low-pressure experimental data in aminefunctionalized materials.17−19,21

Following the procedure explained in the computational section,12 we calculated the CO2 uptake for the samples at pressures of up to 20 bars by GCMC (Figure 4). At low loadings, the CO2 capture was higher than for the raw material for the entire pressure range studied. For I04, the maximum uptake was 110% compared to that of the raw material whereas for G04 it was 103%. The functionalization of MCM-41 with 0.4 mmol/g APTES slightly increases the uptake over that of the raw material. Counterintuitively, the created microporosity and not the chemisorbed CO2 is the main reason for the increase in adsorption in the material functionalized with 0.4 mmol/g APTES. However, the isotherm shape changed drastically when the amine loading was increased. The isotherms of the functionalized material saturated at lower CO2 density because of the space occupied by the chains and the carbamates. The adsorption below 500 kPa was much higher for both materials, indicating the feasibility of using amine-functionalized materials to separate CO2 from flue gas streams at low pressure. This behavior was also observed by Ho et al.36 The authors observed that the size and polarity of the impregnated chains arranged the molecules in a fashion that favored the distribution of CO2 molecules. 203

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

captured by the composite in adsorption processes. This behavior can be better seen by the use of probability distribution plots, where the position of the CO2 and the amino moieties inside the pore is recorded during the simulation and normalized by the total positions. This is seen in the density profiles of the distance of the CO2 molecules from the center of the pore at different pressures, as depicted in Figure 5. For these density profiles, we calculated the distance of the carbon atom of the CO2 molecules from the center of the MCM-41 pore during the GCMC simulations. This distance was plotted as the probability density of the adsorbed CO2 molecules on MCM-41. For the calculation of the probability densities in MCM-41, we used radial distance points that contained an equal volume instead of using equidistant radial points to obtain a more consistent probability distribution. That way, it was ensured that all of the points in the distribution could, in practice, store the same number of CO2 molecules. The distribution of the fluid layers around the pore space indicates how the molecules organize in a particular fashion. This information can be used to characterize the different behaviors of G and I and obtain better insight into the adsorption process of these materials. Figure 5 shows that, for G, the CO2 molecules are attracted to the open end of the grafted chains. At 500 kPa for the materials with a lower degree of functionalization, I04 and G04, the CO2 molecule distributions are very similar, except for a small number of molecules adsorbed around the center of the pore; this is also seen in the smaller amount of CO2 adsorbed by I04 compared to that adsorbed by G04 at 500 kPa (Figure 4). Although materials with a low degree of functionalization behave similarly, materials with a greater number of surface groups have very distinct behavior. The CO2 molecules on G20 are preferentially located farther from the pore surface than on I20. The high density of grafted chains blocks the accessibility to a large section of the pore space close to the surface, and thus the CO2 molecules are attracted to the open end of the grafted chains. For I20, the fluid molecules are preferentially

Figure 4. Adsorption isotherms of carbon dioxide at 298 K on raw MCM-41 (dashed line), amine-impregnated MCM-41 (circles), and amine-grafted MCM-41 (squares) for amine densities of 0.4 (top) and 2.0 mmol/g (bottom).

The microporosity created in I20 greatly increases the interactions of CO2 with the adsorbent for pressures below 1000 kPa. This enhancement, which is not observed in G for any amine loading, might enable a much larger capture of CO2 from diluted point sources. The different adsorption characteristics between G and I are a result of the different functionalization methods. The latter is not bonded to the surface and can distribute freely around the pore space whereas the movements of the former are constricted by the bonds to the surface. Hence, the distribution of the amine groups in the composite plays a significant role in the adsorption of CO2. The high dispersion of the amines combined with a high accessibility allows for more CO2 to be

Figure 5. Density profiles of the distance of the carbon atom (C) in CO2 to the center of MCM-41 at 298 K. Snapshots of the adsorption of CO2 molecules on G20 and I20 at 1000 kPa (right bottom). 204

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

end of the organic chains, which should enhance the solid−fluid interactions. The different configurations of the chains of G and I can produce very different results. The conformation changes of the surface of the functionalized chain can have an effect on the effective pore size and also on the hydrophilic/hydrophobic characteristics. Dacquin et al.37 observed, by molecular simulations, that the introduction of different end groups on the grafted chains changed the preferential orientation of the functionalized chains, hence changing the material from hydrophilic to hydrophobic. In the amine profiles, it is seen that for I04 and G04 the presence of CO2 does not change the behavior of the chains. In G20, the chains change from a very sharp distribution to a narrower one. Because at low pressure the chains have a similar aligned preferential orientation, once the pressure is increased the alignment is broken, allowing more CO2 molecules to reside close to the pore surface (Figure 7).

attracted to the pore surface, with a fraction of the gas molecules distributed homogeneously around the pore space. Figure 5 shows a snapshot comparing G20 and I20. When the pressure increases to 1000 kPa, the behavior is similar to that observed in the density profiles at 500 kPa; the general shape of the profiles is maintained while the values increase because of the pressure change. However, when the pressure is increased to 2000 kPa the shape of the profiles is modified by the higher density of fluid molecules inside the materials. The CO2 molecules close to the surface in G04 reach a maximum, and a negatively skewed distribution is seen. This is caused by more CO2 molecules forming successive layers around the pore. In I04, we see a larger uptake due to the increased interactions near the center of the pore caused by the impregnated chains. The high loading of organic chains in I20 causes the profile at 2000 kPa to change only slightly with respect to those at 500 and 1000 kPa; the cavities created by the impregnated chains attract the CO2 molecules strongly even at low pressure. Thus, the main difference among the profiles when increasing the pressure is a slight change in the density. The largest change in the density profiles is observed for G20, where the distribution changes from normal to a nonsymmetric bimodal distribution. The peak of the initial normal distribution is displaced to the right when the distribution changes to bimodal. This drastic change in the profiles is due to a reorganization of the grafted chains. This change from a low-energy position of the grafted molecules to a more disordered one is caused by the pressure that supplies the energy needed to reaccommodate the grafted chains and allow the adsorption of more CO2. This change in the grafted molecules can be seen in the amine density profiles (Figure 6). Using a similar approach to the CO2 distributions, we calculated the density profiles for the N atom in the functionalized organic chains. The N atom is used instead of the center of mass, as in the case of CO2, because the N atom is located in one of the ends of the APTES chains. This position allows for the approximate locating of the

Figure 7. Scheme of the adsorption of CO2 (a) with the configuration of the chains aligned and (b) with a broad angle distribution.

Figure 7a shows a scheme of how the aligned molecules in G can block the CO2 molecules adsorbed close to the surface. In Figure 7b, the molecules are displaced by the increase in the pressure, as observed in the broader density distribution in Figure 6, and the CO2 molecules can access the pore surface. Although the positions close to the surface should be energetically favorable for adsorption in G20, adsorption close to the surface occurs only at pressures large enough to provide energy to displace the functionalized chains.

■

CONCLUSIONS We have presented here results concerning N2 and CO2 adsorption in amine-functionalized MCM-41. Two different functionalization methods, impregnation and grafting, were studied. Molecular simulations were used to allow a direct comparison between the impregnation and grafting of organic chains, highlighting the different adsorption behavior. The

Figure 6. Density profiles of the distance of the nitrogen atom (N) of the grafted chains to the center of the pore in MCM-41 at 298 K. 205

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206

Langmuir

Article

(12) Lee, K. B.; Beaver, M. G.; Caram, H. S.; Sircar, S. Ind. Eng. Chem. Res. 2008, 47, 8048. (13) Sircar, S. Ind. Eng. Chem. Res. 2006, 45, 5435. (14) Yue, M. B.; Chun, Y.; Cao, Y.; Dong, X.; Zhu, J. H. Adv. Funct. Mater. 2006, 16, 1717. (15) Leal, O.; Bolívar, C.; Ovalles, C.; García, J. J.; Espidel, Y. Inorg. Chim. Acta 1995, 240, 183. (16) Knowles, G. P.; Graham, J. V.; Delaney, S. W.; Chaffee, A. L. Fuel Process. Technol. 2005, 86, 1435. (17) Knöfel, C.; Descarpentries, J.; Benzaouia, A.; Zelenák, V.; Mornet, S.; Llewellyn, P. L.; Hornebecq, V. Microporous Mesoporous Mater. 2007, 99, 79. (18) Knöfel, C.; Martin, C. l.; Hornebecq, V.; Llewellyn, P. L. J. Phys. Chem. C 2009, 113, 21726. (19) Schumacher, C.; Gonzalez, J.; Pérez-Mendoza, M.; Wright, P. A.; Seaton, N. A. Ind. Eng. Chem. Res. 2006, 45, 5586. (20) Ho, L. N.; Perez Pellitero, J.; Porcheron, F.; Pellenq, R. J. M. Langmuir 2011, 27, 8187. (21) Builes, S.; Vega, L. F. J. Phys. Chem. C 2012, 116, 3017. (22) Materials Studio Database Accelrys Software Inc.: San Diego, 2010. (23) Coasne, B. J. Chem. Phys. 2004, 120, 2913. (24) Builes, S.; López-Aranguren, P.; Fraile, J.; Vega, L. F.; Domingo, C. J. Phys. Chem. C 2012, 116, 10150. (25) Frenkel, D.; Smit, B. Understanding Molecular Simulation: From Algorithms to Applications; Academic Press: San Diego, 2002. (26) Potoff, J. J.; Siepmann, J. I. AIChE J. 2001, 47, 1676. (27) Harris, J. G.; Yung, K. H. J. Phys. Chem. 1995, 99, 12021. (28) Frierson, M. R.; Allinger, N. L. J. Phys. Org. Chem. 1989, 2, 573. (29) Brodka, A.; Zerda, T. W. J. Chem. Phys. 1996, 104, 6319. (30) MacElroy, J. M. D. Mol. Phys. 2002, 100, 2369. (31) Groen, J. C.; Peffer, L. A. A.; Pérez-Ramírez, J. Microporous Mesoporous Mater. 2003, 60, 1. (32) Düren, T.; Millange, F.; Férey, G.; Walton, K. S.; Snurr, R. Q. J. Phys. Chem. C 2007, 111, 15350. (33) Yue, M. B.; Sun, L. B.; Cao, Y.; Wang, Y.; Wang, Z. J.; Zhu, J. H. Chem.Eur. J. 2008, 14, 3442. (34) McKittrick, M. W.; Jones, C. W. Chem. Mater. 2003, 15, 1132. (35) Meth, S.; Goeppert, A.; Prakash, G. K. S.; Olah, G. A. Energy Fuels 2012, 26, 3082. (36) Ho, N. L.; Perez-Pellitero, J.; Porcheron, F.; Pellenq, R. J. M. J. Phys. Chem. C 2012, 116, 3600. (37) Dacquin, J.-P.; Cross, H. E.; Brown, D. R.; Düren, T.; Williams, J. J.; Lee, A. F.; Wilson, K. Green Chem. 2010, 12, 1383.

analysis was focused on the change in the adsorption properties as a function of the type and degree of functionalization. The results show that impregnation consistently gives a higher adsorption uptake than grafting because of the mobility of the functionalized chains. The impregnated molecules can create a network that resembles microporosity inside the mesopores of MCM-41, provided that a large enough density of functionalized chains is used. This network enhances the lowpressure adsorption properties of the solid material, increasing the potential use of this technique for the separation and capture of gases. However, it should be kept in mind that, experimentally, impregnated materials can be leached when desorbing the gases, whereas grafted molecules can have a higher stability. The behavior of the grafted chains at a high density of functionalization can change with the gas pressure. The position of the chains tends to be arranged in an orderly aligned fashion at low concentrations. However, the chains can be bent once the pressure increases, allowing for the gas molecules to adsorb closer to the surface.

■

ASSOCIATED CONTENT

S Supporting Information *

Simulation parameters for the functionalized aminosilane chains. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Support for this work was provided by Carburos Metálicos from the Air Products Group and the Spanish Government (project CEN2008-1027 CENIT SOST-CO2). Additional support from the Spanish Government (project CTQ200805370/PPQ) and from the Catalan Government (2009SGR666 and a TALENT grant to S.B.) is also acknowledged.

■

REFERENCES

(1) Veltman, K.; Singh, B.; Hertwich, E. G. Environ. Sci. Technol. 2010, 44, 1496. (2) Lepaumier, H.; Picq, D.; Carrette, P.-L. Ind. Eng. Chem. Res. 2009, 48, 9061. (3) López-Aranguren, P.; Saurina, J.; Vega, L. F.; Domingo, C. Microporous Mesoporous Mater. 2012, 148, 15. (4) Rana, S.; Bajaj, A.; Mout, R.; Rotello, V. M. Adv. Drug Delivery Rev. 2012, 64, 200. (5) Kumar, P.; Guliants, V. V. Microporous Mesoporous Mater. 2010, 132, 1. (6) Builes, S.; Roussel, T.; Ghimbeu, C. M.; Parmentier, J.; Gadiou, R.; Vix-Guterl, C.; Vega, L. F. Phys. Chem. Chem. Phys. 2011, 13, 16063. (7) Choi, S.; Drese, J. H.; Jones, C. W. ChemSusChem 2009, 2, 796. (8) Wang, W.; Lofgreen, J. E.; Ozin, G. A. Small 2010, 6, 2634. (9) D’Alessandro, D. M.; Smit, B.; Long, J. R. Angew. Chem., Int. Ed. 2010, 49, 6058. (10) Xu, X.; Song, C.; Andresen, J. M.; Miller, B. G.; Scaroni, A. W. Energy Fuels 2002, 16, 1463. (11) Versteeg, G. F.; Van Dijck, L. A. J.; Van Swaaij, W. P. M. Chem. Eng. Commun. 1996, 144, 113. 206

dx.doi.org/10.1021/la3038507 | Langmuir 2013, 29, 199−206