Ionic liquid functionalization – an effective way to tune carbon dioxide adsorption properties of carbon nanotubes

Abstract

In this research, the influence of non-covalent functionalization by ionic liquids on carbon dioxide (CO₂) adsorption–desorption properties of multi-walled carbon nanotubes (MWNTs) and partially exfoliated MWNTs (PEMWNTs) has been studied. In addition, the effect of polymerization of the ionic liquid on CO₂ adsorption–desorption properties has also been studied at low pressures (<100 kPa) and at different temperatures. The thermodynamic parameters were also determined. It was found that ionic liquid functionalization significantly improved the adsorption capacity through weak CO₂ complex formation with the substituted nitrogen of the imidazolium cation. Furthermore, the adsorption isotherm analysis suggested that the residual functional groups were more affine compared to the ionic liquid moieties. The heat of adsorption behaves differently upon polymerization of the ionic liquid on both substrates. The influence of polymerization on the entropy change was much more significant with the MWNTs substrate, whereas it was negligible with PEMWNTs.

---

1. Introduction

Carbon nanomaterials (CNMs) have attained immense importance as adsorbents for various adsorbates, because of their unique physical and chemical properties.^1–5 As solid state adsorbents, CNMs have comprehensive advantages, particularly in terms of thermal stability and adsorbent regeneration. The adsorption properties depend on the lateral surface properties, such as local surface chemistry and density of surface anchoring sites. Recently, the effect of partial exfoliation on carbon dioxide (CO₂) adsorption properties of multi-walled carbon nanotubes (MWNTs) has been studied, and it was concluded that the surface functional groups significantly influence the adsorption–desorption behavior.⁶ The carbon hexagonal lattice allows surface functionalization with a variety of covalent and non-covalent functionalizations. Suitable functionalization with a task specific functional moiety was expected to tune the adsorption properties, considerably.⁷ The non-covalent functionalization included surface modification by metal oxide nanoparticles⁸ and polymer chains,⁹ whereas structural defects, attached functional groups¹⁰ and doped heterogeneous atoms¹¹ were classified as covalent functionalities. Thus, it was of interest to study the influence of surface functionalization on adsorption properties.

CO₂ has good solubility in ionic liquids (ILs), which were identified as being good CO₂ capture media.^12–14 In particular, imidazolium-based ILs made weak intermediates with CO₂ through the nitrogen atoms present in the cation.¹⁵ The advantages of an IL include its low vapor pressure, high thermal stability, wide liquid temperature range and good chemical stability.¹⁶ The solubility enhancement of CO₂ in ILs has been observed through confinement studies.¹⁷ For example, IL incorporated carbon nanotubes show good CO₂ solubility when the IL is confined inside the tube because of a favorable negative transfer energy.¹⁸

The high pressure CO₂ adsorption studies on IL functionalized graphene showed that IL functionalization doubles the adsorption capacity of graphene, while the heat of adsorption remains in physisorption range even after functionalization.¹⁹ Surface functionalization of graphitic nanomaterials with IL results in a uniform distribution of anchoring sites, which leads to a higher CO₂ adsorption capacity. Furthermore, CO₂ molecules interact with IL sites through reversible physical interaction, which facilitates easy desorption. Several other researchers have also reported CO₂ adsorption capacity enhancement in supported ILs, where porous solid state supports have been used.^20,21

ILs, particularly with imidazolium cations, are prone to make solid-like, short range ordering on atomically flat surfaces, such as graphitic materials.²² This crystallization reduces the ratio of accessible anchoring sites. Recently, Tang et al. have reported that the polymerized ionic liquids (PILs) have an even higher CO₂ sorption capacity together with faster sorption/desorption rates than IL monomers with the same backbone.²³ These polymers are porous in nature, which facilitates better accessibility, which can also be supported with high surface area carbon nanostructures in order to achieve a high adsorption capacity. Recently, IL and PIL functionalized graphene have been synthesized and evaluated for their CO₂ adsorption properties and used as the cathode catalyst support for CO₂ conversion.^24,25 Here, PIL functionalization improved the adsorption capacity more significantly than IL functionalization.

Drawing inspiration from previous studies that determined the effect of partial exfoliation of MWNTs on CO₂ adsorption properties, MWNTs and partially exfoliated MWNTs (PEMWNTs) have been used as support materials for IL and PIL moieties in this study.⁶ Furthermore, the low pressure (<100 kPa) CO₂ adsorption properties of IL and PIL functionalized MWNTs and PEMWNTs have been determined experimentally. It is believed that this is the first study on low pressure CO₂ adsorption properties of IL or PIL functionalized MWNTs (one-dimensional, 1D) and PEMWNTs (1D and two-dimensional).

2. Experimental section

2.1 Material synthesis

MWNTs and PEMWNTs were synthesized by the method reported in a previous study.²⁶ These materials were functionalized by IL as follows: 100 mg of either material was thoroughly dispersed in 10 ml of ethanol followed by the addition of 100 mg of 1-butyl-3-methylimidazolium tetrafluoroborate ([BMIM][BF₄]) ionic liquid. The mixture was thoroughly stirred for six hours at ambient conditions. The dispersion was filtered using a Nylon-66 membrane and dried under reduced pressure without further washing.

PIL surface functionalized MWNTs and PEMWNTs were synthesized using the free radical polymerization of 1-vinyl-3-methylimidazolium tetrafluoroborate ([VMIM][BF₄]).²⁷ Typically, 100 mg of MWNTs (or PEMWNTs) were dispersed in 25 ml of dimethylformamide. A round-bottomed flask (50 ml) equipped with a reflux condenser was charged with this mixture followed by the addition of [VMIM][BF₄] (200 mg) and 2,2′-azobisisobutyronitrile (7 mg) and refluxed for 16 h at 353 K under vigorous stirring in a nitrogen atmosphere. Then 100 ml of methanol was added to the mixture to quench the reaction and precipitate the PIL functionalized nanostructures together with a certain amount of free polymer chains. The precipitate was washed with deionized water and methanol repeatedly to remove loosely attached polymer chains and unreacted monomer. Finally, the product was filtered through a Nylon-66 membrane and dried for 12 h at 333 K under reduced pressure. A second generation IL with respective side chains was chosen, where fluorinated anions were commonly chosen for both monomeric and polymeric ILs in order to understand the effect of polymerization.

2.2 Characterization techniques

The surface morphology of synthesized materials was recorded using transmission electron microscopy (TEM) using an FEI Tecnai G-20 instrument. A PerkinElmer Fourier-transform infrared (FTIR) spectrometer was used to measure the vibrational characteristics of the synthesized materials. Carbon dioxide adsorption properties of CNMs were determined at various temperatures using a Micromeritics ASAP 2020 V3.00 H surface area analyzer with a water bath.

2.3 Adsorption analysis

Low pressure CO₂ adsorption–desorption studies were carried out up to 100 kPa pressure by same method used in a previous study using a surface area analyzer. Briefly, 150–200 mg of sample was loaded into the sample tube together with the filling rod and closed by a seal frit (20 μm pore size). The sample was unconventionally degassed at 373 K for 3 h (conventional activation for CNMs: 450–500 K, 8–12 h) in order to remove moisture, because a possible change in the chemical properties of IL and PIL was expected at high temperatures. The mass difference between the empty sample cell (sample tube and filling rod and seal frit) and the cell with the degassed sample gives the actual sample mass (m) with high accuracy (±0.001 mg).

Prior to the analysis, available free space (V) inside the sample cell was calculated using highly pure helium gas at each temperature. The adsorption–desorption analysis was carried out using a vacuum-swing adsorption technique at multiple sub-ambient pressures and temperatures, where the adsorbate was degassed at reduced pressure. In a typical analysis, the sample cell was filled with a known amount (n) of highly pure CO₂ gas and the pressure drop was continuously recorded.

At equilibrium, the number of moles of gaseous CO₂ (n_eq) inside the tube was calculated using the van der Waals equation:

where P_eq is the equilibrium pressure of n_eq moles of gaseous CO₂ in volume V at temperature T. Thus, the difference n − n_eq gives the adsorbed moles of CO₂ (n_ad) at equilibrium:

n_ad = n − n_eq (2)

The pressure inside the sample tube was reduced by removing a known amount of gaseous CO₂ (n*) from the total amount of gas (n = n_ad + n_eq) in the desorption analysis. The number of moles of gaseous CO₂ (n^*_eq) at equilibrium desorption pressure in volume (V) was calculated using eqn (1). The adsorbed amount of CO₂ molecules on the surface of the adsorbent (n^*_ad) was determined using the equation:

n^*_ad = n − n* − n^*_eq (3)

The specific adsorption capacity at equilibrium (Q_eq) was reported to be normalized to the absolute mass of the sample (m). The adsorption isotherms were obtained for IL and PIL functionalized MWNTs and PEMWNTs at multiple temperatures.

The adsorption process at low pressures can be closely modeled using the Langmuir and Freundlich adsorption isotherm equations.

The Langmuir isotherm²⁸ is represented as:

while the Freundlich equation²⁹ is:where Q_eq (mol g⁻¹) is the amount of adsorbed CO₂ molecules at equilibrium pressure P (Pa), Q_max (mol g⁻¹) is the maximum adsorption capacity at a complete monolayer coverage, K_L is the Langmuir isotherm coefficient, K_F is the Freundlich isotherm constant (mol g⁻¹) and n is the adsorption intensity that defines the favorability of adsorption.

The isosteric heat of adsorption was calculated from the slope of the adsorption isosteres using the Clausius–Clapeyron equation:

where ΔH is the differential enthalpy of adsorption, T is the temperature and P is the pressure of isosteric points on the adsorption isotherms. The absolute value of differential enthalpy (|ΔH|) gives the “isosteric heat” of adsorption, while the intercept (ΔS/R) of isosteres with the ln P-axis in the van't Hoff plot gives the entropy change in adsorption (ΔS).³⁰

Typically, experiments were carried out with a 20 s equilibrium interval (the time interval between successive pressure readings) with 2% relative pressure and 200 Pa absolute pressure tolerances (whichever is less is considered at each equilibrium pressure).

3. Results and discussions

3.1 Morphological analysis

The TEM micrographs (Fig. 1) reveal the morphological information about IL or PIL functionalized MWNTs and PEMWNTs. The TEM image of MWNTs-IL (Fig. 1(a)) shows the presence of IL, where the inner diameter of MWNTs is partially screened by IL layers. The presence of IL on the PEMWNTs surface (Fig. 1(c)) is not clear to see, but can be observed from the inhomogeneous shades on the surface. Because the surface of the PEMWNTs is not as smooth as that of the MWNTs, the presence of IL moieties is not clear enough.

Fig. 1 Transmission electron micrographs of (a) MWNTs-IL, (b) MWNTs-PIL, (c) PEMWNTs-IL and (d) PEMWNTs-PIL.

Similarly, the TEM image of MWNTs-PIL (Fig. 1(b)) shows that the polymeric substance is wrapped on the surface. Because MWNTs were introduced in the free radical polymerization process, radicals are accumulated on the surface of the CNMs and grow as long chains. The 1D nature of MWNTs allows the long polymer chains to wrap around the surface. But, PIL was observed on the surface of PEMWNTs as deposited islands (Fig. 1(d)) because of the roughly unraveled structure. Here, structural defects (dangling bonds, heptagons and pentagons) and residual functional groups act as anchoring sites.³¹ These anchoring sites arise from acid treatment (purification) in MWNTs, while those result on unraveled, wrinkled layers of PEMWNTs from partial exfoliation.

The surface areas of MWNTs and PEMWNTs were determined to be 67 and 147 m² g⁻¹, respectively, in a previous report.²⁶ But, IL or PIL functionalized MWNTs and PEMWNTs were expected to show the wrong surface area and porosity. Here, IL (or PIL) moieties fill the pores and cover the surface of the material upon functionalization as shown in the TEM images. Because the ILs have good selectivity towards CO₂ in a mixture of gases, the pores are inaccessible to nitrogen atoms.³²

3.2 Molecular vibrational analysis

The surface covalent functional groups on purified and partially exfoliated MWNTs were identified in a previous study.²⁶ In this study, the presence of IL or PIL moieties on the surface of MWNTs, resulting from additional non-covalent functionalization, was confirmed by FTIR spectroscopy (Fig. 2). Moisture and surface hydroxyl functionalities produced stretching vibrations at 3431 cm⁻¹, while the in-plane bending vibrations of the carboxylic group occurred at ∼1405 cm⁻¹ in all samples. The hexagonal honeycomb lattice results in a signal corresponding to stretching vibration of aromatic rings (∼1636 cm⁻¹), while the ring breathing mode was identified at 1037 cm⁻¹ in all samples. The peaks at the fingerprint region (500–2000 cm⁻¹) could be assigned to various stretching and bending modes of residual functional groups, such as hydroxyl, carboxyl, epoxy and carbonyl.³³

Fig. 2 Fourier-transform infrared spectra of (a) MWNTs,⁶ (b) MWNTs-IL, (c) MWNTs-PIL, (d) PEMWNTs,⁶ (e) PEMWNTs-IL and (f) PEMWNTs-PIL.

The FTIR spectra of IL (or PIL) functionalized MWNTs (or PEMWNTs) show stronger anti-symmetric and symmetric C–H vibrations at 2922 and 2853 cm⁻¹ than that of their pure counterparts, which may be attributed to the presence of IL (or PIL) on the surface. Furthermore, the C=N stretching (1750 cm⁻¹) and imidazolium ring stretching vibrations (1584 cm⁻¹) were the result of IL (or PIL) functionalization. It has to be noted that the region 600–700 cm⁻¹ is populated with more peaks for both PIL functionalized substrates, because of the signal from the vinyl group of PIL functionality.

3.3 Adsorption isotherm analysis

The low pressure adsorption–desorption isotherm properties of ionic liquid functionalized CNMs were recorded at multiple temperatures up to 100 kPa equilibrium pressure (Fig. 3). Adsorption isotherms showing the equilibrium adsorption capacity of MWNTs-IL at ∼100 kPa were 110, 96, 87 and 81 μmol g⁻¹ at 283, 288, 293 and 298 K, respectively. Ionic liquid functionalization improves ∼25% of the specific adsorption capacity of MWNTs at ∼100 kPa. Once again, the adsorption isotherms show almost straight line behavior and do not follow the Langmuir model as observed for pure MWNTs because of the partial contribution from CO₂ trapped inside the tubes (Table 1).^6,34,35 Similarly, PEMWNTs-IL have also shown ∼38% improved performance compared with PEMWNTs at 283 K. The adsorption analysis revealed the maximum adsorption capacity (at ∼100 kPa) of 410, 378, 327 and 286 μmol g⁻¹ at 283, 288, 293 and 298 K sample temperatures, respectively. It is important to note that IL on PEMWNTs enhances the CO₂ adsorption capacity better than that on MWNTs surface. This can be attributed to the better IL confinement on the highly disordered PEMWNTs surface, which results in higher CO₂ capturing capacity through density enhancement.³⁶ Here, wrinkles and grooves provide better IL confinement sites.

Fig. 3 CO₂ adsorption–desorption isotherms of (a) MWNTs-IL, (b) PEMWNTs-IL, (c) MWNTs-PIL and (d) PEMWNTs-PIL at different sample temperatures.

Table 1 Comparison of adsorption isotherm parameters of pure and IL or PIL functionalized MWNTs and PEMWNTs^a

Material	T (K)	Qeq @ 100 kPa	kH (bar)	Langmuir			Freundlich
				Qmax	KL	R^2	n	KF	KF/n	R^2
a Units: Qeq (μmol g^−1), Qmax (μmol g^−1), KL (kPa^−1), n (AU) and KF (μmol g^−1 kPa^−(1/n)).b From a previous study.^6
MWNTs^b	283	88	253	Fit did not converge			1.096	1.305	1.190	0.998
	288	78	277				1.100	1.207	1.097	0.999
	293	68	318				1.096	1.038	0.946	0.999
	298	63	350				1.148	1.123	0.978	0.998
PEMWNTs^b	283	297	13	387	0.0270	0.979	2.070	32.184	15.544	0.999
	288	271	16	357	0.0269	0.981	2.036	28.763	14.122	0.999
	293	250	20	338	0.0245	0.983	1.968	24.407	12.398	0.999
	298	214	26	292	0.0242	0.986	1.934	20.306	10.49	0.998
MWNTs-IL	283	110	209	Fit did not converge			0.947	0.856	0.903	0.998
	288	96	232				1.001	0.977	0.976	0.998
	293	87	257				0.973	0.787	0.808	0.999
	298	81	278				0.936	0.609	0.650	0.999
PEMWNTs-IL	283	410	12	875	0.0083	0.990	1.447	16.885	11.667	0.999
	288	378	15	834	0.0080	0.990	1.422	14.944	10.505	0.999
	293	327	20	683	0.0089	0.991	1.455	13.956	9.587	0.999
	298	286	27	621	0.0083	0.991	1.424	11.435	8.027	0.999
MWNTs-PIL	283	137	161	Fit did not converge			1.088	1.990	1.828	0.998
	288	120	181				1.072	1.664	1.550	0.998
	293	100	219				1.054	1.292	1.224	0.998
	298	79	277				1.047	0.994	0.949	0.998
PEMWNTs-PIL	283	491	13	847	0.0133	0.991	1.615	28.456	17.614	0.999
	288	436	15	747	0.0128	0.991	1.618	25.828	15.961	0.999
	293	406	18	728	0.0120	0.992	1.567	21.777	13.893	0.998
	298	364	22	683	0.0111	0.994	1.522	18.097	11.889	0.999

---

However, PIL functionalization improves the adsorption capacity of MWNTs and PEMWNTs at 283 K by 56% and 65%, respectively. Although the adsorption sites (imidazolium in the cation and fluorine in the anion) in both IL and PIL are same, the adsorption capacity is significantly improved upon polymerization. This certainly cannot be attributed to the amount of IL or PIL loaded on the support, because PIL with a similar molecular structure shows only ∼70 μmol g⁻¹ of adsorption capacity at similar conditions.^37,38 Thus, the improvement must be attributed to the porous structure of the PIL alone. Briefly, the PIL produces randomly entangled chains of repeated monomeric units, which are wrapped or reside as islands on the surface. But, the IL produces solid-like, short range ordering on the substrate.²² In the present case, such layering was not observed using X-ray diffraction analysis at ambient conditions (not given), because of the hydrophilic nature. Such solid-like layering of hydrophobic IL on the polymer or graphitic carbon surface at similar conditions was found in previous studies.^39,40 Thus, it was expected that IL would make short range ordering in a moisture free atmosphere or vacuum, which would certainly reduce the accessible anchoring sites for the CO₂ molecules.

In addition, the improvement in the materials studied may be attributed to the synergic effect of the combined performance of the substrate and functional moiety, because IL or PIL with nearly same molecular structure gives <100 μmol g⁻¹ adsorption capacity.^12,23

Henry's constant describes the affinity of the surface towards the adsorbate, and is defined as:

where k_H is Henry's constant, x is the amount of gas adsorbed (g per g adsorbent) and P is the CO₂ pressure. Because isotherms are not linear, in some cases at pressures higher than 15 kPa, the Henry's constants are calculated by fitting the experimental data with <15 kPa equilibrium pressure. The calculated Henry's constants are presented in Table 1. As the quantitative information (number of anchoring sites per unit mass of sample) of functional groups is not available, Henry's constant is defined as the pressure required for the adsorption of equal mass of CO₂ on the adsorbent.

The value of Henry's constant of MWNTs is decreased approximately 20 times after partial exfoliation. This is strong evidence for the presence of residual functional groups on the surface of PEMWNTs, which act as anchoring sites for CO₂ molecules. Functionalization of the MWNT surface with IL reduces the k_H values because of its high affinity towards CO₂ molecules. However, the k_H value is not changed much with the PEMWNTs surface. This can be directly attributed to the residual functional groups. Briefly, PEMWNTs themselves have residual functional groups which already largely reduce the k_H values. Additional functionalization with IL may not change the k_H values, if the residual functional groups are more affine than IL and PIL moieties.

Furthermore, it is found that the k_H values of MWNTs have been decreased upon PIL functionalization. This may be because of the amorphous nature of the PIL phase, which offers a large amount of highly affine anchoring sites unlike the solid-like IL phase. Once again, PIL functionalization did not make a significant change in the k_H value of the PEMWNTs. In general, the Henry's constant values have obviously increased with temperature because of the fact that CO₂ molecules have a high kinetic energy at high temperatures, which assists them to escape from the anchoring sites.

Adsorption isotherms of all materials closely follow the Freundlich model rather than the Langmuir model, which suggests the distribution of adsorption energy. In particular, isotherms of MWNT-based samples do not converge when modeled with the Langmuir equation, as was observed for pure MWNTs in a previous study.⁶ It was also noted that this model fits more closely for IL or PIL functionalized PEMWNTs than non-functionalized ones. This can be attributed to the adsorption energy distribution window. In pure PEMWNTs, the adsorption energy distribution is mainly contributed from structural defects and residual functional groups, which may show a large distribution window. But, in IL (or PIL) functionalized PEMWNTs the major contributors are residual functional groups and IL (or PIL) moieties, which have a higher adsorption energy than structural defects.

The Langmuir coefficient, K_L, is related to the initial adsorption rate and has been significantly reduced after IL or PIL functionalization. Because the surface residual functional groups are partially covered by IL (or PIL) moieties, the contribution from them must be low, at low pressures. In addition to this, the decrease in the K_L value suggests that residual functional groups are more affine than IL and PIL functionalities.

The Freundlich constant, n, defines the favorability of adsorption (related to adsorption at low pressures).⁴¹ In general, the n value and thus favorability of adsorption was increased upon partial exfoliation because of the high surface area and residual functional groups. But, IL or PIL functionalization slightly reduces the n value with both base materials, i.e., the adsorption at low pressure is decreased. This is in good agreement with the behavior of k_H and K_L values. Nevertheless, adsorption at near ambient pressure is improved upon IL (or PIL) functionalization as IL and PIL moieties are quantitatively higher than residual functional groups. Thus, Henry's constant (k_H), the Freundlich constant (n), and the Langmuir constant (K_L) are in mutual agreement for the materials studied, which suggests that the residual functional groups are more affine than the IL and PIL moieties.

The ratio of Freundlich coefficients (K_F/n) has a strong correlation with the molecular orbital energy of adsorbate and adsorbent system (adsorption energy). The K_F/n values suggest that the PIL moieties are more favorable than the IL moieties for CO₂ adsorption, which is in close agreement with results found in the literature.^23,37,38

3.4 Thermodynamics of adsorption

Isosteric heats of adsorption of IL or PIL functionalized carbon nanostructures have been determined from their corresponding adsorption isosteres at different adsorbed amounts (Fig. 4). The van't Hoff plots of MWNT-IL and MWNT-PIL were plotted in a range of isosteric points from 10 to 60 μmol g⁻¹, while those of PEMWNT-IL and PEMWNT-PIL were presented for constant adsorbed amounts of 50 to 200 μmol g⁻¹. The isosteric points follow Arrhenius behavior, suggesting that the adsorption rate is temperature-dependent.

Fig. 4 CO₂ adsorption isosteres (van't Hoff plot) of IL and PIL functionalized MWNTs and PEMWNTs with different adsorbed amounts.

Here, the slope of the adsorption isosteres is equivalent to ΔH/R, while the intercept with the pressure (ln P) axis gives the ΔS/R values. The van't Hoff plots clearly show negative slopes, resulting in negative values of the differential enthalpy of adsorption. This implies that the adsorption is exothermic in nature (physisorption) and also suggests that IL or PIL moieties form no chemical interaction, such as formamide formation. The interaction is completely physical through the weak acid–base interaction between CO₂ and the imidazolium cation.⁴²

Strength of adsorbate–adsorbent interaction, i.e., the difference between the activation energies for adsorption and desorption at equilibrium, can be represented by the isosteric heat of adsorption. The absolute value of differential enthalpy (|ΔH|) was calculated from the slope of isosteres, using the Clausius–Clapeyron equation (eqn (6)).

The carbon dioxide physisorption reaction on the MWNT-IL surface has a |ΔH| value of 15 ± 1 kJ mol⁻¹ at an adsorbed amount of 60 μmol g⁻¹. This value is slightly higher than that observed with pure MWNTs, suggesting the good affinity of the adsorbent towards the adsorbate.⁶ Similarly, the CO₂ adsorption studies on PEMWNT-IL shows the |ΔH| value of 23 ± 2 kJ mol⁻¹ at an adsorbed amount of 200 μmol g⁻¹ and was found to be higher than that of PEMWNTs for the obvious reason.

The calculated isosteric heats of adsorption (|ΔH|) are 27 ± 2 kJ mol⁻¹ (at 60 μmol g⁻¹) and 30 ± 3 kJ mol⁻¹ (at 200 μmol g⁻¹) for MWNT-PIL and PEMWNT-PIL, respectively. It is clear from |ΔH| values that the PIL functionalization improves the differential enthalpy of adsorption more significantly than IL, which is responsible for the high adsorption. The good affinity towards CO₂ molecules and its porous nature make PIL a more promising functional moiety than the IL monomers.

The isosteric heat of adsorption as a function of adsorbed amount is given for all the materials studied in Fig. 5. In general, continuous variations of the isosteric heat of adsorption with surface coverage for IL or PIL functionalized MWNTs and PEMWNTs were observed, because of the heterogeneity of the surface.⁴³

Fig. 5 Comparison of isosteric heat of adsorption of IL or PIL functionalized MWNTs and PEMWNTs as a function of adsorbed amount.

But, the change in |ΔH| value with surface coverage follows different trends for IL and PIL functionalized supports. It is interesting to note that IL functionalized supports follow a trend similar to that of non-functionalized supports. Here, the trend of the |ΔH| value with surface coverage shows multiple plateaus because of the ideal dependence of gas density on adsorbent surface with pressure and heterogeneity of the adsorbent surface. Briefly, the adsorbate can be approximated to ideal gas at low pressures, where adsorption takes place only because of the interaction between the adsorption site and the CO₂ molecule and no adsorbed CO₂ interacts with the new one. Because identical adsorption sites show the same strength of adsorption, the region becomes a plateau. Once these sites are occupied, adsorption takes place at low affine sites. In the present case, at least two well defined plateaus were found, which may be assigned to residual functional groups and IL sites (layered IL). The conclusion derived from K_F/n values suggests that the high |ΔH| value plateau corresponds to the residual functional groups.

But, PIL functionalized supports show an entirely different trend compared to pure and IL functionalized supports. Here, the |ΔH| value gradually decreases with surface coverage, suggesting a continuous distribution of adsorption energy. Although the adsorption sites in IL and PIL moieties are the same, the porous nature of amorphous PIL results in different adsorption energies for sites on the surface and inside pores. In general, PIL functionalized supports show higher |ΔH| values than the IL functionalized supports, while the |ΔH| values of CO₂ adsorption on IL or PIL functionalized materials are higher than those of their pure counterparts. According to Al-Muhtaseb and Ritter, the behaviors of |ΔH| with surface coverage suggests that residual functional groups and IL functionalization results in moderate heterogeneity to the adsorbent surface, while PIL moieties make the surface strongly heterogeneous.⁴³

The change in entropy (ΔS) of the adsorbent by adsorption of CO₂ was calculated from the ln P-axis intercepts (ΔS/R) of the van't Hoff plot at multiple adsorbed amounts. Fig. 6 shows the variation in ΔS values with surface coverage for IL and PIL functionalized MWNTs and PEMWNTs. In general, the ΔS values are decreased with surface coverage, suggesting that the adsorption is through physical interaction. The calculated values of ΔS are presented in Table 2.

Fig. 6 Comparison of entropy change upon adsorption of CO₂ on MWNTs-IL and PEMWNTs-IL as a function of the adsorbed amount.

Table 2 Summary of Q_eq, |ΔH| and ΔS values of IL or PIL functionalized MWNTs and PEMWNTs^d

Material	Qeq^a (μmol g^−1)	|ΔH| (kJ mol^−1)	ΔS (J mol^−1 K^−1)
a At 100 kPa, 283 K.b At 60 μmol g^−1.c At 200 μmol g^−1.d Inside brackets – % increase from base materials (MWNTs and PEMWNTs).
MWNTs^6	88	12.1 ± 0.2^b	−149 ± 9^b
MWNTs-IL	110 (25%)	15 ± 1 (25%)^b	−174 ± 1 (16%)^b
MWNTs-PIL	137 (56%)	27 ± 2 (125%)^b	−395 ± 34 (165%)^b
PEMWNTs^6	297	19.1 ± 0.8^c	−240 ± 20^c
PEMWNTs-IL	410 (38%)	23 ± 2 (21%)^c	−334 ± 31 (39%)^c
PEMWNTs-PIL	491 (65%)	30 ± 3 (58%)^c	−350 ± 9 (46%)^c

---

Fig. 6 clearly shows that the polymerization of IL functionalities on MWNTs increase the ΔS values significantly. As discussed in the adsorption isotherm analysis, IL moieties have a tendency to make solid-like, short range ordering on a smooth graphitic surface. In such cases, CO₂ molecules could access only the ordered surface and no CO₂ dissolution/penetration can be expected. But, polymerization results in the porous nature of the IL, which enables CO₂ to access the inner part of the PIL matrix. The entropy is increased because of the accommodation of CO₂ inside the PIL matrix (wrapped on the MWNTs), which increases the disorder in the system. It is also important to notice that IL with a similar structure has the dimension of ∼0.6 nm (size of imidazolium ring), which is comparable to the size of the CO₂ molecule (0.33 nm).²² Thus, accommodation of CO₂ inside the PIL matrix results in a large change in entropy.

The increase in ΔS value of PEMWNT-IL compared to MWNT-IL must arise from the base materials as observed in a previous study.⁶ Surprisingly, the polymerization of IL on the PEMWNT surface has not changed the ΔS values of CO₂ adsorption. It is believed that the surface roughness of the PEMWNTs must be the major reason for this phenomenon. Briefly, PIL moieties make islands on the PEMWNTs as shown in Fig. 1(d), and are not wrapped as on the MWNT surface, because of the highly wrinkled, disordered nature of the surface with a large amount of residual functional groups. This leads to deposition of largely disordered islands of PIL on the PEMWNT surface. Upon CO₂ adsorption, the disorder in the PIL phase may not change significantly, which results in an unchanged ΔS value of CO₂ adsorption. Further research is needed to demonstrate the nature of CO₂ molecules in the supported IL or PIL phase. Table 2 presents the summary of Q_eq, |ΔH| and ΔS values of IL or PIL functionalized MWNTs and PEMWNTs.

The advantages of IL functionalization includes their low vapour pressure compared to conventional amines, which prevents the loss of functional moieties in desorption. Here, the IL reversibly captures CO₂ molecules through simple dissolution, rather than the formation of stable molecules, such as carbamides. In this study, desorption isotherm was observed with a very low adsorbate retention and residue (∼2% at 1 kPa) even at sub-ambient temperatures. Thus, desorption/regeneration of CO₂ is relatively easy compared to an aqueous amine. Furthermore, the thermal and chemical stability of ILs also allows the high temperature desorption.

4. Conclusion

The carbon dioxide adsorption–desorption analysis of IL and PIL functionalized MWNTs and PEMWNTs was carried out at low pressures (<100 kPa). It was found that the surface of MWNTs and PEMWNTs were functionalized by IL moieties. PIL was wrapped on the surface of MWNTs, while it forms islands on the PEMWNT surface because of its roughness. In general, the adsorption isotherm fits better with the Freundlich model compared to the Langmuir model, suggesting heterogeneity of the adsorbent surface. Henry's constant (k_H), favorability (n) and adsorption energy suggest that residual functional groups are more highly energetic than IL and PIL functionalities. The improvement in adsorption capacity upon IL functionalization is relatively effective on the PEMWNT surface. But, polymerization of IL shows almost similar improvement on both substrates. The heat of adsorption is significantly increased upon IL functionalization with both substrates. But, polymerization has more of an influence on the heat of adsorption of MWNT-PIL adsorbent compared to that of PEMWNT-PIL. The change in entropy upon CO₂ adsorption changed more with PEMWNT-IL than with MWNT-IL. However, polymerization of IL changed ΔS values more with MWNTs-PIL, but it showed a negligible increase with the PEMWNTs-PIL adsorbent. This preferential uptake of atmospheric CO₂ makes IL or PIL functionalized nanostructures suitable for adsorption and separation applications.

Acknowledgements

The authors acknowledge the Sophisticated Analytical Instrument Facility of the Indian Institute of Technology Madras (SAIF-IITM) for the FTIR analysis. One of the authors (Tamilarasan) acknowledges the IITM for financial support (senior research fellowship).

References

1. S. D. Kenarsari, D. Yang, G. Jiang, S. Zhang, J. Wang, A. G. Russell, Q. Wei and M. Fan, RSC Adv., 2013, 3, 22739–22773.
2. M. Zhang, B. Gao, X. Cao and L. Yang, RSC Adv., 2013, 3, 21099–21105.
3. M. Barathi, A. S. Krishna Kumar, C. U. Kumar and N. Rajesh, RSC Adv., 2014, 4, 53711–53721.
4. S. Vasudevan and J. Lakshmi, RSC Adv., 2012, 2, 5234–5242.
5. I. Cabasso, S. Li, X. Wang and Y. Yuan, RSC Adv., 2012, 2, 4079–4091.
6. P. Tamilarasan and S. Ramaprabhu, J. Appl. Phys., 2014, 116, 124314.
7. R. S. Madyal and J. S. Arora, RSC Adv., 2014, 4, 20323–20333.
8. Y. Kang, F. Cai, H. Chen, M. Chen, R. Zhang and Q. Li, RSC Adv., 2015, 5, 6136–6141.
9. A. Kumar Mishra and S. Ramaprabhu, RSC Adv., 2012, 2, 1746–1750.
10. B. Yang, H. Hu, Q. Yu, X. Zhang, Z. Li and L. Lei, RSC Adv., 2014, 4, 56224–56234.
11. R. Zhu, Q. Chen, X. Wang, S. Wang, J. Zhu and H. He, RSC Adv., 2015, 5, 7522–7528.
12. X. Zhang, X. Zhang, H. Dong, Z. Zhao, S. Zhang and Y. Huang, Energy Environ. Sci., 2012, 5, 6668–6681.
13. H. Peng, Y. Zhou, J. Liu, H. Zhang, C. Xia and X. Zhou, RSC Adv., 2013, 3, 6859–6864.
14. P. C. Hillesheim, J. A. Singh, S. M. Mahurin, P. F. Fulvio, Y. Oyola, X. Zhu, D.-e. Jiang and S. Dai, RSC Adv., 2013, 3, 3981–3989.
15. C. Cadena, J. L. Anthony, J. K. Shah, T. I. Morrow, J. F. Brennecke and E. J. Maginn, J. Am. Chem. Soc., 2004, 126, 5300–5308.
16. M. Freemantle, An Introduction to Ionic Liquids, The Royal Society of Chemistry, Cambridge, UK, 2010.
17. J. Zhang, Q. Zhang, X. Li, S. Liu, Y. Ma, F. Shi and Y. Deng, Phys. Chem. Chem. Phys., 2010, 12, 1971–1981.
18. W. Shi and D. C. Sorescu, J. Phys. Chem. B, 2010, 114, 15029–15041.
19. P. Tamilarasan, T. S. Remya and S. Ramaprabhu, Graphene, 2013, 1, 3–10.
20. M. Ramdin, T. W. de Loos and T. J. H. Vlugt, Ind. Eng. Chem. Res., 2012, 51, 8149–8177.
21. S. Hanioka, T. Maruyama, T. Sotani, M. Teramoto, H. Matsuyama, K. Nakashima, M. Hanaki, F. Kubota and M. Goto, J. Membr. Sci., 2008, 314, 1–4.
22. S. Bovio, A. Podestà, C. Lenardi and P. Milani, J. Phys. Chem. B, 2009, 113, 6600–6603.
23. J. Tang, W. Sun, H. Tang, M. Radosz and Y. Shen, Macromolecules, 2005, 38, 2037–2039.
24. P. Tamilarasan and S. Ramaprabhu, J. Mater. Chem. A, 2015, 3, 101–108.
25. P. Tamilarasan and S. Ramaprabhu, J. Mater. Chem. A, 2015, 3, 797–804.
26. P. Tamilarasan and S. Ramaprabhu, J. Mater. Chem. A, 2014, 2, 14054–14063.
27. B. Wu, D. Hu, Y. Kuang, B. Liu, X. Zhang and J. Chen, Angew. Chem., Int. Ed., 2009, 48, 4751–4754.
28. I. Langmuir, J. Am. Chem. Soc., 1916, 38, 2221–2295.
29. H. Freundlich, Kapillarchemie: eine Darstellung der Chemie der Kolloide und verwandter Gebiete, Akademische verlagsgesellschaft mbh, 1922.
30. W. Gao, D. Butler and D. L. Tomasko, Langmuir, 2004, 20, 8083–8089.
31. C. Xu, X. Wang, J. Zhu, X. Yang and L. Lu, J. Mater. Chem., 2008, 18, 5625–5629.
32. B. J. Adzima, S. R. Venna, S. S. Klara, H. He, M. Zhong, D. R. Luebke, M. S. Mauter, K. Matyjaszewski and H. B. Nulwala, J. Mater. Chem. A, 2014, 2, 7967–7972.
33. N. Colthup, J. Opt. Soc. Am., 1950, 40, 397–400.
34. C. Matranga, L. Chen, M. Smith, E. Bittner, J. K. Johnson and B. Bockrath, J. Phys. Chem. B, 2003, 107, 12930–12941.
35. C. Matranga and B. Bockrath, J. Phys. Chem. B, 2004, 108, 6170–6174.
36. W. Shi and D. R. Luebke, Langmuir, 2013, 29, 5563–5572.
37. J. Tang, H. Tang, W. Sun, M. Radosz and Y. Shen, J. Polym. Sci., Part A: Polym. Chem., 2005, 43, 5477–5489.
38. J. Tang, H. Tang, W. Sun, H. Plancher, M. Radosz and Y. Shen, Chem. Commun., 2005, 3325–3327.
39. P. Tamailarasan and S. Ramaprabhu, J. Phys. Chem. C, 2012, 116, 14179–14187.
40. P. Tamilarasan and S. Ramaprabhu, Mater. Chem. Phys., 2014, 148, 48–56.
41. Y. Otake, N. Kalili, T. H. Chang and E. Furuya, Sep. Purif. Technol., 2004, 39, 67–72.
42. M. C. Kroon, E. K. Karakatsani, I. G. Economou, G.-J. Witkamp and C. J. Peters, J. Phys. Chem. B, 2006, 110, 9262–9269.
43. S. A. Al-Muhtaseb and J. A. Ritter, J. Phys. Chem. B, 1999, 103, 2467–2479.

---