Combined DFT/CC and IR spectroscopic studies on carbon dioxide adsorption on the zeolite H-FER

Abstract

Adsorption of carbon dioxide on H-FER zeolite (Si:Al = 8:1) was investigated by means of a combined methodology involving variable-temperature infrared spectroscopy and DFT/CC calculations on periodic zeolite models. The experimentally found value of adsorption enthalpy was ΔH⁰ = −30 kJ mol⁻¹. According to calculations, adsorption complexes on isolated Si(OH)Al Brønsted acid sites (single sites) involve an adsorption enthalpy in the range of −33 to −36 kJ mol⁻¹, about half of which is due to weak intermolecular interactions between CO₂ and the zeolite framework. Calculations show clearly the significant role played by weak intermolecular interactions; adsorption enthalpies calculated with standard GGA type exchange-correlation functionals are about 13 kJ mol⁻¹ underestimated with respect to experimental results. Good agreement was also found between calculated and experimentally observed stretching frequencies for these complexes. Calculations revealed that CO₂ adsorption complexes involving two neighbouring Brønsted acid sites (dual sites) can be formed, provided that the dual site has the required geometry. However, no clear evidence of CO₂ adsorption complexes on dual sites was experimentally found.

---

Broader context

Despite current efforts to develop alternative energy sources, a fast move away from (carbon based) fossil fuels is unlikely to take place. Hence, in the short term there is a need to implement cost effective means for CO₂ sequestration (e.g., from flue gases of cold-fired power stations and industrial manufacturing plants) and underground storage. For that purpose, porous solids capable of reversibly adsorbing CO₂ could be used, provided that they have a large adsorption capacity, chemical stability and favourable thermodynamics for the adsorption-desorption process. Metal organic frameworks, porous carbons and zeolites are among CO₂ adsorbents being currently investigated for this purpose. Zeolites, by virtue of their easy ion exchange and well known crystal structure, are amenable to comprehensive studies which should help understanding of the finer details of the gas–solid interaction energy and adsorption thermodynamics. The study reported herein combines theoretical calculations at the DFT/CC level and experimental measurements by variable-temperature FT-IR spectroscopy; it shows that CO₂ adsorbed on zeolite H-FER forms hydrogen-bonded complexes involving a standard enthalpy change of about −30 kJ mol⁻¹. The main finding, however, is that about one half of this corresponds to weak intermolecular interactions between the adsorbed CO₂ molecules and zeolite framework atoms.

1. Introduction

Fossil fuels, which currently provide about 80% of the world's energy needs,¹ are a main source of increasing levels of carbon dioxide in the atmosphere, which aggravates the greenhouse effect and its global adverse consequences. Despite recent alternatives proposed to curtail the use of such carbon based fuels² a fast move away from them is unlikely to occur, if anything because of the long lifetime (for economic reasons) of the energy supply infrastructure. Therefore, there is a need to develop an efficient means for CO₂ sequestration, alternatively referred to as carbon capture and storage (CCS).³ Implementation of CCS from the flue gas of stationary sources, such as coal-fired power stations or industrial manufacturing plants (e.g., cement and iron and steel production) can be accomplished by cryogenic technology, or by using liquid amine-based chemical absorbents; but these technologies are energy-intensive, meaning that a large energy penalty has to be paid^4–6 (in the case of chemical absorbents because of the energy intensive nature of the regeneration process). Besides, alkanolamines currently used as chemical solvents for treating gas streams having a low CO₂ partial pressure pose corrosion problems, as well as environmental and health hazards derived from reclaimer waste, unintentional emissions and accidental release;⁷ although some of these problems could perhaps be overcome by grafting the amine absorbent to mesoporous silica^8–11 or other convenient support.

An alternative to chemical sorbents is the use of porous solids that can reversibly capture and release CO₂ by means of pressure or temperature swing (adsorption) cycles (PSA or TSA, respectively). Porous adsorbents capable of separating CO₂ from flue gas by PSA or TSA could (in principle) be cost effective, provided that the thermodynamics of the adsorption-desorption cycle is favourable, and they pose less environmental problems than liquid (amine based) chemical absorbents. Among such solid CO₂ adsorbents, porous coal and activated carbons,^12–16 metal–organic frameworks and related microporous coordination polymers,^17–20 and zeolites^21–29 are currently under very active investigation.

For fundamental studies, zeolites have several advantages over other porous CO₂ adsorbents. First, their crystal structure is well known, and it can be resistant to repeated adsorption-desorption cycles. Secondly, easy cation exchange makes zeolites ideal materials for systematic studies on the interaction of CO₂ with a large variety of adsorption centres and, in the third place, the rich variety of framework topology and different pore size offered by zeolites facilitates analysis of the influence of these factors on relevant thermodynamic aspects of CO₂ adsorption. From the experimental side, previous detailed studies dealt mainly with alkali-metal exchanged zeolites,^21–28 although some consideration was also given to alkaline-earth metal exchanged samples^27–30 and to the protonic forms of faujasite-type and ZSM-5 zeolites.^24,31,32Carbon dioxide adsorption on alkaline faujasite-type (X and Y) zeolites was also analysed by means of quantum chemical calculation on cluster models,^27,33 and by Grand Canonical Monte Carlo and molecular dynamics simulations.^34,35 Relevant reported results in relation to the present work will be referred to in the discussion.

We report on a combined theoretical and IR spectroscopic study of carbon dioxide adsorption on the zeolite H-FER. Theoretical DFT calculations on a periodic model of the zeolite are used to (i) determine the location and relative stability of the zeolite Brønsted acid Si(OH)Al groups that constitute the CO₂ adsorption sites, (ii) find out the geometry of the corresponding CO₂ adsorption complexes and their binding energy, and (iii) determine the characteristic stretching frequencies of each type of adsorption complex. Variable temperature IR spectroscopy is used to (i) obtain the spectroscopic features of the CO₂ adsorption complexes and (ii) determine the corresponding adsorption enthalpy and entropy. Finally, from a combined analysis of experimental results and theoretical calculations, pertinent conclusions are drawn.

2. Materials and methods

2.1. Experimental details

A sample of ferrierite in the ammonium form (NH₄-FER) having a nominal Si:Al ratio of 8:1 was obtained from a commercial firm and checked by powder X-ray diffraction, which showed good crystallinity and confirmed the expected FER structure type. From this parent material, H-FER was obtained by thermal treatment inside an IR cell as described below. Total conversion of NH₄-FER into H-FER (during thermal treatment) was checked by the absence of IR absorption bands corresponding to the ammonium ion. For IR spectroscopy, a thin self-supported wafer of the zeolite sample was prepared and heated at 650 K for 3 h under a dynamic vacuum (residual pressure < 10⁻⁴ Torr) inside a home made cell, described elsewhere,³⁶ which allowed in situ thermal treatment, gas dosage and variable temperature IR spectroscopy to be carried out. After thermal activation of the sample wafer the cell was dosed with 0.1 Torr of helium (to improve thermal contact between the cell body and the zeolite wafer) and cooled. After recording the sample background spectrum the cell was dosed with CO₂ and closed, and a series of IR spectra was recorded within the temperature range of 240–290 K, while simultaneously registering temperature and gas equilibrium pressure inside the cell. In order to check reproducibility, and also to improve accuracy, the cell was then outgassed and dosed again with CO₂, and a new series of variable temperature IR (VTIR) spectra was recorded. For temperature and pressure measurements a platinum resistance thermometer (placed close to the sample wafer) and a capacitance pressure gauge (MKS, Baratron) were used. Precision of these measurements was better than ±10⁻² Torr for pressure and ±2 K for temperature. Pressure correction (for helium inside the cell) was determined from a calibration plot as described elsewhere.³⁷ Transmission FTIR spectra were recorded, at 3 cm⁻¹ resolution, by using a Bruker IFS66 spectrometer; 64 scans were accumulated for each spectrum.

2.2. Experimental determination of standard adsorption enthalpy and entropy

As discussed in detail elsewhere,^38,39 a series of spectra recorded over a sufficiently large temperature range, while simultaneously measuring temperature and equilibrium pressure of the adsorbed (physisorbed) gas can be used to determine the standard adsorption enthalpy and entropy by following the variable-temperature IR (VTIR) method.^38,40 In this method, absorbance (A) of a characteristic IR band of either the solid adsorbent or the adsorbed gas, temperature (T) and equilibrium pressure (p) are considered to be interrelated by the Langmuir-type equation:

θ = A/A_M = K(T)p/[1 + K(T)p] (1)

where θ stands for the fractional coverage of the adsorption sites, A_M is the integrated IR absorbance corresponding to full coverage (θ = 1) and K is the adsorption equilibrium constant at temperature T. Combination of eqn (1) with the well known van't Hoff eqn (2) yields eqn (3) below:

K(T) = exp(−ΔH⁰/RT) exp(ΔS⁰/R) (2)

ln[A/(A_M − A)p] = (−ΔH⁰/RT) + (ΔS⁰/R) (3)

Note that eqn (3) can also be written as:

ln[θ/(1– θ)p] = (−ΔH⁰/RT) + (ΔS⁰/R) (4)

After determining θ (or relative absorbance) as a function of T and p for IR spectra taken over a relatively large temperature range, a plot of the left-hand side of eqn (4) or (3)versus the reciprocal of the temperature gives direct access to the thermodynamic quantities ΔH⁰ and ΔS⁰ involved in the gas–solid adsorption process under study.

2.3. Computational methods and models

Ferrierite has an orthorhombic unit cell (Immm space group) with four tetrahedral (Tn) sites and eight framework oxygen atoms, O(m), that are symmetrically independent.⁴¹ The numbering scheme for the framework T and O atoms introduced by Vaughan is used;⁴¹ note that this numbering scheme differs from that used in the Database of Zeolite Structures,⁴² where T1 and T4 positions are switched. The structure of FER is depicted in Fig. 1, where the numbering scheme for framework atoms is also given for convenience.

Fig. 1 Structure of FER viewed along the perpendicular channel direction (A) and the framework atom numbering scheme (B). Framework T and O atoms are depicted and , respectively.

Two models of H-FER zeolite were used: (i) High-silica H-FER was modelled with a double unit cell (UC) having the composition H₁Al₁Si₇₁O₁₄₄ (corresponding to Si/Al = 71) using single UC dimensions optimized previously,⁴³ and (ii) H-FER(Si/Al = 8) model having a composition H₈Al₈Si₆₄O₁₄₄ with a re-optimized double UC (described below). The following Brønsted acid site labelling was introduced: H_n,m denotes the Brønsted site on O(m) framework atom in the vicinity of framework Al in Tn position. Equilibrium geometries of CO₂ adsorption complexes were obtained with the fixed double UC volume at the periodic DFT level and electronic interaction energies, ΔE_DFT, were calculated for the following process:

CO₂(g) + H-FER → CO₂/H-FER (5)

The framework deformation energy, E_def, was evaluated as the difference between the energy of the H-FER system at the geometry optimized with and without adsorbed CO₂:

H-FER → [H-FER]^≠ (6)

where [H-FER]^≠ denotes the H-FER system at the geometry optimized for CO₂/H-FER. The deformation energy thus describes the energy needed to displace H⁺ cation and framework atoms from their equilibrium geometry to bind the CO₂ molecule.

The calculations were performed using a periodic DFT model implemented in the VASP program^44–46 employing the Perdew–Burke–Ernzerhof (PBE) exchange-correlation functional,⁴⁷ the projector augmented wave approximation (PAW) of Blöchl^48,49 and the plane wave basis set with a kinetic energy cut-off of 400 eV, with the Brillouin-zone sampling restricted to the Γ-point. Zero-point energy (ZPE) corrections for CO₂ adsorption complexes were calculated within the harmonic approximation, using 9 degrees of freedom for the CO₂ molecule and 6 for each O–H group involved in the adsorption complex. The CO₂ stretching frequencies (ω) were evaluated within the harmonic approximation, whereas the O–H stretching frequencies of Brønsted acid OH groups (ν_OH) were evaluated using the ν_OH/r_OH correlation described previously.⁵⁰

The distribution of 8 framework Al atoms and 8 H⁺ ions in the H-FER(Si/Al = 8) model was obtained as follows. (i) Eight Al atoms were randomly distributed within the H-FER double UC respecting the Löwenstein rule and assuming an statistical distribution of Al within individual T sites (1:2:2:3 ratio for the occupancy of T1, T2, T3, and T4 sites). (ii) Each AlO₄ tetrahedron was charge-compensated by H⁺, respecting the energy preferences found for high-silica H-FER.⁵⁰ Thus, O(1):O(2):O(3):O(4):O(5):O(6):O(7):O(8) occupation of the Brønsted acid sites was 0:0:1:0:0:2:2:3, respectively, with the exception of H_4,6 site that was replaced by the H_4,8 site due to the close contact of two Brønsted acid OH groups. (iii) The volume of the double UC for this H-FER(Si/Al = 8) model was then fitted using the Birch–Murnaghan equation of state,^51,52 with a fixed cell shape, at the periodic DFT level (a = 19.1297, b = 14.3625 and c = 15.1688 Å, and V = 4167.62 ų). (iv) Since the most stable Brønsted acid sites in the vicinity of isolated framework AlO₄ tetrahedra may not represent energetically the most favourable distribution of H⁺ cations for a given distribution of Al atoms, the distribution of H⁺ cations was further refined. Theoretically there are 4⁸ (65536) distinguishable H⁺ configurations, too many to be explicitly treated. Therefore, the energetically stable H⁺ configuration was determined using the following iteration method. Individual AlO₄ tetrahedra were taken one at a time, the energy of each one of the four possible Brønsted acid sites on that tetrahedron was calculated (constant volume lattice energy minimization) and H⁺ was placed in the energetically most stable position. This procedure was repeated until self-consistency was reached (i.e., until no change in H⁺ position was found for 8 subsequent iterations). Due to the large number of calculations required, the most stable H⁺ distribution was determined at the inter-atomic potential functions level, employing the core–shell model potential parameterized at the DFT level by Sierka and Sauer;⁵³ calculations were performed with the GULP code.⁵⁴ The resulting H⁺ distribution gave the Brønsted acid site population 0:1:1:1:0:2:2:1 for H⁺ on O(1):O(2):O(3):O(4):O(5):O(6):O(7):O(8) atoms, respectively (Fig. 2). (v) The double UC volume was re-optimized for the new H⁺ site configuration at the periodic DFT level (a = 19.1460, b = 14.3747 and c = 15.1816 Å, and V = 4178.26 ų). This new H⁺ configuration is 19 kJ mol⁻¹ more stable than the initial H⁺ configuration (described in item (ii) above) at the periodic DFT level.

Fig. 2 Distribution of 8 framework Al atoms and the most stable Brønsted acid site configuration in the H-FER(Si/Al = 8) model viewed along the [001] (A) and the [010] (B) crystallographic direction. Framework Al, Si, and O atoms depicted , and , respectively, Brønsted acid H atoms depicted ○.

For reliable description of the interaction of the zeolite framework with a molecule of the size of CO₂ it is necessary to account for the effect of weak inter-molecular interactions that are not described properly at the DFT level. The DFT/CC correction scheme recently developed by the authors has been used for this purpose.⁵⁵ This method is based on the correction of the DFT error defined as the difference between DFT and accurate coupled clusters calculations, CCSD(T), performed at the complete basis set limit. The underlying assumption of the method is the pairwise representability of the DFT error, ΔE_DFT/CC, in terms of the interatomic distances, R_ij,

where ε_ij are correction functions, which were obtained from calculations on reference systems that included CO₂⋯H₂O, CO₂⋯H₃O⁺, and CO₂⋯Si(OH)₄. More details about the DFT/CC method used are given in ESI†, together with the corresponding correction functions, ε_ij.

3. Results

3.1. Structure and stability of CO₂ adsorption complexes: DFT investigation

Adsorption enthalpies and vibrational frequencies (O–H stretching mode, ν_OH, and CO₂ asymmetric stretching, ω₃) for CO₂ adsorption complexes formed over isolated Brønsted acid sites (single sites) and for those bridged between two Brønsted acid sites (dual sites) in H-FER zeolite are summarized in Table 1, together with corresponding geometrical parameters. The results obtained for the high-silica model are discussed first. The CO₂ adsorption complexes were investigated for the most stable Brønsted acid sites (determined previously⁵⁰) in the vicinity of each of four framework Al atom. As shown in Table 1, calculated values of ΔH⁰ depend significantly on the Brønsted acid site considered, being in the range of −28 to −37 kJ mol⁻¹. The most stable adsorption complex is formed on the H_3,8 Brønsted acid site for which also the largest shift in O–H stretching frequency, Δν_OH, was found. The same Brønsted acid site was also reported to form the most stable adsorption complex with CO and with N₂.⁵⁰ The least stable adsorption complex among those investigated with the high-silica model is formed on the H_1,3 Brønsted acid site, which gives ΔH⁰ = −28 kJ mol⁻¹. This lower stability is due to the fact that in the non-perturbed H-FER the corresponding H atom is involved in intra-zeolite hydrogen bonding, and it has to move towards the void space (see Fig. 2) in order to interact with CO₂ (as reflected in the larger E_def for the H_1,3 Brønsted acid site, Table 1).

Table 1 CO₂ adsorption complexes on H⁺ sites in high-silica H-FER model and in H-FER(Si/Al = 8) model (all energy terms in kJ mol⁻¹)

H^+ site	CO2 location	r(H⋯O) (Å)	Ob–H⋯O^a (°)	H⋯O=C^b (°)	ν OH (cm^−1)	ω 3 (CO2) (cm^−1)	E def	ΔEDFT	ΔZPVE	ΔEDFT/CC^c	ΔEDFT-D^d	ΔH^0 (0 K)^e
a Angle between framework oxygen atom belonging to Brønsted OH group, Ob, H^+ atom and oxygen atom of adsorbed CO2 molecule. b Angle defined by H^+ atom, and O and C atoms of CO2 molecule. c Interaction energy corrections evaluated with DFT/CC method. d Interaction energy corrections evaluated with the DFT-D method using PBE functionals and s6 = 0.75 (the interaction between CO2 and Brønsted acid H atoms excluded). e Adsorption enthalpy calculated as ΔH^0(0 K) = ΔEDFT + ΔZPVE + ΔEDFT/CC. f Distance between H^+ sites involved in the dual Brønsted acid sites is shown between brackets (in Å).
High- Silica ( Si / Al = 71) single sites
H3,8	M	1.798	174	130	3433	2366	1.2	−20.6	2.3	−18.7	−20.6	−37.0
H2,7	M	1.947	157	155	3507	2374	2.1	−18.2	2.1	−18.5	−20.1	−34.6
H4,6	P-cage	1.852	172	156	3470	2375	1.6	−15.2	1.9	−21.9	−23.6	−35.2
H1,3	P-cage	2.050	149	156	3514	2372	6.5	−10.1	1.2	−19.5	−20.6	−28.4
Al-rich ( Si / Al = 8) single sites
H4,8	M	1.870	174	141	3480	2370	4.0	−17.8	1.8	−20.3	−21.9	−36.3
H4,6	M	1.952	152	141	3461	2370	8.0	−15.9	1.7	−19.6	−20.7	−33.8
H2,7	M	1.912	156	128	3490	2365	1.7	−15.8	2.4	−20.5	−22.2	−33.9
H3,2	P-cage	1.892	175	177	3497	2378	1.5	−17.8	2.0	−19.5	−20.9	−35.3
H3,4	P-cage	1.956	163	170	3516	2375	1.2	−16.9	1.5	−17.8	−19.3	−33.2
Al-rich ( Si / Al = 8) dual sites
H3,4-H2,7 (5.3)^f	P-cage	1.962/1.994	171/156	135/135	3514/3545	2370	4.8	−22.1	2.5	−22.7	−25.5	−42.3
H4,6-H4,6 (7.6)^f	P-cage	1.923/3.460	175/163	162/156	3515/3584	2377	1.1	−19.4	1.8	−19.5	−21.0	−37.1

---

For the low silica H-FER(Si/Al = 8) model, the CO₂ adsorption complexes were investigated for the Brønsted acid site configuration depicted in Fig. 2. In the majority of adsorption complexes formed, the CO₂ molecule interacts with only a single Brønsted acid site. The CO₂ adsorption complexes formed on the single Brønsted acid sites H_4,8 and H_3,4 are depicted in Fig. 3. The adsorption enthalpies are in the range of −33 to −36 kJ mol⁻¹; i.e., they are similar to those in high-silica H-FER. A nearly linear O_b–H⋯O=C=O arrangement was found for the CO₂ adsorption complex on H_3,2 Brønsted acid site; as a consequence, this is among the most stable CO₂ adsorption complexes in H-FER(Si/Al = 8).

Fig. 3 CO₂ adsorption complex in H-FER(Si/Al = 8) model formed on (a) isolated H_4,8, (b) isolated H_3,4, (c) dual site formed by H_3,4 and H_2,7 and (d) dual site formed by a pair of H_4,6 sites. CO₂ molecule and H⁺ involved in the adsorption complex are depicted as spheres (see caption of Fig. 2 for framework atom colouring scheme).

In two cases the CO₂ molecule was found to interact simultaneously with two Brønsted acid sites (dual Brønsted acid sites, Fig. 3c and 3d). In these complexes each O atom of CO₂ interacts with one Brønsted acid site. These complexes involving a pair of Brønsted acid sites were found to be more stable than those formed on isolated Brønsted acid sites. For instance, CO₂ adsorption complexes on isolated H_2,7 and H_3,4 sites have adsorption enthalpies of −34 and −33 kJ mol⁻¹, respectively, while the bridged CO₂ complex between those same sites, H_3,4–H_2,7 has ΔH⁰ = −42 kJ mol⁻¹. Significant stabilization of the CO₂ adsorption complex due to the second Brønsted OH group depends on the dual site geometry, meaning not only the right distance apart between the two Brønsted acid sites concerned but also that none of them should be involved in intra-zeolite hydrogen bonding. The stabilization of −8 kJ mol⁻¹ found for adsorption complex on H_3,4–H_2,7 dual site is likely due to the fact that the corresponding r(H–H) distance is only 5.3 Å. In the case of the H_4,6–H_4,6 dual site, where r(H–H) is 7.6 Å, the stabilization due to the second Brønsted OH group is only −3 kJ mol⁻¹. Note also that none of the bridged CO₂ complexes on dual Brønsted acid sites has a linear O_b–H⋯O=C=O⋯H–O_b arrangement (Fig. 3 and Table 1).

3.2. Variable-temperature IR spectroscopy

The blank IR spectrum of H-FER (Si:Al = 8:1) in the O–H stretching region, recorded at 263 K, is shown in Fig. 4a (⋯). The weak IR absorption band seen at 3747 cm⁻¹ corresponds to silanols, while that appearing at 3603 cm⁻¹ is the characteristic O–H stretching band of the zeolite Brønsted acid sites. This latter band is rather broad and shows a marked tail in the low frequency side. Both of these features stem from a heterogeneity of the acidic Si(OH)Al groups, which was discussed in detail elsewhere.⁵⁰ However, because of inherent uncertainty about how the band should be decomposed, attempts at band resolution did not give reliable results regarding CO₂ adsorption enthalpy. Therefore, we decided to use integrated intensity of the 3603 cm⁻¹ band as it appears in the spectra.

Fig. 4 (a) Representative VTIR spectra (O–H stretching region) of CO₂ adsorbed on H-FER. Temperature (K) and equilibrium pressure (Torr, in brackets) as follows: 1, 244 (0.36); 2, 253 (0.50); 3, 260 (0.67); 4, 265 (0.82); 5, 269 (0.96); 6, 274 (1.09); 7, 279 (1.21). The ⋯ is the zeolite blank spectrum at 263 K. (b) Representative VTIR spectra (ν₃ region) of CO₂ adsorbed on H-FER. Temperature (K) and equilibrium pressure (Torr, in brackets) as follows: 1, 252 (0.12); 2, 265 (0.13); 3, 271 (0.15); 4, 275 (0.17); 5, 279 (0.20); 6, 283 (0.22); 7, 286 (0.23); 8, 289 (0.25). The zeolite blank spectrum was subtracted.

Upon interaction of the zeolite with adsorbed CO₂ the 3603 cm⁻¹ band was found to decrease to an extent which was a function of temperature, as shown in Fig. 4a for some representative variable-temperature spectra. Simultaneously, a new and much broader band corresponding to OH species hydrogen-bonded to CO₂ appeared, showing a maximum at about 3500 cm⁻¹. The small band at about 3705 cm⁻¹, which most likely corresponds to a combination mode (ν₁ + ν₃) of adsorbed carbon dioxide,^56,57 is of no concern here.

In the CO₂IR spectroscopic region, the only feature observed was a broad band at 2346 cm⁻¹, shown in Fig. 4b. This band corresponds to the asymmetric stretching vibration of CO₂ (ν₃ mode) perturbed by interaction with the zeolite Brønsted acid OH groups; note that in the free molecule⁵⁶ this ν₃ mode appears at 2349.3 cm⁻¹. For CO₂ adsorbed on alkali and alkaline-earth cation exchanged zeolites, several authors reported (besides a main band corresponding to the ν₃ mode) weak IR absorption bands appearing in the range of 1300 to 1700 cm⁻¹, which were assigned to carbonate species,^27,28,58,59 our IR spectra of CO₂ adsorbed on H-FER did not shown any absorption bands in that wavenumber region. Fig. 4b clearly shows that the band at 2346 cm⁻¹ has a complex nature, thus revealing some heterogeneity of the adsorbed species. However, for the same reasons as given above, resolution of the band into its components was not carried out, and the overall integrated intensity was used for deriving CO₂ adsorption enthalpy.

From two independent series of VTIR spectra in the O–H stretching region, the van't Hoff plot depicted in Fig. 5a was obtained. Note that the integrated intensity of the 3603 cm⁻¹ band divided by its maximum value (i.e., corresponding to the zeolite blank spectrum) gives the fraction (1–θ) of free OH sites, from which the corresponding θ value needed for eqn (4) was obtained. The linear plot in Fig. 5a gave the values ΔH⁰ = −30.1 kJ mol⁻¹ and ΔS⁰ = −124 J mol⁻¹K⁻¹ for the standard adsorption enthalpy and entropy of CO₂ on H-FER, respectively. Similarly, by using integrated intensity of the IR absorption bands corresponding to the ν₃ mode of adsorbed CO₂ (Fig. 4b) the van't Hoff plot depicted in Fig. 5b was obtained, applying eqn (3). The needed value of A_M, for which only an approximation was experimentally known, was obtained as that giving the best linear plot of eqn (3) for the whole set of experimental data, following an iteration procedure explained in detail elsewhere.^37,38 From the linear plot in Fig. 5b, the corresponding values of standard adsorption enthalpy and entropy resulted in ΔH⁰ = −30.3 kJ mol⁻¹ and ΔS⁰ = −126 J mol⁻¹K⁻¹. As expected, these values coincide (within experimental error) with those derived from the O–H stretching band; which gives further confidence in the results obtained. The average of both sets of results yields the final values of ΔH⁰ = −30.2 kJ mol⁻¹ and ΔS⁰ = −125 J mol⁻¹K⁻¹. The estimated error limits are ±1 kJ mol⁻¹ for enthalpy and ±10 J mol⁻¹K⁻¹ for entropy.

Fig. 5 (a) Plot of the left-hand side of eqn (4) against reciprocal temperature for CO₂ adsorbed on H-FER; data obtained from the O–H stretching band at 3603 cm⁻¹. (b) Plot of the left-hand side of eqn (3) against reciprocal temperature for CO₂ adsorbed on H-FER; data obtained from the ν₃ stretching band at 2346 cm⁻¹. ■ and ● refer to two independent series of measurements for each case.

4. Discussion

For the H-FER sample having Si:Al = 8:1, calculations show that ΔH⁰ is in the range of −33 to −36 kJ mol⁻¹ for CO₂ adsorption on single Brønsted acid sites, while for CO₂ adsorption on dual Brønsted acid sites ΔH⁰ takes values from −37 to −42 kJ mol⁻¹. Experimentally an overall value was obtained giving ΔH⁰ = −30 kJ mol⁻¹. Adsorption enthalpies calculated for single Brønsted acid sites are on average 4 kJ mol⁻¹ larger than the experimentally observed value while those calculated for dual Brønsted acid sites are 7–12 kJ mol⁻¹ overestimated. Calculations clearly show that the CO₂ adsorption complexes on the dual Brønsted acid sites would be formed providing that a pair of Brønsted acid sites characterized by r(H–H) in the range of 5–8 Å exists; the stability of such complexes decreases with increasing r(H–H).

The experimentally observed IR absorption bands corresponding to free Brønsted OH groups and to those involved in the adsorption complex with CO₂ show maxima at 3603 and 3500 cm⁻¹, respectively. Corresponding calculated frequencies are in the range of 3600–3607 cm⁻¹ for free OH,⁵⁰ while for OH involved in adsorption complexes they are in the range of 3461–3516 cm⁻¹ for single sites and 3514–3584 cm⁻¹ for dual sites. For CO₂ complexes on single sites the calculated ν_OH values match very well the corresponding experimental value (broad band centred at 3500 cm⁻¹). However, for complexes on dual sites the calculated frequencies are too high when compared with experimental results. Moreover, the ν/r correlation used for ν_OH calculations tends to slightly underestimate the frequencies of Brønsted OH groups involved in H-bonded complexes, as shown for the case of N₂ in H-FER.⁵⁰ Based on these arguments, it seems safe to conclude that the experimentally observed band at 3500 cm⁻¹ corresponds mainly to the CO₂ adsorption complexes on single Brønsted acid sites. We cannot rule out a small contribution from adsorption complexes on dual sites, however, the fact that the band maxima does not show any significant shift at low CO₂ coverage provides additional evidence that the contribution of dual sites (if any) is small.

Contrary to the arguments presented above the calculations on H-FER(Si/Al = 8) model show the existence of dual Brønsted acid sites. However, this may be related to the particular Al distribution being considered in the model that has been generated assuming the Löwenstein rule⁶⁰ but not respecting the Dempsey rule.⁶¹ Note that ²⁹Si NMR studies on NH₄-FER(Si/Al = 8.4)⁶² show that the fraction of Si atoms having two adjacent AlO₄ tetrahedra, Si(2Al), was only 2.2% of the total framework Si atoms, while the H-FER model used in our calculations implies a much larger proportion (7.8%) of such Si(2Al) units in the framework (and even a 1.5% of Si(3Al) units). Calculations show that dual Brønsted acid sites do stabilize CO₂ adsorption complexes, however, on account of the above argument they could be scarce in the actual H-FER sample investigated.

DFT calculations underestimate adsorption enthalpy of CO₂ on H-FER by about 13 kJ mol⁻¹ (Table 1) mainly due to an inherent deficiency of DFT for the description of weak inter-molecular interactions. It is clear that a reliable correction scheme should be used for accurate description of inter-molecular interactions in zeolites. The DFT/CC method used here proved to be quite appropriate, since it leads to significantly better results (overestimation of ΔH⁰ by only 4 kJ mol⁻¹). A small overestimation of the interaction energies at the DFT/CC level has been also reported for other adsorption complexes.^63,64 In addition to DFT/CC calculations, the DFT-D method described in ref. 65 was also used for the correction of intermolecular (dispersion) interactions. ΔE_DFT-D calculations were performed with the MOLDRAW program,⁶⁶ using PBE functional and s₆ = 0.75 (also reported in Table 1). The ΔE_DFT-D corrections were 1–3 kJ mol⁻¹ larger than those obtained at the DFT/CC level. Thus, ΔH⁰ values obtained at the DFT-D level were about 6 kJ mol⁻¹ overestimated with respect to experimental ΔH⁰. The performance of the DFT-D method for the description of molecular crystals has been recently discussed.⁶⁷

Regarding the asymmetric stretching frequency (ν₃) of the adsorbed CO₂ molecule, the experimentally found value for CO₂ on H-FER is ν₃ = 2346 cm⁻¹, which is 3 cm⁻¹ smaller than the corresponding value for free CO₂. However, it should be noticed that CO₂ adsorbed on silicalite (a pure silica zeolite analogue) shows ν₃ = 2341 cm⁻¹, hence the experimentally found value for CO₂ in H-FER is blue-shifted with respect to this ref. 23. For free CO₂ the calculated harmonic frequency ω₃ is 2365 cm⁻¹, while for CO₂ in purely siliceous FER we found ω₃ = 2357 cm⁻¹. Calculated ω₃ values for CO₂ adsorption complexes on H-FER range from 2365 to 2378 cm⁻¹, in fair agreement with the experimental value. Calculations show that CO₂ adsorption complexes on single and dual Brønsted acid sites in H-FER cannot be experimentally distinguished based on corresponding ν₃ values. In addition, calculations also show that ω₃ values do not correlated with the O_b–H⋯O angle, neither with ν_OH or ΔH⁰. However, ω₃ does appear to correlate with the H⋯O=C angle in the sense that ω₃ increases when the H⋯O=C angle approaches linearity.

In a broader context, the ΔH⁰ value found in this work for CO₂ adsorption on H-FER can be compared with corresponding values for other protonic and basic zeolites. For that purpose, Table 2 summarizes relevant data reported by several authors.^{22–24,26,31,32,68–72} Briefly, main points worth of notice are: (i) for protonic zeolites the vast majority of reported ΔH⁰ values are within the range of −25 to −30 kJ mol⁻¹, suggesting that the CO₂ adsorption enthalpy in such zeolites is not largely dependent on structure type; (ii) with the exception of caesium, alkali-metal exchanged zeolites show ΔH⁰ values significantly higher than protonic zeolites and, as expected, such values increase (from Rb⁺ to Li⁺) when the polarising power (charge/radius ratio) of the cation increases; (iii) the maximum ΔH⁰ value shown in Table 2 corresponds to Ca-CHA, again showing the effect of cation polarizing power. Altogether it is clear that, by choosing the right zeolite, ΔH⁰ can be conveniently tuned within a wide range spanning well over 40 kJ mol⁻¹; and that should facilitate the choice of a suitable adsorbent for materials based CCS. Note that the optimum value of ΔH⁰ would depend on specific parameters (such as temperature and pressure) relevant to the actual CCS process being engineered.

Table 2 Adsorption enthalpy of CO₂ on protonic and basic zeolites

Zeolite	−ΔH^0 (kJ mol^−1)	Method^b	Ref.
a Depending on Si/Al ratio. b VTIR: Variable-temperature IR spectroscopy; CP: chromatography pulse techniques; Qst: isosteric heat of adsorption; Cal: calorimetry.
H-FER	30.0	VTIR	This work
H-ZSM-5	26.5	CP	68
	28.8	Qst	69
	32–35^a	CP	24
H-Y	21–30^a	CP	24
	27	Qst	31
Li-ZSM-5	58.9	Qst	69
Na-FER	45–52^a	Qst	26
Na-ZSM-5	50.0	Qst	32
	49.0	Cal	22,23
	46.3	Qst	69
Na-Y	36	CP	24
Na-X	49.1	Qst	32
	48.1	Cal	70
	50	Qst	71
K-ZSM-5	44.1	Cal	70,23
	36.0	Qst	69
Rb-ZSM-5	34.9	Qst	69
Cs-ZSM-5	33.0	Qst	69
Ca-CHA	70	Qst	72

---

Conclusions

Adsorption of CO₂ on the zeolite H-FER was investigated by a combination of variable temperature IR spectroscopy and calculations at the periodic DFT level employing the DFT/CC correction. The main conclusions can be summarized as follows:

(i) The experimentally found value of adsorption enthalpy for CO₂ on H-FER (ΔH⁰ = −30 kJ mol⁻¹) is within the range of corresponding values reported for other protonic zeolites, which suggests that for such zeolites CO₂ adsorption enthalpy does not depend significantly on zeolite structure type.

(ii) Calculations show that two types of the CO₂ adsorption complexes can be formed on H-FER, complexes on isolated Brønsted acid sites (single sites) characterized by ν_OH in the range of 3461–3516 cm⁻¹, and complexes where the CO₂ molecule bridges two nearby Brønsted acid sites (dual sites) characterized by ν_OH values in the range of 3514–3584 cm⁻¹. The experimentally observed IR absorption band is centred at about 3500 cm⁻¹ and hence should correspond to CO₂ adsorbed on single sites.

(iii) Adsorption enthalpies calculated at the DFT level are significantly underestimated (by about 13 kJ mol⁻¹); a much better agreement with the experimentally found ΔH⁰ value was obtained at the DFT/CC level, which gives an overestimation of about 4 kJ mol⁻¹. About half of the overall interaction energy is due to weak intermolecular interactions between CO₂ and zeolite framework atoms.

Acknowledgements

Work in Prague was supported by research projects 203/09/0143 (GA ČR), MSM0021620857 and LC512 (MŠMT ČR). Access to the METACentrum computing facilities provided by the research intent MSM6383917201 is acknowledged. The Spanish MICINN is gratefully acknowledged for financial support: PCI2006-A7-0618 and for a post-doctoral fellowship to AP.

References

1. A. W. C. van den Berg and C. Otero Arean, Chem. Commun., 2008, 668.
2. M. Z. Jacobson, Energy Environ. Sci., 2009, 2, 148.
3. F. M. Orr, Energy Environ. Sci., 2009, 2, 449.
4. J. Davidson, Energy, 2007, 32, 1163.
5. E. J. Stone, J. A. Lowe and K. P. Shine, Energy Environ. Sci., 2009, 2, 81.
6. K. Z. House, C. F. Harvey, M. J. Aziz and D. P. Schrag, Energy Environ. Sci., 2009, 2, 193.
7. B. Thitakamal, A. Veawab and A. Aroonwilas, Int. J. Greenhouse Gas Control, 2007, 1, 318.
8. T. C. Drage, A. Arenillas, K. M. Smith and C. E. Snape, Microporous Mesoporous Mater., 2008, 116, 504.
9. V. Zelenak, M. Badanicova, D. Halamova, J. Cejka, A. Zukal, N. Murafa and G. Goerigk, Chem. Eng. J., 2008, 144, 336.
10. J. Wei, J. Shi, H. Pan, W. Zhao, Q. Ye and Y. Shi, Microporous Mesoporous Mater., 2008, 116, 394.
11. M. B. Yue, L. B. Sun, Y. Cao, Y. Wang, Z. J. Wang and J. H. Zhu, Chem.–Eur. J., 2008, 14, 3442.
12. J. Prezepiorski, M. Skrodzewicz and A. W. Morawski, Appl. Surf. Sci., 2004, 225, 235.
13. L. Zhou, X. Liu, J. Li, N. Wang, Z. Wang and Y. Zhou, Chem. Phys. Lett., 2005, 413, 6.
14. C. Lu, B. Hsunling, B. Wu, F. Su and J. F. Hwang, Energy Fuels, 2008, 22, 3050.
15. S. Ottiger, R. Pini, G. Storti and M. Mazzotti, Langmuir, 2008, 24, 9531.
16. M. Mastalerz, A. Drobniak and J. Rupp, Energy Fuels, 2008, 22, 4049.
17. A. R. Millward and O. M. Yaghi, J. Am. Chem. Soc., 2005, 127, 17998.
18. P. L. Llewellyn, S. Bourrelly, C. Serre, Y. Filinchuk and G. Férey, Angew. Chem., Int. Ed., 2006, 45, 7751.
19. P. L. Llewellyn, S. Bourrelly, C. Serre, A. Vimont, M. Daturi, L. Hamon, G. De Weireld, J. S. Chang, D. Y. Hong, Y. K. Hwang, S. H. Jhung and G. Férey, Langmuir, 2008, 24, 7245.
20. S. R. Caskey, A. G. Wong-Foy and A. J. Matzger, J. Am. Chem. Soc., 2008, 130, 10870.
21. R. S. Franchi, P. J. E. Harlick and A. Sayari, Ind. Eng. Chem. Res., 2005, 44, 8007.
22. B. Bonelli, B. Onida, B. Fubini, C. Otero Arean and E. Garrone, Langmuir, 2000, 16, 4976.
23. B. Bonelli, B. Civalleri, B. Fubini, P. Ugliengo, C. Otero Arean and E. Garrone, J. Phys. Chem. B, 2000, 104, 10978.
24. P. J. E. Harlick and F. H. Tezel, Microporous Mesoporous Mater., 2004, 76, 71.
25. S. K. Wirawan and D. Creaser, Microporous Mesoporous Mater., 2006, 91, 196.
26. A. Pulido, P. Nachtigall, A. Zukal, I. Domínguez and J. Cejka, J. Phys. Chem. C, 2009, 113, 2928.
27. P. Galhotra, J. G. Navea, S. C. Larsen and V. H. Grassian, Energy Environ. Sci., 2009, 2, 401.
28. J. Pawlesa, A. Zukal and J. Cejka, Adsorption, 2007, 13, 257.
29. T. Montanari and G. Busca, Vib. Spectrosc., 2008, 46, 45.
30. R. W. Stevens, R. V. Siriwardane and J. Logan, Energy Fuels, 2008, 22, 3070.
31. J. Pires, M. Brotas de Carvalho, F. Ramoa Ribeiro and E. G. Derouane, J. Mol. Catal., 1993, 85, 295.
32. J. A. Dunne, M. Rao, S. Sircar, R. J. Gorte and A. L. Myers, Langmuir, 1996, 12, 5896.
33. D. F. Plant, G. Maurin, I. Deroche, L. Gaverova and P. L. Llewellyn, Chem. Phys. Lett., 2006, 426, 387.
34. G. Maurin, P. L. Llewellyn and R. G. Bell, J. Phys. Chem. B, 2005, 109, 16084.
35. D. F. Plant, G. Maurin, I. Deroche and P. L. Llewellyn, Microporous Mesoporous Mater., 2007, 99, 70.
36. A. A. Tsyganenko, P. Yu. Storozhev and C. Otero Arean, Kinet. Catal., 2004, 45, 530.
37. C. O. Arean, O. V. Manoilova, A. A. Tsyganenko, G. T. Palomino, M. P. Mentruit, F. Geobaldo and E. Garrone, Eur. J. Inorg. Chem., 2001, 1739.
38. E. Garrone and C. Otero Arean, Chem. Soc. Rev., 2005, 34, 846.
39. C. O. Arean, D. Nachtigallova, P. Nachtigall, E. Garrone and M. R. Delgado, Phys. Chem. Chem. Phys., 2007, 9, 1421.
40. C. Otero Arean, O. V. Manoilova, G. Turnes Palomino, M. Rodriguez Delgado, A. A. Tsyganenko, B. Bonelli and E. Garrone, Phys. Chem. Chem. Phys., 2002, 4, 5713.
41. P. A. Vaughan, Acta Crystallogr., 1966, 21, 983.
42. http://www.iza-structure.org/databases/ .
43. O. Bludsky, M. Silhan, P. Nachtigall, T. Bucko, L. Benco and J. Hafner, J. Phys. Chem. B, 2005, 109, 9631.
44. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 48, 13115.
45. G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 49, 14251.
46. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169.
47. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865.
48. P. E. Blochl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953.
49. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758.
50. P. Nachtigall, O. Bludsky, L. Grajciar, D. Nachtigallova, M. R. Delgado and C. O. Arean, Phys. Chem. Chem. Phys., 2009, 11, 791.
51. F. D. Murnaghan, The compressibility of Media under Extreme Pressures, National Academy of Science, 1944.
52. F. Birch, Phys. Rev., 1947, 71, 809.
53. M. Sierka and A. Sauer, Faraday Discuss., 1997, 106, 41.
54. J. D. Gale, J. Chem. Soc., Faraday Trans., 1997, 93, 629.
55. O. Bludsky, M. Rubes, P. Soldan and P. Nachtigall, J. Chem. Phys., 2008, 128, 114102.
56. G. Herzberg, Molecular Spectra and Molecular Structure, van Nostrand Reinhold: New York, 1950.
57. K. E. Dierenfeldt, J. Chem. Educ., 1995, 72, 281.
58. P. A. Jacobs, F. M. Van Cauwelaert and E. F. Vansant, J. Chem. Soc., Faraday Trans. 1, 1973, 69, 2130.
59. J. C. Lavalley, Catal. Today, 1996, 27, 377.
60. W. Löwenstein, Am. Mineral., 1954, 39, 92.
61. E. Dempsey, J. H. Kühl and D. H. Olson, J. Phys. Chem., 1969, 73, 387.
62. P. Sarv, B. Wichterlova and J. Cejka, J. Phys. Chem. B, 1998, 102, 1372.
63. M. Rubes, P. Nachtigall, J. Vondrasek and O. Bludsky, J. Phys. Chem. C, 2009, 113, 8412.
64. M. Rubes and O. Bludsky, ChemPhysChem, 2009, 10, 1868.
65. S. Grimme, J. Comput. Chem., 2006, 27, 1787.
66. P. Ugliengo, D. Viterbo and G. Chiari, Z. Kristallogr., 1993, 207, 9.
67. B. Civalleri, C. Zicovich-Wilson, L. Valenzano and P. Ugliengo, CrystEngComm, 2008, 10, 405.
68. V. R. Choudhary and S. Mayadevi, Sep. Sci. Technol., 1993, 28, 2197.
69. T. Yamazaki, M. Katoh, S. Ozawa and Y. Ogino, Mol. Phys., 1993, 80, 313.
70. R. M. Barrer and R. M. Gibbons, Trans. Faraday Soc., 1965, 61, 948.
71. E. A. Vasileva, S. S. Khvoshchev and I. V. Karetina, Izv. Akad. Nauk SSSR, Ser. Khim., 1984, 9, 1942.
72. S. S. Khvoshchev and A. V. Zverev, Zeolites, 1991, 11, 742.

---

Footnote

† Electronic supplementary information (ESI) available: Details of the DFT/CC method are provided together with information on the reference set and correction functions ε_ij(R_ij). See DOI: 10.1039/b911253g

---