DOI: 10.1002/cssc.

201200270

Controlling the Adsorption Enthalpy of CO2 in Zeolites by

Framework Topology and Composition
Lukš Grajciar,[a] Jiř Čejka,[b] Arnošt Zukal,[b] Carlos Otero Aren,[c] Gemma Turnes Palomino,[c]
and Petr Nachtigall*[a]

Zeolites are often investigated as potential adsorbents for CO2 structural type MFI, FER, FAU, LTA, TUN, IMF, and -SVR are dis-
adsorption and separation. Depending on the zeolite topology cussed in light of results of periodic density functional theory
and composition (Si/Al ratio and extra-framework cations), the calculations corrected for the description of dispersion interac-
CO2 adsorption heats at low coverages vary from 20 to tions. Key factors influencing the stability of CO2 adsorption
60 kJ mol1, and with increasing surface coverage adsorption complexes are identified and discussed at the molecular level.
heats either stay approximately constant or they quickly drop A general model for CO2 adsorption in zeolites and related ma-
down. Experimental adsorption heats obtained for purely sili- terials is proposed and data reported in literature are evaluat-
ceous porous solids and for ion-exchanged zeolites of the ed with regard to the proposed model.

Introduction

Oil, natural gas, and coal together account for, at present, the potential to be less energy intensive than regeneration of
about 80 % of the primary energy supply. It is worthwhile chemical CO2 absorbents. The main types of porous materials
noting that these three carbon-based energy sources (known currently under active research for CO2 separation are porous
as fossil fuels) made the rapid development of our technologi- carbons,[6] metal–organic frameworks (MOFs),[6a, 7] zeolites,[8]
cal civilization possible, leading to the present level of well- and amine-functionalized ordered mesoporous silica.[6a, 8a, 9]
being and comfort. However, CO2 produced by burning fossil Propelled by the need to find a cost-effective solution for CCS,
fuels, together with that generated by several industrial pro- research on CO2 sorbents is currently attracting considerable
cesses (cement manufacturing, iron and steel production, and attention,[9a, b, 10] and several excellent reviews on this research
petrochemical plants, among others), results in it being pres- field (including both physical and chemical sorbents for CO2)
ently vented into the atmosphere at a rate of about 28 billion have appeared.[8a, d, 11] Among porous materials for CO2 capture,
tons per year. The consequent increase in the greenhouse nowadays MOFs constitute the main focus of very active re-
effect, which can adversely affect climate, is causing worldwide search.[12] Nevertheless, for several reasons (see below), zeolites
concern. Replacing fossil fuels with renewable, and cleaner, continue to be attractive in this field. Herein, we focus on the
energy sources could provide a way out of this problem in the mechanism of CO2 adsorption in zeolites and mesoporous
long run, but we still need a mid-term solution to allow the silicas.
humanity to continue using fossil fuels until cost-effective re- The main advantages of zeolites are low cost and high ther-
newable energy can be implemented on a large scale. Carbon mal stability, as well as easy cation exchange, which facilitate
dioxide capture and sequestration (CCS)[1] could constitute part the tuning of gas–solid interaction energy. Such interaction
of that mid-term solution, particularly if current research in this energy is the main factor to consider when screening porous
area brings about a significant reduction of cost.
Implementation of CCS from the flue gases of stationary
[a] L. Grajciar, Prof. P. Nachtigall
sources (such as fossil-fuel power plants) can be accomplished Department of Physical and Macromolecular Chemistry
by using liquid amine-based (or ammonia-based) chemical ab- Faculty of Science, Charles University in Prague
sorbents,[2] but, besides being energy intensive and expen- Hlavova 2030, CZ-128 40 Prague 2 (Czech Republic)
Fax: (+ 420) 221-951-289
sive,[3] that technology can pose environmental hazards de-
E-mail: [email protected]
rived from accidental spills;[4] hence, it is convenient to search
[b] Prof. J. Čejka, Dr. A. Zukal
for less expensive and safer CO2-capturing media. Among J. Heyrovský Institute of Physical Chemistry
them, porous solids capable of separating CO2 from flue gases Academy of Sciences of the Czech Republic
by physical adsorption (instead of chemical absorption) consti- Dolejškova 3, Prague 8, 182 23 (Czech Republic)
tute a main topic of current research. [c] Prof. C. Otero Aren, G. Turnes Palomino
Department of Chemistry
Besides separating CO2 from N2 in flue gases of power sta-
University of the Balearic Islands
tions, porous adsorbents are also used to separate CO2 from Palma de Mallorca (Spain)
methane in natural gas and landfill gas. In both cases pressure Supporting Information for this article is available on the WWW under
swing adsorption (PSA) processes can be used,[5] which have http://dx.doi.org/10.1002/cssc.201200270.

ChemSusChem 2012, 5, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 1
These are not the final page numbers! ÞÞ
 P. Nachtigall et al.

adsorbents for gas separation by PSA processes; the differen- peated four times). Li, Cs, and K forms were then prepared from
tial stability of adsorption complexes formed by different com- the parent Na forms under the same conditions by using the re-
ponents of a gas mixture should be large enough to enhance spective metal nitrates or chlorides. After ion exchange, the sam-
ples were thoroughly washed with distilled water and dried over-
separation selectivity, while keeping the stability low enough
night.
to facilitate regeneration of the adsorbent with low energy ex-
penditure. To find the most suitable adsorbent for a particular
application, it is essential to understand the mechanism of CO2 Adsorption measurements
adsorption at the molecular level.
Adsorption isotherms of nitrogen at 77 K and carbon dioxide in
The interaction of CO2 with alkali-metal-exchanged zeolites
the temperature range from 273 to 333 K were recorded by using
has been investigated computationally by employing DFT and an ASAP 2020 (Micromeritics) static volumetric apparatus. To attain
various models of adsorbent (e.g., bare metal cations,[13] large sufficient accuracy in the accumulation of the adsorption data, the
cluster models,[14] periodic models of zeolites[15]) or interatomic ASAP 2020 instrument was equipped with pressure transducers
potentials and grand canonical Monte Carlo simulations covering the 0.133, 1.33, and 133 kPa ranges. Generally, before the
(GCMC).[16] While GCMC simulations give access to thermody- adsorption measurements, the individual samples were outgassed
namic features relevant to higher coverage and temperature, under a turbomolecular pump vacuum by using a heating pro-
the accuracy of these simulations depends heavily on the qual- gram that afforded slow removal of humidity at a relatively low
temperature. Starting at ambient temperature, the samples were
ity of the interatomic potential parameters employed. Several
outgassed at 383 K until a residual pressure of 0.5 Pa was attained.
interatomic potentials were used to describe the interaction of After further heating at 383 K for 1 h, the temperature was in-
CO2 with zeolite adsorbents, and in the majority of cases these creased up to 623 K and maintained for 8 h. A homemade thermo-
interatomic potentials were adjusted to reproduce experimen- stat capable of maintaining the temperature of the sample within
tal adsorption isotherms or adsorption heats. The use of DFT an accuracy of  0.01 K was used for the measurements of carbon
(employing standard generalized-gradient approximation dioxide adsorption at 273, 293, 313, and 333 K. The corresponding
(GGA)-type exchange-correlation functionals) is even more nitrogen adsorption isotherm was obtained prior to measurements
problematic: first, because only data relevant for the zero cov- with carbon dioxide, for which the sample was regenerated. The
outgassing procedure was performed at 523 K for 12 h under a tur-
erage limit is obtained and, second, because of the inability of
bomolecular pump vacuum. The only exception to this sample ac-
GGA functionals to account properly for dispersion interac- tivation procedure was that used for the Na-A zeolite, which, like
tions, which constitutes a major drawback. other low-silica zeolites, tended to form (irreversibly) surface carbo-
The main goal of the investigation reported herein was to nates.[8g, 24] For this reason the sample was saturated with CO2 prior
combine the highly accurate ab initio description of the inter- to the experiments and left to stand until surface carbonate forma-
action of CO2 with zeolites with experimentally determined ad- tion ended. Subsequently, free CO2 was outgassed and volumetric
sorption heats to understand the mechanism of CO2 interac- adsorption measurements were started. Further details about this
tion with zeolites at the molecular level. For that purpose, we activation procedure can be found elsewhere.[19]
combined experimentally determined isosteric heats of adsorp-
tion with DFT/coupled cluster (CC)[17] calculations (recently Calculations
shown to give a very accurate description of adsorbate–ad-
sorbent interactions[18]) on periodic zeolite models. The role of Results reported herein for CO2 interaction with Silicalite-1, Na-
ZSM-5, siliceous LTA, and Na-Y adsorbents were all obtained by
different factors that determine the gas–solid interaction
using a periodic model and the DFT method implemented in the
energy, such as zeolite topology and zeolite chemical composi- VASP program package[25] and augmented for the description of
tion, is thus analyzed. Purely siliceous polymorphs and several dispersion interactions[17] (described below). The exchange-correla-
alkali-metal-exchanged zeolites (sodium-exchanged zeolites, in tion functional of Perdew, Becke, and Ernzerhof (PBE)[26] was em-
particular) with different framework topologies and Si/Al ratios ployed together with the projector-augmented wave approxima-
were both investigated. tion (PAW) of Blçchl[25c, 27] and the plane wave basis set with a kinet-
ic energy cutoff of 400 eV. Brillouin-zone sampling was restricted
to the G point. The same level of theory was also used in recent
theoretical investigations of CO2 adsorption in alkali-metal-ex-
Experimental Section changed FER zeolites[15b, 20, 28] and in zeolite Na-LTA.[19]
Calculations on Silicalite-1 were performed with the orthorhombic
Materials
unit cell (UC) optimized previously (Si96O192, a = 20.241 , b =
Zeolite Na-A (Si/Al ratio = 1) reported in this contribution was ob- 20.001 , c = 13.514 ).[29] A UC of the same volume was also used
tained from the National Bureau of Standards (US).[19] Zeolites Li-Y, as a model of the high-silica Na-ZSM-5, represented by
K-Y, and CsY were prepared by ion exchange of a parent Na-Y Si96nAlnNanO192 (n = 1 and 2). The Y zeolite was represented by a re-
sample with an Si/Al ratio of 2.5.[15a] Zeolite ZSM-5 was purchased duced UC with the composition Si35Al13Na13O96 (Si/Al = 2.69:1) and
from Zeolyst. Other zeolites, namely, ferrierite,[20] MCM-22,[21] ZSM- cell parameters a = b = g = 608 and a = b = c = 17.37 . The experi-
11,[22] TNU-9,[22] IM-5,[22] and SSZ-74,[22] as well as the mesoporous mental cubic UC was used for calculations on purely siliceous LTA
molecular sieve SBA-15,[23] were synthesized in the J. Heyrovsky In- (a = b = c = 11.87 ).[30] Calculations on Na-A and various ferrierites
stitute of Physical Chemistry (Prague). All molecular sieves were were described previously.[19, 20, 28]
transformed into the sodium form by ion exchange with sodium Dispersion forces contribute substantially to the total interaction
nitrate (0.5 m, 100 mL of the aqueous solution per 1 g of material; energy between CO2 and microporous adsorbents,[16a, b, d] and
ion exchange was carried out at room temperature for 4 h and re- hence, they have to be accounted for in calculations. It was recent-

&2& www.chemsuschem.org  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 13
ÝÝ These are not the final page numbers!
Adsorption Enthalpy of CO2 in Zeolites

ly shown that standard exchange-correlation functionals (not ac- work cation positions can be distinguished:[32] 1) channel wall sites,
counting for nonlocal interactions) strongly underestimate the where cations are located on top of, or inside, a ring located on
total interaction energy between zeolites and adsorbed CO2. the surface of the channel; and 2) intersection sites, where cations
Hence, an appropriate correction for dispersion interactions must are located on the edge formed by two intersecting channels.
necessarily be applied.[20, 28] A combined DFT/CC method recently Compared with cations located at intersection sites, those located
developed was shown to give results in a very good agreement in channel wall sites typically interact with more framework
with experimental data,[20, 28] and for this reason it was adopted in oxygen atoms, are buried deeper into the channel surface, and
the present investigation. The DFT/CC method is based on the have a lower ability to bind molecules of adsorbate. Examples of
pairwise representability of the DFT error, DEDFT/CC, defined as the channel wall sites and intersection sites can be found in Figure 1
difference between the precise CCSD(T) and DFT interaction (e.g., Z6/T10 and I2/T6 sites, respectively). Some necessary details
energy [Eq. (1)]:[18a, 31] about specific zeolites investigated herein are briefly reviewed
X below.
DEDFT=CC ¼ DECC  DEDFT ¼ eij ðRij Þ ð1Þ
ij
Na-ZSM-5: The CO2 adsorption complexes formed on the Na +
cation located in the vicinity of framework AlO4 tetrahedra in the
T4, T6, T10, T11, and T12 positions (adopting the numbering
in which DEDFT and DECC are interaction energies calculated for the scheme from Ref. [33]) were considered (Figure 1). A detailed de-
reference system at the DFT and coupled cluster CCSD(T) levels, re- scription of these Na + sites in Na-ZSM-5 can be found in Ref. [29].
spectively, eij are the correction functions depending on the intera- Notation for cation sites introduced previously was adopted:[34]
tomic distances Rij, and i and j runs over atoms in adsorbent and intersection sites, sites in the main channel, and sites in the zigzag
adsorbate, respectively. The correction functions for CO2 interaction channel are denoted as “I”, “M”, and “Z” sites, respectively
with zeolites can be found in the Supporting Information of (Figure 1).
Ref. [31], together with more details about this method.
The geometry of each CO2 adsorption complex was optimized at (Li-, Na-, K-)FER: All details about the models and methods used
the DFT/CC level; atomic positions of both adsorbate and adsorb- for the investigation of FER zeolites can be found in Refs. [20] (Li-
ent were fully relaxed, while the size and shape of the UC were FER, Na-FER, and K-FER) and [28] (siliceous FER).
constrained. The zero-point vibrational energy corrections, DEZPVE, Na-Y: Framework Al distribution was generated randomly within
were evaluated within the harmonic approximation considering the constraints defined by the Lçwenstein and Dempsey rules.[35]
nine degrees of freedom for the CO2 molecule; the second deriva- The distribution that matched the results of 29Si NMR spectroscopy
tives were calculated numerically by using  0.005  displace- obtained for the Y zeolite with Si/Al = 2.7:1 was selected for calcu-
ments. An almost constant value of DEZPVE was found for CO2 inter- lations. Na + cations were distributed in the following way: 3, 8,
actions with FER zeolites;[20] therefore, a constant value of DEZPVE = and 2 in sites I, II, and I’, respectively.[36] Note that this is a unique
1 and 2 kJ mol1 was used for Silicalite-1 and Na-MFI, respectively. cation distribution where all sites II are fully occupied and one of
Adsorption enthalpies DH0 [Eq. (3)] were calculated for the process four sites I is replaced by a pair of I’ sites. Note also that only eight
shown in Equation (2) as a sum of interaction energies calculated Na + cations in sites II are accessible for CO2 molecules.
at the DFT level (DEDFT), DFT/CC correction term (DEDFT/CC), and cor-
rection for the zero-point vibrational energy (DEZPVE): Na-A: Calculations on the CO2/Na-A system were described in
detail in Ref. [19]; the following notation for Na + sites was adopt-
ðCO2 Þg þ Zeolite ! CO2    Zeolite ð2Þ ed: Na + cations located inside the 6-ring, in the 8-ring, and on top
of the 4-ring were denoted S1, S2, and S3, respectively.
DH0 ¼ DE DFT þ DE DFT=CC þ DE ZPVE ð3Þ

It must be stressed that all calculations discussed herein (both new Generalized model for CO2 adsorption in zeolites
results reported for Silicalite-1, Na-MFI, Na-FAU, and siliceous LTA
and recently published results for Na-LTA, M-FER, and siliceous FER) Based on a systematic investigation of carbon monoxide adsorp-
were carried out at the same level of theory (DFT/CC) and consis- tion in metal-exchanged zeolites by a combination of experimental
tent model (periodic model of
each particular UC). Therefore, the
results obtained can be directly
compared with each other.

Notation used for adsorption

sites
The cationic CO2 adsorption sites
in zeolites are discussed, in partic-
ular, for zeolites belonging to the
structural type MFI, FER, FAU, and
LTA, for which, first, there are
both experimental and theoretical Figure 1. Extra-framework cation sites for Na + in Na-ZSM-5, viewed along the main channel (left) and the zigzag
results available and, second, channel (right). The Na + cation located in the Z8, Z6, M6, and I2 site is depicted as a green, violet, orange, and
these zeolites cover the entire blue ball, respectively. The Z8 and Z6 sites are located in the zigzag channel, whereas the M6 site is in the main
range of allowed Si/Al ratios (from channel; the I2 site is in the channel intersection. Framework Si, Al, and O atoms are shown in gray, black, and
1 to 1). Two types of extra-frame- red, respectively.

ChemSusChem 0000, 00, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org &3&
These are not the final page numbers! ÞÞ
 P. Nachtigall et al.

and theoretical methods, a general model for adsorption in micro- Dispersion interactions: Dispersion interactions are maximal when
porous materials was proposed by some of us.[37] This model was interacting atoms are at a distance equal to the sum of van der
recently verified for CO2 interacting with alkali-metal-exchanged Waals radii and they decay quickly (r6 in the asymptotic region)
zeolites.[20] It was shown that dispersion interactions between the with increasing atom separation. Thus, dispersion interactions are
adsorbate and the adsorbent played a much greater role in the particularly important for adsorption in microporous channels,
case of CO2 (where they accounted roughly for about one half of since the molecule of the adsorbate (CO2) is surrounded by frame-
the overall interactions) than in the case of CO. The original model work atoms.
described elsewhere[37] was adopted herein, emphasizing features In the case of CO2 adsorption, both effects, from the top and
relevant to the CO2 molecule: increased importance of dispersion bottom, are predominantly of electrostatic origin, and hence, they
interactions and electrostatic interactions due to the electric quad- are described by standard DFT functionals with good accuracy.
rupole moment. Note that both oxygen atoms bear a negative Conversely, dispersion interactions are not correctly described
charge, an ideal situation for simultaneous interaction of the CO2 when using common exchange-correlation functionals within gen-
molecule with two extra-framework cations. The interaction of CO2 eralized-gradient approximation and the correction for the accu-
with microporous adsorbents can be formally divided into three rate description of dispersion interactions is required. The DFT/CC
contributions, as shown in Figure 2; these are described in more correction scheme described above was adopted herein. It should
detail below. be mentioned that the DFT/CC method is defined as a global cor-
rection of DFT (not only dispersion energy correction). However,
for the systems investigated herein the dominant part of the DFT/
CC correction accounted for the lack of dispersion energy. There-
fore, in the discussion below the DEDFT and DEDFT/CC contributions
are considered to be a measure of electrostatic and dispersion in-
teractions, respectively, between the adsorbent and adsorbate.

Results and Discussion

The strength and mechanism of CO2 interaction with zeolites
depend on many parameters, and different types of interac-
tions dominate isosteric heats of adsorption for zeolites with
different concentrations of extra-framework cations. It is there-
fore advantageous to classify the zeolitic materials based on
their composition: 1) purely siliceous materials (Si/Al!1),
Figure 2. Generalized model of CO2 interacting with a zeolite. CO2 interacts 2) high-silica zeolites (Si/Al > 12), 3) zeolites with intermediate
with the primary extra-framework metal cation (violet ball at the bottom of Si/Al ratios (12 > Si/Al > 4), and 4) Al-rich zeolites (Si/Al < 4). The
the figure) and it can also interact with a secondary extra-framework metal
cation (violet ball at the top of the figure) if there happens to be one.
results reported herein, along with abundant literature reports,
Dispersion interactions between the CO2 molecule and the zeolite also play are discussed below following this classification.
an important role, as there are many framework atoms all around the CO2
molecule, both within a distance corresponding to the sum of atomic van
der Waals radii and also at a longer distance. CO2 adsorption in purely siliceous molecular sieves
There are no extra-framework cations present in siliceous ma-
Effect from bottom: This effect accounts for the interaction of CO2 terials. Therefore, the analysis of the results obtained for these
with the primary extra-framework metal cation. In the case of alkali
materials should reveal the character and importance of CO2
or alkaline-earth-metal cations in zeolites, effects from bottom are
interactions with the zeolitic framework. Isosteric heats of CO2
dominated by the electrostatic interaction between the metal
cation and the CO2 quadrupole. Therefore, the importance of this adsorption in purely siliceous microporous materials have
effect increases with increasing cation charge, decreasing cation been reported for several structures, including MFI, BEA, and
size, and decreasing coordination of extra-framework cations to LTA topologies.[30, 38] They are summarized in Table 1 together
framework oxygen atoms. The adsorption site where the CO2 mole- with the results for the mesoporous purely siliceous material
cule interacts with only one extra-framework cation was denoted SBA-15. Experimental results can be compared with adsorption
as the single cation (SC) site. enthalpies, DHads (0 K), calculated at the DFT/CC level for LTA
(Si/Al!1) and MFI (Si/Al!1) and with those previously re-
Effect from top: When two extra-framework cations are separated ported for FER (Si/Al!1).[28] A more detailed description of
by a distance between 5 and 10  (depending on cation size and the calculated results is given in the Supporting Information.
charge), the CO2 molecule can interact with both cations simulta- Calculated and experimental results are in good agreement
neously. Such a site, where CO2 interacts with two cations at the and, importantly, their relative values are correctly reproduced.
same time, was denoted as the dual cation (DC) site. An example
Calculations show that the interaction between CO2 and silica-
of CO2 adsorption on a DC site is shown in Figure 2. For zeolites
with very high concentration of accessible extra-framework cations lite is dominated by dispersion interactions that account for at
(e.g., LTA), the CO2 molecule can simultaneously interact even with least 80 % of the overall interaction and in some cases (e.g.,
more than two extra-framework cations;[19] the corresponding ad- CO2 interaction with the 8-ring in FER reported in Ref. [28]) dis-
sorption sites were denoted as multiple cation (MC) sites. persion interactions account for almost 100 % of adsorbate–ad-

&4& www.chemsuschem.org  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 13
ÝÝ These are not the final page numbers!
Adsorption Enthalpy of CO2 in Zeolites

It follows that adsorption heats of CO2 in purely siliceous

Table 1. CO2 adsorption heats in pure-silica zeotypes.
zeotypes should increase with decreasing size of the channel
Density[a] Qst (0 q)[b] DHads (0 K)[c] or cavity and with increasing thickness of the channel wall.
!
Zeotype
topology [SiO2 per 1000 3] [kJ mol1] [kJ mol1] Indeed, data reported in Table 1 shows that Qst correlates with
MFI 18.4 27,[d] 29[e] 26–28 the zeolite density; nonetheless, this correlation does not hold
FER 17.6 23–25[f] for mesoporous SBA-15. The framework density of SBA-15 is
BEA 15.3 21[g] –
7.9 SiO2 per 1000 3 ; however, it is more meaningful to consid-
LTA 14.2 21[h] 20–22
SBA-15 11.0[i] 26[j] – er a density of 11.0 SiO2 per 1000 3, which is half of the densi-
ty of the SBA-15 wall (amorphous silica): CO2 adsorbs on the
[a] From the Database of Zeolite Structures.[39] [b] Experimental isosteric
heats of adsorption at the low coverage limit. [c] Adsorption enthalpies wall of SBA-15 mesopores and it interacts with the atoms of
calculated at the DFT/CC level. [d] From Refs. [38a, c]. [e] From Ref. [38b]. the wall on one side of the CO2 molecule, whereas there are
[f] From Ref. [28]. [g] From Ref. [38c]. [h] From Ref. [30]. [i] Half of the den- no atoms to interact with on the other side. Clearly, the CO2
sity of amorphous silica. [j] From Ref. [10b]. adsorption heat on a smooth SBA-15 mesopore wall should
not exceed 20 kJ mol1; a significantly larger experimentally de-
termined value thus indicates that the surface of the wall is
sorbent interactions. Consequently, the CO2 molecules are ad- corrugated and that CO2 is not adsorbed on the flat silica sur-
sorbed in the vicinity of the channel or cavity wall at the dis- face. This is in line with recent reports evaluating the surface
tance given approximately by the sum of van der Waals roughness of SBA-15 molecular sieves.[41]
radii,[40] as shown in Figure 3 a for CO2 adsorbed in LTA (Si/Al!
1). CO2 is located inside the LTA supercage close to the D4R
CO2 adsorption in high-silica zeolites
unit to maximize dispersion interactions (geometry with the
The concentration of extra-framework cations in high-silica
zeolites is expected to be too low to allow the formation of
DC sites suitable for adsorption of CO2 (this is true only for
samples with a homogeneous concentration of framework alu-
minum throughout the crystal; this point is addressed in
a greater detail below). Assuming there are no DC sites in
a high-silica sample, the interaction of CO2 with the zeolite is
driven by the (specific) electrostatic interaction with the extra-
framework cations at SC sites and it is further increased by the
(nonspecific) dispersion interaction with the zeolite framework.
Therefore, adsorption heats are driven by the effect from
Figure 3. Adsorption of CO2 in purely siliceous materials: LTA (a) and MFI (b).
Atoms of the CO2 molecule are depicted as balls; framework Si and O atoms bottom, which reflects the coordination of the extra-framework
are shown as gray and red wires, respectively. Atoms of the 6-ring and D4R cation with the framework.
of LTA closest to adsorbed CO2 are depicted in tube mode (a). MFI is viewed Interactions of CO2 with alkali-metal cations in a high-silica
along the zigzag channel direction; Si and O atoms on the main channel
FER zeolite were investigated previously by a combination of
surface are shown in a tube mode (b).
experimental and theoretical approaches.[20] It was shown that
the effect from bottom on the adsorption enthalpy of CO2 de-
creased with increasing cation size. It is also known from a pre-
maximum number of framework atoms at the van der Waals vious investigation of carbon monoxide interaction with lithi-
contact with atoms of the adsorbate). The structure of the CO2 um-exchanged zeolites that the effect from bottom is larger in
adsorption complexes found in the main channel of MFI (Si/ ZSM-5 than that in FER (see, e.g., Refs. [42] and [29] for Na-FER
Al!1) is depicted in Figure 3 b; the CO2 molecule is located and Na-ZSM-5, respectively, or Refs. [43] and [44] for Li-FER and
in the main channel, as far as possible from the channel inter- Li-ZSM-5, respectively). For this reason, we have investigated
sections and the orientation of the CO2 molecule is such that CO2 adsorption in a high-silica Na-ZSM-5 zeolite computation-
the number of framework atoms in its vicinity is maximal. ally by employing the DFT/CC method and a periodic DFT
Calculations also indicate that the potential energy surface model; results are summarized in Table 2.
(PES) for CO2 diffusion through the channels is relatively flat The Na + cation at the Z6/T10 site is located on top of a 6-
corresponding to the fact that the PES is mostly determined ring on the wall of the zigzag channel (Figure 4 a) and it is co-
by the dispersion interactions. It should also be mentioned ordinated to four framework oxygen (Of) atoms. Consequently,
that based on DFT/CC periodic calculations a somewhat pecu- the interaction of CO2 with this Na + cation is the weakest
liar structure has been reported for FER (Si/Al!1), where CO2 among all investigated Na + sites in Na-ZSM-5 (DEDFT =
is located in the middle of an 8-ring window (see Figure 6 a in 22 kJ mol1). The strongest CO2 interaction is found for the I2/
Ref. [28]). A similar structure was also found here for LTA (Si/ T6 intersection site (Figure 4 b), where the Na + cation is coordi-
Al!1); the stability of this structure is driven by an optimal nated to only two Of atoms. The electrostatic component of
van der Waals distance between the carbon atom of CO2 and the interaction energy (corresponding mostly to the CO2 inter-
the eight oxygen atoms of the 8-ring window. action with the extra-framework cation) for the I2/T6 site is

ChemSusChem 0000, 00, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org &5&
These are not the final page numbers! ÞÞ
 P. Nachtigall et al.

plexes are formed) can be taken

Table 2. Calculated characteristics of CO2 adsorption complexes formed in the Na-ZSM-5 zeolite.[a]
as a measure of the effect from
Al position Na + site[b] CO2 location[c] r(NaOf)[d] r(NaOCO) Eint(DFT)[e] DEDFT/CC[f] DH0(0 K)[g] bottom. The difference of
[] [] [kJ mol1] [kJ mol1] [kJ mol1] 5 kJ mol1 observed experimen-
SC site tally is in perfect agreement
T10 Z6 CH 2.36, 2.39, 2.40, 2.68 2.326 22.4 24.2 45 with the results reported in
T4 Z6 CH 2.29, 2.34, 2.63, 2.69 2.324 25.7 24.2 48
Table 2. Experimental and com-
T11 M6 I 2.31, 2.32, 2.49, 2.51 2.320 27.6 19.2 45
T12 Z8 I 2.33, 2.33, 2.63 2.347 31.0 17.4 46 putational results are in very
T6 I2 W 2.28, 2.29 2.345 32.2 20.3 51 good agreement, not only qual-
itatively but also quantitatively
DC site
(difference of about 2 kJ mol1).
T10 I2 2.24, 2.27 2.397 47.1 18.4 64
T12 Z8 2.31, 2.33, 2.55 2.424 Experimental results reported in
Reference [38b] also showed
[a] Calculated at the DFT/CC level by employing a periodic model of the MFI UC. [b] For notation, see Figure 1.
[c] Location of the CO2 molecule within the zeolite channel system; CH, I, and W represent the CO2 molecule in a rather large effect from
the channel, on the intersection, and on the wall at the channel crossing, respectively. [d] Distance between bottom for Li-ZSM-5; about
the Na + cation and framework oxygen atoms. [e] Interaction energy between CO2 and Na-ZSM5 calculated at 10 kJ mol1 measured as the dif-
the DFT level. [f] DFT/CC correction of DFT interaction energy, mostly due to the dispersion interaction. [g] Ad- ference between Qst at low cov-
sorption enthalpy at 0 K calculated as Eint(DFT) + DEDFT/CC + DZPVE; a zero-point vibrational energy correction of
1
2 kJ mol calculated previously for the CO2/Na-FER system was used. [15b] erage and at q = 1, which is
almost twice as much as the dif-
ference observed for Na-ZSM-5
and K-ZSM-5. The isosteric heat
of CO2 on Li-ZSM-5 (Si/Al = 30)
10 kJ mol1 larger than that for the Z6/T10 site (Table 2). Calcu- drops down to 27 kJ mol1 (value reported for Silicalite-1),
lations thus show that the effect from bottom is indeed signifi- which is already below the coverage 1.5:1 = CO2/Li[38b] and is
cant in MFI-type zeolites. The results summarized in Table 2 much faster than in the case of Na- and K-ZSM-5 (Si/Al = 30)
also show a good correlation between cation coordination samples. This indicates that the adsorption capacity of Li + sites
with the framework and its ability to bind CO2 ; channel wall in Li-ZSM-5 is smaller, which is due to the preference of small
sites (Z6/T10, Z6/T4, M6/T11, and Z8/T12) are all characterized Li + cations for channel wall sites[32, 34] that do not allow the for-
by a smaller value of DEDFT than that of the intersection site mation of geminal adsorption complexes;[29, 44] with increasing
(I2/T6). However, the situation is more complex when the dis- cation size there is a growing preference for intersection sites
persion component of the CO2–framework interaction is also where geminal complexes can be easily formed. Consequently,
considered. When CO2 interacts with the Na + cation at the the increasing preference for intersection sites found for larger
channel wall site (where the electrostatic interaction with CO2 cations[34] leads to increased population of geminal complexes
+
is weaker), and when the localization of Na within the zeolite and an increased number of CO2 molecules directly interacting
channel system allows, the CO2 molecule is located as far as with extra-framework cations. Note also that the CO2 adsorp-
possible from the channel intersection to maximize the disper- tion complexes on DC sites are not formed in Na-ZSM-5 (Si/
sion component of its interaction with the framework. The dis- Al = 30), although calculations showed that such complexes
persion interaction is maximal when CO2 is in the zigzag chan- were about 10 kJ mol1 more stable than those of CO2 adsorp-
nel (Figure 4 a), where there is the maximum number of frame- tion complexes on the SC sites.
work atoms around the adsorbate. As seen from the results The detailed description of CO2 adsorption in ZSM-5 zeolites
given in Table 2, the dispersion interaction (DEDFT/CC) is com- given above is remarkably similar to that reported for high-
parable with, or even larger than, the electrostatic interaction silica ferrierites.[20, 28] We believe that this is a general model
(DEDFT) for sites Z6/T10 and Z6/T4. The dispersion interaction that can be used for the description of CO2 adsorption in any
for these sites is up to 7 kJ mol1 larger than that for intersec- high-silica zeolite. Experimental results for several sodium-ex-
tion sites. Consequently, the difference in adsorption enthalpy changed high-silica zeolites are collected in Figure 5, showing
(last column of Table 2) for the channel wall and intersection the dependence of isosteric heats on CO2 loading (expressed
sites is not as large. as the number of CO2 molecules per extra-framework cation).
The isosteric heats of adsorption of CO2 on M-ZSM-5 (Si/Al = The results for Na-ZSM-5 (Si/Al = 35), Na-ZSM-11 (Si/Al = 35),
30; M = Li, Na, K) were recently investigated.[38b] The reported Na-TNU-9 (Si/Al = 35), Na-IM-5 (Si/Al = 30), and Na-SSZ-74 (Si/
results for Na-ZSM-5 can be directly compared with those Al = 100) are reported. All of these zeolites have relatively simi-
quoted in Table 2. Experimental isosteric heats reported for Na- lar topologies: the channel system of all of them is formed by
ZSM-5 (Si/Al = 30) start at 49 kJ mol1 and they gradually de- a three-dimensional network of 10-ring channels (limited in
crease to 46 and 44 kJ mol1 at CO2/Na ratios of 0.5:1.0 and the third dimension for IM-5). All of these high-silica zeolites
1.0:1.0, respectively.[38b] The difference between the isosteric are also available with similar Si/Al ratios (except for SSZ-74,
heats at the low coverage limit (which correspond to the ener- which can only be synthesized with a very high Si/Al ratio).
getically most favorable sites) and those obtained for CO2/ The exact location of alkali-metal cations in the channel sys-
Na = 1:1 (q = 1 indicates the least stable CO2 adsorption com- tems is not known. Isosteric adsorption heats at low coverages

&6& www.chemsuschem.org  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 13
ÝÝ These are not the final page numbers!
Adsorption Enthalpy of CO2 in Zeolites

Figure 5. Isosteric heats of carbon dioxide adsorption in high-silica Na-zeo-

lites. The adsorbed amount is expressed as the number of CO2 molecules
per Na + cation.

able here (intersection and channel wall sites differing in metal

cation/framework coordination) and, in addition, with increas-
ing extra-framework cation concentration (decreasing Si/Al
ratio) the number of DC sites increases. Therefore, the isosteric
heat of adsorption measured for such zeolites does not
depend only on the cation size and coordination (effect from
bottom), but also on the cation concentration and attendant
presence of DC sites (effect from top). It should be noted that
the ratio of SC and DC cation sites depends, in addition to
cation concentration, on the zeolite topology and even (in
some cases) on the synthetic method (see below).
Experimental isosteric heats obtained for Li-, Na-, and K-ex-
changed ferrierites with two different Si/Al ratios [FER (Si/Al =
8.6) and FER (Si/Al = 27) samples] are shown in Figure 6. Re-
Figure 4. CO2 adsorption complexes on Na-ZSM-5. CO2 atoms and Na + cat- sults reported for the high-silica Na-FER sample are similar to
ions are depicted as balls, see caption of Figure 1 for color scheme. ZSM-5 is those reported in Figure 5 for different high-silica zeolites.
viewed along the main channel; the zigzag channel surface is shown in tube Conversely, results reported for the Na-FER sample with Si/Al =
mode. The CO2 adsorption complexes on the Na + cation in the channel wall
8.6:1 are distinctly different (Figure 6). The higher isosteric
site Z6/T10 (a) and in the intersection site I2/T6 (b) are shown, together with
the adsorption complex on a DC site (c). heats obtained at low coverage of CO2 were recently ascribed
to CO2 adsorption complexes on DC sites.[20] Formation of CO2
adsorption complexes on DC sites was also supported by IR
obtained for TNU-9, IM-5, ZSM-5, and ZSM-11 are between 40 spectroscopy, which showed a shift of Dn3 (asymmetric stretch-
and 45 kJ mol1. The values of adsorption heats at low cover- ing vibration) to a higher value than that observed for CO2 ad-
ages indicate the absence (or only a very limited concentra- sorption on SC sites, in agreement with theoretical predic-
tion) of DC sites in these zeolites. For higher coverages, the ad- tions.[15b, 20, 28, 45]
sorption heats decrease to about 30 kJ mol1, which is typical Adsorption of CO2 in alkali-metal-exchanged FER zeolites
for CO2 adsorption in siliceous microporous solids. was also investigated recently by using calorimetric measure-
ments, which allowed the determination of isosteric heats for
very low amounts of adsorbed CO2 (about 1 CO2 per 100 extra-
CO2 adsorption in zeolites with intermediate Si/Al ratios
framework cations, q = 0.01 of a monolayer).[46] Isosteric heats
The largest heterogeneity of CO2 adsorption sites can be obtained at low coverage for Li-FER (Si/Al = 8.6) and Li-FER (Si/
found in zeolites with intermediate Si/Al ratios. The same set Al = 27) were equal, indicating that the amount of DC sites
of SC sites as those found in high-silica materials are also avail- was negligible in both samples. Conversely, for Na-FER and K-

ChemSusChem 0000, 00, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org &7&
These are not the final page numbers! ÞÞ
 P. Nachtigall et al.

ent templates, showed a qualitatively different Qst dependence

of the amount of CO2 adsorbed. In particular, one of the sam-
ples showed a constant value of Qst irrespective of the amount
of CO2 adsorbed, which was explained in terms of a very regu-
lar distribution of aluminum atoms in the zeolite framework
(which are situated as far as possible from one another).

CO2 adsorption in Al-rich zeolites

A combination of volumetric gas adsorption measurements,
variable-temperature (VTIR) spectroscopy,[49] and periodic DFT
calculations was used to investigate CO2 adsorption on the
zeolite Na-A (Si/Al = 1:1).[19] The main findings can be summar-
ized as follows: Due to the large number of Na + cations acces-
sible for adsorbate molecules in each zeolite supercage, all of
the adsorbed CO2 molecules interact with more than one
extra-framework cation, thus forming adsorption complexes on
DC and MC sites. A possible SC site was also found computa-
tionally; however, the calculated adsorption enthalpy
Figure 6. Isosteric heats of carbon dioxide adsorption in ferrierites. The ad-
sorbed amount is expressed as the number (a) of CO2 molecules per cation. (23 kJ mol1) was only about half of that found for DC and
MC sites (44 to 55 kJ mol1). The CO2 molecule in the SC
site would be directed towards the center of the supercage,
which results in a small dispersion interaction energy
FER zeolites, adsorption heats at the q!0 limit measured for (6 kJ mol1) compared with that shown by CO2 adsorbed on
FER (8.6) samples were about 10 kJ mol1 larger than those ob- DC and MC sites (21 to 23 kJ mol1). Therefore, and in
tained for FER (27) samples. The difference between Li-FER and agreement with experimental results, CO2 only adsorbs on DC
Na- and K-FER zeolites is understandable in light of a previous or MC sites in Na-A.
investigation of alkali-metal-cation coordination in zeolites.[43–44] Qualitatively different conclusions were drawn, based on
First, the optimum distance between a pair of extra-framework combined experimental and computational results for CO2 ad-
cations to set up a DC site for the CO2 molecule is 6.5, 7.3, and sorption on Na-Y.[15a] In this case, no CO2 adsorption complexes
8.3  for Li, Na, and K cations, respectively (as determined by on DC sites were found. This observation was explained by the
cluster model calculations). Second, the Li + cation preferably fact that Na + cations in adjacent SII sites (inside the zeolite su-
occupies the channel wall sites, in particular the 6-ring, where percage) were 9.9  apart from each other, a significantly
it can fit in the middle of the ring, and thus, maximize its inter- longer distance than the optimum separation (7.3 ) for a DC
action with the zeolite framework; the probability of formation site.
of a DC site is thus further decreased. The calculations referred to above were performed by Pirn-
Experimental data reported for CO2 adsorption on Na-FER gruber et al. for a model of the Y zeolite with Si/Al = 5:1.[15a]
and K-FER zeolites indicate the presence of a small amount As reported in the Experimental Section, we performed period-
(however, experimentally detectable) of DC sites even for the ic DFT/CC calculations for CO2 adsorption on Na-Y (Si/Al =
FER (Si/Al = 27) zeolites.[20, 46] The presence of DC sites in some 2.7:1); the results obtained are summarized in Table 3. There
high-silica zeolites leads to a large increase of Qst at low are eight accessible Na + cations in our model, which is the
amounts of CO2 adsorbed. Large values of Qst were reported same as in the model used by Pirngruber et al.[15a] Our calcula-
for q!0 for other zeolites with a large Si/Al ratio; for example, tions confirmed that there were no DC sites available for CO2
Qst = 49 kJ mol1 for K-ZSM-5 (Si/Al = 14)[13a] and 61 kJ mol1 for adsorption, not even when the Si/Al ratio was as small as 2.7:1.
Na-ZSM-5 (Si/Al = 14).[47] Therefore, we also investigated the The most stable CO2 adsorption complex (denoted B in
possible formation of CO2 adsorption complexes on DC sites in Table 3) is depicted in Figure 7; the CO2 molecule interacts
Na-ZSM-5. Instead of a systematic investigation of DC sites in with a Na + cation in site II and the dispersion interaction with
Na-ZSM-5 (which is computationally very demanding), we only the framework is maximized when the molecule is tilted to-
considered one particular pair of Na + cations described recent- wards a 4-ring in the supercage wall. Note that an adsorption
ly for CO adsorption in Na-ZSM-5.[29] A very stable CO2 adsorp- complex with a structure similar to that reported in Ref. [15a],
tion complex (DH0 = 64 kJ mol1) was found (Table 2) for this which points towards the center of the supercage, was also
DC site in agreement with experimental reports; the corre- found (C in Table 3) to be a local minimum (only 1 kJ mol1 less
sponding structure is depicted in Figure 4 c. stable than complex B). Further investigations showed that the
It has been reported recently that the dependence of Qst on relative stability of adsorption complexes in Na-Y (Na/Al =
the amount of CO2 adsorbed was also influenced by the tem- 2.7:1) depended on the particular configuration of framework
plate synthesis procedure.[48] Two FER samples with almost Al and Na + atoms. It is not possible to say which of these com-
identical Si/Al ratios (Si/Al = 15.6), but synthesized with differ- plexes represents the global minimum on the PES without fur-

&8& www.chemsuschem.org  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 13
ÝÝ These are not the final page numbers!
Adsorption Enthalpy of CO2 in Zeolites

increase in Qst observed for NaY

Table 3. Calculated characteristics of CO2 adsorption complexes formed in FAU-type zeolites.[a]
for a > 0.8 should be attributed
CO2 site CO2 location[b] r(NaOf)[c] r(NaOCO) Eint(DFT)[d] DEDFT/CC[e] DH0(0 K)[f] to the increasing importance of
[] [] [kJ mol1] [kJ mol1] [kJ mol1] lateral interactions.
A supercage wall 2.25, 2.26, 2.36 2.399 20.0 17.2 35
B supercage wall 2.25, 2.26, 2.35 2.381 21.3 19.8 39
C supercage center 2.25, 2.26, 2.37 2.413 22.1 17.5 38
CO2 adsorption in micro- and
[a] Calculated at the DFT/CC level by employing a periodic model of the FAU primitive cell. [b] See Figure 7.
[c] Distance between the Na + cation and framework oxygen atoms; O atoms of AlO4 tetrahedra are in italic.
mesoporous materials:
[d] Interaction energy between CO2 and FAU calculated at the DFT level. [e] DFT/CC correction of DFT interac- General aspects
tion energy, mostly due to the dispersion interaction. [f] Adsorption enthalpy at 0 K calculated as Eint(DFT) +
DEDFT/CC + DZPVE; a zero-point vibrational energy correction of 2 kJ mol1 calculated previously for the CO2/Na- The results reported above
FER system was used.[15b] show that a clear understanding
of the mechanisms of CO2 ad-

Figure 8. Isosteric heats of CO2 adsorption in NaA, NaX, and NaY zeolites.
The adsorbed amount is expressed as the number of CO2 molecules per
Na + cation.

sorption in zeolites can only be attained when the framework

Figure 7. CO2 adsorption complex on Na-Y. CO2 atoms and Na + cations are
depicted as balls, see caption of Figure 1 for color scheme. The CO2 mole- topology on the one side, and the extra-framework cation size,
cule interacts with a Na + cation at site II and is tilted towards a 4-ring situat- charge, concentration, and distribution, on the other, are si-
ed in the supercage wall. multaneously analyzed. Some general aspects of the interac-
tion of CO2 with micro- and mesoporous materials are critically
reviewed below.
ther investigation. However, it can be safely concluded that no The strongest electrostatic interaction is expected for the
DC sites are available in Na-Y (Si/Al = 2.7:1). The adsorption of smallest alkali-metal cation. However, in some cases, the same
CO2 on DC sites in Na-Y can be expected only when sites III zeolite sample ion-exchanged to the Na, K, or Cs form gives
become occupied by Na + cations and that can only be expect- higher Qst than that of the Li form, see, for example, Figure 6
ed for Si/Al < 2:1.[36] for FER (Si/Al = 27). This behavior can be explained by one of
The dependence of isosteric heats on CO2 loading for zeo- the following effects: First, the concentration of extra-frame-
lites NaA, NaX, and NaY is shown in Figure 8. In the case of work cations is such that the amount of DC sites in the Li form
NaA zeolite, Qst quickly decreases with increasing a; a relatively of the zeolite is negligible, whereas there is an experimentally
low pore volume and high concentration of Na + cations result detectable amount of DC sites in another alkali-metal form of
in only about 0.6 CO2/Na + ratio at 105 Pa. High and quickly de- this zeolite. Second, Li + cations are only located at the channel
creasing Qst observed for the NaX zeolite can be attributed to wall sites, whereas larger cations are located at the intersection
a large concentration of Na + in SIII’ sites. Upon the saturation sites as well.
of Na + cations in SIII’ sites, the isosteric heats reported for NaX For a particular framework topology, the Qst value increases
become very similar to those found for the NaY zeolite; CO2 with increasing cation concentration. First, increased cation
molecules interact dominantly with Na + cations at SII sites. An concentration may result in occupation of intersection sites

ChemSusChem 0000, 00, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org &9&
These are not the final page numbers! ÞÞ
 P. Nachtigall et al.

that are not populated at lower cation concentrations (e.g.,

sites III vs. sites II in FAU). However, even this generalization is
not 100 % valid; the synthetic procedure can influence the
framework Al distribution, and thus, isosteric heats also
depend on zeolite synthetic conditions.[48]
Zeolite topology has a large effect on isosteric heats. First,
zeolites in which extra-framework cations occupy sites on the
channel intersection are characterized by large Qst values;
a larger Qst value is found for the intersection of two 10-rings
(e.g., MFI) than that for the intersection of 10- and 8-ring chan-
nels (e.g., FER). Second, DC sites are already formed for Si/Al >
10 (e.g., FER) in zeolites with narrow channels, whereas for
zeolites with a large cavity (e.g., FAU) a value of Si/Al < 2 is
required.
The arguments given above should help us to understand
the differences in isosteric heats reported for zeolites and relat-
ed materials. The fact that FAU zeolites with a relatively high
concentration of extra-framework cations are characterized by
rather low Qst values[15a, 16c, 50] is still somewhat puzzling and is Figure 9. Isosteric heats of carbon dioxide adsorption on alkali-metal-cation-
doped Al-SBA-15. The adsorbed amount is expressed as the number of CO2
discussed in greater detail below. There are only DC and MC
molecules per cation.
sites in Na-A (Si/Al = 1:1), and there is even a significant frac-
tion of DC sites in Na-FER (Si/Al = 8.6) and Na-ZSM-5 (Si/Al =
14:1) zeolites. Conversely, there are no DC sites in Na-FAU (Si/ Conclusions
Al = 2.7). This rather surprising result can be understood as fol-
lows: 1) Only some Na + cations in Na-FAU (Si/Al = 2.7:1) are ac- Based on the analysis of numerous experimental and theoreti-
cessible for the CO2 molecules (8 SII sites in the model adopted cal results collected for zeolites with different structures and
herein); hence the effective Na + concentration for CO2 adsorp- compositions, key factors influencing the CO2 interaction with
tion in FAU-type zeolites (considering only the accessible Na + zeolites were identified. There are three dominant effects de-
cations) can be significantly lower than the actual Si/Al ratio termining heats of adsorption of CO2 :
corresponding to the zeolite chemical composition. 2) For
a broad range of Si/Al ratio, Na + cations are regularly distribut- 1) Effects from bottom: The strongest interaction can be ex-
ed among sites II, I, and I’ (with nearly 100 % occupancy of site- pected with extra-framework cations with a large charge/
s II); sites I and I’ are not accessible to CO2, and nearest SII sites ionic radius ratio; however, the cation must also be suffi-
are almost 10  apart from each other, which is too large a dis- ciently exposed in the zeolite channel (or cavity) to effi-
tance for a CO2 molecule to interact with both Na + cations si- ciently interact with CO2. Increasing coordination of the
multaneously. The CO2 adsorption mechanism is qualitatively extra-framework metal cation with the framework leads to
different in K-FAU (Si/Al = 2.7:1). The K + ions are too large to a decreased interaction energy with CO2. It follows that
fit inside the 6-ring; thus, site II is shifted above the ring to- extra-framework cations on channel intersections usually
wards the supercage, and consequently, adjacent K + cations in show larger values of DH0 than those located on channel
sites II are only 8.6  apart,[15a] which is close enough to consti- walls. The number of framework Al atoms in the vicinity of
tute a DC site for CO2. the extra-framework cation being considered can also play
Differences between micro- and mesoporous materials are a role.
discussed below based on the CO2 isosteric heats measured 2) Effects from top: The charge distribution in CO2 makes this
for alkali-metal-exchanged Al-SBA-15 (Figure 9). Interpretation molecule an ideal adsorbate to interact simultaneously
of these results is complicated by the fact that there is no in- with two extra-framework cations in DC sites. Adsorption of
formation about the structure of the surface of the mesopores. CO2 on these sites is stronger than that on SC sites. The DC
The relatively low values of isosteric heats reported in Figure 9 site is formed when two extra-framework cations are at
indicate that there are no DC sites in any of the Al-SBA-15 sam- a suitable distance from each other to form a linear
ples. The lower value of the isosteric heat observed for the M + ···O=C=O···M + adsorption complex (the optimum dis-
Al-SBA-15/Li + sample (relative to that of Al-SBA-15/Na + and tance for such a CO2 adsorption complex is 6.5, 7.3, and
Al-SBA-15/K + ) indicates that Li + cations are relatively well co- 8.3  for Li + , Na + , and K + , respectively). The concentration
ordinated with the framework oxygen atoms, while Na + and of dual cation (DC) sites depends mainly on the Si/Al ratio,
K + cations are more exposed towards the mesopore free which dictates the average concentration of extra-frame-
space and more accessible for molecules of CO2. work cations. However, it also depends on zeolite topology
and on the distribution of Al atoms in the zeolite frame-
work.

&10& www.chemsuschem.org  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 13
ÝÝ These are not the final page numbers!
Adsorption Enthalpy of CO2 in Zeolites

3) Dispersion interactions: van der Waals dispersion interac- [7] a) P. L. Llewellyn, S. Bourrelly, C. Serre, A. Vimont, M. Daturi, L. Hamon,
tions between CO2 and framework atoms are behind the G. De Weireld, J. S. Chang, D. Y. Hong, Y. K. Hwang, S. H. Jhung, G. Ferey,
Langmuir 2008, 24, 7245 – 7250; b) A. O. Yazaydin, R. Q. Snurr, T. H. Park,
relatively large CO2 adsorption enthalpy observed, even for K. Koh, J. Liu, M. D. LeVan, A. I. Benin, P. Jakubczak, M. Lanuza, D. B. Gal-
purely siliceous zeolites. Dispersion interactions depend on loway, J. J. Low, R. R. Willis, J. Am. Chem. Soc. 2009, 131, 18198 – 18199;
the zeolite topology (channel diameter) and on channel c) O. K. Farha, A. O. Yazaydin, I. Eryazici, C. D. Malliakas, B. G. Hauser,
wall thickness (framework density). In some cases, these in- M. G. Kanatzidis, S. T. Nguyen, R. Q. Snurr, J. T. Hupp, Nat. Chem. 2010, 2,
944 – 948; d) J. R. Li, J. Sculley, H. C. Zhou, Chem. Rev. 2012, 112, 869 –
teractions can account for more than 50 % of the overall in- 932; e) Y. S. Bae, R. Q. Snurr, Angew. Chem. 2011, 123, 11790 – 11801;
teraction energy between CO2 and zeolites, even for rela- Angew. Chem. Int. Ed. 2011, 50, 11586 – 11596.
tively low Si/Al ratios. [8] a) D. M. D’Alessandro, B. Smit, J. R. Long, Angew. Chem. 2010, 122,
6194 – 6219; Angew. Chem. Int. Ed. 2010, 49, 6058 – 6082; b) S. Cavenati,
C. A. Grande, A. E. Rodrigues, J. Chem. Eng. Data 2004, 49, 1095 – 1101;
It follows that for applications requiring large CO2 adsorp- c) P. J. E. Harlick, F. H. Tezel, Microporous Mesoporous Mater. 2004, 76,
tion heats, zeolites showing either the maximal effect from 71 – 79; d) S. Choi, J. H. Drese, C. W. Jones, ChemSusChem 2009, 2, 796 –
bottom (e.g., Li + cations on the intersection site formed by 854; e) A. Ghoufi, L. Gaberova, J. Rouquerol, D. Vincent, P. L. Llewellyn,
two 10-ring channels) or zeolites with a large number of DC G. Maurin, Microporous Mesoporous Mater. 2009, 119, 117 – 128; f) S. T.
Yang, J. Kim, W. S. Ahn, Microporous Mesoporous Mater. 2010, 135, 90 –
sites (e.g., a pair of Na + cations in two neighboring 8-ring in 94; g) T. Montanari, E. Finocchio, E. Salvatore, G. Garuti, A. Giordano, C.
FER) should be used. For cases where constant adsorption Pistarino, G. Busca, Energy 2011, 36, 314 – 319; h) M. R. Delgado, C. O.
heats are required, it becomes more complicated to find a suit- Arean, Energy 2011, 36, 5286 – 5291.
able zeolite; a lower cation concentration should be used (to [9] a) P. J. E. Harlick, A. Sayari, Ind. Eng. Chem. Res. 2006, 45, 3248 – 3255;
b) V. Zeleňk, M. Badaničov, D. Halamov, J. Čejka, A. Zukal, N. Murafa,
avoid the formation of DC sites) and as homogeneous a distri- G. Goerigk, Chem. Eng. J. 2008, 144, 336 – 342; c) A. Zukal, J. Jagiello, J.
bution of extra-framework cations as possible is desirable. Mayerova, J. Cejka, Phys. Chem. Chem. Phys. 2011, 13, 15468 – 15475.
[10] a) P. J. E. Harlick, A. Sayari, Ind. Eng. Chem. Res. 2007, 46, 446 – 458; b) A.
Zukal, I. Dominguez, J. Mayerova, J. Cejka, Langmuir 2009, 25, 10314 –
Acknowledgements 10321.
[11] a) G. Frey, C. Serre, T. Devic, G. Maurin, H. Jobic, P. L. Llewellyn, G.
De Weireld, A. Vimont, M. Daturi, J. S. Chang, Chem. Soc. Rev. 2011, 40,
The work of J.C. and A.Z. was supported by the Grant Agency of 550 – 562; b) Q. A. Wang, J. Z. Luo, Z. Y. Zhong, A. Borgna, Energy Envi-
the Czech Republic (GACR 203/08/0604). The work of P.N. and ron. Sci. 2011, 4, 42 – 55; c) L. Valenzano, B. Civalleri, K. Sillar, J. Sauer, J.
L.G. was supported by GACR (203/09/0134) and by grant Phys. Chem. C 2011, 115, 21777 – 21784.
[12] a) A. R. Millward, O. M. Yaghi, J. Am. Chem. Soc. 2005, 127, 17998 –
MSM0021620857 from ME CR.
17999; b) S. Keskin, T. M. van Heest, D. S. Sholl, ChemSusChem 2010, 3,
879 – 891; c) P. D. C. Dietzel, V. Besikiotis, R. Blom, J. Mater. Chem. 2009,
Keywords: adsorption · carbon dioxide · density functional 19, 7362 – 7370; d) L. Grajciar, A. D. Wiersum, P. L. Llewellyn, J.-S. Chang,
P. Nachtigall, J. Phys. Chem. C 2011, 115, 17925 – 17933.
theory · zeolites [13] a) B. Bonelli, B. Civalleri, B. Fubini, P. Ugliengo, C. O. Arean, E. Garrone, J.
Phys. Chem. B 2000, 104, 10978 – 10988; b) E. Garrone, B. Bonelli, C.
[1] a) R. B. Alley, T. Berntsen, N. L. Bindoff, Z. Chen, A. Chidthaisong, P. Fried- Lamberti, B. Civalleri, M. Rocchia, P. Roy, C. O. Arean, J. Chem. Phys.
lingstein, J. M. Gregory, G. C. Hegerl, M. Heimann, B. Hewitson, B. J. Hos- 2002, 117, 10274 – 10282; c) P. Galhotra, J. G. Navea, S. C. Larsen, V. H.
kins, F. Joos, J. Jouzel, V. Kattsov, U. Lohmann, M. Manning, T. Matsuno, Grassian, Energy Environ. Sci. 2009, 2, 401 – 409.
M. Molina, N. Nicholls, J. Overpeck, D. Qin, G. Raga, V. Ramaswamy, J. [14] D. F. Plant, G. Maurin, I. Deroche, L. Gaberova, P. L. Llewellyn, Chem.
Ren, M. Rusticucci, S. Solomon, R. Somerville, T. F. Stocker, P. A. Stott, Phys. Lett. 2006, 426, 387 – 392.
R. J. Stouffer, P. Whetton, R. A. Wood, D. Wratt Summary for Policymak- [15] a) G. D. Pirngruber, P. Raybaud, Y. Belmabkhout, J. Cejka, A. Zukal, Phys.
ers, in Climate Change: The Physical Science Basis (Eds.: S. Solomon, D. Chem. Chem. Phys. 2010, 12, 13534 – 13546; b) A. Pulido, P. Nachtigall, A.
Qin, M. Manning, Z. Chen, M. Marquis, K. B. Averyt, M. Tignor, H. L. Zukal, I. Dominguez, J. Cejka, J. Phys. Chem. C 2009, 113, 2928 – 2935.
Miller), IPCC 2007: Cambridge University Press, Cambridge, 2007; [16] a) E. D. Akten, R. Siriwardane, D. S. Sholl, Energy Fuels 2003, 17, 977 –
b) S. A. Rackley, Carbon Capture and Storage, Elsevier, Amsterdam, 2010; 983; b) A. Hirotani, K. Mizukami, R. Miura, H. Takaba, T. Miya, A. Fahmi,
c) J. M. Klara, J. E. Plunkett, Int. J. Greenhouse Gas Control 2010, 4, 112 – A. Stirling, M. Kubo, A. Miyamoto, Appl. Surf. Sci. 1997, 120, 81 – 84; c) G.
118. Maurin, P. L. Llewellyn, R. G. Bell, J. Phys. Chem. B 2005, 109, 16084 –
[2] a) D. Aaron, C. Tsouris, Sep. Sci. Technol. 2005, 40, 321 – 348; b) G. Pelle- 16091; d) D. F. Plant, G. Maurin, I. Deroche, P. L. Llewellyn, Microporous
grini, R. Strube, G. Manfrida, Energy 2010, 35, 851 – 857; c) R. Strube, G. Mesoporous Mater. 2007, 99, 70 – 78.
Pellegrini, G. Manfrida, Energy 2011, 36, 3763 – 3770; d) A. A. Olajire, [17] O. Bludský, M. Rubeš, P. Soldn, P. Nachtigall, J. Chem. Phys. 2008, 128,
Energy 2010, 35, 2610 – 2628. 114102.
[3] a) E. J. Stone, J. A. Lowe, K. P. Shine, Energy Environ. Sci. 2009, 2, 81 – 91; [18] a) M. Rubeš, J. Kysilka, P. Nachtigall, O. Bludský, Phys. Chem. Chem. Phys.
b) K. Z. House, C. F. Harvey, M. J. Aziz, D. P. Schrag, Energy Environ. Sci. 2010, 12, 6438 – 6444; b) L. Grajciar, O. Bludský, P. Nachtigall, J. Phys.
2009, 2, 193 – 205; c) R. Strube, G. Manfrida, Int. J. Greenhouse Gas Con- Chem. Lett. 2010, 1, 3354 – 3359.
trol 2011, 5, 710 – 726; d) J. Davison, Energy 2007, 32, 1163 – 1176. [19] A. Zukal, C. O. Arean, M. R. Delgado, P. Nachtigall, A. Pulido, J. Mayerov,
[4] a) B. Thitakamol, A. Veawab, A. Aroonwilas, Int. J. Greenhouse Gas Con- J. Cejka, Microporous Mesoporous Mater. 2011, 146, 97 – 105.
trol 2007, 1, 318 – 342; b) M. Karl, R. F. Wright, T. F. Berglen, B. Denby, Int. [20] A. Zukal, A. Pulido, B. Gil, P. Nachtigall, O. Bludsky, M. Rubes, J. Cejka,
J. Greenhouse Gas Control 2011, 5, 439 – 447. Phys. Chem. Chem. Phys. 2010, 12, 6413 – 6422.
[5] a) A. L. Chaffee, G. P. Knowles, Z. Liang, J. Zhany, P. Xiao, P. A. Webley, [21] A. Zukal, J. Pawlesa, J. Cejka, Adsorption 2009, 15, 264 – 270.
Int. J. Greenhouse Gas Control 2007, 1, 11 – 18; b) R. T. Yang, Gas Separa- [22] A. Zukal, J. Mayerova, M. Kubu, Top. Catal. 2010, 53, 1361 – 1366.
tion by Adsorption Processes, Butterworths, Boston, 1987. [23] A. Zukal, J. Mayerova, J. Cejka, Phys. Chem. Chem. Phys. 2010, 12, 5240 –
[6] a) N. Hedin, L. J. Chen, A. Laaksonen, Nanoscale 2010, 2, 1819 – 1841; 5247.
b) R. P. Ribeiro, T. P. Sauer, F. V. Lopes, R. F. Moreira, C. A. Grande, A. E. [24] T. Montanari, G. Busca, Vib. Spectrosc. 2008, 46, 45 – 51.
Rodrigues, J. Chem. Eng. Data 2008, 53, 2311 – 2317; c) C. Lu, H. Bai, B. [25] a) G. Kresse, J. Hafner, Phys. Rev. B 1994, 49, 14251 – 14269; b) G. Kresse,
Wu, F. Su, J. F. Hwang, Energy Fuels 2008, 22, 3050 – 3056; d) L. Zhou, X. J. Furthmuller, Comput. Mater. Sci. 1996, 6, 15 – 50; c) G. Kresse, D. Jou-
Liu, J. Li, N. Wang, Z. Wang, Y. Zhou, Chem. Phys. Lett. 2005, 413, 6 – 9. bert, Phys. Rev. B 1999, 59, 1758 – 1775.

ChemSusChem 0000, 00, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org &11&
These are not the final page numbers! ÞÞ
 P. Nachtigall et al.

[26] J. P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 1996, 77, 3865 – [41] a) A. Zukal, H. Siklova, J. Cejka, Langmuir 2008, 24, 9837 – 9842; b) C. J.
3868. Gommes, H. Friedrich, M. Wolters, P. E. de Jongh, K. P. de Jong, Chem.
[27] P. E. Blçchl, Phys. Rev. B 1994, 50, 17953 – 17979. Mater. 2009, 21, 1311 – 1317; c) I. G. Shenderovich, G. Buntkowsky, A.
[28] C. Otero Aren, M. R. Delgado, G. F. Bibiloni, O. Bludský, P. Nachtigall, Schreiber, E. Gedat, S. Sharif, J. Albrecht, N. S. Golubev, G. H. Findenegg,
ChemPhysChem 2011, 12, 1435 – 1443. H. H. Limbach, J. Phys. Chem. B 2003, 107, 11924 – 11939; d) A. A. Guri-
[29] C. O. Arean, M. R. Delgado, K. Frolich, R. Bulanek, A. Pulido, G. F. Bibiloni, nov, Y. A. Rozhkova, A. Zukal, J. Cejka, I. G. Shenderovich, Langmuir
P. Nachtigall, J. Phys. Chem. C 2008, 112, 4658 – 4666. 2011, 27, 12115 – 12123.
[30] M. Palomino, A. Corma, F. Rey, S. Valencia, Langmuir 2010, 26, 1910 – [42] P. Nachtigall, M. R. Delgado, K. Frolich, R. Bulanek, G. T. Palomino, C. L.
1917. Bauca, C. O. Arean, Microporous Mesoporous Mater. 2007, 106, 162 – 173.
[31] A. Pulido, M. R. Delgado, O. Bludsky, M. Rubes, P. Nachtigall, C. O. Arean, [43] P. Nachtigall, R. Bulanek, Appl. Catal. A 2006, 307, 118 – 127.
Energy Environ. Sci. 2009, 2, 1187 – 1195. [44] D. Nachtigallov, P. Nachtigall, O. Bludský, Phys. Chem. Chem. Phys.
[32] P. Nachtigall, K. Frolich, H. Drobna, O. Bludsky, D. Nachtigallova, R. Bula- 2004, 6, 5580 – 5587.
nek, J. Phys. Chem. C 2007, 111, 11353 – 11362. [45] O. Bludský, P. Nachtigall, V. Spirko, Collect. Czech. Chem. Commun. 2011,
[33] D. H. Olson, G. T. Kokotailo, S. L. Lawton, W. M. Meier, J. Phys. Chem. 76, 669 – 682.
1981, 85, 2238 – 2243. [46] R. Bulnek, K. Frolich, E. Frydova, P. Cicmanec, Top. Catal. 2010, 53,
[34] J. Kučera, P. Nachtigall, Phys. Chem. Chem. Phys. 2003, 5, 3311 – 3317. 1349 – 1360.
[35] a) W. Loewenstein, Am. Mineral. 1954, 39, 92 – 96; b) W. Dempsey, G. H. [47] B. Bonelli, B. Onida, B. Fubini, C. O. Arean, E. Garrone, Langmuir 2000,
Kuhl, D. H. Olson, J. Phys. Chem. 1969, 73, 387 – 390. 16, 4976 – 4983.
[36] T. Frising, P. Leflaive, Microporous Mesoporous Mater. 2008, 114, 27 – 63. [48] P. Nachtigall, L. Grajciar, J. Perez-Pariente, A. B. Pinar, A. Zukal, J. Cejka,
[37] a) D. Nachtigallov, O. Bludský, C. Otero Aren, R. Bulnek, P. Nachtigall, Phys. Chem. Chem. Phys. 2012, 14, 1117 – 1120.
Phys. Chem. Chem. Phys. 2006, 8, 4849 – 4852; b) C. Otero Aren, M. R. [49] a) C. Otero Aren, O. V. Manoilova, G. T. Palomino, M. R. Delgado, A. A.
Delgado, C. L. Bauca, L. Vrbka, P. Nachtigall, Phys. Chem. Chem. Phys. Tsyganenko, B. Bonelli, E. Garrone, Phys. Chem. Chem. Phys. 2002, 4,
2007, 9, 4657 – 4661. 5713 – 5715; b) P. Nachtigall, M. R. Delgado, D. Nachtigallova, C. O.
[38] a) J. A. Dunne, R. Mariwala, M. Rao, S. Sircar, R. J. Gorte, A. L. Myers, Arean, Phys. Chem. Chem. Phys. 2012, 14, 1552 – 1569.
Langmuir 1996, 12, 5888 – 5895; b) R. Bulnek, K. Frolich, E. Frydova, P. [50] G. Maurin, Y. Belmabkhout, G. Pirngruber, L. Gaberova, P. Llewellyn, Ad-
Cicmanec, J. Therm. Anal. Calorim. 2011, 105, 443 – 449; c) S. Bourrelly, sorption 2007, 13, 453 – 460.
G. Maurin, P. L. Llewellyn in Molecular Sieves: From Basic Research to In-
dustrial Applications, Parts A and B, Vol. 158 (Eds.: J. Cejka, N. Zilkova, P.
Nachtigall), Elsevier, Amsterdam, 2005, pp. 1121 – 1128.
[39] http://www.iza-struture.org/databases/. Received: April 15, 2012
[40] A. Bondi, J. Phys. Chem. 1964, 68, 441 – 451. Published online on && &&, 0000

&12& www.chemsuschem.org  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim ChemSusChem 0000, 00, 1 – 13
ÝÝ These are not the final page numbers!
FULL PAPERS
Adsorb or not adsorb? The role of dif- L. Grajciar, J. Čejka, A. Zukal,
ferent factors that determine the gas– C. Otero Aren, G. Turnes Palomino,
solid interaction energy, such as zeolite P. Nachtigall*
topology and zeolite chemical composi-
&& – &&
tion, is analyzed based on a combination
of accurate ab initio descriptions of CO2 Controlling the Adsorption Enthalpy
interactions with zeolites and experi- of CO2 in Zeolites by Framework
mental investigations of adsorption Topology and Composition
heats (see scheme).

ChemSusChem 0000, 00, 1 – 13  2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemsuschem.org &13&
These are not the final page numbers! ÞÞ