You are on page 1of 27

Accepted Manuscript

Title: Adsorption Equilibrium of Methane and Carbon Dioxide


on Zeolite 13X: Experimental and Thermodynamic Modeling

Author: Fatemeh Gholipour Masoud Mofarahi

PII: S0896-8446(16)30008-0
DOI: http://dx.doi.org/doi:10.1016/j.supflu.2016.01.008
Reference: SUPFLU 3549

To appear in: J. of Supercritical Fluids

Received date: 25-11-2015


Revised date: 11-1-2016
Accepted date: 12-1-2016

Please cite this article as: <doi>http://dx.doi.org/10.1016/j.supflu.2016.01.008</doi>

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Adsorption Equilibrium of Methane and Carbon Dioxide

on Zeolite 13X: Experimental and Thermodynamic

Modeling

t
ip
cr
Fatemeh Gholipour, Masoud Mofarahi1

Chemical Engineering Department, Persian Gulf University, Bushehr, Iran

us
Abstract

A static, volumetric method has been used to determine the adsorption equilibrium of CH4 and

an
CO2 and their binary mixtures on 13X molecular sieves at various temperatures between 273 and 343

K. Pressures for the pure component data extend up to 10 bar, while all binary data were obtained at 4
M
bar and 6 bar. The measured pure isotherms were regressed using different isotherm equations and the

regressed parameters were applied to different thermodynamic models such as ideal adsorbed solution
d

theory (IAST), vacancy solution theory (VST) and real adsorbed solution theory (RAST). CH4/CO2
te

system that deviate from ideality is not well represented by IAST and VST, whereas RAST which

include activity coefficient in the adsorbed phase as characterizing parameter for non-ideality, show a
p

much better representation of the binary equilibria. Experimental selectivities CO2/CH4 range from 26
ce

to 2 at different pressure and temperatures. It is concluded that zeolite 13X suitable for natural gas

industrial separation with yCO2 < 0.2 at 303 K and 4 bar. For landfill gas upgrading, where yCO2 < 0.45,
Ac

zeolite 13X can be successfully applied at 303 K, 4 bar and 303 K, 6 bar and 323 K, 4 bar.

Key words: methane and carbon dioxide separation, zeolite 5A, non-ideal solution, vacancy

solution model, ideal adsorbed solution theory, real adsorbed solution theory

1. Introduction

Since the beginning of industrialization, the amount of carbon dioxide (CO2) in the atmosphere

has increased by about 30% and the current level is estimated to double or triple before the end of this

1
Corresponding author. E-mail: mofarahi@pgu.ac.ir. Fax: +98 7733441495.

Page 1 of 26
century [1]. Methane is the most important non-CO2 greenhouse gases responsible for global warming

with more than 10 % of total greenhouse gases emissions [2]. Natural gas systems, biogas and

landfills are the main sources of methane emissions to atmosphere. Nowadays, natural gas supplies

one-fourth of the energy required in the world’s homes, vehicles, industries, and power plants. The

t
consumption of natural gas is expected to grow by 50% over the next 20 years [3].

ip
Natural gas consists mainly of methane, about 80 to 95 %, and often nitrogen and carbon dioxide

as the impurities. The heating value of clean natural gas is around 11.0 kWh/m3, but the presence of

cr
CO2 decreases heating value to a half of the one in natural gas (6.5 kWh/m3). To satisfy the pipeline

us
quality, the maximum amount of carbon dioxide in natural gas cannot exceed 2%. So the separation of

CO2 from methane is essential for the upgrading of natural gas and reducing pipeline corrosion

an
induced by acid CO2 gas.

A typical landfill gas contains by volume basis an average of approximately 55% methane, 40%
M
carbon dioxide, 2.3% nitrogen, 0.6% oxygen, 2% water vapor, less than 100 parts per million (PPM)

of hydrogen sulfide and other insignificant smaller amounts of sulfur and hydrocarbon compounds. In
d
many countries, which have to import natural gas or fossil fuels, landfill gas can be treated as an

important national resource of directly available methane. This reason, together with a tighter control
te

in methane emissions to meet Kyoto Protocol targets, puts landfill gas into consideration for energy
p

production [4].
ce

During anaerobic digestion (i.e. digestion in the absence of oxygen) organic material is broken

down in several steps by different types of microorganisms. The end-products are a gas containing
Ac

mainly methane and carbon dioxide, referred to as biogas. Biogas can be utilized as a renewable

energy source in combined heat and power plants, as a vehicle fuel, or as a substitute for natural gas.

The methane in the biogas can also be utilized in industrial processes and as a raw material in the

industry. Upgrading of biogas has gained increased attention due to rising oil and natural gas prices

and increasing targets for renewable fuel quotes in many countries.

Absorption with amine aqueous solutions [5], membranes [6], and adsorption using porous solids

have been proposed for the separation of CO2/CH4 mixtures. Among these methods, adsorption

processes have become increasingly competitive and already favorable because of its low energy

Page 2 of 26
requirement, easy operation, and low maintenance. The adsorption phenomenon has the additional

degree of freedom that is related to choice of the adsorbent, whenever the adsorbent is changed, all of

the adsorption behavior data will change accordingly. One of the most important factors for gas

adsorptive separation processes by microporous materials is pore size. When the pore size of a

t
o o
material is located between the kinetic diameters of two gas molecules (CO2 : 3.3 A ; CH4 :3.8 A ),

ip
one can separate the two gases by a molecular sieving effect (or a steric effect). If the pores are the

cr
right size, only the smaller molecule (CO2) can diffuse into the pores, whereas the larger molecule

(CH4) is totally excluded. If the pore size is slightly larger than the kinetic diameter of the larger

us
molecule (CH4), one can separate the two gases by a kinetic separation, which is achieved by the

difference in the diffusion rates. In this case, the larger molecule (CH4) diffuses slower than the

an
smaller molecule (CO2). When the pore size is large enough that both molecules can readily diffuse

into the pores, the two molecules may be separated by differences in their equilibrium adsorption,
M
which is used in a large majority of adsorptive separation processes. Even for separation processes

based on differences in equilibrium adsorption, the pore size may play a role in dictating the amount
d

adsorbed [6].
te

With the development of technology, the separation of CO2/CH4 in porous materials received great

attention from researcher. Among these materials, Zeolite 13X is the most commonly and most
p

effective adsorbent used because of its polar surface, high specific surface area and big pore volume.
ce

Cavenati et al. [3] measured pure adsorption equilibrium data for CO2 and CH4 in zeolite 13X at

pressure up to 5 MPa at temperature 298, 308 and 323 K. They reported that the ideal selectivity of
Ac

CO2/CH4 can reach a value of 7 at 298K and 1atm using single component isotherms. Mulgundmath

et al. [7] compared CECA zeolite 13X to silicallite for the CO2/CH4 separation. They used pure and

binary adsorption isotherms at two different temperatures 40 and 100°C and pressures up to 5 atm and

showed that CECA zeolite 13X is more effective than silicallite. Silva et al [8] studied the binary

adsorption of CO2 and CH4 in binderless pellets of zeolite 13X at 313 and 423K and pressure up to

5atm. They reported that experimental selectivity of CO2/CH4 range from 37 at low pressure and

313K and to approximately 5 at 423K. Joe MCEwen et al. [9] measured the pure adsorption isotherms

Page 3 of 26
of CO2, CH4 and N2 at 25°C and pressure up to 1bar on Zeolite 13X, Zeolitic imidazolate frameworks

(ZIFs) and activated carbon. The results showed that zeolite 13X has higher gas adsorption capacity

compared to activated carbon and ZIFs.

In this study, adsorption equilibrium data of pure CH4 and CO2 and their binary mixtures at five

t
temperatures of 273, 283, 303, 323, and 343 K and pressure up to 10 bar on Zeochem zeolite 13X are

ip
presented. The textural properties of the adsorbent were characterized by an Accelerated Surface Area

and Porosimetry Apparatus (ASAP 2020). Equilibrium selectivities and x-y diagrams were obtained

cr
experimentally at different pressures and temperatures and compared with the calculated ones.

us
In this work, due to the measured data in a constant-pressure path, the spreading pressure of the

mixture was evaluated for each data point of binary system, after that the experimental activity

an
coefficient for adsorbed phase was achieved. These experimental activity coefficient values reveal

nonideal behavior of CH4/CO2 mixture. Such interesting data have never been previously published.
M
Thus, the goal of this study is several-fold. First, the surface area and pore volume of Zeolite 13X

were characterized by N2 adsorption at 77K in a Micromeritics ASAP2020 volumetric apparatus.


d
Second, the adsorption isotherm for pure CO2 and binary CH4/CO2 system on Zeolite 13X at different

pressure and temperature were obtained experimentally. Third, the detailed adsorption behavior was
te

evaluated using Ideal Adsorbed Solution Theory (IAST), Vacancy Solution Theory (VST) and Real
p

Adsorbed Solution Theory (RAST). Finally, the optimum operating condition for natural gas
ce

industrial separation and landfill gas upgrading using zeolite 13X were achieved.
Ac

2. Model Description

2.1 Vacancy Solution Theory (VST)

Suwanayan and Danner [10] developed a thermodynamic model for the prediction of gas

adsorption in 1980. In this model, both gas and adsorbed phases are considered to be solutions of

adsorbates in a hypothetical solvent called “vacancy”. A vacancy is a vacuum space that can be filled

with adsorbate molecules. So the pure gas adsorption system is considered as a binary vacancy

solution and the pure component adsorption isotherm is:

Page 4 of 26
n is ,∞ θ 1 − (1 − Λ 3 i )θ Λ (1 − Λ 3 i )θ (1 − Λ i 3 )θ
p =[ ][ Λ i 3 ] exp[ − 3i − ] (1)
bi 1 − θ Λ i 3 + (1 − Λ i 3 )θ 1 − (1 − Λ 3i )θ Λ i 3 + (1 − Λ i 3 )θ

The first term on the right-hand side is the Langmuir equation. The second term represents the

nonideality of the system in terms of the activity coefficient, as defined by the Wilson equation

parameters ( Λ i3 and Λ 3i ). In this equation the subscript 3 represent the vacancy, nis ,∞ is the number

t
ip
of adsorbed molecules at saturation, bi is the Henry's law constant, θ is the fractional surface

cr
coverage. By equating the chemical potentials of an adsorbate in the gas and adsorbed phases, an

equation for the distribution of that adsorbate between the two phases can be obtained as:

us
__
n is , ∞ Λ i 3 π αi
y iφ i p = x i γ is n ms exp( Λ 3 i − 1) exp( ) (2)
n ms , ∞ bi RT

an
where φi is the fugacity coefficient of component i in the gas phase. In this study, these fugacity

coefficients are neglected because for these pressures the correction is negligible.
M
2.2 Ideal Adsorbed Solution Theory (IAST)

The equilibrium between the ideal gas or vapor phase and an adsorbed phase can be expressed
d

similar as for vapor-liquid equilibria (VLE):


te

Py i = Pi0 (π ) xiγ i (π ) (3)

Where y i and xi are the mole fraction of component i in the gas phase and the adsorbed phase,
p

respectively. γ i (π ) is the activity coefficient of the adsorbed phase and Pi0 (π ) is the equilibrium gas
ce

phase pressure corresponding to the solution temperature and solution spreading pressure ( π ) for the

adsorption of pure component i. In the case of an ideal solution, the activity coefficient is equal to
Ac

unity for all values of temperature, spreading pressure and adsorbed phase mole fraction of

component i.

The ideal adsorbed solution theory was presented by Myers and Prausnitz [11]. In this theory the

standard state is defined by the equality of the spreading pressure for each component with the

mixture spreading pressure, so π i0 = π . The spreading pressure of component i can be calculated by the

Gibbs adsorption isotherm:

Page 5 of 26
Pi0
π i0 A qi
= ∫ P dP
0
i (4)
Rg T i
0

Where qi and Pi are related to pure adsorption and pressure. Indeed, Gibbs adapted the classical

thermodynamics of the bulk phase and applied it to the adsorbed phase. In doing this the concept of

t
volume in the bulk phase is replaced by the area, and the pressure is replaced by the spreading

ip
pressure. The spreading pressure is the negative of the surface potential. Surface potential indicates

cr
the minimum work required to load the adsorbent to a certain level in an isothermal path. So, lower

equilibrium capacity results in lower (less negative) surface potential than higher equilibrium

us
capacity.

2.3 Real adsorbed solution theory (RAST)

an
Because of the interactions between the adsorbed molecules and the adsorbed molecules with the

solid surface, IAST is inadequate to predict the nonideal adsorption behavior. To describe nonideal
M
adsorption equilibria, adsorbate phase activity coefficient have to be taken into account [12]. No

models are available for prediction of activity coefficients in adsorbate mixtures. Costa et al. [13] used
d

the Wilson and UNIQUAC equation to correlate their data on binary and ternary mixture adsorbates.
te

Although these correlations were reasonably successful, the empirical binary constants in these

equations differed considerably from those reported for bulk solutions [14]. Usually the adsorbate
p

phase activity coefficient has to confirm three fundamental thermodynamic restrictions.


ce

1) The activity coefficient of component i must approach unity when the composition of

component i approaches unity, e.g. lim γ i = 1


xi →1
Ac

2) The activity coefficient of component i must approach the infinite dilution value γ i∞ as the

composition of this component approaches zero, e.g. lim γ i = γ i∞


xi →0

3) As the total surface coverage approaches zero, e.g. the spreading pressure approaches zero, the

activity coefficient of every adsorbate component must approach unity, e.g. lim γ i = 1
π →0

Real adsorbed solution theory was presented by Costa et al. [13]. According to this theory,

adsorbed phase activity coefficient must be taken into account. The basic equation of RAST is:

Page 6 of 26
yi P = xiγ i Pi 0 (5)

Pi 0 is the pressure that the pure component i should have in the gas phase to give rise. It has to be

evaluated from single component data at the same spreading pressure and temperature as the

equilibrium mixture. To evaluate the spreading pressure at any point in the multi-component

t
adsorption, Talu [15] presented the following equation:

ip
P y1
x1 x2
π = ∫ n pure d ln P + ∫ nmixture ( − )dy1 (6)

cr
0 0
y1 y 2

Therefore, the spreading pressure can evaluate by integration in a single-component plane from

us
P=0 (π=0) to the pressure of the mixture followed by integration in the constant pressure plane to the

mixture composition. In order to use equation 6, mixture experimental data must be collected in

an
constant pressure.

In this study, we used two well-known activity coefficient models, Wilson and NRTL, to
M
determine the activity coefficient of the adsorbed components. The Wilson equation is:

Λ 12 Λ 21
ln γ 1 = − ln( x1 + Λ 12 x 2 ) + x 2 ( − ) (7)
x1 + Λ 12 x 2 Λ 21 x1 + x 2
d

where the binary constants, Λ12 and Λ 21 , must be determined experimentally.


te

The NRTL equation is:


p

G21 τ 12G12
ln γ 1 = x 22 [τ 21 ( )2 + ]
x1 + x2 G21 ( x2 + x1G12 ) 2
ce

(8)
G12 = exp( −α12τ 12 )
G21 = exp( −α12τ 21 )
Ac

where the binary constants, τ 12 and τ 21 must be determined experimentally.

3. Materials

CO2 (99.99%), CH4 (99.95%) and Helium (99.999%) were used for the adsorption isotherm

measurements. Surface area and pore volume of Zeolite 13X (provided by Zeochem Co. (Switzerland))

were characterized by N2 adsorption at 77K in a Micromeritics ASAP2020 volumetric apparatus. The

sample was degassed at 573K for 6 hr under a vacuum before analysis. The Brunauer-Emmett-Teller

Page 7 of 26
(BET) surface area was calculated using adsorption data in a relative pressure ranging from 0.04 to

0.20. The t-plot method was used to calculate micropore volumes and micropore surface areas.

4. Apparatus and procedure

The adsorption equilibrium data were measured using a static volumetric apparatus used in our

t
previous works [16]. The apparatus was tested to be leak-proof for pressures up to 20 bar. The

ip
adsorption cell consisted of a stainless-steel column with a diameter of 2.8 cm. The column was

located in a water bath (MC 12, Julabo Tech.) to supply the isothermal condition. A pressure

cr
transducer with ±0.5 mbar uncertainty was used for direct measurement of pressure and a circulation

us
pump was used to achieve equilibrium conditions in a shorter period of time.

Before the start of each experimental run, the adsorbent was regenerated at 573K under a 0.25 bar

an
helium purge provided by a vacuum pump with 0.05 mbar vacuum levels, for approximately 6 hr.

Prior to the first adsorption run, the exact volume in the column was determined using Helium
M
expansion. The initial and equilibrium gas phase composition were determined by Gas

Chromatograph (GC) BEIFEN 3420 calibrated with standard mixtures of known compositions. The
d
thermal conductivity detector (TCD) with a packed column was initially calibrated under the injector

temperature of 120 °C, detector temperature of 180 °C, column temperature of 70 °C and a 20 ml/min
te

helium flow rate. These operating conditions were kept fixed in all measurements. It must be noted
p

that prior to each mixture adsorption data point, the adsorbents were regenerated in situ in adsorption
ce

measuring cell. Also, we have weighted the adsorbents, after the experiment finished. No mass loss

was observed.
Ac

When adsorption occurs at the gas‐solid interface, the weight of the solid increases and the

pressure of the gas decreases. Thus, the amount adsorbed can be determined by measuring the change

in pressure of the gas in an accurately known volume. Measurement of gas adsorption uptake at a

certain pressure and temperature requires two steps. In the first step, a specific amount of gas is

Page 8 of 26
confined in the reference vessel. The exact volume of the reference vessel is measured using helium

expansion before. The total amounts of gas that will be available for adsorption in the second step

determine using an equation of state by known temperature, pressure and volume. Then the dosing

valve is opened and the gas is expanded into the adsorption vessel, where it is exposed to the

t
adsorbent. The pressure of gas decreases until the equilibrium occurs. After equilibration, the pressure

ip
is recorded and the final amount of gas remaining in the gas phase measured using an equation of

state. So the amount of adsorbed gas can be determined using mole balance before and after

cr
equilibrium.

us
5. Results and discussion

The adsorption-desorption isotherm of zeolite 13X measured with nitrogen gas at 77 K presented

an
in Figure 1 showed a typical type I behavior. The hysteresis loop indicates the presence of

mesoporosity. The low-pressure isotherm provides information about the microporosity. The pore
M
size distribution of the micropores has been determined by using the Horvath---Kawazoe equation

showed in Figure 1. Textural properties of zeolite 13X are listed in Table 1.

5.1. Pure carbon dioxide and methane adsorption isotherms


d

Pure gas adsorption isotherms of CO2 have been measured at five temperatures of 273, 283, 303,
te

323, and 343 K and pressure up to 10 bar and shown in Figure 2. Pure methane adsorption data from
p

our previous work [17] shown in Figure 3. Equilibrium loading of CO2 is higher than methane in all
ce

temperatures and pressures. The pore structure of zeolite 13X is large enough to neglect the steric

effect of the adsorbate with the adsorbent structure. The different adsorption capacity of each
Ac

component is due to cationic nature of adsorbent surface [18]. Carbon dioxide possesses a large

quadruple moment (-13.71×1040) which produces strong attraction to the electrostatic field of the

cationic site and results in high capacity.

The CO2 adsorption data obtained experimentally in the current study shows good agreement with

the data found in the literature at same temperature and pressure [3, 9, 19, 20]. To have a full

description of the adsorption, temperature- dependent version of Langmuir, Freundlich, Langmuir-

Freundlich, and Toth equations have been applied. The parameters of different models have been

obtained via non- linear regression and RMS (Root Mean Square) deviation as bellow:

Page 9 of 26
0.5
 N
exp 2 

 ∑i =1
(qical −q
i ) 
RMS =   (9)
 N 
 
 

where N is the number of data points and q ical , and qiexp are calculated and experimental

t
adsorbed amounts, respectively. The temperature dependent form of the isotherm models, the

ip
regressed parameters and corresponding error values for carbon dioxide and methane are tabulated in

cr
Table 2.

According to the RMS values, the best results can obtain with Toth equation for CH4 and CO2

us
adsorption isotherms. The parameters and corresponding errors of vacancy solution model are listed

in Table 3. As shown in this table, the isotherm data are well correlated by vacancy solution model.

5.2 Heat of Adsorption

an
The isosteric heat of adsorption is a measure of the interaction between adsorbate molecules and
M
adsorbent lattice atoms, and represents the energetic heterogeneity of a solid surface. Because of

difficulty in Calorimetric methods, estimation of heat of adsorption using adsorption isotherms is of a


d

great interest. Isosteric heat of adsorption of pure methane and carbon dioxide are estimated using
te

Clausius-Clapeyron equation as:

d ln Pi ∆H ad
p

( )q = (10)
d (1 / T ) R
ce

For this purpose, logarithm of pressure versus the inverse absolute temperature at constant amount

adsorbed was plotted and the heat of adsorption for CH4 and CO2 were obtained 15 (at loading 1.8
Ac

mol/kg) and 32 kJ/mol( at loading 2.8 mol/kg), respectively. These reported values for Carbon dioxide

and Methane are in agreement with previously published data [3, 22]. If different levels of surface

energy exist, the isosteric heat of adsorption varies with the surface coverage. As the isosteric heats of

CH4 and CO2 are independent of the amount adsorbed, the surface is energetically homogeneous [21].

5.3 Binary carbon dioxide and methane adsorption data on zeolite 13X

Design of an industrial adsorptive separation process requires the reliable multicomponent

adsorption equilibrium data. The Gibbs phase rule identifies the degree of freedom for the adsorption

Page 10 of 26
of a pure gas, one, when the temperature is constant. Therefore, the pure data of adsorption always

shows in a 2-dimension plot at constant temperature. In the binary case, the degree of freedom

increases and the data should be shown in a 3-dimensions curve including gas composition, pressure,

and amount adsorbed. In the gas mixture adsorption, data are usually depicted at constant pressure or

t
gas phase composition. It is difficult to accurately control the equilibrium pressure and the gas phase

ip
composition in volumetric method. So many experiments are required to achieve specified point with

constant pressure.

cr
In the current study, all data of gas mixture of methane and carbon dioxide on zeolite 13X

us
designed at a relatively constant pressure path. Experimental data of binary gas mixture have been

determined at two constant pressures of 4 and 6 bar and temperatures of 303 and 323 K. The gas

an
phase mole fraction was measured using Gas Chromatograph (GC) analyser, then the adsorbed phase

mole fraction and adsorbed amount of each component determined applying mole balance before and
M
after equilibrium. As mentioned, all data obtained point by point. It means that after one equilibrium

test, the adsorbent has been regenerated and then used in the next test.
d
Binary equilibrium data and adsorbed phase properties are tabulated in Table 4 and 5.

Experimental selectivity of adsorption was obtained using equilibrium molar fraction of the bulk gas
te

phase measured directly utilizing GC and molar fraction of the adsorbed phase determined applying
p

mole balance:
ce

xCO2 / yCO2
SCO2 / CH 4 = (11)
xCH 4 / yCH 4
Ac

The measured selectivity of CO2/CH4 system varies between 26 at pressure of 4bar and

temperature of 303 K to 2 at pressure of 6bar and temperature of 323 K, according to Table 4 and 5.

IAST and VST models only need pure data to predict mixture adsorption behaviour, but RAST

require also the activity coefficients parameters. In this section, at first, experimental activity

coefficients using all mixture equilibrium data have been evaluated and then prediction of the models

will be compared.

In this research, the binary experimental data are designated in four case studies. The data are

collected in 303 K and 4 bar, 303K and 6 bar, 323 K and 4 bar, 323 K and 6 bar. Firstly, all binary

Page 11 of 26
experimental data are used to determine the activity coefficient of adsorbed phase. After that each

case study will be considered separately. The extended pressure is calculated using pure and binary

data and equation 6. Then, values of Pj0 are obtained using Gibbs integral, Langmur-Frendlich

isotherm and extended pressures values.

t
The calculated extended pressure and experimental activity coefficient at 303 and 323 K are

ip
presented in Tables 6 and 7 respectively. The values of activity coefficient fitted using NRTL and

Wilson models, along with error presents can be found in Tables 8 and 9. The results show more

cr
accurate fitting using NRTL model. Therefore, this model is applied in RAST. The agreement of data

us
with NRTL model at 303 K is depicted in Figures 4.

In this study, equilibrium pressure has been controlled to be constant by adjustment of initial

an
pressure and composition of the inlet gas. The binary experimental data were obtained at temperatures

of 303 and 323 K and pressures of 4 and 6 bar over much of the gas phase composition from Y= 0 to
M
1. Figure 5 shows the experimental points for the isobaric X-Y equilibrium diagram of the mixture

CH4/CO2 at different pressures and temperatures. Predicted equilibrium diagrams by IAST, VST and
d
RAST are included. It is clear from Figure 5 that CH4/CO2 mixture exhibits nonideal behavior. The

large distance between the adsorbed and the gas phase composition indicates that the selectivity is
te

quite high.
p

This system that deviate from ideality is not well represented by IAST and VST, whereas RAST
ce

which include activity coefficient in the adsorbed phase as characterizing parameter for non-ideality,

show a much better representation of the binary equilibria. Use of the activity coefficient for the
Ac

CH4/CO2 system caused the RAST to predict real behaviour qualitative and quantitative.

It is interesting to observe the effect of pressure and temperature on adsorption behaviour of

CH4/CO2 system on zeolite 13X. As seen from Figure 5 (a-d), there are some differences between the

behaviors of the system at different operating conditions. At 303 K and 4 bar, owing to more

adsorption of CO2 than methane, in all points, the composition of CO2 in the adsorbed phase (x), is

higher than the composition of CO2 in the gas phase (y). The equilibrium curve shows that high purity

CO2 can be obtained at low CO2 feed gas concentrations. This property makes zeolite 13X suitable for

natural gas industrial separation with yCO2 < 0.2 at 303 K and 4 bar. For landfill gas upgrading, where

Page 12 of 26
yCO2 < 0.45, zeolite 13X can be successfully applied at 303 K, 4 bar and 303 K, 6 bar and 323 K, 4

bar. Comparing a, b or c, d in Figure 5, it can be observed that the equilibrium curve approaches y=x

line by increasing pressure at constant temperature. Since in all cases RAST is more accurate than

VST and IAST, the following, results of RAST will be presented only.

t
The amount of adsorbed methane and carbon dioxide predicted from RAST and the percentage

ip
absolute deviation from experimental data were presented in Table 4 and 5 for each data point. The

percentage average absolute errors are about 20%. So Model agreement with experimental data for

cr
predicting the adsorbed amount is not good.

us
The adsorbed amount of each component and the total amount of adsorption with the RAST results

an
have been shown in Figure 6. As seen in Figure 6, there is a large adsorption capacity differences

between components. CO2 dominated this system over more than half of the gas phase composition.
M
Only at high CH4 gas phase mole fraction, the amount of methane adsorbed were appreciable. This

result was consistent with the single component capacities. CO2 which was preferentially adsorbed
d
from the gas mixture exhibited the large single component adsorption capacity. The increase of the

loading with decreasing temperature shows that the adsorption process is exothermic, as expected.
te

6. Conclusion
p

In this work, pure adsorption equilibrium data for methane and carbon dioxide on zeolite 13X were
ce

measured using a volumetric method over a wide range of pressure and temperature. The binary

equilibrium data were modeled with vacancy solution theory, ideal adsorbed solution theory and real
Ac

adsorbed solution theory. Experimental selectivities CO2/CH4 range from 26 at a pressure of 4bar and

temperature of 303 K to 2 at pressure of 6bar and temperature of 323 K. it can be observed that the

equilibrium curve approaches y=x line by increasing pressure at constant temperature. These results

show the important role of the pressure and temperature in the adsorption equilibrium, particularly on

selectivity and ideal/nonideal behavior.

The ideality assumptions inherent in the IAST and VST preclude them from predicting real

adsorption behavior observed in the present study. It is concluded that the best theoretical model for

the prediction of binary CH4/CO2 adsorption equilibrium is the RAST. Further development in

Page 13 of 26
calculating the activity coefficients in this theory could lead to improved quantitative results.

References

[1] S. Oddy, J. Poupore, F.H. Tezel, Separation of CO2 and CH4 On CAX zeolite for use in landfill

gas separation, The Canadian J. Chemical Engineering, 19 (2013) 1031–1039.

t
[2] P. Li, F.H. Tezel, Pure and binary adsorption of methane and nitrogen by silicalite, J.

ip
Chemical & Engineering Data, 54 (2009) 8---15.

[3] S. Cavenati, C.A. Grande, A.E. Rodrigues, Adsorption equilibrium of methane, carbon dioxide

cr
and nitrogen on zeolite 13X at high pressures, J. Chemical & Engineering Data, 49 (2004) 1095-1101.

us
[4] S. Cavenati, C.A. Grande, A.E. Rodrigues, Upgrade of methane from landfill gas by pressure

swing adsorption, Energy & Fuels, 19 (2005) 2545-2555.

an
[5] C. Yu, C.H. Huang, C.S. Tan, A review of CO2 capture by absorption and adsorption, Aerosol

and Air Quality Research, 12 (2012) 745–769.


M
[6] Y.S. Bae, R.Q. Snurr, Angew, Development and evaluation of porous materials for carbon

dioxide separation and capture, Angewandte Chemie International Edition, 50 (2011) 1586 – 11596.
d
[7] V.P. Mulgundmath, F.H. Tezel, T. Saatcioglu, T.C. Golden, Adsorption and separation of

CO2/N2 and CO2/CH4 by 13X zeolite, The Canadian J. Chemical Engineering, 90 (2012) 730- 738.
te

[8] J. A.C. Silva, A. F. Cunha, K. Schumann, A.E. Rodrigues, Binary adsorption of CO2/CH4 in
p

binderless beads of 13X zeolite, Microporous and Mesoporous Materials, 187 (2014) 100–107.
ce

[9] J. McEwen, J.D. Hayman, A.O. Yazaydin, A comparative study of CO2, CH4 and N2 adsorption

in ZIF-8, zeolite-13X and BPL activated carbon, Chemical Physics, 412 (2013) 72–76.
Ac

[10] E. Costa, J.L. Sotelo, G. Calleja, C. Marron, Adsorption of binary and ternary hydrocarbon

gas mixtures on activated corbon: experimental determination and theoretical prediction of the ternary

equilibrium data, The American Institute of Chemical Engineers J., 27 (1981) 5–12.

[11] A.L. Myers, J.M. Prausnitz, Thermodynamics of mixed-gas adsorption, The American

Institute of Chemical Engineers J., 11 (1965) 121–127.

[12] M. Sakuth, J. Meyer, J. Gmehling, Measurement and prediction of binary adsorption

equilibria of vapors on dealuminated Y-zeolites (DAY), Chemical Engineering and Processing, 37

(1998) 267–277.

Page 14 of 26
[13] E. Costa, J.L. Sotelo, G. Calleja, C. Marron, Adsorption of binary and ternary hydrocarbon

gas mixtures on activated carbon: Experimental determination and theoretical prediction of the ternary

equilibrium data, The American Institute of Chemical Engineers J., 27 (1981) 5–12.

[14] R. Vaart, C. Huiskes, H. Bosch, T. Reith, Single and mixed gas adsorption equilibria of

t
carbon dioxide/methane on activated carbon, Adsorption 6 (2000) 311–323.

ip
[15] O. Talu, I. Zwiebel, Multicomponent adsorption equilibria of nonideal mixtures, The

American Institute of Chemical Engineers J., 32 (1986) 1263–1276.

cr
[16] M. Mofarahi, F. Gholipour, Gas adsorption separation of CO2/CH4 system using zeolite 5A,

us
Microporous and Mesoporous Materials, 200 (2014) 1–10.

[17] M. Mofarahi, A. Bakhtyari, Pure and binary adsorption equilibria of methane and nitrogen on

an
zeolite 5A, J. Chemical & Engineering Data, 60 (2015) 683–696.

[18] O. Talu, Needs, status, techniques and problems with binary gas adsorption experiments,
M
Advances in Colloid and Interface science, 76-77 (1998) 227-269.

[19] Z. Zhang, W. Zhang, X. Chen, Q. Xia, Z. Li, Adsorption of CO2 on Zeolite 13X and
d
Activated Carbon with Higher Surface Area, Separation Science and Technology, 45 (2010) 710–719.

[20] H. Deng, H. Yi, X. Tang, Q. Yu, P. Ning, L. Yang, Adsorption equilibrium for sulfur dioxide,
te

nitric oxide, carbon dioxide, nitrogen on 13X and 5A zeolites, The Chemical Engineering J., 188
p

(2012) 77– 85.


ce

[21] D.D. Do, Adsorption Analysis: Equilibria and Kinetics, Imperial College Press, London,

1998.
Ac

[22] S.U. Rege, R.T. Yang, M.A. Buzanowski, Sorbents for air pre-purifcation in air separation

Chemical Engineering Science, 55 (2000) 4827-4838.

Table 1. Textural properties of zeolite 13X (Zeochem Co.)

BET surface Langmuir surface Micropore Total pore

area (m2/g) area (m2/g) volume volume

(cm3/g) (cm3/g)

Page 15 of 26
564 744 0.250 0.305

Table 2. Equations, Parameters and RMS Deviation of Adsorption Models

Models Equations Parameters CO2 CH4

t
ip
( )
Langmuir bP −1
C µ = C µs C µ s , 0 mol .kg 5.2765 6.3357
[21] 1 + bP

cr
 Q 
b = b ∞ exp  
 RT 
(
b ∞ bar −1
) 2.14×10−4
0.2063

us
  T  
C µ s = C µ s , 0 exp  χ  1 −  Q / R (K ) 3.11 × 10 3
144.45
  T 0  

χ (dimensionless) 0.3429 2.9635

an
RMS(dimensionless)
0.3417 0.1027
M
Freundlich  −1 
k0  mol.kg− 1.bar n 
1
q = kp n
  4.4695 0.9119
[21]  
d
   
k = k
0
exp  χ 1 − T
  T




χ (dimensionless) 1.6354 5.0530
  0  
te

1 1  T 
= + α 1 − 0  n 0 (dimensionless) 7.6965 1.727
n n0  T 
p

α (dimensionless) 0.5428 1.5672


ce

0.2378 0.0985
RMS(dimensionless)
Ac

Langmuir-

( )
1

Cµs,0 mol.kg −1
n
( bP )
Freundlich C µ = C µs 1 6.5925 4.7789
1 + ( bP ) n

[21]

 Q  T0 
b = b 0 exp   − 1   χ (dimensionless)
 RT 0  T  0.1089 -2.4134

1 1  T 
= + α 1 − 0  b0 (bar −1 ) 8.5278 0.2109
n n0  T 

  T   Q / R (K ) 3.1622 ×10 2.63 × 10


C µ s = C µ s , 0 exp  χ  1 − 

  T 0 

Page 16 of 26
n 0 (dimensionless) 2.3596 1.037

α (dimensionless) 0.5799 0.3808

0.1068 0.0472
RMS(dimensionless)

t
ip
Toth [21] (
q t
qm
) =
( bP ) t
1 + ( bP ) t
(
qm,0 mol.kg −1 ) 6.8664 5.0974

cr
  T  
q m = q m , 0 exp  χ  1 −  χ (dimensionless)
  T 0   -0.5717 -0.3166

us
 Q  T0 
b = b 0 exp   − 1  
 RT 0  T  b0 (bar −1 ) 463.138 0.2131

an
 T  4.3066 × 10 2.2066 × 10
t = t0 + α 1 − 0  Q / R (K )
 T 
M
t 0 (dimensionless) 0.3294 0.8933

α (dimensionless) 0.0133 2.4215


d

P/P0 0.1022 0.0334


te

RMS(dimensionless)
p
ce

Table 3. Parameters and RMS Deviations of Vacancy Solution Model for CH4 and CO2

Model's Parameters
Ac

nis ,∞ bi Λ i3 Λ 3i
Temperat G
RMS
ure as

C 8.393 215.9
273Ka 0.1020 10.2328 2.3491
O2 5 491

C 8.785 1.131
0.0478 4.6235 1.0231
H4 5 4

Page 17 of 26
C 8.005 120.9
283K 0.1135 8.7379 4.0546
O2 0 200

C 5.593 0.824
0.0321 2.0044 0.4989
H4 9 6

C 9.072 126.5

t
303K 0.0616 10.387 8.9227

ip
O2 6 9

C 4.405 0.449

cr
0.0397 0.6239 1.0292
H4 5 8

C 13.04 134.6

us
323K 0.0882 12.4934 14.4359
O2 00 7

C 3.198 0.348

an
0.0134 1.2725 0.1297
H4 0 7

C 6.974 131.3
M
343K 0.0673 2.6685 2.0028
O2 5 2

C 3.558 0.249
0.0433 0.3145 3.1796
d

H4 1 7
a
te

Measured with 0.1 K uncertainty.


p

Table 4. Binary Experimental Data in 303K

qCH4 qCO2 Errorb Errorb


ce

Pa(bar) yCH4 qtotal qCH4 qCO2 SCO2/CH4


(RAST) (RAST) (CH4) (CO2)

4.021 0.182 4.254 0.054 4.200 17 0.048 4.773 11 13.6


Ac

4.169 0.472 4.079 0.136 3.943 26 0.045 4.529 66.9 14.8

4.078 0.797 3.600 0.549 3.051 22 0.426 3.451 22.4 0.19

4.033 0.911 2.254 1.236 1.018 9 1.323 1.377 7.0 35.2

5.996 0.149 4.291 0.085 4.206 9 0.050 4.972 41.1 18.2

6.058 0.507 4.28 0.291 3.994 14 0.190 4.569 34.7 14.3

6.035 0.762 3.900 0.988 2.912 9 0.989 3.311 0.04 13.7

6.025 0.869 3.484 1.836 1.648 6 1.857 2.012 1.10 22.0

Page 18 of 26
a
Measured with 0.5 mbar uncertainty. Average

b
|qexp-qRAST|×100/qexp 23.03 16.49
Error

Table 5. Binary Experimental Data in 323K

t
ip
qCH4 qCO2 Errorb Errorb
Pa(bar) yCH4 qtotal qCH4 qCO2 SCO2/CH4
(RAST) (RAST) (CH4) (CO2)

cr
4.064 0.300 3.729 0.091 3.638 17 0.040 3.979 56 9.3

4.116 0.584 3.221 0.236 2.985 18 0.140 3.371 40.6 12.9

us
4.086 0.753 2.311 0.851 1.459 5 0.896 2.193 5.2 50.3

4.003 0.867 1.670 1.064 0.605 4 1.254 0.946 17.8 56.3

an
6.031 0.331 3.984 0.358 3.626 5 0.311 4.138 13.1 14.12

6.099 0.488 3.779 1.020 2.759 3 0.871 3.278 14.6 18.81


M
6.065 0.723 2.542 1.449 1.093 2 1.530 1.470 5.5 34.4
Average
21.8 28.01
a
Measured with 0.5 mbar uncertainty.
Error
d
te

Table 6. Spreading Pressure and Experimental Activity Coefficient

Values for 303K


p

xCH4 0.012 0.033 0.152 0.253 0.53 0.55


ce

γ
20.95 3.181 0.816 0.722 0.3 0.198
Ac

CH4

π 3.510 4.655 7.166 6.961 8.14 7.99

Table 7. Spreading Pressure and Experimental Activity

Coefficient Values for 323K

xCH4 0.07 0.27 0.36 0.57 0.63

Page 19 of 26
γ
3.066 1.356 0.503 0.729 0.262
CH4

π 3.038 2.331 4.26 2.896 5.037

t
Table 8. Parameters of NRTL and

ip
WILSON Activity Coefficient Models at 303K

cr
for CO2-CH4

Models Parameters RMS

us
τ 21 τ12
NRTL
7.864 2.510 0.173

an
Λ 21 Λ12
WILSON -
2.099 0.367
M
0.0008
d
te

Table 9. Parameters of NRTL and

WILSON Activity Coefficient Models at


p

323K for CO2-CH4


ce

Models Parameters RMS

τ 21 τ12
Ac

NRTL -
6.525 0.214
2.525

Λ 21 Λ12
WILSON -
2.078 0.423
0.0008

Page 20 of 26
t
ip
cr
us
Figure 1: a) N2 Adsorption Isotherm of Zeolite 13X at 77K, b) Pore Size Distribution of Zeolite 13X

an
M
d
p te
ce
Ac

Figure 2: Pure Gas Adsorption Isotherm of Carbon dioxide on Zeolite 13X

Page 21 of 26
t
ip
cr
us
an
Figure 3: Pure Gas Adsorption Isotherm of Methane on Zeolite 13X [17]
M
d
p te
ce
Ac

Page 22 of 26
t
ip
cr
us
an
Figure 4: The Agreement of Experimental Activity Coefficient Data with NRTL Model at 303 K
M
d
p te
ce
Ac

Page 23 of 26
t
ip
cr
us
an
M
d
te

Figure 5: isobaric X-Y Equilibrium Diagram of the Mixture CH4/CO2 at Different Pressures and
p

Temperatures, Experimental Data, ……. RAST Results, - - IAST Results, --- . VST Results
ce
Ac

Page 24 of 26
Ac
ce
pte
d
M
an
us
cr
ip
t

Page 25 of 26
t
ip
cr
us
Figure 6: The Adsorbed Amount of Methane and Carbon Dioxide and the Total Amount of

Adsorption with the RAST Results, Exp total, ■ Exp CH4, ♦Exp CO2, …… RAST Total, --- .

an
RAST CH4, - - RAST CO2
M
Highlights:
d

Carbon dioxide and Methane binary adsorption equilibrium data were measured
te

Adsorption selectivity was measured at different temperatures and pressures


p
ce

Experimental data were compared against results of thermodynamic models


Ac

Page 26 of 26

You might also like