Adsorption (2015) 21:419–432

DOI 10.1007/s10450-015-9683-7

Adsorption kinetics of CO2, CO, N2 and CH4 on zeolite LiX pellet

and activated carbon granule
Youngsan Ju1 • Yongha Park1 • Dooyoung Park1 • Jae-Jeong Kim1 •

Chang-Ha Lee1

Received: 3 April 2015 / Revised: 22 June 2015 / Accepted: 23 June 2015 / Published online: 30 June 2015
Ó Springer Science+Business Media New York 2015

Abstract The adsorption uptake curves of CO2, CO, N2 Dc/rc Micropore diffusion time constant (s-1)
and CH4 on zeolite LiX and activated carbon were mea- Dp/Rp Effective macropore diffusion time constant (s-1)
sured using a volumetric method at 293, 308 and 323 K h Overall heat transfer coefficient (J m-1 s-1 K-1)
and pressure up to 100 kPa. The experimental uptake K Henry constant (–)
curves were correlated with a non-isothermal kinetic model P Pressure (kPa)
because the adsorption kinetics was controlled by heat Q Adsorbed amount (mol kg-1)
generation and transfer, and an isothermal model showed Qst Isosteric heat of adsorption (kJ mol-1)
large deviation from experimental uptake. The adsorption R Ideal gas constant (J mol-1 K-1)
rates of the gases on zeolite LiX and activated carbon were R2 Coefficient of determination (R2) (–)
affected by the isosteric heat of adsorption, heat transfer T Time (s)
rate and adsorption affinity. At the same pressure and T Temperature (K)
temperature, the sequence of effective diffusion time con- Vg Volume occupied by the gas (m3)
stants was CO2  CO \ N2 \ CH4 for zeolite LiX and Vs Volume occupied by the adsorbent (m3)
CO2  CH4 B N2 \ CO for activated carbon. The a Dimensionless parameter defined by non-
adsorption rate of CO showed the largest difference isothermal model (–)
between the two adsorbents. Effective diffusion time con- b Dimensionless parameter defined by non-
stants (D/R2) for all cases were provided, which depended isothermal model (–)
on pressure and temperature. qs Density of the adsorbent (g m-3)

Keywords Activated carbon  Zeolite LiX  Adsorption Subscript

kinetics  Heat of adsorption  Non-isothermal model 0 Initial state
? Equilibrium state
List of symbols
a External surface area divided by the volume of
the adsorbent (m-1) 1 Introduction
cp Concentration in the particle (mol kg-1)
Cs Heat capacity of the sample (J g-1 K-1) Recently, the reduction of greenhouse gas (GHG) emissions
2
D/R Effective diffusion time constant (s-1) has been emphasized in various industrial fields in order to
D Diffusivity (m2 s-1) prevent global warming. Therefore, many adsorptive tech-
nologies have been reported to recover clean fuel sources (H2,
CH4, CO, etc.) and to capture CO2 from various effluent gases.
& Chang-Ha Lee Since H2, CH4 and CO are used as chemical raw
[email protected]
materials or as clean fuel sources, great needs from various
1
Department of Chemical and Biomolecular Engineering, industrial fields lead to strong economic motivation for the
Yonsei University, Seoul, Korea development of adsorption processes to recover these gases

123
420 Adsorption (2015) 21:419–432

from reforming gas, syngas, coke oven gas, bio-gas and wide pore size distribution from micropores to macropores
more traditional sources (Lee et al. 2008; Grande and (Gray and Do 1991; Himeno et al. 2005).
Rodrigues 2007; Ahn et al. 2001; Yang and Lee 1998). For In this study, the adsorption kinetics of four pure com-
example, efficient H2 recovery from the off-gas of the ponents (CO2, CO, N2 and CH4) on commercial activated
integrated gasification combined cycle (IGCC) is an carbon granules and zeolite LiX pellets is described
important issue for green power generation (Lee et al. because these components are generally found in H2 off-
2014b). After carbon capture in the IGCC, recovered H2 gases and various effluent gases. The adsorption of H2 and
can be used to power hydrogen turbines and as a renewable Ar was too fast to evaluate their adsorption rates on the
energy source. adsorbents. The adsorption uptake curves were measured
Pressure or vacuum swing adsorption (PSA or VSA) using a volumetric method at three different temperatures
process using porous adsorbents is a promising process for of 293, 308 and 323 K and pressures up to 100 kPa. The
such purpose which can meet the requirement for high effective diffusion time constants (D/R2) of each compo-
separation efficiencies with respect to energy and envi- nent on each adsorbent at various pressure and temperature
ronmental targets. To select proper adsorbents and to conditions were obtained by correlating a non-isothermal
design efficient adsorption processes, adsorption equilib- kinetic model with experimental uptake curves. The effect
rium and kinetic data for each component on a porous of adsorption heat on adsorption kinetics was evaluated by
adsorbent are the most important factors. The effluent gases comparing the results of isothermal and non-isothermal
from traditional and non-traditional processes consist of kinetic models. The controlling mechanism of diffusion in
3–6 major components after various pretreatment processes zeolite LiX was determined by comparison of the effective
including particle removal, drying and/or large molecule diffusion time constants of different sizes of zeolite pellets.
removal. For example, the coal gasifier in the IGCC gen- The results allow optimized design of various adsorptive
erates a mixture gas containing CO2, CO, N2, CH4, Ar and processes for H2 recovery and CO2 capture.
H2 (Lee et al. 2014a). Bio-gas and natural gas are mainly
composed of CH4, CO2 and N2 (Cavenati et al. 2006;
Grande and Rodrigues 2007). Coke oven gas and other 2 Theory
effluent gases from iron and steel industries contain H2,
CH4, CO2, N2, and CO, and reforming gas in petrochemical In a volumetric system, a certain amount of adsorbate gas
industries consists of H2, CO2, CH4 and CO as a main is injected into the cell, and the pressure is monitored until
component (Yang and Lee 1998; Ahn et al. 2001). In many it reaches an equilibrium state. The pressure change was
studies of adsorption processes, two different adsorbents calculated based on the amount adsorbed using the virial
such as activated carbons and zeolites are simultaneously coefficients and known volume of the adsorption cell.
applied to one adsorptive unit for optimal process design The adsorbents, activated carbon and/or zeolite, packed in
(You et al. 2012; Ahn et al. 2001, 2012; Cavenati et al. the adsorption bed have a certain range of granule or pellet
2006; Siriwardane et al. 2001). Naturally, to meet the size. In the study, a certain number of adsorbents were used
proper selection criteria for designing the processes, the in the adsorption cell as received from manufacturers. Since
kinetic behaviors and thermodynamic characteristics of both micropore and macropore diffusions are described by
various gases in adsorbents should be understood. the same form of Fick’s diffusion law, the effective diffusion
In adsorptive processes, pelletized adsorbents are gen- time constant could be interpreted as a micropore diffusivity
erally packed into an adsorption bed to minimize pressure (Dc/r2c ) or effective macropore diffusivity (Dp/R2p), depend-
drop in the bed. Zeolite pellets consist of micropores in ing on the controlling diffusion mechanism of each system.
zeolite crystal, macropores among zeolite crystals and In the study, the effective diffusion time constant (D/R2)
binder. Therefore, studies have analyzed the contributions obtained from the experimental uptake curve was assumed to
of macropores and micropores in the rate-determining step reflect the overall diffusion mechanism of an adsorbate in
of N2, CH4 and CO2 in zeolite 4A, 5A, CaX and 13X various pores of the adsorbent pellet.
pellets (Ahn et al. 2002, 2004; Silva et al. 2012; Hu et al. The adsorption kinetic model for experimental uptake is
2014). In a carbon molecular sieve, controlled micropores derived from Fick’s diffusion law in the cell assuming
significantly affect the adsorption kinetics, and the kinetic constant diffusivity (D). To analyze the adsorption kinetics,
difference among adsorbates strongly depends on molec- the simplest analytical solution for a constant volume
ular size and adsorption affinity (Bae and Lee 2005; Giesy system, which assumes isothermal condition and a linear
and LeVan 2013; Yang et al. 2014). There are few studies equilibrium relationship, was derived as follows: (Crank
of adsorption kinetics of activated carbon granules with 1956)

123
Adsorption (2015) 21:419–432 421

     
M1  Mt X1
6a0 ða0 þ 1Þ 2 D
ha D
¼ exp bn 2 t ð1Þ a¼ = 2 ð9Þ
M1  M0 n¼1 9a0 þ 9 þ a02 b2n R qs Cs R
 
DH oq
where b¼ ð10Þ
qs Cs oT c0 ;T0
3bn
tan bn ¼ ð2Þ where a is the external surface area divided by the volume
3 þ a0 b2n
of the adsorbent; qs and Cs represent the density and heat
3V capacity of the sample, respectively; h is the overall heat
a0 ¼ ð3Þ
4pR3 K transfer coefficient; and DH is the heat of adsorption.
In Eqs. (1)–(3), Mt, M0 and M? represent the amount Parameter a describes the relative ratio of heat transfer
adsorbed at ambient time t, initial time 0 and equilibrium, (ha/qsCs) to effective diffusion time constant (D/R2).
respectively, V is the volume of the system, R is the radius Therefore, a large value of a implies that the heat generated
of the adsorbent sample and K is the effective Henry’s law during adsorption transferred into the surroundings rather
constant at a corresponding pressure interval in a stepwise than increasing the temperature of the adsorbent. A small
change. value of a, on the other hand, implies that heat generation
When the isothermal model is applied to experimental is concentrated at the initial part of the uptake because of
data, a key factor that generates significant error is the heat the rapid uptake of molecules. Therefore, temperature
effect generated by the heat of adsorption. The heat gen- increases during the adsorption, which reduces the equi-
erated leads to increased temperature in the system, and librium capacity. The slope of the uptake curve changes
heat is dissipated into the surroundings. Therefore, the with time until heat is fully dissipated into the atmosphere.
adsorbate molecules diffuse faster at higher temperatures Parameter b represents the heat of adsorption (DH/qsCs)
while adsorption of molecules on the adsorbent surface is and the temperature dependence of the equilibrium
retarded due to thermal resistance and equilibrium shift capacity on the adsorbent (qq*/qT). When the equilibrium
(Ahn et al. 2002, 2004; Do 1998). Due to the heat of shift with temperature increase is minor, the kinetic result
adsorption, the former isothermal model was modified to is close to that predicted by the isothermal model. On the
reflect the non-isothermal condition induced by heat contrary, if an adsorbate has strong temperature depen-
release (Kočiřı́k et al. 1984). The following equations are dence with a large value of b, the uptake curve deviates
the analytical solution for a non-isothermal condition in a from the isothermal model, resulting in a heat transfer-
volumetric system: controlled system.
 h i2  The reliability of simultaneous non-linear regression for

X 1 9 1 þ K Vs Yn
exp b2n RD2 t three parameters (a, b and D/R2) in Eq. (4) is not guar-
M1  Mt Vg b2n
¼ h   i anteed in general. Therefore, to obtain (ha/qsCs) and b as a
M1  M0 n¼1 1 þ 3 b0 b cot b Yn2 þ 1 þ 3 K Vs 14 An
un 2 un n n b
n
2 V g b n un first step, the kinetic model of an entirely heat-controlled
ð4Þ system was presented in the asymptotic form of kinetic
models as follows: (Hu et al. 2014; Lee and Ruthven 1979;
in which bn are the roots of: Ruthven et al. 1980)


Vs 1 M1  Mt b hat
ðbn þ aÞ þ 3bYn  3K ðbn þ aÞYn ¼ 0 ð5Þ ¼ exp ð11Þ
Vg b2n M1  M0 1 þ b qs Cs ð1 þ bÞ

and where In the corresponding system, the uptake curve is linear

  on a semi-log scale for time t.
An ¼ Yn b2n  a bn cot bn  2a þ b2n ðb2n  aÞ ð6Þ As a result, a and b were obtained first from the latter
part of the experimental uptake curve by using Eq. (11),
Yn ¼ bn cot bn  1 ð7Þ
! and then the effective diffusion time constant was obtained
1 1 Vs Y n by fitting the entire uptake curve using the non-isothermal
¼ 1  3K ð8Þ kinetic model, Eq. (4). In this approach, the secant method,
un b Vg b2n
incremental search method (ISM) and least-squares method
In Eqs. (4)–(8), Vs and Vg are the volume occupied by were used for non-linear regression using MATLAB
the solid and the gas, respectively. (Mathworks, Inc.). In addition, a few sets of uptake data
In this model, two parameters (a, b) regarding the rate of were fitted using the isothermal model in order to evaluate
heat transfer and heat of adsorption are defined as follows: the thermal effects on adsorption.

123
422 Adsorption (2015) 21:419–432

In this study, experimental deviation from the model small and the effective diffusion time constants were
was estimated by the following coefficient of determination almost the same within the experimental error range.
(R2): However, a smaller amount of sample showed slightly
P more fluctuation in the isotherm measurement. In the study,
2
ðqexp  qni Þ2
R ¼1 P ð12Þ 0.8779 g of zeolite LiX with the average pellet size of
ðqexp  qÞ2 1.5 mm and 1.3811 g of activated carbon were used for the
where qexp was the experimental point; qni was the value experiments.
from the non-isothermal kinetic model; and q was the mean The adsorbent was placed into the adsorption cell with a
value of the experimental data. When the model agreed ceramic cap (Swagelok SS-8-VCR-2-GR-.5M) with 0.5 lm
well with the data, the value of R2 approached 1. pores after two He purge. Then, it was assumed that the flow
resistance into the adsorption cell is negligible. To minimize
error caused by moisture and any trace of pollutants adsorbed
3 Experimental section during the installation of the adsorption cell, the zeolite LiX
and activated carbon in the cell were reactivated at 623 and
3.1 Material 393 K, respectively, under vacuum for longer than 12 h.
The system, composed of gas storage tanks, lines, valves
Spherical zeolite LiX pellets and cylindrical activated and gauges, was set in an air bath, while the temperature of
carbon granules were supplied by ZEOCHEM Co. (Z10- the adsorption cell was controlled by a water bath circu-
05-03) and KURARAY CHEMICAL Co. (coal-derived lator. Two temperature controllers for each air bath and
activated carbon; 2GA-H2J), respectively. The zeolite LiX water bath were used in order to keep the temperature of
and activated carbon were analyzed using an automatic the internal system and adsorption cell constant. The
volumetric sorption analyzer (BELSORP-Max) with CO2 internal volume of the system and the dead volume in the
at 298 K. Table 1 presents the measured and supplied adsorption system were measured by using helium gas at
physical properties of the adsorbents. Gases used as an experimental temperature.
adsorbates (CO2, CO, N2 and CH4) were of high purity The internal system was purged with helium gas before
([99.99 %). experiments, followed by an evacuation process with vac-
uum pumps. After repeating the purge/vacuum step three
3.2 Experimental procedure times, a specified amount of adsorptive gas was released into
the internal system. The experiment was carried out by
A high-pressure volumetric system in Fig. 1 (BELSORP- injecting a small amount of adsorbate into the adsorption cell
HP) was used to measure adsorption equilibrium and at a constant temperature, and the pressure variation was
uptake. Prior to the experimental run, zeolite LiX was monitored with an electrical pressure sensor (Full scale:
regenerated at 623 K in an oven and activated carbon at 0.133 MPa (abs) 9 1 unit, Accuracy: ±0.08 % of full
393 K in a vacuum oven for longer than 8 h. The mass of scale). Once equilibrium was reached, a certain amount of
adsorbent was measured by a microbalance with an accu- adsorbate was again injected into the adsorption cell, and this
racy of ±10 lg. procedure was repeated with a stepwise pressure change. For
Before deciding on the sample mass in the volumetric each experiment, all thermodynamic and uptake data were
system, two sample masses (0.3922 and 0.8779 g) were recorded automatically on an interfaced computer. The
tested for the N2/zeolite LiX system at 293 K in a pressure amount of gas adsorbed was calculated from the measured
range of 0–100 kPa. The difference between the uptake data on temperature, volume and pressure based on the
curves obtained from two different sample masses was very compressibility factors from NIST thermodynamic isotherm
properties (Linstrom and Mallard 2001). The recorded
pressure change was converted into fractional uptake data at
Table 1 Physical properties of adsorbents
the equilibrium pressure.
Property Zeolite LiX Activated carbon

Type Pellet Cylindrical

Particle size (mm) 1.5–1.7 1.7–2.36 4 Results and discussion
Particle porosity 0.64 0.43
BET surface area (m2/g) 755 800
4.1 Comparison of isothermal and non-isothermal
3 kinetic models
Pore volume (cm /g) 0.326 0.205
Pellet density (g/cm3) 2.4 0.85
Five to six sets of uptake data for CO2, CO, N2 and CH4 on
Heat capacity (J/g K) 1.76 1.05
zeolite LiX and activated carbon at 298, 308 and 323 K

123
Adsorption (2015) 21:419–432 423

Fig. 1 Schematic diagram of volumetric apparatus

were compiled at similar pressure conditions for each

adsorbate over a pressure range of 15–90 kPa. Since the
kinetic models assume a linear equilibrium relationship, it
was assumed for the linear relationship between pressures
of a stepwise change in the range of 3–9 kPa, depending on
the adsorbate and the pressure condition. The effective
diffusion time constant was obtained by fitting each
experimental uptake curve using a non-isothermal kinetic
model.
Figures 2 and 3 represent the experimental uptake
curves for CO2, CO, N2 and CH4 at 293 K in zeolite LiX
and activated carbon, respectively. The non-isothermal
kinetic model predicted reasonably well the experimental
uptake curve for each adsorbate with R2 values higher than
0.98 for all gases. As a representative case, the isothermal
Fig. 2 Experimental and theoretical uptake curves of various gases
model results for CH4 on zeolite LiX and CO on activated on zeolite LiX at 293 K: short dashed line non-isothermal diffusion
carbon are shown in Figs. 2 and 3. The same effective model; solid line isothermal diffusion model
diffusion time constant obtained from the non-isothermal
model was applied to the isothermal model.
Over the range of measurement time, the isothermal equilibrium. The result implied that isothermal diffusion at
model predicted the uptake curves during the initial period the initial period time was changed to the non-isothermal
time, but the deviation from the experimental results diffusion by the heat release from the heat of adsorption
increased with time even though their heats of adsorption with time. Although the non-isothermal model accurately
were smaller than that of CO2, as shown in Fig. 4. The predicted the experimental uptake curves, the deviation
adsorption of the adsorbates on zeolite LiX and activated from the CO2 uptake curve was relatively higher than that
carbon was disturbed by a thermal effect and was com- of the other curves. The assumptions of constant diffusivity
pletely controlled by the heat when approaching and instantaneous equilibration of adsorbate in the

123
424 Adsorption (2015) 21:419–432

experimental range might lead to greater deviation in the

CO2 uptake curve than in those of the other components.

4.2 Adsorptive kinetic behaviors

The isosteric heat of adsorption and the adsorption iso-

therm are key factors that affect adsorption rate. Because
the high isosteric heat of adsorption arises from a strong
affinity, it increases the molecular diffusion rate but pre-
vents adsorption on the adsorbent surface. In general, a
slower adsorption rate is expected at a higher isosteric heat
of adsorption. To explain the adsorptive kinetic behaviors,
the adsorption isotherms and the isosteric heats of
adsorption for all gases on zeolite LiX and activated carbon
are presented in the same experimental pressure range in
Fig. 3 Experimental and theoretical uptake curves of various gases
on activated carbon at 293 K: short dashed line non-isothermal Figs. 4a and b, respectively.
diffusion model; solid line isothermal diffusion model

Fig. 4 Adsorption isotherms at 293 K and isosteric heats of adsorption on adsorbents with respect to the adsorbed amount on a, b zeolite LiX
and c, d activated carbon: (circle CO2; up-pointing triangle CO; square N2; down-pointing triangle CH4) (Data from Park et al. 2014)

123
Adsorption (2015) 21:419–432 425

As shown in Fig. 4, the adsorption isotherm and heat of addition, Fig. 4a implies that N2 and CO had low adsorp-
adsorption of N2 on activated carbon were the lowest in all tion affinity on activated carbon, resulting in low isosteric
the adsorbate/adsorbent systems. In Figs. 2 and 3, the heats of adsorption. As a result, a slower than expected
uptake curve of N2 on activated carbon was curved around adsorption rate for N2 and CO was observed on activated
5–7 s more steeply than the other curves. This uptake curve carbon.
indicates that the adsorption of N2 occurred mainly in the In the case of zeolite LiX, CO2 had an a of 1.8, while the
initial time period, but the adsorption rate of N2 with rel- others (CO, N2 and CH4) had values of 0.81, 0.58 and 0.56,
atively weak adsorption affinity was affected by a thermal respectively. Although the relative heat transfer rate for
effect even though the heat of adsorption was relatively CO2 was faster than those of the others, the approximately
low. However, the result could be expected for fast thermal 2–3 times higher isosteric heat of adsorption shown in
equilibrium in the N2/activated carbon system. Fig. 4 generated high thermal resistance during CO2
The effective diffusion time constants obtained from the adsorption and slow CO2 adsorption. The other gases
uptake curves are listed in Tables 2 and 3, and presented in showed significant differences between the isosteric heats
Fig. 5. The order of adsorption rate on zeolite LiX was of adsorption but not between relative heat transfer rates.
CO2  CO \ N2 \ CH4 at 293 K over the whole pressure As a result, since the heat transfer rates were not suffi-
range, as shown in Fig. 5a. In Fig. 4b, the isosteric heats of ciently high to quickly release the produced heat, the order
adsorption of CO2, CO, N2 and CH4 on zeolite LiX were of adsorption rates on zeolite LiX followed the decreasing
CH4 \ N2 \ CO  CO2, which is opposite the adsorption order of isosteric heat of adsorption.
rate order. For adsorption on activated carbon, the order of
adsorption rate was CO2  CH4 B N2 \ CO. According 4.3 Dependence of pressure and temperature
to Fig. 4b, the sequence of adsorption rates of gases was on adsorption kinetics
not perfectly matched with the order of the isosteric heat of
adsorption: N2 B CO \ CH4 \ CO2. Figure 5 represents the effective diffusion time constants of
Because the heat generation is concentrated at the early gases on zeolite LiX and activated carbon at 293 K as a
stage, the influence of heat of adsorption on adsorption rate function of pressure. The effective diffusion time constants
is quite strong at this stage. On the other hand, when the for the gases were on the same order of magnitude except for
system approaches equilibrium, the adsorption rate is more that of the CO2/zeolite LiX system. The effective diffusion
affected by heat transfer rather than heat generation. The time constants of some gases, especially N2 on zeolite LiX
contributions of these two effects to the adsorption rate and CO on activated carbon, increased slightly with pressure,
differ among adsorbate/adsorbent systems. while CO2 and CO on zeolite LiX and CO2 and CH4 on
As mentioned in the non-isothermal kinetic model in activated carbon remained nearly constant. The pressure
Eq. (4), the effective diffusion time constant was affected dependences may have arisen from changes in the slope of
by the two parameters of heat of adsorption (b) and rate of the adsorption isotherm and the isosteric heat of adsorption.
heat transfer (a). The values used for the non-isothermal The diffusion in pelletized zeolites is affected by both
kinetic model are listed in Tables 2 and 3. The relative heat micropore and macropore diffusions, but it is hard to
transfer rates of gases, a, represents the influence of the determine the controlling diffusion mechanism of the
heat transfer rate (ha/qsCs) compared to the diffusion rate. zeolite system by the effective diffusion time constant.
Therefore, if the system has a large a value, it is controlled Thus, a controlling mechanism in zeolite pellet was indi-
by adsorption affinity due to immediate heat release. In the rectly examined by measuring the adsorption rate by
opposite case, the contribution of slow heat transfer to changing the macropore diffusion length (Lee and Ruthven
adsorption rate is increased because the generated heat 1979). Figure 6 is the comparison between the effective
precludes additional adsorption on the surface. diffusion time constants in different size pellets of zeolite
The adsorption rate of CO2 with the largest isosteric heat LiX (1.5 and 0.8 mm). The effective diffusion time con-
of adsorption on activated carbon was slower than that of stant of CO2 in Fig. 6a, that has the highest adsorption
the other gases. However, the adsorption rate of N2, which affinity to zeolite LiX, was not affected by the pellet size,
was expected to be the fastest, was similar to that of CH4. whereas CO and CH4 showed a little difference at a rela-
When comparing approximate heat transfer rates in terms tively low pressure range. It implied that the contribution
of a, CO2 and CH4 were greater than 1, whereas N2 and CO of micropore diffusion to the effective diffusion time
were 0.24 and 0.50 respectively. Therefore, due to rela- constants in CO2, CO and CH4/zeolite LiX systems is
tively slow heat transfer, the adsorption of N2 on the much higher than that of macropore diffusion.
activated carbon surface is disturbed by the generated heat. The effective diffusion time constants of N2 in zeolite
Consequently, the adsorption uptake curve for activated LiX showed a certain difference between two different size
carbon is drastically changed, as shown in Fig. 3. In pellets in Fig. 6b. The uptake at the smaller size pellet was

123
426 Adsorption (2015) 21:419–432

Table 2 Effective diffusion time constants and non-isothermal model parameters (ha/qsCs and b) of CO2, CO, N2 and CH4 on zeolite LiX
obtained from the experimental uptake curves and the non-isothermal model
CO2
293 K
P (kPa) 19.9 30.6 44.4 52.4 61.3 80.7
D/R2 (s-1) 7.71 9 10-3 7.52 9 10-3 7.84 9 10-3 8.01 9 10-3 8.14 9 10-3 7.98 9 10-3
ha/qsCs (s-1) 0.0125 0.0129 0.0141 0.0141 0.0150 0.0156
b (–) 0.220 0.294 0.441 0.467 0.586 0.693
308 K
P (kPa) 22.7 35.4 42.7 59.2 68.4 77.9
D/R2 (s-1) 1.06 9 10-2 8.81 9 10-3 1.17 9 10-3 1.08 9 10-2 9.77 9 10-3 9.46 9 10-3
ha/qsCs (s-1) 0.0142 0.0141 0.0142 0.0146 0.0154 0.0157
b (–) 0.203 0.221 0.380 0.445 0.579 0.628
323 K
P (kPa) 25.2 47.4 56 65.1 74.5 84.3
D/R2 (s-1) 1.38 9 10-2 1.29 9 10-2 1.32 9 10-2 1.25 9 10-2 1.12 9 10-2 1.18 9 10-2
-1
ha/qsCs (s ) 0.0157 0.0160 0.0161 0.0164 0.0178 0.0185
b (–) 0.232 0.328 0.397 0.421 0.566 0.605
CO
293 K
P (kPa) 23.3 32.8 43.2 54.2 65.7 77.7
D/R2 (s-1) 3.02 9 10-2 3.47 9 10-2 3.51 9 10-2 3.48 9 10-2 3.41 9 10-2 3.51 9 10-2
ha/qsCs (s-1) 0.0242 0.0241 0.0283 0.0245 0.0290 0.0261
b (–) 0.208 0.250 0.371 0.343 0.506 0.439
308 K
P (kPa) 21.2 30.3 40.4 51.5 63.1 75.1
D/R2 (s-1) 2.94 9 10-2 5.04 9 10-2 4.37 9 10-2 4.31 9 10-2 5.05 9 10-2 4.10 9 10-2
ha/qsCs (s-1) 0.0251 0.0277 0.0273 0.0271 0.0272 0.0321
b (–) 0.213 0.268 0.290 0.322 0.347 0.536
323 K
P (kPa) 19 28.3 38.7 50 62.1 74.1
D/R2 (s-1) 2.76 9 10-2 2.63 9 10-2 4.96 9 10-2 4.11 9 10-2 5.19 9 10-2 4.91 9 10-2
-1
ha/qsCs (s ) 0.0231 0.0273 0.0286 0.0285 0.0301 0.0327
b (–) 0.155 0.232 0.276 0.332 0.360 0.493
N2
293 K
P (kPa) 18.9 29.4 40.7 52.2 63.9 76.1
D/R2 (s-1) 3.03 9 10-2 3.93 9 10-2 4.99 9 10-2 5.33 9 10-2 5.39 9 10-2 5.43 9 10-2
ha/qsCs (s-1) 0.0279 0.0287 0.0287 0.0327 0.0331 0.0326
b (–) 0.103 0.161 0.192 0.295 0.339 0.358
308 K
P (kPa) 23.2 35.1 47.4 59.9 72.5 85.2
D/R2 (s-1) 4.12 9 10-2 5.04 9 10-2 5.98 9 10-2 6.83 9 10-2 7.54 9 10-2 7.24 9 10-2
ha/qsCs (s-1) 0.0306 0.0290 0.0301 0.0356 0.0353 0.0359
b (–) 0.083 0.139 0.180 0.276 0.280 0.310
323 K
P (kPa) 26.7 40 53.3 66.8 80.4
D/R2 (s-1) 5.03 9 10-2 6.60 9 10-2 7.20 9 10-2 7.24 9 10-2 7.17 9 10-2
-1
ha/qsCs (s ) 0.0287 0.0267 0.0304 0.0314 0.0366
b (–) 0.0699 0.0989 0.157 0.213 0.299

123
Adsorption (2015) 21:419–432 427

Table 2 continued
CH4
293 K
P (kPa) 19 29 39.5 50.2 61.3 83.6
D/R2 (s-1) 4.73 9 10-2 6.31 9 10-2 6.03 9 10-2 6.94 9 10-2 6.77 9 10-2 6.41 9 10-2
-1
ha/qsCs (s ) 0.0300 0.0329 0.0341 0.0349 0.0379 0.0400
b (–) 0.081 0.124 0.177 0.226 0.295 0.375
308 K
P (kPa) 22.6 33.9 45.6 57.2 69 80.4
D/R2 (s-1) 6.01 9 10-2 7.23 9 10-2 7.03 9 10-2 7.44 9 10-2 6.33 9 10-2 5.78 9 10-2
-1
ha/qsCs (s ) 0.0335 0.0323 0.0345 0.0353 0.0415 0.0453
b (–) 0.073 0.110 0.185 0.207 0.370 0.496
323 K
P (kPa) 25.8 38.6 51.3 63.9 76.6
D/R2 (s-1) 7.38 9 10-2 8.20 9 10-2 8.21 9 10-2 8.39 9 10-2 6.87 9 10-2
-1
ha/qsCs (s ) 0.0327 0.0333 0.0375 0.0369 0.0442
b (–) 0.056 0.107 0.143 0.183 0.346

faster in the whole pressure range. Therefore, the macro- CO2 and CH4 with activated carbon. The other systems
pore diffusion worked as a controlling step in the effective showed relatively constant heats of adsorption.
diffusion time constant. However, the difference at a low In Fig. 5, the adsorption rate increased because the
pressure was relatively small when compared with the influence of adsorption potential decreased with pressure.
difference in N2/zeolite 10X system (Ahn et al. 2002). It However, such phenomena were observed only in the low-
may stem from the relatively strong adsorption affinity of pressure range, and the CO2/adsorbent systems did not
N2 with zeolite LiX. Then, the difference between two show any significant changes over the entire pressure
different size pellets decreased with an increase in pressure, range. Considering the slope change of the isotherms and
which could be interpreted as the transition of the con- the isosteric heat variation with pressure, the two phe-
trolling step from macropore diffusion to micropore dif- nomena had offsetting effects on adsorption rate. The CO/
fusion. It was also confirmed that the effective diffusion activated carbon system, which had relatively fast
time constants increased as the slope of isotherm adsorption compared to the other systems, increased the
decreased, which implied the decrease in macropore dif- effective diffusion time constant at high pressures. Because
fusion contribution with pressure (Hu et al. 2014; Ahn et al. of the stepwise increment in pressure for the volumetric
2002). device, sufficiently high pressure guaranteed higher surface
The diffusion rate in macropore controlled system is coverage on the adsorbent, which interrupted diffusion and
easily affected by pressure because the slope of isotherm adsorption of molecules on the adsorbent surface. The
decreases with pressure. If the slope of the isotherm is pressure effect on the adsorption rate is significant at low
steep, adsorption potential is high near the adsorbent pressures and becomes weaker with increasing pressure.
crystal surface when adsorbate molecules are trapped near The comparison of uptake curves between the two
the micropores, which resist diffusion in macropores (Ahn pressures is shown in Fig. 7. As mentioned before,
et al. 2004). As shown in Fig. 4a, the slopes of adsorption adsorption was rapid in the early stages of all the systems
isotherms decreased with increasing pressure. In zeolite and then approached equilibrium as the system became
LiX, the slope changes of adsorption isotherms from heat-controlled. At higher pressures, the uptake curve was
pressures greater than 20 kPa became very small in CO2 smooth during the initial period of time, and the adsorption
and CO, and those in N2 and CH4 were not significant over rate also was faster.
the entire pressure range. In activated carbon, the slope The uptake curves of the CO2/zeolite system in Fig. 7a
changes of adsorption isotherms were relatively smaller almost overlapped in the initial period time, and their dif-
than those in zeolite LiX over the entire pressure range. ference gradually increased with time. The uptake curves
The isosteric heats of adsorption in Fig. 4b decreased with of CO showed a similar trend. On the other hand, since the
increasing loading amount, resulting from decreasing N2/zeolite LiX system with slow heat transfer rates resulted
adsorption affinities of CO2 and CO with zeolite LiX and in a heat-controlled system at higher pressures earlier than

123
428 Adsorption (2015) 21:419–432

Table 3 Effective diffusion time constants and non-isothermal model parameters (ha/qsCs and b) of CO2, CO, N2 and CH4 on activated carbon
obtained from the experimental uptake curves and the non-isothermal model
CO2
293 K
P (kPa) 17.5 30.1 39.1 54.9 63.5 74.7
D/R2 1.49 9 10-2 2.23 9 10-2 1.93 9 10-2 1.73 9 10-2 1.63 9 10-2 1.54 9 10-2
ha/qsCs (s-1) 0.0268 0.0244 0.0236 0.0231 0.0267 0.0240
b (–) 0.175 0.221 0.281 0.367 0.545 0.562
308 K
P (kPa) 18.3 31.6 44.5 58.2 72.8 85.4
D/R2 1.80 9 10-2 2.01 9 10-2 2.38 9 10-2 1.84 9 10-2 1.20 9 10-2 1.27 9 10-2
ha/qsCs (s-1) 0.0234 0.0248 0.0234 0.0261 0.0207 0.0210
b (–) 0.153 0.248 0.279 0.440 0.477 0.491
323 K
P (kPa) 16.6 27.9 42.9 53.3 69.8 84.1
D/R2 1.61 9 10-2 1.93 9 10-2 2.47 9 10-2 2.17 9 10-2 1.95 9 10-2 2.10 9 10-2
-1
ha/qsCs (s ) 0.0260 0.0249 0.0244 0.0244 0.0286 0.0272
b (–) 0.152 0.190 0.231 0.327 0.480 0.511
CO
293 K
P (kPa) 24.4 34 48.4 58 72 86.2
D/R2 6.48 9 10-2 7.62 9 10-2 8.68 9 10-2 8.87 9 10-2 8.74 9 10-2 9.19 9 10-2
ha/qsCs (s-1) 0.0482 0.0457 0.0434 0.0441 0.0495 0.0470
b (–) 0.0856 0.119 0.139 0.186 0.276 0.266
308 K
P (kPa) 26.6 37.3 47.5 57.6 72.5 87.2
D/R2 8.30 9 10-2 8.92 9 10-2 9.38 9 10-2 9.38 9 10-2 1.01 9 10-1 9.70 9 10-2
ha/qsCs (s-1) 0.0470 0.0400 0.0455 0.0417 0.0480 0.0511
b (–) 0.0592 0.0868 0.123 0.150 0.177 0.254
323 K
P (kPa) 22.9 34.1 45.1 55.7 71.5 86.8
D/R2 7.67 9 10-2 9.93 9 10-2 1.00 9 10-1 1.04 9 10-1 1.20 9 10-1 1.15 9 10-1
-1
ha/qsCs (s ) 0.0328 0.0368 0.0427 0.0396 0.0428 0.0441
b (–) 0.0423 0.0518 0.0633 0.127 0.129 0.209
N2
293 K
P (kPa) 18.8 32.5 46.2 55.4 69 82.7
D/R2 4.42 9 10-2 5.44 9 10-2 6.45 9 10-2 6.38 9 10-2 5.68 9 10-2 5.73 9 10-2
ha/qsCs (s-1) 0.0120 0.0147 0.0158 0.0180 0.0194 0.0226
b (–) 0.144 0.183 0.206 0.250 0.321 0.572
308 K
P (kPa) 20.8 31.1 41.2 56.2 70.7 89.6
D/R2 5.96 9 10-2 7.18 9 10-2 7.76 9 10-2 6.99 9 10-2 7.30 9 10-2 5.91 9 10-2
ha/qsCs (s-1) 0.0150 0.0183 0.0174 0.0168 0.0215 0.0257
b (–) 0.145 0.201 0.243 0.269 0.355 0.736
323 K
P (kPa) 22.4 33 43.7 54.2 69.6 84.9
D/R2 7.45 9 10-2 7.79 9 10-2 7.49 9 10-2 7.71 9 10-2 8.04 9 10-2 7.87 9 10-2
-1
ha/qsCs (s ) 0.0138 0.0173 0.0182 0.0196 0.0206 0.0212
b (–) 0.182 0.237 0.296 0.365 0.466 0.463

123
Adsorption (2015) 21:419–432 429

Table 3 continued
CH4
293 K
P (kPa) 22.4 34.4 46.8 59.3 71.9 84.7
D/R2 6.10 9 10-2 5.71 9 10-2 5.29 9 10-2 5.64 9 10-2 5.87 9 10-2 5.13 9 10-2
-1
ha/qsCs (s ) 0.0622 0.0660 0.0696 0.0748 0.0792 0.0823
b (–) 0.140 0.268 0.364 0.524 0.606 0.753
308 K
P (kPa) 25.6 38.4 51.3 64.6 77.8
D/R2 7.74 9 10-2 7.16 9 10-2 6.07 9 10-2 5.90 9 10-2 6.13 9 10-2
-1
ha/qsCs (s ) 0.0659 0.0633 0.0708 0.0799 0.0809
b (–) 0.119 0.182 0.319 0.515 0.522
323 K
P (kPa) 28.4 42.3 56.1 69.7 83
D/R2 8.76 9 10-2 8.19 9 10-2 7.29 9 10-2 5.59 9 10-2 6.05 9 10-2
-1
ha/qsCs (s ) 0.0597 0.0583 0.0609 0.0656 0.0626
b (–) 0.0802 0.146 0.186 0.429 0.391

Fig. 5 Effective diffusion time constants (D/R2) at 293 K obtained by non-isothermal diffusion model: a zeolite LiX and b activated carbon
(circle CO2; up-pointing triangle CO; square N2; down-pointing triangle CH4)

at lower pressures, the uptake curves showed a parallel loading amount, while those of CO and N2 were almost
decrease along the predicted lines by the heat controlled constant. This implies that the initial uptake was strongly
limit cases after passing certain period of time shown in affected by the heat generated from adsorption.
Fig. 7b. CH4 showed an uptake curve similar to that of N2. Figure 8 represents the effect of temperature on the
As shown in Fig. 4b, the heats of adsorption of CO2 and uptake curves of CO2 as a representative case. The results
CO in zeolite LiX were relatively high and decreased with at other pressure conditions are shown in Tables 2 and 3.
increasing loading amount, while those of N2 and CH4 The adsorption rate increased with temperature due to
were almost constant. faster molecular diffusion with higher kinetic energy. The
In the activated carbon systems, the uptake curves of temperature dependence of the amount adsorbed was
CO2 and CH4 had the same trend as those of CO2 and CO highest for N2 on zeolite LiX and CH4 on activated carbon
on zeolite LiX, while CO and N2 showed similar behavior (Park et al. 2014). The isotherm results according to tem-
to N2 and CH4 on zeolite LiX during the initial period. As perature were in agreement with the dependence of the
shown in Fig. 4b, the heats of adsorption of CO2 and CH4 adsorption rates because lower surface coverage at higher
were relatively high and decreased with an increasing temperature could result in an increased adsorption rate.

123
430 Adsorption (2015) 21:419–432

Fig. 6 Comparison of diffusion time constants for adsorption of a CO2, CO, CH4 and b N2 in two different size of zeolite LiX pellet at 293 K:
(solid symbol for 0.8 mm; blank symbol for 1.5 mm; circle CO2; up-pointing triangle CO; square N2; down-pointing triangle CH4)

Fig. 7 Effect of pressure on uptake curves of a CO2/zeolite LiX, b N2/zeolite LiX, c CO2/activated carbon, and d CO/activated carbon at 293 K.
(short dashed line non-isothermal kinetic model; solid line heat transfer control limit)

123
Adsorption (2015) 21:419–432 431

Fig. 8 Effect of temperature on uptake curves of CO2: a zeolite LiX and b activated carbon (circle T = 293 K; up-pointing triangle T = 308 K;
square T = 323 K; solid symbol for about 20 kPa; blank symbol for about 80 kPa; short dashed line non-isothermal diffusion model)

The level of temperature dependence depended on affected the adsorption rate if the heat transfer rates were
adsorbate/adsorbent system as well as applied pressure. not sufficiently high to quickly release the generated heat.
Generally, the temperature effect on adsorption rate is In addition, the initial uptake was strongly affected by the
stronger in a system with higher temperature dependence heat generated from adsorption.
on equilibrium capacity. As shown in Fig. 8, the shapes of The slope change in the isotherm and the isosteric heat
uptake curves of CO2 at similar pressure were similar variation with pressure produced an off-set effect on
regardless of temperature and adsorbent. However, the adsorption rate. The contribution of micropore diffusion on
effects of temperature on effective diffusion constant on the zeolite LiX system seemed to be stronger than that of
zeolite LiX were similar regardless of pressure, while those macropore diffusion. The effect of pressure on adsorption
on activated carbon were higher at higher pressure than at rate was significant at low pressure and became weaker
lower pressure. with increasing pressure because of the decreased slope of
the adsorption isotherm, decreased isosteric heat of
adsorption and high surface coverage with pressure. The
5 Conclusion dependence of adsorption rate on temperature was in
agreement with the isotherm results, showing that the
The adsorption uptakes for CO2, CO, N2 and CH4 on adsorption rate increased with temperature.
zeolite LiX and activated carbon were measured at 293, Activated carbon granules and zeolite LiX pellets are
308 and 323 K up to 100 kPa. The effective diffusion time widely used in adsorptive cyclic processes with respect to
constants (D/R2) of each component on each adsorbent equilibrium separation for simple gas mixture. Even though
were obtained by fitting a non-isothermal kinetic model adsorption equilibrium and selectivity work as a key factor
with experimental uptake curves. in the separator design, the order of magnitude in the value
At the same pressure and temperature, the order of adsorp- of adsorption rate also contributes to improving the design
tion rates on zeolite LiX was CO2  CO \ N2 \ CH4, but accuracy in adsorptive separation processes. Therefore, the
the isosteric heats of adsorption showed the opposite order adsorption kinetic results obtained from the volumetric
(CH4 \ N2 \ CO  CO2). The order of adsorption rates on method in the study contribute to improving the design
activated carbon was arranged as CO2  CH4 B N2 \ CO, accuracy of various adsorptive processes using zeolite LiX
but the sequence of adsorption rates was not perfectly mat- and activated carbon for various effluent gases with CO2,
ched with the order of isosteric heats of adsorption CO, N2 and CH4.
(N2 B CO \ CH4 \ CO2).
The adsorption rates of the gases on zeolite LiX and Acknowledgments We would like to acknowledge the financial
support from the R&D Convergence Program of MSIP (Ministry of
activated carbon were affected by the isosteric heats of Science, ICT and Future Planning) and NST (National Research
adsorption, heat transfer rate and adsorption affinity. Even Council of Science & Technology) of Republic of Korea (CRC-14-1-
though the heat of adsorption was small, it significantly KRICT).

123
432 Adsorption (2015) 21:419–432

References volume/variable-pressure conditions. J. Chem. Soc. Faraday

Trans. 1 80(8), 2167–2174 (1984)
Ahn, H., Moon, J.-H., Hyun, S.-H., Lee, C.-H.: Diffusion mechanism Lee, H.-H., Lee, J.-C., Joo, Y.-J., Oh, M., Lee, C.-H.: Dynamic
of carbon dioxide in zeolite 4A and CaX pellets. Adsorption modeling of Shell entrained flow gasifier in an integrated
gasification combined cycle process. Appl. Energy 131, 425–440
10(2), 111–128 (2004)
Ahn, H., Yang, J., Lee, C.-H.: Effects of feed composition of coke (2014a)
oven gas on a layered bed H2 PSA process. Adsorption 7(4), Lee, J.C., Lee, H.H., Joo, Y.J., Lee, C.H., Oh, M.: Process simulation
339–356 (2001) and thermodynamic analysis of an IGCC (integrated gasification
combined cycle) plant with an entrained coal gasifier. Energy 64,
Ahn, H., Yoo, H.-K., Shul, Y., Hyun, S., Lee, C.-H.: Diffusion
mechanism of N2 and CH4 in pelletized zeolite 4A, 5A and CaX. 58–68 (2014b)
J. Chem. Eng. Jpn. 35(4), 334–345 (2002) Lee, J.J., Kim, M.K., Lee, D.G., Ahn, H., Kim, M.J., Lee, C.H.: Heat-
Ahn, S., You, Y.-W., Lee, D.-G., Kim, K.-H., Oh, M., Lee, C.-H.: exchange pressure swing adsorption process for hydrogen
Layered two-and four-bed PSA processes for H2 recovery from separation. AIChE J. 54(8), 2054–2064 (2008)
coal gas. Chem. Eng. Sci. 68(1), 413–423 (2012) Lee, L.-K., Ruthven, D.M.: Analysis of thermal effects in adsorption
Bae, Y.-S., Lee, C.-H.: Sorption kinetics of eight gases on a carbon rate measurements. J. Chem. Soc. Faraday Trans. 1(75),
molecular sieve at elevated pressure. Carbon 43(1), 95–107 2406–2422 (1979)
(2005) Linstrom, P., Mallard, W.: NIST Chemistry Webbook. NIST Standard
Cavenati, S., Grande, C.A., Rodrigues, A.E.: Separation of CH4/CO2/ Reference Database No. 69 (2001)
N2 mixtures by layered pressure swing adsorption for upgrade of Park, Y., Moon, D.-K., Kim, Y.-H., Ahn, H., Lee, C.-H.: Adsorption
isotherms of CO2, CO, N2, CH4, Ar and H2 on activated carbon and
natural gas. Chem. Eng. Sci. 61(12), 3893–3906 (2006)
Crank, J.: The Mathematics of Diffusion. Oxford University Press, zeolite LiX up to 1.0 MPa. Adsorption 20(4), 631–647 (2014)
London (1956) Ruthven, D.M., Lee, L.K., Yucel, H.: Kinetics of non-isothermal
Do, D.D.: Adsorption Analysis: Equilibria and Kinetics, vol. 2. sorption in molecular sieve crystals. AIChE J. 26(1), 16–23
Imperial College Press, London (1998) (1980)
Giesy, T.J., LeVan, M.D.: Mass transfer rates of oxygen, nitrogen, Silva, J.A., Schumann, K., Rodrigues, A.E.: Sorption and kinetics of
and argon in carbon molecular sieves determined by pressure- CO2 and CH4 in binderless beads of 13X zeolite. Microporous
swing frequency response. Chem. Eng. Sci. 90, 250–257 (2013) Mesoporous Mater. 158, 219–228 (2012)
Grande, C.A., Rodrigues, A.E.: Layered vacuum pressure-swing Siriwardane, R.V., Shen, M.-S., Fisher, E.P., Poston, J.A.: Adsorption
adsorption for biogas upgrading. Ind. Eng. Chem. Res. 46(23), of CO2 on molecular sieves and activated carbon. Energy Fuels
7844–7848 (2007) 15(2), 279–284 (2001)
Yang, J., Lee, C.H.: Adsorption dynamics of a layered bed PSA for
Gray, P.G., Do, D.D.: Dynamics of carbon dioxide sorption on
activated-carbon particles. AIChE J. 37(7), 1027–1034 (1991) H2 recovery from coke oven gas. AIChE J. 44(6), 1325–1334
Himeno, S., Komatsu, T., Fujita, S.: High-pressure adsorption (1998)
equilibria of methane and carbon dioxide on several activated Yang, Y., Ribeiro, A.M., Li, P., Yu, J.-G., Rodrigues, A.E.:
Adsorption equilibrium and kinetics of methane and nitrogen
carbons. J. Chem. Eng. Data 50(2), 369–376 (2005)
Hu, X., Mangano, E., Friedrich, D., Ahn, H., Brandani, S.: Diffusion on carbon molecular sieve. Ind. Eng. Chem. Res. 53(43),
mechanism of CO2 in 13X zeolite beads. Adsorption 20(1), 16840–16850 (2014)
121–135 (2014) You, Y.-W., Lee, D.-G., Yoon, K.-Y., Moon, D.-K., Kim, S.M., Lee,
Kočiřı́k, M., Struve, P., Bülow, M.: Analytical solution of simulta- C.-H.: H2 PSA purifier for CO removal from hydrogen mixtures.
neous mass and heat transfer in zeolite crystals under constant- Int. J. Hydrogen Energy 37(23), 18175–18186 (2012)

123