Investigation of CO 2 Adsorption Performance and Fluidization Behavior of

Mesoporous Silica Supported Polyethylenimine

Dang Viet Quang, Mustapha Soukri, Jak Tantana, Pradeep Kumar Sharma,
Thomas O. Nelson, Marty Lail, Luke J.I. Coleman, M.R.M. Abu Zahra

PII: S0032-5910(16)30365-5
DOI: doi: 10.1016/j.powtec.2016.06.027
Reference: PTEC 11735

To appear in: Powder Technology

Received date: 11 January 2016

Revised date: 19 April 2016
Accepted date: 14 June 2016

Please cite this article as: Dang Viet Quang, Mustapha Soukri, Jak Tantana,
Pradeep Kumar Sharma, Thomas O. Nelson, Marty Lail, Luke J.I. Coleman, M.R.M.
Abu Zahra, Investigation of CO2 Adsorption Performance and Fluidization Behav-
ior of Mesoporous Silica Supported Polyethylenimine, Powder Technology (2016), doi:
10.1016/j.powtec.2016.06.027

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.
 ACCEPTED MANUSCRIPT

Investigation of CO 2 Adsorption Performance and Fluidization Behavior of

Mesoporous Silica Supported Polyethylenimine

Dang Viet Quanga, Mustapha Soukrib, Jak Tantanab, Pradeep Kumar Sharmab Thomas O.

PT
Nelsonb, Marty Lailb, Luke J.I. Colemanb, and M.R.M. Abu Zahraa,*
a
The Institute Center for Energy (iEnergy), Masdar Institute of Science and Technology, P.O.Box

RI
54224, Masdar city, Abu Dhabi, UAE
b
Energy Technology Division, RTI International, Research Triangle Park, NC 27709-2194, USA

SC
*Tel.: +97128109181
*Email:[email protected]

NU
Abstract MA
A thorough and comprehensive study of different parameters affecting the sorbents

prepared by impregnating polyethyleneimine (PEI) on silica for CO2 capture has been
D

conducted. The CO2 capture performances of resulting sorbents were evaluated on both
TE

packed-bed and fluidized-bed reactors. Obtained results indicated that CO2 adsorption-
P

desorption performances of sorbents are greatly affected by the operation temperature in both
CE

simulated flue gas from coal fired power plant (SCF) and natural gas combined cycle power

plant (NGCC). The optimal adsorption temperatures are from 50 to 80oC for NGCC flue gas
AC

and from 70 to 90 oC for SCF flue gas, while the optimal regeneration temperatures are from

110 to 130oC for both flue gas. Adsorbent containing 30 to 35 wt% PEI fluidized well with

both simulated flue gas and steam. Results from this study revealed that PEI impregnated

silica is a promising sorbents for CO2 capture process, using a fluidized-bed reactor, from

both coal-fired power and natural gas combined cycle power plants.

Key words: CO2 capture; Adsorption; polyethyleneimine; CO2 loading; Packed-bed

reactor; Fluidization.

1
 ACCEPTED MANUSCRIPT

1. Introduction

CO2 capture technologies have undergone significant development in the last several years,

and today large-scale Carbon Capture & Sequestration (CCS) demonstrations are underway.

PT
Several concepts for capturing CO2 in industrial processes and power plants are being developed.

RI
Among these, amine scrubbing based on CO2 absorption using ethanolamine (MEA) solutions

[1, 2] is considered as a current state-of-the-art industry process. However, the liquid phase

SC
amine scrubbing technology suffers from high regeneration cost, high degradation, and

NU
equipment corrosion [3-5]. As a result, there has been a strong interest to develop alternative

more stable and less energy-intensive CO2 removal technologies. Adsorption using solid sorbents
MA
[6-9], particularly amine-functionalized sorbents, is appealing due to their potentially lower

energy requirements for sorbent regeneration [1, 2, 10-22]. By immobilizing amine

D
TE

functionalities on high-surface porous solid supports, the amino solid sorbents offer higher amine

efficiency and thus superior sorbent performance [18, 19, 23, 24]. Among various solid sorbent
P

supports including zeolites [25] activated carbons [26-29], activated alumina [30-32], and
CE

membranes [33], polyethyleneimine (PEI) impregnation on mesoporous silicas have been

AC

attracting great attention [6-9, 34-42], because of its simple preparation procedure which can be

easily translated to commercial scale as well as the potential to reduce the overall costs related to

CO2 capture by reducing the energy required to release the CO2 during regeneration [10].

Previous studies indicated that the heat of CO2 adsorption reduced considerably by

immobilizing amine on solid substrate. For example, the heat of CO2 adsorption of sorbent

prepared by immobilizing PEI on CARiACT G10 silica is around 50.0 kJ/mole [43]. 3-

Aminopropyltriethoxysilane functionalized mesoporous silica generates the heat of CO2

adsorption about 60.0 kJ/mole [44]. Meanwhile, the heat of CO2 absorption of a conventional

2
 ACCEPTED MANUSCRIPT

aqueous MEA solution is much higher; it is about 85 kJ/mole [10]. Generally, the heat of CO2

adsorption is reduced due to the impregnation of amine solvents on a solid support. Because of

low heat capacity, solid sorbent also requires less the sensible heat to raise temperature of

PT
adsorption bed to the regeneration temperature, however, the most significance of solid sorbent

RI
is the avoidance of solvent vaporization which could account for 30–50% of total regeneration

energy [10, 22, 42].

SC
Nevertheless, several factors may influence the amine-containing sorbents performance,

NU
in particular, adsorption capacity, kinetics, and the stability of the sorbents under different

operating conditions such as adsorption and desorption temperature, pressure and number of
MA
cycles as well as their interactions with different species that may occur in industrial gases.

Typical species include not only impurities such as water vapor, SO2, H2S, NOx and O2 [45-52],
D
TE

but also CO2 itself, which deactivates amine-containing materials under dry conditions [34, 51,

53].
P

The aim of the current study is to investigate the CO2 adsorption performance and
CE

fluidization behavior of mesoporous silica-supported PEI under different operating conditions.

AC

Different types of silicas with various physical properties has been studied to evaluate their effect

on the CO2 capacity of the sorbent. In addition, different solvents and impregnating conditions

were also thoroughly evaluated. The CO2 adsorption performances of the resulting sorbents have

been investigated in a packed-bed reactor (PBR) and their fluidization has been visually studied

in fluidized-bed reactor (FBR). Finally, the influence of impurities on CO2 adsorption

performance during the long-term operation has been performed including humidity, oxygen and

SO2.

3
 ACCEPTED MANUSCRIPT

2. Experimental

2.1. Chemical reagents

Silicagels with different structural properties including 8 samples of irregular shape

PT
particles having different particles sizes and porosity, namely Silica from I to VIII and 1

RI
spherical bead sample, namely Silica-IX were purchased from Sigma Aldrich. The irregular

shape silica samples have surface area ranging from 155 to 353 m2/g, pore volume ranging from

SC
1.1 to 1.2 cm3/g, and pore size ranging from 113 to 320 Å. All the silica substrates, used as

NU
support in this study, were dried at 130oC for 2 hours prior to impregnation with PEI. Branched

polyethyleneimine (PEI, 99 % and average Mw ≈ 600) was obtained from Alfa Easer
MA
(Massachusetts, USA). Methanol (MeOH, 99.9%), ethanol (EtOH, 99.9%), and isopropanol

(iPrOH, 99.9%) were provided by Fisher Scientific (Massachusetts, USA). Deionized (DI) water
D
TE

was collected from a centralized water deionizing and filtering system at RTI International.

2.2. Sorbent preparation

P

Typical adsorbent preparation procedure involved a wet modification method [35, 54]. First,
CE

a desired amount of PEI was dissolved in a solvent, which is MeOH, EtOH, iPrOH, or water, in a
AC

1-L flask before adding a desired amount of silica under vigorous mixing in order to produce

homogeneous slurry. The solvent was evaporated using a rotary evaporator (IKA RV 10

Rotovapor, USA). The resulting PEI-impregnated silica was dried in vacuum oven at 80oC over

night before used for testing.

2.3. Sorbent characterization

Characterization of the prepared adsorbents involved the methods of surface area

measurement, pore volume measurement, pore size measurement, nitrogen content analysis, and

Thermal Gravimetric Analysis (TGA). The surface area, pore volume, and pore sizes were

4
 ACCEPTED MANUSCRIPT

measured by a nitrogen adsorption-desorption method using a Micromeritics Surface and

Porosity Analyzer (ASAP 2020). First, sample was transferred into an analyzing tube with small

bulb, whose mass was previously determined using analytical balance and degassed at 110oC for

PT
10 h. The mass of analyzing tube containing sample was determined again after degassing and

RI
the tube was mounted on Micromeritics Surface and Porosity Analyzer for analyzing. The mass

of sample was determined by the mass difference between the tube with sample after degassing

SC
and empty tube. The pore volume and pore size were calculated by the Barrett-Joyner-Halenda

NU
(BJH) method using the desorption data. Nitrogen content within the sorbents was analyzed on a

Thermo Scientific CNS elemental analyzer (Flash 2000). TGA was conducted on a
MA
thermogravimetric analyzer (SDT Q600), which involved heating sorbent samples from room

temperature to 600oC in nitrogen gas at a rate of 10oC/min.

D
TE

2.4. CO2 adsorption-desorption on packed-bed reactor

The CO2 adsorption – desorption performance of the adsorbent was investigated using
P

packed-bed reactor system (PBR) designed and constructed by RTI International (RTI, North
CE

Carolina, USA) as shown in Figure 1. The composition of the feed gas was adjusted by changing
AC

the flow rate of individual gases controlled by mass flow controllers. Water vapor was

introduced to the gas stream by flowing the mixture of air and N2 to the temperature controlled

humidifier (A). The temperature of the humidifier can be adjusted to vary the moisture content in

the simulated flue gas stream.

The CO2 stream combined with the N2/Air/H2O mixed gas at the outlet of the humidifier to

avoid the CO2 from dissolving in the water in humidifier, which would result in the increase

acidity of the water and inaccurate CO2 content in the simulated flue gas stream. The effluent of

the PBR (B) entered a condenser (C) and water collector (D) to remove water vapor prior to

5
 ACCEPTED MANUSCRIPT

entering the CO2 analyzer. The CO2 concentration in the PBR effluent was analyzed using a

Horiba CO2 analyzer (VA-3000). PBR operated autonomously by National Instruments Lookout

program coupled with customized PLC control developed by RTI’s engineers. The software

PT
allows the PBR to be controlled according to preset test conditions (defined by operator) and to

RI
collect data points every second. The data including operating temperatures, the flow rate of

input gases and CO2 concentration in the output gas are recorded. The PBR is capable of

SC
performing multiple adsorptions-regeneration test cycles with varying test conditions without the

NU
need for operator attendance. In a typical experiment, approximately 2 g of adsorbent was mixed

with silicon carbide beads, a filler and heat transfer material, then loaded into the PBR, which
MA
was made of stainless steel column with 1.27 cm inner diameter and 20 cm length.
D

A typical test undergoes 5 stages including 1) feed test, 2) purge, 3) adsorption, 4)

TE

regeneration, and 5) cooling. The feed test stage allows system to adjust the temperature of the

process lines, humidifier, and reactor to defined values for the adsorption stage and to analyze
P
CE

and adjust the feed gas composition at the same flow rate being used in the adsorption stage. The

feed gas can be tailored to various simulated flue gas conditions, in these tests in particular, flue
AC

gas from coal fired power plant (SCF) or natural gas combined cycle power plant (NGCC). The

SCF flue gas has following composition: CO2 = 15 vol%, O2 = 4.5 vol%, and water vapor = 5.65

vol% in balance with N2. The NGCC flue gas has following composition: CO2 = 3 vol%, O2 = 14

vol%, and water vapor = 5.65 vol% in balance with N2. When the CO2 concentration in the feed

gas is stable, the system will enter the purge stage by flowing N2 through the system until no

CO2 is detected in the outlet gas of the reactor. The adsorption stage is initiated by switching a

valve to allow CO2-laden feed gas to enter the reactor with a flow rate of 150 mL/min. The end

of the adsorption process was determined by the breakthrough point for CO2 which occurs when

6
 ACCEPTED MANUSCRIPT

CO2 concentration in outlet gas reached 99.95% of that in the feed gas. The end of the adsorption

stage may also be determined by the minimum and maximum adsorption times, which was

defined as 30 and 45 min, respectively, for the SCF flue gas and 60 and 90 min for NGCC flue

PT
gas. The regeneration stage follows the adsorption and is carried out by flowing humidified N2

RI
through the reactor at the flow rate of 150 mL/min at a desired temperature. The regeneration

end point was defined by the regeneration off-gas (outlet gas) having concentration of CO2 of

SC
less than 0.1%. Otherwise, it was determined by the minimum or maximum desorption times of

NU
30 and 45 min, respectively. The cooling of the system is performed by flowing N2 at the flow

rate of 350 mL/min to reduce reactor temperature to either the temperature of the adsorption
MA
stage or to room temperature (if the tests are complete). The CO2 loading (wt%) was calculated

from regeneration data by dividing the total amount of CO2 desorbed by the mass of the
D
TE

adsorbent and multiplying with 100. The CO2 loading is defined as the amount of CO2, which

desorbs at a specific regeneration condition and does not include CO2 lean loading. Each test
P

consisted of at least 3 adsorption-regeneration cycles and the CO2 loading that is reported was
CE

taken as the average between all cycles.

AC

2.5. Fluidization and adsorption-desorption studies using the prepared sorbents

and a fluidized bed reactor

The fluidization and adsorption-desorption performance of the prepared adsorbents were studied

in a fluidized-bed reactor system (FBR) designed by RTI. A schematic illustration of the FBR is

shown in Figure 2.

The FBR allows for adsorption-desorption cycling tests to be conducted at ambient pressure

and at temperatures ranging from room temperature to 140oC. The composition of the feed gas

was controlled by adjusting the flow rate of two mass flow controllers, MFC5 for N2 flow and

7
 ACCEPTED MANUSCRIPT

MFC6 for CO2 flow, and the feed rate of a water pump. A micro pump was used to supply a

precise amount of water into steam generator (A), where the temperature was held at

approximately 300oC to ensure that all supplied water was converted to steam. The steam was

PT
combined with a mixture of N2/CO2 gases from MFC to form the simulated flue gases used in

RI
the adsorption tests. The simulated flue gases then entered a glass fluidization column (B), 80 cm

in height and 2.54 cm in diameter. Noted that the steam was used solely as the CO2-stripping

SC
agent in fluidization column. The temperature of the column (B) is controlled by heating oil from

NU
circulation bath. The gas exiting column (B) flowed through a cyclone (C), condenser (D), and

water collector (E) to remove fine particles and water vapor before entering a CO2 analyzer and
MA
then being vent. In a typical experiment, a predefined amount of sorbent was loaded into the

column (B) in order to obtain a bed height of about 15 cm. The sorbent bed was first fluidized
D
TE

with N2 until fluidization stabilized and the desired setting temperature was reached. For each

adsorbent sample, the fluidization performance was tested by using 3 different gas compositions:
P

N2, simulated flue gas, and steam. The pressure drop across the column at various gas velocities
CE

was recorded to analyze the fluidization characteristics of the adsorbent.

AC

3. Results and discussion

3.1. Adsorbent characterization

The properties of the silica substrate (Silica-I) and its sorbents containing various PEI

concentrations are listed in Table 1. In this study, PEI impregnation was conducted in MeOH.

Adsorption was conducted at 65oC in the SCF flue gas and regeneration was conducted at 110 oC

in humidified N2 containing 6.65 vol% of water vapor. The porous properties of the silica support

decreased upon impregnation with PEI from zero to 45 wt% PEI loading. In fact, both the

surface area and pore volume of the sorbent decreased almost linearly as a function of increased

8
 ACCEPTED MANUSCRIPT

PEI concentration (Figure 3). However, the pore size decreases from 150 Å to about 110 Å

irrespective of the PEI concentration (Figure S1). In contrast to the porous properties, the bulk

density of the adsorbent increases along with the increase in the PEI concentration (Figure 4),

PT
since the empty volume inside silica structure is being filled by the PEI molecules.

RI
The elemental analyses (Table 1) revealed that mesoporous silica impregnated with 30–

45 wt% PEI contains 8.77–13.16 wt% of organic bound nitrogen, respectively. These results

SC
confirm that the PEI was successfully impregnated onto the silica substrate by the wet

NU
impregnation method.

The thermal stability of the solid sorbent under a range of operating temperatures
MA
(between 50 to 130oC for adsorption and regeneration, respectively) are of critical importance.

Thus, TGA analyses were performed on representative samples and the results are shown in
D
TE

Figure 5.

The TGA analyses confirmed that silica substrate is strongly resistant to thermal
P

treatment as it lost only 1.18 % of its weight up to about 100oC (caused by moisture evaporation)
CE

and then remained stable when temperature increased to 600oC. However, the TGA profile of
AC

these two different adsorbents differs significantly from the silica substrate. The initial weight

loss at 100oC is attributed to the evaporation of absorbed water, and significant weight reduction

was observed around 250–425oC with an overall weight loss of 29.1 and 36 wt% for 30 and 37.5

wt% PEI impregnation, respectively, which is likely due to the decomposition of the PEI. These

results clearly indicate that these sorbents are suitable for the desired CO2 capture process, in

which the operating temperatures are lower than 150oC.

3.2. Effect of PEI concentration on CO2 loading performance

9
 ACCEPTED MANUSCRIPT

The CO2 adsorption on silica supported PEI is mainly based on the reaction of CO2 with

amino groups. Since the branched PEI contains primary, secondary, and tertiary amino groups,

there are several mechanisms that account for the CO2 adsorption on the adsorbent [55]. First, a

PT
chemical adsorption occurs through reactions between CO2 molecules and primary and

RI
secondary amino groups to form carbamate species (Eq. 1) or possible carbonate species in the

presence of water (Eq. 2). Second, chemical adsorption takes place via a reaction between CO2

SC
molecules and tertiary amino groups in the presence of water to form bicarbonate products (Eq.

NU
3) [56]. These are reversible reactions; the carbamate and carbonate species can be decomposed

at high temperature to regenerate amine and free the CO2 gas. In general, the chemical
MA
adsorption usually occurs with a relatively high rate at the initial stage, about 15-30 min and then

slows down due to less availability of active and exposed amino groups as well as slow
D
TE

dispersion rate of CO2 into the adsorbent structure. Therefore, in the later adsorption stage, in

addition to the chemical adsorption, the physical adsorption of CO2 onto the adsorbent should be
P

considered as well [35, 57]. PEI-based adsorbent contains numerous polar sites on the surface
CE

and pore wall. Those polar sites can attract CO2 molecules by forming a weak Van der Waal
AC

interaction to induce a physical adsorption. The physical adsorption is particularly enforced at

low temperature and high pressure. In this study, the SCF and NGCC flue gas have relatively

low CO2 concentration at ambient pressure; the CO2 physical adsorption could be negligible.

Thus, the CO2 adsorption capacity of the sorbent is mostly contributed by the chemical

adsorption.

CO2 + 2R2NH ↔ R2NH2+ + R2NCOO- (1)

CO2 + H2O + R2NH ↔ R2NH2 +·HCO3- (2)

CO2 + H2O + R3N ↔ R3NH+·HCO3- (3)

10
 ACCEPTED MANUSCRIPT

Theoretically, the CO2 adsorption capacity should be linearly increased as a function of

the amino group density (PEI concentration in the sorbent). According to the Reaction (1), (2),

and (3), higher PEI concentration should result in an increase in the CO2 adsorption capacity due

PT
to their right shift. However, the experimental results showed a different scenario as seen in

RI
Figure 6.

The CO2 loading capacity is initially increased with higher PEI concentration (from 30 to

SC
40 wt%), then appears to stabilize at a maximum CO2 loading with even higher PEI

NU
concentration (up to 45 wt%). According to previous studies [35, 58], the CO2 adsorption

capacity of silica-supported sorbents increases with increasing PEI concentration, but only to a
MA
specific point and then a decrease in performance actually occurs with even higher PEI loading.

In the light of this finding, we examined the porosity of sorbents and as already highlighted in
D
TE

Table 1, a significant reduction of the adsorbent pore volume and pore size suggest that the

pores were filled with the PEI. At higher PEI concentrations, as the polymeric amine packs into
P

the substrate pores, it is likely that numerous pores become full. Thus, the PEI molecules packed
CE

inside the pores likely become condense and display as solid particles, in which, the majority of
AC

the amino groups allocated inside particles [35, 58]. Hence, they become inaccessible and do not

take part in the adsorption of CO2. This assumption is asserted by experimentally molar ratio of

CO2/N shown in Table 1. The molar ratio of CO2/N reduced as the N content increased by the

higher PEI impregnated concentration.

3.3. Effect of solvent on CO2 loading

According to a literature survey, PEI impregnation is usually conducted using methanol

(MeOH) as a solvent to produce PEI/silica slurry. MeOH is frequently used because it is miscible

with PEI, easy to evaporate due to low boiling point, and relatively inexpensive. However, it is a

11
 ACCEPTED MANUSCRIPT

toxic solvent with relatively high vapor pressure and may not make an ideal choice as a

commercial solvent. Therefore, we investigated other solvents including ethanol (EtOH),

isopropanol (iPrOH), and water (H2O) as possible replacements for MeOH in the sorbent

PT
preparation. The effect that different solvent have on the CO2 loading was examined and the

RI
results are presented in Figure 7.

The prepared adsorbent with the lowest CO2 loading (9.1 wt%) was actually the one

SC
prepared using MeOH. When MeOH was replaced by other solvents, the CO2 loading were

NU
consistently improved, with the EtOH-prepared sorbent having the highest CO2 loading (10.9

wt%). By using EtOH as a replacement to MeOH, CO2 loading was enhanced 19.3 % and EtOH
MA
does not have the same toxicity and environmental concern as MeOH. It is not clear for us the

reason for this improvement, however, it is likely due to the different conformation of PEI inside
D
TE

silica structure causing by using different solvents. Because CO2 loading of the sorbent was

improved by replacing MeOH with EtOH, the EtOH solvent will be used to prepare sorbent for
P

the following experiments.

CE

3.4. Effect of adsorption and regeneration temperature on CO2 loading

AC

performance

Several studies indicated that the CO2 adsorption rate of amine supported mesoporous silica

is regulated by chemical reaction between CO2 and amino groups in the sorbent and undergoes

two typical stages, before and after breakthrough time [18, 59, 60]. The first stage is controlled

by reactions among CO2 molecules and the ready-accessible active sites (amino groups). In the

second stage, the adsorption rate is dominated by the diffusion rate of CO2 to active sites, which

are located inside the condense structure of PEI in the sorbent. The first stage occurs in few

12
 ACCEPTED MANUSCRIPT

minutes, while the second stage continues until it reaches the thermodynamic equilibrium, which

may take several hours or days.

For this reason, the CO2 adsorption capacity measured at the equilibrium is usually much

PT
higher than the CO2 loading at the first 10-30 min. Because the measurement of the CO2

RI
adsorption capacity consumes more time, the CO2 loading reported in this study involves the first

stage of the adsorption only. In practical application, the CO2 loading at the first stage is more

SC
significant due to its high adsorption rate. Therefore, the adsorption parameters at the

NU
breakthrough (90% CO2 removal) such as the breakthrough time (BT) and CO2 loading at
MA
breakthrough become very important to evaluate the performance and effectiveness of the

sorbent. Longer breakthrough time and higher CO2 loading at the breakthrough allow to lengthen
D

sorbent retention in the reactor, which helps slow down the sorbent circulation rate. The slow
TE

sorbent circulation rate will help decrease the energy requirement for heating the sorbent bed and

this results in the reduction in energy consumption.

P
CE

The effect of adsorption temperature on the CO2 loading of the sorbent is presented in

Figure 8. Tests conducted on sorbents (Silica-I) containing 37.5 wt% PEI impregnated using
AC

EtOH as solvent. The sorbent was regenerated at 110oC in humidified N2. As the adsorption

temperature elevated from 30 to 50oC, the CO2 loading increased and reached to a maximum at

50oC for both simulated flue gases. The maximum loadings recorded were 13.6 and 9.6 wt% for

SCF and NGCC flue gases, respectively. Once we increased the adsorption temperature above

50oC, the CO2 loading dropped for both cases (SCF flue gas and NGCC flue gas). Even though,

the sorbent loses its adsorption capacity at higher temperature, CO2 loading on the sorbent is

effectively performed at large temperature range (from 30 to 100oC). However, to find the

optimum temperature for the process to enhance CO2 removal efficiency and energy saving, the

13
 ACCEPTED MANUSCRIPT

breakthrough time and CO2 loading at the breakthrough were considered. The optimum

temperature should maximize both breakthrough time and CO2 loading at breakthrough. As seen

in Figure 8, both breakthrough time and CO2 loading at the breakthrough reached maximum

PT
values at certain temperature range and the result suggested that the optimum adsorption

temperatures are 70–90 oC for SCF flue gas and 50–80 oC for NGCC flue gas, respectively.

RI
The reduction in CO2 loading, as the adsorption temperature increased, is expectable due

SC
to the nature of the reactions between amine and CO2. These reversible and exothermic reactions

NU
(Reaction 1, 2, and 3), will shift to the left at a higher temperature, which is more favorable to

desorption than the adsorption and as the results, CO2 loading decreases with the increase in the
MA
adsorption temperature. This explanation, however, only satisfied the CO2 adsorption mechanism

at above 50oC. Below 50oC, the CO2 loading increased with the increase in the adsorption
D
TE

temperature. Apparently, there is another element that dominates the adsorption at low

temperature range. As discussed earlier, the way PEI is packed at low temperature (amino groups
P

are partially confined in particles), causes less accessibility to CO2 gas. As temperature increases,
CE

the particle-like PEI becomes more flexible and expendable, allowing more amino groups to
AC

expose to CO2, which results in higher CO2 adsorption capacity.

The effect of regeneration temperature on the cyclic CO2 loading of the sorbent was also

studied. To investigate this effect, the temperature in the adsorption stage was held constant at

65oC, while temperature in regeneration stage was varied from 90–150oC for both SCF and

SNCC flue gases. The obtained results are shown in Figure 9. From this figure, it is clear that

the CO2 loading was significantly enhanced by the increase in the regeneration temperature in

both cases. In fact, between 90–130oC, the CO2 loading increases rapidly and almost linearly as a

function of regeneration temperature. The rate of CO2 loading increase while increasing the

14
 ACCEPTED MANUSCRIPT

regeneration temperature then slows down at higher temperature (above 130oC) for both flue

gases (SCF and NGCC). These results are in good agreement with those from previous study

where the CO2 desorption from loaded PEI/silica sorbent increased monotonically with

PT
temperature [61].

RI
Nevertheless, the lean CO2 loading of the sorbent (The lean loading is the amount of

absorbed CO2 per gram of sorbent that is not desorbed at regeneration temperature) should be

SC
taken into account. When the regeneration temperature is high enough (~150oC, Figure 9), the

NU
sorbent will be fully desorbed and the lean loading will reach zero. Therefore, the increase in

regeneration temperature will reduce the lean loading and results in higher cyclic CO2 loading.
MA
It is clear that the higher regeneration temperature helps enhance the CO2 loading,

however, the increase in temperature may cause many unexpected problems such as higher
D
TE

energy requirement, sorbent degradation (thermal and oxidative), and shortening the lifecycles of

the sorbent. For example, by increasing regeneration temperature from 130 to 150 oC while
P

keeping adsorption temperature at 80 oC, the CO2 loading improves only 5.2 % but the energy
CE

required for heating sorbent bed from 130 to 150 oC increases 40% for both SCF and NGCC flue
AC

gases. The increments of CO2 loading and sensible heat with raising temperature from 100 to 150
o
C shown in Figure 9 (inset) indicated that sensible heat increases linearly, while CO2 loading

increment slows down with an inflection point at 130 oC. An optimal regeneration temperature

should help maintain a reasonable CO2 loading, while saving energy and avoiding rapid

degradation. The result suggested that the optimal temperature for adsorbent regeneration lies in

the range of 110 to 130 oC.

3.5. Adsorption kinetics and CO2 removal efficiency

3.5.1. SCF flue gas conditions

15
 ACCEPTED MANUSCRIPT

The kinetics of CO2 adsorption onto the adsorbent (Silica-I) was studied for adsorption

under SCF flue gas conditions. The results of these studies are shown in Figure 10. As exhibited

in this figure, the CO2 adsorption was rapid during the first few minutes of adsorption with more

PT
than 99% of CO2 removal efficiency. After few minutes the adsorption rate was slowing down

RI
and CO2 removal efficiency was reduced sharply, as indicated by the increase of CO2

concentration in the effluent which quickly reached the CO2 level in the feed gas. In general, the

SC
CO2 breakthrough time was observed within the first 3 to 5 min, but was longer as the adsorption

NU
temperature was elevated.

The results presented in Figure 10 indicate that the accumulative CO2 loading, which is
MA
the percentage of CO2 loading at specific time in comparison with the loading at the end of test,

as well as the CO2 concentration in the effluent, increased dramatically at the first few minutes
D
TE

and slightly changed after the breakthrough. The CO2 concentration in the effluent mostly

reached its feed gas level, while, the accumulative CO2 loading continued to increase. As
P

discussed earlier, although the breakthrough was observed, the CO2 concentration in effluent was
CE

almost similar to that in influent, the adsorption is still occurring until equilibrium. This results in
AC

the continuous increase in accumulative CO2 loading.

In Figure 10, the CO2 loading percentage at the breakthrough is intercept between the

accumulative CO2 loading curve and a vertical dashed line which originates from the

breakthrough point. If considering the breakthrough (90% CO2 removal) as a criteria to evaluate

the sorbent performance, it was found that the CO2 loading percentage at the breakthrough was

improved significantly with the rise in the adsorption temperature (Table 2).

The CO2 loading at the breakthrough (Table 2) attained a maximum at an adsorption

temperature of 70oC (6.85 wt%). The variation in the CO2 loading at the breakthrough was

16
 ACCEPTED MANUSCRIPT

similar to the change in the breakthrough time. These results further confirmed that the

adsorption rate of the sorbent was enhanced by increasing adsorption temperature. However,

both the CO2 adsorption capacity and breakthrough time tend to decrease as the adsorption

PT
temperature is above 90oC.

RI
3.5.2. NGCC flue gas conditions
The results of kinetic studies of CO2 adsorption from NGCC flue gas at 65oC is shown in

SC
Figure 11. The results are similar to the kinetic observation in the SCF flue gas. CO2 adsorption

NU
is rapid at the beginning and the CO2 removal efficiency reached > 98% within the first 10 min.

The percentage of accumulative CO2 loading at the breakthrough, CO2 loading at the
MA
breakthrough, the breakthrough time, and total CO2 loading are shown in Table 3.

The breakthrough time as well as the CO2 loading at the breakthrough reached maximum
D

value at 50oC. The change in the CO2 loading at the breakthrough is analogous to that in the total
TE

CO2 loading, which increases to a maximum at 50oC (7.35 wt%) and reduced at higher
P

temperature. Though, the sorbent operating in the NGCC flue gas has a longer breakthrough time
CE

and higher CO2 loading at the breakthrough compared to the SCF flue gas case; the maximum
AC

CO2 loading at the breakthrough was 7.35 wt% in the NGCC flue gas compared to 6.85 wt% in

the SCF flue gas, the overall CO2 loading in the NGCC flue gas is lower than the loading in the

SCF flue gas.

As discussed above, the CO2 loading and adsorption kinetic are temperature-dependent, therefore

there is a need to determine an optimum operating temperature for the adsorption process. The

optimum temperature should be beneficial in regard to both the CO2 loading and breakthrough

time performance. It is evident from the experimental results presented (Figure 8) that the

prepared sorbent can operate at a wide temperature range. However, the optimum working

temperature varies with the CO2 concentration in the flue gas. The best operating temperature for

17
 ACCEPTED MANUSCRIPT

the NGCC flue gas ranges from 50 to 80 oC (Figure 8, dash line) which achieved a CO2 loading

at the breakthrough from 6.56 to about 7.35 wt% and the BT from 22.4 to 23.7 min. By

comparison, the optimum temperature range is higher for the SCF flue gas at about 70 to 90oC

PT
(Figure 8, dotted line) with a CO2 loading at the breakthrough from 6.74 to 6.85 wt% and the BT

RI
from 4.8 to 5.0 min, respectively. The broad and relatively high working temperature has a

significant effect on the CO2 capture process in term of energy saving. It can help reduce the

SC
cooling energy required during adsorption while also makes narrower the temperature gap

NU
between the adsorption and desorption process which will help decrease the sensible heat

requirement. These results indicates that the prepared sorbent is capable of effective CO2
MA
removal from both SCF and NGCC flue gases.

3.6. Effect of silica substrate on CO2 loading performance

D
TE

3.6.1. Effect of silica bead type

In this study, silica was used as substrates to impregnate PEI. Silica is a porous material,
P

which is produced by acidizing a solution of sodium silicate. This type of silica is formed by a
CE

variety of tiny silica particles (primary particles) that connects together through the condensation
AC

of Si-O-Si bonding. Pores in silica are irregularly shaped and formed by gaps and empty

channels within the network of silica primary particles. Two types of silica (silica spherical bead

and silica gel) are usually produced. The former can be used as desiccant without any further

processing as the beads are produced with desired diameters. The latter is produced as a gel and

must be broken into smaller size particles that are required by specific applications. These

smaller particles have irregular shapes in associated with a larger number of opening pores

compared to bead type. In this study, both types of silicagel were used, impregnated with 37.5

wt% PEI and test for CO2 adsorption performance to determine any effect incurred by the silica

18
 ACCEPTED MANUSCRIPT

type used. Adsorption and regeneration were conducted at 65oC in SCF flue gas and 110oC in

humidified N2, respectively. Results from this study are presented in Table 4 showing the

properties of 1 PEI impregnated silica spherical bead (Silica-IX) and 8 PEI impregnated silicagel

PT
samples (Silica from I to VIII).

RI
The sorbent prepared using the silica spherical bead results in a lower CO2 loading (10

wt%) compared to a silicagel (Silica-VII, 11.5 wt%) having similar pore size and particle size. In

SC
another experiment, the CO2 loading of Silica-IX containing 35 wt% PEI is 8.7 wt%, which is

NU
lower that of Silica-I containing the same PEI concentration prepared in the same MeOH solvent

(9.8 wt%). These results indicate that the irregular-shaped silicagel may be a superior substrate
MA
choice for CO2 adsorption application. The lower CO2 loading on the spherical silica sorbent

may be due to the lower surface area and pore volume of that particular substrate. Moreover, the
D
TE

difference in CO2 loading between two silica substrates is most likely due to the difference in

their surface properties, particularly the density and status of silanol groups (Si-OH) which affect
P

the dispersion of the PEI into the silica structure [62, 63]. In addition, the difference in the
CE

surface conformation of the silica spherical beads and those silicagel particles also influences the
AC

impregnation of the PEI. The silica spherical beads have a convex surface; the chemical potential

of atoms on such a surface is higher than that on a flat and concave surface. The silicagel

particles possess irregular shapes, which have mostly flat or rough surface with a lower chemical

potential compared to convex surface of spherical beads. Thus, the impregnation of the PEI on

the silicagel particles is more effective than that on the spherical beads.

3.6.2. Effect of the pore size and particle size

The porous properties of silica i.e. pore volume and pore size in particular, have

significant influence on the PEI impregnation and the CO2 adsorption performance of the

19
 ACCEPTED MANUSCRIPT

resulting sorbent. The above discussion indicated that the silicagel particles are more favorable to

the PEI impregnation than the spherical ones; thus silicagel particles were used to investigate the

effect of the pore size on the CO2 adsorption capacity. Most silicagels selected for this study had

PT
similar pore volume ranging from 1.1–1.2 cm3/g, while BET surface area and pore size vary

from 155–353 m2/g and 113–320 Å, respectively. The pore size distributions of silica

RI
impregnated with 37.5 wt% PEI using EtOH were presented in Figure S2 and 3. Table 4

SC
presents the result obtained from this study and shows no clear relation between the BET surface

NU
area and the CO2 loading of the sorbent. However, the pore size and particle size have a very

pronounced effect on the CO2 loading. Silica-VIII has the largest pore size among the samples
MA
investigated (320 Å) and the associated sorbent has the highest CO2 loading. In general, the

results show that the irregular-shaped silicagels with larger pore size result in sorbent with higher
D
TE

CO2 loading capacity.

To investigate the effect of particle size on an adsorbent’s CO2 adsorption, the silicagel
P

was categorized into three groups of closely related average pore sizes. The first group includes
CE

Silica-II and Silica-III with average pore sizes 113.2 and 122.9 Å, respectively. The second
AC

group consists of 4 samples with average pore sizes ranging from 150 to 155 Å (Silica-I, Silica-

IV, Silica-V, and Silica-VI). The third group has two samples with an average pore size 284 and

320 Å (Silica-VII and Silica-VIII). In the group 1 and 2, test results indicate a similar trend in the

influence of the particle size on the CO2 loading capacity, i.e. the CO2 loading was enhanced by

reduction in the silica particle size. However, this effect was not observed in group 3. Thus, the

variation in the silica particle size has a great effect on the CO2 adsorption performance of the

prepared sorbent. This effect is more apparent and pronounced in the silica with small pore

diameters.

20
 ACCEPTED MANUSCRIPT

Average pore size due to the PEI impregnation was reduced on all silica samples with

average pore size below 155 Å and Silica-VIII with larger average pore size (320 Å) but large

particle size (200-500 µm). The pore size reduction was, however, not observed on Silica-VII

PT
and IX, which have large pore size and small particle size (Table 4). It is not clear, but likely that

RI
the larger fraction of small pores in the small particles was filled with PEI in comparison with

large particles, which results in the shift of pore size distribution toward larger pore sizes.

SC
NU
3.7. Long-term operation and the effect of impurities on CO2 loading performance

In commercial application, sorbents should be durable and tolerate high temperature and
MA
chemical impurities in order to maintain desirable CO2 adsorption capacities, selectivity, and

adsorption kinetics over long term operation under relevant process conditions. In this study, the
D
TE

stability of prepared sorbents (Silica-I) was evaluated after long-term operation in the packed-

bed reactor under SCF and NGCC flue gas conditions. Adsorption and regeneration temperatures
P

were kept at 65 and 110oC, respectively for every cycle. Moreover, the effect of SO2 impurity on
CE

the sorbent performance was evaluated by adding different SO2 concentration into the SCF flue
AC

gas. As exhibited in Figure 12, the adsorption capacity of the sorbent operating in “clean” flue

gas (i.e. without SO2) slowly drops by 16.6 % after 335 cycles in SCF flue gas (Figure 12A) and

by 7.6 % after 265 cycles in NGCC flue gas (Figure 12B).

The adsorption capacity is higher in the SCF flue gas, but the rate of CO2 loading loss is

faster; 13.4 % for the SCF flue gas compared to much lower 7.6 % loss rate for NGCC flue gas

after 265 cycles. Since the CO2 concentration in SCF flue gas (14.7 vol%) is higher than that in

NGCC flue gas (3 vol%), while the O2 content in SCF flue gas (4.5 vol %) is much lower (14 vol

% in NGCC flue gas), the faster degradation rate in the SCF flue gas is most likely due to a

21
 ACCEPTED MANUSCRIPT

higher CO2 concentration in the flue gas which may accelerate the formation of heat-stable urea

[51].

In a commercial power plant, SCF flue gas will contain a certain amount of SO2, the level

PT
of which is usually dependent on the type of coal used and the desulfurized technology (if any)

RI
deployed at the power plant. Similar to aqueous amine solvents, SO2 can deactivate a solid

sorbent by irreversibly reacting with amino sites to form a product which is not regenerable at

SC
desired process conditions and will lead to a reduction of the CO2 loading. The effect that SO2

NU
has on CO2 adsorption performance can be seen in Figure 12C and 12D. CO2 loading was

dramatically reduced as the SO2 impurity was introduced into the flue gas. SCF flue gas
MA
containing 50 ppm SO2 caused 43.4% loss in the CO2 loading after 108 adsorption-regeneration

cycles (Figure 12C). The impact was more severe using a SO2 concentration of 200 ppm in the
D
TE

SCF flue gas, resulting in 60.1% of the CO2 loading lost after only 88 cycles (Figure 12D).

Photos of the sorbent before and after long-term operation test, shown in Figure 12 (inset),
P

provide a visual observation of the sorbent degradation. These pictures show that the original
CE

white sorbent darkens after 335 cycles in “clean” SCF flue, but turned yellowish-brown with the
AC

addition of SO2 into the flue gas. It is apparent from these results that the presence of SO2 in the

flue gas is a critical issue that will affect the solid adsorbent-based CO2 capture process at

commercial scale. The SO2 concentration in flue gas varies and can reach above 200 ppm

depending on the types of coal and selected technology of power plant [64]. Thus, a deep

desulfurization system to eliminate SO2 will likely be required for application in coal-fired

power plants. However, the desulfurization can be omitted for a natural gas combined cycle

power plant because the flue gas from such a power plant contains a negligible amount of SO2.

3.8. Fluidization study

22
 ACCEPTED MANUSCRIPT

3.8.1. Fluidization with N2 gas

Fluidization tests were conducted with N2 to investigate the fluidization behavior of the

CO2 adsorbent. The sorbent used for these experiments is silica-VI impregnated with various PEI

PT
concentrations. Small, smooth bubbles were observed when the minimum fluidization velocity

RI
(µmf) for the sorbent bed was achieved suggesting that this adsorbent falls into the category of

Geldart B particles [65, 66]. Figure 13 exhibits the variation in pressure drop across the column

SC
as a function of increasing superficial gas velocity.

NU
At the beginning of this experiment, the pressure drop increased dramatically with the

superficial gas velocity until the µmf, it is around 0.5 (m/s), was achieved and the fluidization was
MA
occurred. Following this point, the pressure drop became constant irrespective of the increase in

the superficial gas velocity at least until the superficial gas velocity reached 3 m/s then rose with
D
TE

increasing superficial gas velocity. These results indicated that the smooth fluidization only

occurs between minimum fluidization velocity 0.5 and 3 m/s. At higher superficial gas velocities
P

the process is dominated by bubbling-slugging fluidization. Therefore, the superficial gas

CE

velocity of 3 m/s is considered as minimum slugging velocity (µms). The difference in these two
AC

fluidization regimes can visually observed in the pictures shown in Figure 13 (inset A and B).

The same fluidization behavior was observed on the adsorbent containing 35 and 37.5 wt% PEI.

Moreover, the fluidization behavior of the sorbent with N2 is not influenced by a change in the

temperature. As shown in Figure 14, similar fluidization was obtained at 50, 75 and 110oC.

A recent study on the fluidization of PEI impregnated mesoporous silica published by

Zhang et al [67] indicated that the sorbent smoothly fluidized between minimum air flow rate

and up to 10 L/min (equal a superficial gas velocity of 4.73 m/s). This maximum superficial gas

velocity for a smooth fluidization is higher than µms in the current study (3 m/s). The fluidization

23
 ACCEPTED MANUSCRIPT

of the sorbent is mostly affected by particle size, density, interaction between particles and

column wall. It is likely that the change in the fluidization behavior with increasing superficial

gas velocity is resulted from the interaction among PEI impregnated silica particles and

PT
aggregates. The particles tend to stick together due to the cohesiveness of the PEI. At a gas

RI
velocity below µms, the collision is not strong enough to stick a large amount of particles, thus

the process fluidizes smoothly. At gas velocity greater than µms, the particles and aggregates

SC
have high kinetic energy and the collision of such particles and aggregates results in the

NU
formation of larger aggregates. Those larger aggregates enhance the interaction between

aggregates and the wall of fluidization column causing slugging in the fluidizing bed and the
MA
increase in pressure drop. Thus, above µms the fluidization of the sorbent is similar to that of

Geldart D particles [66].

D
TE

3.8.2. Fluidization with simulated flue gas

Above discussion indicated that adsorbent exhibits fluidization at a superficial gas

P

velocities below µms and bubbling slugging fluidization at superficial gas velocities beyond µms
CE

with dry N2 gas. The characteristics of dry N2 gas, however, are much different with that of the
AC

flue gas, which contains a certain amount of water vapor. The PEI impregnated silica can adsorb

the water from flue gas and changes the properties, the density in particular [22]. Increasing in

the sorbent humidity will enhance the interaction among particles as well as the interaction

between particles and column wall, which will cause the change in sorbent’s fluidization

characteristics. Moreover, the fluidization in flue gas usually occurs at the same time with the

adsorption of CO2 on the sorbent, therefore, the fluidization of the sorbent in flue gas needs to be

investigated.

24
 ACCEPTED MANUSCRIPT

In this experiment, fluidization tests were continued by using a simulated flue gas with a

composition of CO2 ~15 vol% and humidity ~8 vol% in balance with N2 at different

temperatures. It was postulated that the sorbent may become more cohesive with higher PEI

PT
concentration, particularly in the presence of water vapor; therefore, the fluidization of sorbent

RI
containing 30, 35, and 37.5 wt% PEI were examined. Initially, the experiments were carried out

at 75oC to determine the variation in the pressure drop with increasing superficial gas velocity.

SC
The tests were started with a gas velocity of 2 m/s since a lower velocity could not be established

NU
due to a limitation of the minimum flow rate of the quantitative pump. Pressure drop change,

with increasing gas velocity, were recorded at 75oC with sorbents of various PEI content as is
MA
depicted in Figure 15.

Results of these experiments revealed that the sorbent fluidization behavior varies with
D
TE

changing PEI content. Interestingly, the increase in the pressure drop for the 30 wt% PEI sorbent

is more rapid than the ones containing 35 and 37.5 wt% PEI. This, however, does not prove that
P

fluidization behavior is better for samples containing higher PEI loadings. Actually, visual
CE

observation during these experiments indicated that the 30 wt% PEI sorbent exhibited superior
AC

fluidization behavior with a smooth bubbling regime at a relatively low gas velocity, which was

then associated with slugging at a higher velocity. In contrast, smooth fluidization was not

visually observed in higher PEI loading sorbents, instead of fluidization, the bed expanded, and

particles became adhesive and connected together to form an aggregate network. The kinetic

energy provided by the gas flow is not strong enough to break down this aggregate network. In

this situation, an extra energy is usually applied to separate the particles from bigger aggregates.

Several researches have tried different methods to improve the fluidization of cohesive particles

including vibration, mixing, and sound assistance [68-72]. In this work, the tests were conducted

25
 ACCEPTED MANUSCRIPT

at different temperatures from 75 to 110oC to investigate the possibility to minimize this effect

and improve the fluidization. Visual observations indicate that the 30 wt% PEI sorbent fluidized

well at this temperature range. However, the 35 wt% PEI sorbent fluidized at a temperature

PT
greater than 90oC, meanwhile 37.5 wt% sorbent did not fluidized at this temperature range.

RI
These results further confirm the effect of water vapor in the flue gas and the PEI content on the

fluidization behavior of the sorbent. In fact, water vapor in the flue gas plays a significantly

SC
important role in a CO2 capture process; it does not only help improve the CO2 adsorption

NU
capacity, but also lessens the degradation of the sorbent [51]. To attain good fluidization, the PEI

concentration of the sorbent should be taken in consideration and the optimal concentration,
MA
based on this studies, should be between 30-35 wt% PEI.

3.8.3. Fluidization with steam

D
TE

As highlighted before, the fluidization of the sorbent is considerably influenced by the

presence of water vapor in the flue gas. In commercial application of this solid sorbent
P

technology, the regeneration process is envisioned to be carried out by flowing steam or water
CE

vapor-enriched gas through regenerator at a desired temperature. It is theorized that fluidization

AC

of PEI impregnated silica under such conditions may prove to be challenging. In this study, the

fluidization behavior of different sorbents containing 30, 35, and 37.5 wt% PEI under steam was

investigated at 110oC, a frequently-used temperature for the regeneration process. The

fluidization results from these experiments are shown in Figure 16.

As seen in this figure, the samples containing 30 and 35 wt% PEI fluidized well, while

the sample containing 37.5 wt% PEI slugged significantly at the range of gas velocities studied.

Obviously, the sorbent containing 37.5 wt% PEI becomes more cohesive in the presence of water

vapor, resulting in stronger interaction among particles and formation of larger aggregates. In

26
 ACCEPTED MANUSCRIPT

contrast to previous fluidization tests with simulated flue gas that form porous channels in the

sorbent bed, fluidization with the steam causes significant slugging, causing the sorbent bed to be

lifted up as the steam flow is applied. These results further confirmed that the prepared sorbents

PT
can be regenerated in a fluidized bed reactor using steam, but PEI impregnated concentration on

RI
the sorbent has an important effect on the fluidization behavior in the presence of water vapor.

3.8.4. Adsorption-regeneration cycle in the fluidized-bed reactor

SC
In addition to performing fluidization experiments, the fluidized-bed reactor (FBR) was

NU
also used for CO2 adsorption-regeneration testing. In a typical adsorption-regeneration cycle, the

sorbent was first fluidized with N2 at a desired velocity to stabilize the fluidization bed while the
MA
bed temperature was raised up to 75oC. Upon reaching the temperature set point, the feed gas

was switched to the simulated flue gas (~18% CO2 and ~10% H2O in balance with N2) and the
D
TE

adsorption stage started. After few minutes, the CO2 concentration in the effluent started rising

and the adsorption stage was complete when the CO2 concentration in the effluent approached an
P

asymptote. After steady-state achieved, the temperature of the fluidization column was elevated
CE

to 110oC for the regeneration stage. In parallel, the feed gas composition was slowly changed by
AC

adding more steam and reducing CO2 and N2. As temperature reached a setting point, the feed

gas was completely replaced by the steam and the CO2 concentration observed in the effluent

was 100 vol% - indicating that the sorbent regeneration was occurring. Finally, the feed gas was

slowly switched to N2 at the same temperature for about 20 min to dry the adsorbent and begin

another cycle. Variation in the CO2 concentration in the effluent and CO2 removal efficiency in a

typical adsorption-regeneration cycle is presented in Figure 17.

Within the early stage of the adsorption, no CO2 was detected in the effluent and the

removal efficiency was 100 %, indicating that the sorbent worked well in the fluidized bed

27
 ACCEPTED MANUSCRIPT

reactor. After the breakthrough point (approximately 5 min), the CO 2 concentration in the

effluent increased and nearly reached equilibrium after 30 min. In the regeneration stage, as

temperature was elevated, the CO2 concentration in the effluent increased until 100 % CO2 was

PT
detected. This result confirms that the prepared sorbent is suitable for CO2 capture from flue gas

RI
using the fluidized-bed reactor.

Previous study on technical and economical evaluation of a CO2 capture process based on

SC
PEI-impregnated adsorbent indicated that a process based on fluidized-bed reactors helps reduce

NU
capital cost (21%), cost of electricity (81.7 compared to 85 $/kWh), and cost of CO2 avoided (62

compared to 69.8$/ton) in comparison with a conventional process based on aqueous

MA
ethanolamine (MEA) 30 wt% [73]. Fluidized-bed reactors were designed to capture 90% of CO2

in the NGCC flue gas operating at 70 and 110 oC for adsorption and regeneration, respectively.
D
TE

The obtained results based on the assumption that the sorbent should be fluidizable and have

CO2 adsorption capacity of 8 wt%. The sorbent prepared in the current study has CO2 loading
P

above 8 wt% at the same adsorption and regeneration temperatures of previous research [73] and
CE

fluidize well in both flue gas and steam. This suggests that PEI-impregnated silicagel could be a
AC

promising adsorbent, which may help reduce the cost of CO2 capture in comparison with a

conventional aqueous amine solution (MEA 30 wt%).

4. Conclusion

An advanced solid CO2 sorbent has been successfully prepared by the impregnation of

PEI onto mesoporous silica. The obtained sorbent is relatively stable under the heat treatment

and starts decomposing at a temperature of ~170oC. The sorbent CO2 capacity can be improved

by increasing the PEI impregnated content up to 40 wt%. However, above this concentration, the

CO2 loading is not improved.

28
 ACCEPTED MANUSCRIPT

The CO2 adsorption-regeneration performances are greatly affected by the operation

temperature in both SCF and NGCC flue gases. The optimal adsorption temperatures are from 50

to 80oC for NGCC flue gas and from 70 to 90 oC for SCF flue gas, while the optimal

PT
regeneration temperatures are from 110 to 130oC for both flue gas. At this condition, the sorbent

RI
also has an optimal CO2 adsorption kinetics with longer breakthrough time and higher CO2

loading at the breakthrough.

SC
The sorbent exhibits a relatively stable CO2 adsorption performance for multiple

NU
adsorption-regeneration cycles. The loss in CO2 loading was 16.6 % for 335 cycles operating

with the SCF flue gas and 7.6 % for 265 cycles operating with the NGCC flue gas. The lifetime
MA
of the sorbent is, however, significantly affected by SO2 impurity in the flue gas. SO2 at a

concentration of 50 ppm caused 43.4 % CO2 loading capacity loss after 108 cycles and a
D
TE

concentration of 200 ppm resulted in 60.1 % CO2 loading capacity loss after 88 cycles.

Regardless of the operating temperature, the sorbent displays a bubbling fluidization

P

regime under dry N2 gas, but it exhibits a dependence on temperature and PEI impregnated
CE

concentration while fluidizing with a humid gas. Sorbent with 30 wt% PEI fluidized well at
AC

temperatures from 75 to 110oC, however, sorbent containing 35 wt% PEI, only fluidized at

temperatures above 90oC. The effect of having higher PEI concentration becomes more severe

when fluidizing with water vapor, in fact, no fluidization was observed on a sorbent containing

37.5 wt % PEI.

In conclusion, PEI impregnated mesoporous silicas are promising sorbents for CO2

capture from both coal-fired power and natural gas combined cycle power plants. These

materials have significant scalability potential due to a simple wet impregnation method as well

as cheap raw materials. These sorbents have a relatively high adsorption capacity that can be

29
 ACCEPTED MANUSCRIPT

maintained for a long-term operation. Also, they are fluidizable with both simulated flue gas

(adsorption stage) and steam (regeneration stage) indicating that they could be suitable for a CO2

capture process using a fluidized bed reactor.

PT
Nomenclature

RI
ASAP 2020 Micromeritics Surface and Porosity Analyzer
BJH Barrett-Joyner-Halenda method

SC
BT Breakthrough time
CCS Carbon Capture & Sequestration
DI Deionized water

NU
MeOH Methanol
EtOH Ethanol
iPrOH Isopropanol
MA
MEA Ethanolamine
NGCC Natural gas combined cycle power plant
PBR Fluidized-bed reactor
D

PBR Packed-bed reactor

PEI Polyethyleneimine
TE

SCF Coal fired power plant

TGA Thermal Gravimetric Analysis
P

µmf Minimum fluidization velocity

CE

µms Minimum slugging velocity

AC

Acknowledgement

This study was supported by grant from Abu Dhabi Clean Energy, Masdar Company,

UAE (12KZOA1) and Department of Energy, USA (DE-FE0007707).

References

[1] A.B. Rao, E.S. Rubin, A Technical, Economic, and Environmental Assessment of Amine-
Based CO2 Capture Technology for Power Plant Greenhouse Gas Control, Environmental
Science & Technology, 36 (2002) 4467-4475.

30
 ACCEPTED MANUSCRIPT

[2] A.M. Wolsky, E.J. Daniels, B.J. Jody, CO2 Capture from the flue gas of conventional fossil-
fuel-fired power plants, Environmental Progress, 13 (1994) 214-219.
[3] B.P. Mandal, A.K. Biswas, S.S. Bandyopadhyay, Absorption of carbon dioxide into aqueous
blends of 2-amino-2-methyl-1-propanol and diethanolamine, Chemical Engineering Science, 58

PT
(2003) 4137-4144.
[4] B.P. Mandal, S.S. Bandyopadhyay, Absorption of carbon dioxide into aqueous blends of 2-

RI
amino-2-methyl-1-propanol and monoethanolamine, Chemical Engineering Science, 61 (2006)

SC
5440-5447.
[5] G. Maurin, P.L. Llewellyn, R.G. Bell, Adsorption Mechanism of Carbon Dioxide in

NU
Faujasites: Grand Canonical Monte Carlo Simulations and Microcalorimetry Measurements, The
Journal of Physical Chemistry B, 109 (2005) 16084-16091.
MA
[6] R. Serna-Guerrero, Y. Belmabkhout, A. Sayari, Modeling CO2 adsorption on amine-
functionalized mesoporous silica: 1. A semi-empirical equilibrium model, Chemical Engineering
Journal, 161 (2010) 173-181.
D

[7] Z.-l. Liu, Y. Teng, K. Zhang, Y. Cao, W.-p. Pan, CO2 adsorption properties and thermal
TE

stability of different amine-impregnated MCM-41 materials, Journal of Fuel Chemistry and

Technology, 41 (2013) 469-475.
P

[8] S.-T. Yang, J.-Y. Kim, J. Kim, W.-S. Ahn, CO2 capture over amine-functionalized MCM-22,
CE

MCM-36 and ITQ-2, Fuel, 97 (2012) 435-442.

[9] M. Bhagiyalakshmi, L. Yun, R. Anuradha, H. Jang, Synthesis of chloropropylamine grafted
AC

mesoporous MCM-41, MCM-48 and SBA-15 from rice husk ash: their application to CO2
chemisorption, J Porous Mater, 17 (2010) 475-484.
[10] D.V. Quang, A.V. Rabindran, N. El Hadri, M.R. Abu-Zahra, Reduction in the regeneration
energy of CO2 capture process by impregnating amine solvent onto precipitated silica, European
Scientific Journal, 9 (2013).
[11] Y. Fan, R.P. Lively, Y. Labreche, F. Rezaei, W.J. Koros, C.W. Jones, Evaluation of CO 2
adsorption dynamics of polymer/silica supported poly(ethylenimine) hollow fiber sorbents in
rapid temperature swing adsorption, International journal of greenhouse gas control, 21 (2014)
61-71.

31
 ACCEPTED MANUSCRIPT

[12] D.H. Jo, H. Jung, D.K. Shin, C.H. Lee, S.H. Kim, Effect of amine structure on CO 2
adsorption over tetraethylenepentamine impregnated poly methyl methacrylate supports,
Separation and Purification Technology, 125 (2014) 187-193.
[13] M.A. Alkhabbaz, R. Khunsupat, C.W. Jones, Guanidinylated poly(allylamine) supported on

PT
mesoporous silica for CO2 capture from flue gas, Fuel, 121 (2014) 79-85.
[14] Y.G. Ko, H.J. Lee, H.C. Oh, U.S. Choi, Amines immobilized double-walled silica

RI
nanotubes for CO2 capture, Journal of Hazardous Materials, 250–251 (2013) 53-60.

SC
[15] Y.G. Ko, H.J. Lee, J.Y. Kim, U.S. Choi, Hierarchically Porous Aminosilica Monolith as a
CO2 Adsorbent, ACS Applied Materials & Interfaces, 6 (2014) 12988-12996.

NU
[16] K. Wang, H. Shang, L. Li, X. Yan, Z. Yan, C. Liu, Q. Zha, Efficient CO2 capture on low-
cost silica gel modified by polyethyleneimine, Journal of Natural Gas Chemistry, 21 (2012) 319-
MA
323.
[17] D.W.F. Brilman, R. Veneman, Capturing Atmospheric CO2 Using Supported Amine
Sorbents, Energy Procedia, 37 (2013) 6070-6078.
D

[18] G. Qi, Y. Wang, L. Estevez, X. Duan, N. Anako, A.-H.A. Park, W. Li, C.W. Jones, E.P.
TE

Giannelis, High efficiency nanocomposite sorbents for CO2 capture based on amine-
functionalized mesoporous capsules, Energy & Environmental Science, 4 (2011) 444-452.
P

[19] G. Qi, L. Fu, B.H. Choi, E.P. Giannelis, Efficient CO2 sorbents based on silica foam with
CE

ultra-large mesopores, Energy & Environmental Science, 5 (2012) 7368-7375.

[20] C. Chen, W.-J. Son, K.-S. You, J.-W. Ahn, W.-S. Ahn, Carbon dioxide capture using
AC

amine-impregnated HMS having textural mesoporosity, Chemical Engineering Journal, 161

(2010) 46-52.
[21] Y. Labreche, R.P. Lively, F. Rezaei, G. Chen, C.W. Jones, W.J. Koros, Post-spinning
infusion of poly(ethyleneimine) into polymer/silica hollow fiber sorbents for carbon dioxide
capture, Chemical Engineering Journal, 221 (2013) 166-175.
[22] D.V. Quang, A. Dindi, A.V. Rayer, N.E. Hadri, A. Abdulkadir, M.R. Abu‐Zahra, Effect of
moisture on the heat capacity and the regeneration heat required for CO2 capture process using
PEI impregnated mesoporous precipitated silica, Greenhouse Gases: Science and Technology,
(2014).
[23] J.-Y. Kim, J. Kim, S.-T. Yang, W.-S. Ahn, Mesoporous SAPO-34 with amine-grafting for
CO2 capture, Fuel, 108 (2013) 515-520.

32
 ACCEPTED MANUSCRIPT

[24] B. Dutcher, M. Fan, S. Cui, X.-D. Shen, Y. Kong, A.G. Russell, P. McCurdy, M. Giotto,
Characterization and stability of a new, high-capacity amine-functionalized CO2 sorbent,
International journal of greenhouse gas control, 18 (2013) 51-56.
[25] O. Cheung, Z. Bacsik, Q. Liu, A. Mace, N. Hedin, Adsorption kinetics for CO2 on highly

PT
selective zeolites NaKA and nano-NaKA, Applied Energy, 112 (2013) 1326-1336.
[26] N.A. Rashidi, S. Yusup, H.L. Lam, Kinetic studies on carbon dioxide capture using

RI
activated carbon, Chemical Engineer Transaction, 35 (2013) 361-366.

SC
[27] B. Guo, L. Chang, K. Xie, Adsorption of Carbon Dioxide on Activated Carbon, Journal of
Natural Gas Chemistry, 15 (2006) 223-229.

NU
[28] T.L.P. Dantas, F.M.T. Luna, I.J. Silva Jr, D.C.S. de Azevedo, C.A. Grande, A.E. Rodrigues,
R.F.P.M. Moreira, Carbon dioxide–nitrogen separation through adsorption on activated carbon in
MA
a fixed bed, Chemical Engineering Journal, 169 (2011) 11-19.
[29] M.S. Shafeeyan, W.M.A.W. Daud, A. Houshmand, A. Shamiri, A review on surface
modification of activated carbon for carbon dioxide adsorption, Journal of Analytical and
D

Applied Pyrolysis, 89 (2010) 143-151.

TE

[30] M. Auta, N.D. Amat Darbis, A.T. Mohd Din, B.H. Hameed, Fixed-bed column adsorption
of carbon dioxide by sodium hydroxide modified activated alumina, Chemical Engineering
P

Journal, 233 (2013) 80-87.

CE

[31] C. Chen, W.-S. Ahn, CO2 capture using mesoporous alumina prepared by a sol–gel process,
Chemical Engineering Journal, 166 (2011) 646-651.
AC

[32] J.A. Thote, R.V. Chatti, K.S. Iyer, V. Kumar, A.N. Valechha, N.K. Labhsetwar, R.B.
Biniwale, M.K.N. Yenkie, S.S. Rayalu, N-doped mesoporous alumina for adsorption of carbon
dioxide, Journal of Environmental Sciences, 24 (2012) 1979-1984.
[33] V. Nafisi, M.-B. Hägg, Development of dual layer of ZIF-8/PEBAX-2533 mixed matrix
membrane for CO2 capture, Journal of Membrane Science, 459 (2014) 244-255.
[34] T.O. Nelson, L.J.I. Coleman, A. Kataria, M. Lail, M. Soukri, D.V. Quang, M.R.M.A. Zahra,
Advanced Solid Sorbent-Based CO2 Capture Process, Energy Procedia, 63 (2014) 2216-2229.
[35] X. Xu, C. Song, J.M. Andresen, B.G. Miller, A.W. Scaroni, Novel Polyethylenimine-
Modified Mesoporous Molecular Sieve of MCM-41 Type as High-Capacity Adsorbent for CO2
Capture, Energy & Fuels, 16 (2002) 1463-1469.

33
 ACCEPTED MANUSCRIPT

[36] X. Xu, C. Song, J.M. Andrésen, B.G. Miller, A.W. Scaroni, Preparation and characterization
of novel CO2 “molecular basket” adsorbents based on polymer-modified mesoporous molecular
sieve MCM-41, Microporous and Mesoporous Materials, 62 (2003) 29-45.
[37] C. Song, X. Xu, J.M. Andresen, B.G. Miller, A.W. Scaroni, Novel Nanoporous “Molecular

PT
Basket” Adsorbent for CO2 Capture, in: J.-S.C. Sang-Eon Park, L. Kyu-Wan (Eds.) Studies in
Surface Science and Catalysis, Elsevier2004, pp. 411-416.

RI
[38] X. Xu, C. Song, B.G. Miller, A.W. Scaroni, Influence of Moisture on CO2 Separation from

SC
Gas Mixture by a Nanoporous Adsorbent Based on Polyethylenimine-Modified Molecular Sieve
MCM-41, Industrial & Engineering Chemistry Research, 44 (2005) 8113-8119.

NU
[39] C. Song, Global challenges and strategies for control, conversion and utilization of CO2 for
sustainable development involving energy, catalysis, adsorption and chemical processing,
MA
Catalysis Today, 115 (2006) 2-32.
[40] X. Ma, X. Wang, C. Song, “Molecular Basket” Sorbents for Separation of CO2 and H2S
from Various Gas Streams, Journal of the American Chemical Society, 131 (2009) 5777-5783.
D

[41] X. Wang, V. Schwartz, J.C. Clark, X. Ma, S.H. Overbury, X. Xu, C. Song, Infrared Study of
TE

CO2 Sorption over “Molecular Basket” Sorbent Consisting of Polyethylenimine-Modified

Mesoporous Molecular Sieve, The Journal of Physical Chemistry C, 113 (2009) 7260-7268.
P

[42] D.V. Quang, A. Dindi, A.V. Rayer, N.E. Hadri, A. Abdulkadir, M.R.M. Abu-Zahra,
CE

Impregnation of Amines Onto Porous Precipitated Silica for CO2 capture, Energy Procedia, 63
(2014) 2122-2128.
AC

[43] A.D. Ebner, M.L. Gray, N.G. Chisholm, Q.T. Black, D.D. Mumford, M.A. Nicholson, J.A.
Ritter, Suitability of a Solid Amine Sorbent for CO2 Capture by Pressure Swing Adsorption,
Industrial & Engineering Chemistry Research, 50 (2011) 5634-5641.
[44] G.P. Knowles, J.V. Graham, S.W. Delaney, A.L. Chaffee, Aminopropyl-functionalized
mesoporous silicas as CO2 adsorbents, Fuel Processing Technology, 86 (2005) 1435-1448.
[45] R.A. Khatri, S.S.C. Chuang, Y. Soong, M. Gray, Thermal and Chemical Stability of
Regenerable Solid Amine Sorbent for CO2 Capture, Energy & Fuels, 20 (2006) 1514-1520.
[46] G. Calleja, R. Sanz, A. Arencibia, E.S. Sanz-Pérez, Influence of Drying Conditions on
Amine-Functionalized SBA-15 as Adsorbent of CO2, Top Catal, 54 (2011) 135-145.
[47] Y. Belmabkhout, R. Serna-Guerrero, A. Sayari, Amine-bearing mesoporous silica for CO2
removal from dry and humid air, Chemical Engineering Science, 65 (2010) 3695-3698.

34
 ACCEPTED MANUSCRIPT

[48] P. Bollini, S. Choi, J.H. Drese, C.W. Jones, Oxidative Degradation of Aminosilica
Adsorbents Relevant to Postcombustion CO2 Capture, Energy & Fuels, 25 (2011) 2416-2425.
[49] A. Heydari-Gorji, A. Sayari, Thermal, Oxidative, and CO2-Induced Degradation of
Supported Polyethylenimine Adsorbents, Industrial & Engineering Chemistry Research, 51

PT
(2012) 6887-6894.
[50] C.S. Srikanth, S.S.C. Chuang, Spectroscopic Investigation into Oxidative Degradation of

RI
Silica-Supported Amine Sorbents for CO2 Capture, ChemSusChem, 5 (2012) 1435-1442.

SC
[51] A. Sayari, Y. Belmabkhout, Stabilization of Amine-Containing CO2 Adsorbents: Dramatic
Effect of Water Vapor, Journal of the American Chemical Society, 132 (2010) 6312-6314.

NU
[52] A. Sayari, A. Heydari-Gorji, Y. Yang, CO2-Induced Degradation of Amine-Containing
Adsorbents: Reaction Products and Pathways, Journal of the American Chemical Society, 134
MA
(2012) 13834-13842.
[53] T.C. Drage, A. Arenillas, K.M. Smith, C.E. Snape, Thermal stability of polyethylenimine
based carbon dioxide adsorbents and its influence on selection of regeneration strategies,
D

Microporous and Mesoporous Materials, 116 (2008) 504-512.

TE

[54] F. Rezaei, C.W. Jones, Stability of Supported Amine Adsorbents to SO2 and NOx in
Postcombustion CO2 Capture. 2. Multicomponent Adsorption, Industrial & Engineering
P

Chemistry Research, 53 (2014) 12103-12110.

CE

[55] M.W. Hahn, M. Steib, A. Jentys, J.A. Lercher, Mechanism and Kinetics of CO 2 Adsorption
on Surface Bonded Amines, The Journal of Physical Chemistry C, 119 (2015) 4126-4135.
AC

[56] R. Sanz, G. Calleja, A. Arencibia, E.S. Sanz-Pérez, CO2 adsorption on branched

polyethyleneimine-impregnated mesoporous silica SBA-15, Applied Surface Science, 256
(2010) 5323-5328.
[57] D.V. Quang, A.V. Rayer, N. El Hadri, M.R.M. Abu-Zahra, Preparation of Polyethylenimine
Impregnated Mesoporous Precipitated Silica for CO2 Capture, in: M.J.A.-M.T. Eljack (Ed.)
Proceedings of the 4th International Gas Processing Symposium, Elsevier, Oxford, 2015, pp. 21-
37.
[58] A. Heydari-Gorji, Y. Belmabkhout, A. Sayari, Polyethylenimine-Impregnated Mesoporous
Silica: Effect of Amine Loading and Surface Alkyl Chains on CO2 Adsorption, Langmuir, 27
(2011) 12411-12416.

35
 ACCEPTED MANUSCRIPT

[59] A. Zhao, A. Samanta, P. Sarkar, R. Gupta, Carbon Dioxide Adsorption on Amine-

Impregnated Mesoporous SBA-15 Sorbents: Experimental and Kinetics Study, Industrial &
Engineering Chemistry Research, 52 (2013) 6480-6491.
[60] A. Heydari-Gorji, A. Sayari, CO2 capture on polyethylenimine-impregnated hydrophobic

PT
mesoporous silica: Experimental and kinetic modeling, Chemical Engineering Journal, 173
(2011) 72-79.

RI
[61] E.R. Monazam, J. Spenik, L.J. Shadle, CO2 Desorption Kinetics for Immobilized

SC
Polyethylenimine (PEI), Energy & Fuels, 28 (2013) 650-656.
[62] R.K. Iler, The chemistry of silica, Wiley, New York, 1979.

NU
[63] E.F. Vansant, P. Voort, K.C. Vrancken, Characterization and Chemical Modification of the
Silica Surface, Elsevier1995. MA
[64] Y. Tan, E. Croiset, M.A. Douglas, K.V. Thambimuthu, Combustion characteristics of coal
in a mixture of oxygen and recycled flue gas, Fuel, 85 (2006) 507-512.
[65] D. Geldart, Types of gas fluidization, Powder Technology, 7 (1973) 285-292.
D

[66] D. Kunii, O. Levenspiel, Fluidization Engineering, Butterworth-Heinemann1991.

TE

[67] W. Zhang, H. Liu, C. Sun, T.C. Drage, C.E. Snape, Performance of polyethyleneimine–
silica adsorbent for post-combustion CO2 capture in a bubbling fluidized bed, Chemical
P

Engineering Journal, 251 (2014) 293-303.

CE

[68] J.A. Quevedo, J. Flesch, R. Pfeffer, R. Dave, Evaluation of assisting methods on fluidization
of hydrophilic nanoagglomerates by monitoring moisture in the gas phase, Chemical Engineering
AC

Science, 62 (2007) 2608-2622.

[69] F. Raganati, V. Gargiulo, P. Ammendola, M. Alfe, R. Chirone, CO2 capture performance of
HKUST-1 in a sound assisted fluidized bed, Chemical Engineering Journal, 239 (2014) 75-86.
[70] Q. Guo, Y. Li, M. Wang, W. Shen, C. Yang, Fluidization Characteristics of SiO 2
Nanoparticles in an Acoustic Fluidized, Chemical Engineering & Technology, 29 (2006) 78-86.
[71] P. Ammendola, F. Raganati, R. Chirone, Effect of operating conditions on the CO 2 recovery
from a fine activated carbon by means of TSA in a fluidized bed assisted by acoustic fields, Fuel
Processing Technology, 134 (2015) 494-501.
[72] F. Raganati, P. Ammendola, R. Chirone, Effect of acoustic field on CO2 desorption in a
fluidized bed of fine activated carbon, Particuology, 23 (2015) 8-15.

36
 ACCEPTED MANUSCRIPT

[73] A. Seif El Nasr, T. Nelson, A. Kataria, M.R.M. Abu-Zahra, Benchmarking of a novel solid
sorbent CO2 capture process for NGCC power generation, International Journal of Greenhouse
Gas Control, 42 (2015) 583-592.

PT
RI
SC
NU
MA
D
P TE
CE
AC

37
 ACCEPTED MANUSCRIPT

Figure captions

Figure 1. Packed-bed reactor (PBR) illustration. A, B, C, and D is humidifier, packed bed

reactor, vapor condenser, and water collector, respectively.

PT
Figure 2. Fluidized-bed reactor (FBR) illustration. A, B, C, D, and E is steam generator,
fluidization column, cyclone, condenser, and water collector, respectively.

RI
Figure 3. Variation in BET surface area and pore volume as a function of PEI concentration

SC
impregnated onto Silica-I
Figure 4. Variation in sorbent bulk density as a function of PEI concentration impregnated on

NU
Silica-I
Figure 5. TGA profiles of silica-I and prepared sorbents
MA
Figure 6. CO2 loading performance of sorbents as a function of PEI concentration impregnated
on Silica-I. PEI impregnation was conducted in MeOH (Silica-I). Adsorption was conducted at
65oC in the SCF flue gas and regeneration was conducted at 110oC in humidified N2.
D

Figure 7. The effect of different solvents on the CO2 loading. Tests was conducted on sorbent
TE

containing 35 wt% PEI (Silica-I). Adsorption was conducted at 65oC in the SCF flue gas and
regeneration was conducted at 110oC in humidified N2.
P

Figure 8. Effect of adsorption temperature on a sorbent CO2 loading. Tests conducted on

CE

sorbents containing 37.5 wt% PEI impregnated on Silica-I using EtOH as solvent and. Sorbent
was regenerated at 110oC in humidified N2.
AC

Figure 9. Effect of regeneration temperature on a sorbent CO2 loading. Adsorption was

conducted at 65oC on sorbents containing 37.5 wt% PEI impregnated on Silica-I using EtOH as
solvent.
Figure 10. CO2 adsorption kinetics at 30oC, 60oC, 80oC and 100oC. Tests conducted with SCF
flue gas on sorbents containing 37.5 wt% PEI impregnated on Silica-I.
Figure 11. CO2 adsorption kinetics at 65oC. Tests conducted with NGCC flue gas on adsorbents
containing 37.5 wt% PEI impregnated on Silica-I.
Figure 12. Long-term CO2 adsorption performance stability of sorbents (Silica-I) containing
37.5 wt% PEI operating in SCF and NGCC flue gas environments.
Figure 13. Fluidization behavior of the adsorbent (Silica-VI) in N2 gas at 75oC
Figure 14. Fluidization behavior of the adsorbent (Silica-VI) in N2 gas at different temperatures.

38
 ACCEPTED MANUSCRIPT

Figure 15. Fluidization behavior of various PEI-loaded sorbents (Silica-VI) in simulated flue gas
at 75oC
Figure 16. Fluidization behavior of various PEI-loaded adsorbents (Silica-VI) in steam at 110oC
Figure 17. A typical adsorption-regeneration cycle in the FBR. Test conducted on adsorbent

PT
containing 30 wt% PEI impregnated on Silica-VI.

RI
SC
N2
MFC1

NU
Air
C
MFC2
To vent
B CO2
MA
A
SO2 analyzer
MFC3

CO2 D
D

MFC4
TE

Figure 1.
P
CE
AC

H2O
Pump
C
N2
MFC5
D
A To vent
B CO2
CO2 analyzer
MFC6

Circulation bath E
Heating oil

Figure 2.

39
 ACCEPTED MANUSCRIPT

300 1.2

250 1.0
BET surface area
Pore volume

BET surface area (m /g)

PT
0.8

2
200

Pore volume (cm )

3
150 0.6

RI
0.4
100

SC
0.2
50

NU
0.0
0 10 20 30 40 50
PEI concentration (wt%)
MA
Figure 3.
D
TE

0.8
P
CE

0.7
AC 3
Bulk density (g/cm )

0.6

0.5

0.4

0.3
0 10 20 30 40 50
PEI concentration ( wt%)

Figure 4.

40
 ACCEPTED MANUSCRIPT

100 1

95

90

85

PT
Percentage (%)
80

75
1 Silica I

RI
70 2 30 wt% PEI/Silica I
2
65 3 37.5 wt% PEI/Silica I

SC
60 3

55

NU
50
0 100 200 300 400 500 600
o
Temperature ( C)
MA
Figure 5.
D
P TE

14
CE

12

10
AC CO2 loading (wt%)

25 30 35 40 45 50
PEI concentration ( wt%)

Figure 6.

41
 ACCEPTED MANUSCRIPT

12

10

Adsorption capacity (wt %)

PT
8

RI
4

SC
2

NU
0
MeOH EtOH Isopropanol water
MA Solvent

Figure 7.
D
TE

14
24
P

12
20
CE
Breakthrough time (BT, min)

10

CO2 loading (wt%)

16
AC

12 6

CO2 loading (SCF flue gas) CO2 loading (NGCC flue gas)
4
8
Breakthrough time (SCF flue gas) Breakthrough time (NGCC flue gas)
CO2 loading at breakthrough (SCF flue gas) CO2 loading at breakthrough (NGCC flue gas)
2
4
0
20 40 60 80 100
o
Temperature ( C)

Figure 8.

42
 ACCEPTED MANUSCRIPT

PT
RI
SC
16 16
CO2 adsorption capacity for SCF flue gas
14 14

NU
12 12
CO2 adsorption capacity for NGCC flue gas

CO2 loading (wt%)

Lean loading (wt%)

10 MA 10
90

8 80
8
70

60
Percentage (%)
6 50 6
40
D

30
4 20
CO2 loading increment
4
Sensible heat increment
10
TE

NGCC lean loading

2 SCF lean loading
0
100 110 120 130 140 150 2
Temperature (oC)
NGCC loading
0 SCF loading
0
P

90 100 110 120 130 140 150

CE

o
Temperature ( C)
AC

Figure 9.

43
 ACCEPTED MANUSCRIPT

PT
RI
20 20
Accumulative CO2 loading 100
CO2 concentration in outlet gas (vol%)

Accumulative CO2 loading 100

CO2 concentration in outlet gas (vol%)

SC
CO2 concentration in outlet gas

and removal efficiency (%)

80

Accumulative CO2 loading

15 CO2 concentration in outlet gas 15 80

Accumulative CO 2 Loading
and removal efficiency (%)
60 60

NU
10 10

T=30oC 40 T=60oC
40

5 5 20
20
MA CO2 removal efficiency
CO2 removal efficiency 0

0 0 0
0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100

Adsorption time (min) Adsorption time (min)

D
TE

20 20
Accumulative CO2 loading 100 Accumulative CO2 loading 100
CO2 concentration in outlet gas (vol%)

CO2 concentration in outlet gas (vol%)

CO2 concentration in outlet gas CO2 concentration in outlet gas

Accumulative CO 2 loading

80
and removal efficiency (%)
P

15 15 80

and removal efficiency (%)

Accumulative CO2 loading
CE

60 60
10 10

T=80oC 40 T=100oC 40

5
AC

5
20
20
CO2 removal efficiency
CO2 removal efficiency
0 0
0 10 20 30 40 50 60 70 80 90 100 0 0
0 10 20 30 40 50 60 70 80 90 100
Adsorption time (min)
Adsorption time (min)

Figure 10.

44
 ACCEPTED MANUSCRIPT

PT
RI
SC
5
Accumulative CO2 loading
CO2 concentration in outlet gas (vol%)

NU
100

4
80

Accumulative CO 2 loading
and removal eficiency (%)
MA
CO2 concentration in outlet gas
3
60
D

2
40
TE

1 20
P

Removal efficiency
CE

0 0
0 10 20 30 40 50 60 70 80 90 100
Time (min)
AC

Figure 11.

45
 ACCEPTED MANUSCRIPT

PT
RI
SC
NU
14

12
MA
A

10 B
CO2 loading (wt%)

8 C
TE

6 D
Origin A C D
P

4
CE

A: SCF flue gas without SO2, loss in CO2 loading: 16.6 % for 335 cycles
2 B: SNGCC flue gas without SO2, loss in CO2 loading: 7.6 % for 265 cycles
C: SCF flue gas with 50ppm SO2, loss in CO2 loading: 43.4 % for 108 cycles
D: SCF flue gas with 200ppm SO2, loss in CO2 loading: 60.1 % for 88 cycles
AC

0
0 50 100 150 200 250 300 350
Number of cycles

Figure 12.

46
 ACCEPTED MANUSCRIPT

10 35 % PEI

PT
37.5 % PEI
9 A
8

RI
Pressure drop (inH2O)

6 B

SC
5

NU
3

2 MA
1
0 2 4 6 8 10 12 14 16
Superficial gas velocity (m/s)

Figure 13.
D
P TE

5
o
CE

50 C
o
75 C
4 o
110 C
Pressure drop (inH2O)
AC

1
0 2 4 6 8 10
Supreficial gas velocity (m/s)

Figure 14.

47
 ACCEPTED MANUSCRIPT

10

30 wt% PEI
35 wt % PEI
8 37.5 wt% PEI

PT
Pressure drop (inH2O)

RI
6

SC
4

NU
2

0 2 4 6 8 10 12 14
MA
Velocity (m/s)

Figure 15.
D
TE

50
P
CE

40
Pressure drop (inH2O)

AC

30 30 wt% PEI
35 wt% PEI
37.5 wt% PEI
20

10

0 2 4 6 8
Superficial gas velocity (m/s)

Figure 16.

48
 ACCEPTED MANUSCRIPT

PT
RI
SC
CO2 concentration in efluent gas (vol%)

100

NU
100
Adsorption
Regeneration

CO2 removal eficiency (%)

80
80
MA
Removal efficiency
60
60
D

40 40
TE

20 20
P
CE

0 0
0 10 20 30 40 50 60 70
Time (min)
AC

Figure 17.

49
 ACCEPTED MANUSCRIPT

Table captions

Table 1. Variation in adsorbent properties as a function of PEI concentration impregnated on

silica (Silica I). PEI impregnation was conducted in MeOH. Adsorption and regeneration were

PT
carried out at 65oC in the SCF flue gas and 110oC in humidified N2, respectively.

RI
Table 2. Effect of adsorption temperature on CO2 loading performance in SCF flue gas. Sorbent
(Silica-I) contains 37.5 wt% PEI impregnated using EtOH. Regeneration were carried out at

SC
110oC in humidified N2.
Table 3. Effect of adsorption temperature on CO2 loading performance in NGCC flue gas.

NU
Sorbent (Silica-I) contains 37.5 wt% PEI impregnated using EtOH. Regeneration were carried
out at 110oC in humidified N2. MA
Table 4. Properties of various silica substrates and their corresponding sorbents prepared by
impregnation of PEI 37.5 wt% using EtOH.
D
P TE
CE
AC

50
 ACCEPTED MANUSCRIPT

Table 1.

PT
PEI concentration CO2 loading BET surface Pore volume Pore size N content Molar ratio Bulk density
(wt%) (wt%) area (m2/g) (cm3/g) (Ǻ) (wt%) of CO2/N (g/cm3)

0 0 300.0 1.15 150 - - 0.39

RI
30 8.48 115.0 0.49 109 8.77 0.31 0.56
35 9.08 88.4 0.39 113 10.20 0.28 0.61

SC
37.5 9.59 76.0 0.32 110 11.20 0.27 0.62
40 10.50 56.0 0.30 113 11.86 0.28 0.65
45 10.47 38.8 0.17 110 13.16 0.25 0.70

NU
MA
Table 2.
CO2 loading CO2 loading at
Temperature Breakthrough CO2 loading
percentage (%) breakthrough
(oC)
D

time (min) (wt%)

at breakthrough (wt%)
TE

30 3.1 33.80 4.39 12.97

40 3.5 40.20 5.46 13.57
50 3.9 43.00 5.86 13.62
P

60 4.3 45.40 6.03 13.29

CE

70 4.8 53.20 6.85 12.88

80 4.9 56.30 6.83 12.14
90 5.0 59.70 6.74 11.29
AC

100 4.7 59.50 6.16 10.36

Table 3.
CO2 loading CO2 loading at CO2
Temperature Breakthrough
percentage (%) breakthrough loading
(oC) time (min)
at breakthrough (wt%) (wt%)
30 16.9 71.60 6.00 8.38
50 23.3 76.33 7.35 9.63
65 22.4 79.93 7.08 8.86
80 23.7 83.79 6.56 7.83
100 19.0 76.43 4.46 5.84

51
 ACCEPTED MANUSCRIPT

Table 4.

Silica substrates Silica-I Silica-II Silica-III Silica-IV Silica-V Silica-VI Silica-VII Silica-VIII Silica-IX

PT
PEI concentration (wt%) 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5

RI
CO2 loading (wt%)* 11.38 11.79 10.65 9.75 10.92 11.69 11.52 11.79 10.03
Silica 150 113.2 122.9 153 152 155 284 320 287

SC
Pore size (Å)
sorbent 110 106.6 115.7 129.5 137.4 116.9 345 158 318
Silica 1.15 1.19 1.23 1.11 1.16 1.12 1.24 1.24 0.81

NU
Pore volume (cm3/g)
sorbent 0.32 0.36 0.38 0.31 0.35 0.27 0.40 0.34 0.22
Silica 300 345.4 353.3 290 305 290 175 155 98
Surface area (m2/g)

MA
sorbent 76 86.27 85.5 63.8 67 61.9 34.2 58.9 20.1
Tap density of sorbent Silica 0.39 0.36 0.36 0.43 0.39 0.44 0.38 0.39 -
(g/cm3) sorbent 0.63 0.56 0.57 0.66 0.61 0.67 0.6 0.61 -

ED
Silica particle size (µm) 75-150 71-144 87-200 250-500 105-175 75-250 40-63 200-500 45-75
Silica pH (suspension 5 wt%) - - - 7.1 7.2 7 7.5 7.3 -

PT
o
Volatile content in silica (%, 160 C) - - - 4.5 4 4.8 3.3 2.9 -
*CO2 loading was determined using SCF flue gas. Adsorption and regeneration were conducted at 65 and 110 oC, respectively.
CE
AC

52
 ACCEPTED MANUSCRIPT

Investigation of CO2 Adsorption Performance and Fluidization Behavior of Mesoporous Silica Supported Polyethylenimine

PT
RI
SC
NU
MA
ED
Graphical abstract

PT
CE
AC

53
 ACCEPTED MANUSCRIPT

Research Highlights

PT
 PEI based sorbent was evaluated on both packed-bed and fluidized-bed reactors

RI
 CO2 adsorption is stable for multiple adsorption/regeneration cycle in humid flue gas without SO2

SC
 Sorbent fluidizes well in both simulated flue gas and steam

NU
MA
ED
PT
CE
AC

54