1-s2.0-S0360544219308072-main

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Energy 179 (2019) 475e489

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Experimental and kinetic study of the catalytic desorption of CO2 from


CO2-loaded monoethanolamine (MEA) and blended
monoethanolamine e Methyl-diethanolamine (MEA-MDEA) solutions
Ananda Akachuku, Priscilla Anima Osei, Benjamin Decardi-Nelson, Wayuta Srisang,
Fatima Pouryousefi, Hussameldin Ibrahim, Raphael Idem*
Clean Energy Technologies Research Institute (CETRI), Faculty of Engineering and Applied Science, University of Regina, Regina, Saskatchewan, S4S 0A2,
Canada

a r t i c l e i n f o a b s t r a c t

Article history: The kinetics of catalyst-aided desorption of CO2 from CO2-loaded MEA (5 mol/dm3) and blended MEA-
Received 6 March 2019 MDEA (7 mol/dm3) solutions at CO2 loadings from 0.3 to 0.5 mol CO2/mol amine were studied over a
Received in revised form Lewis acid catalyst (g-Al2O3) and a Brønsted acid catalyst (H-ZSM-5) in an absorberedesorber CO2
23 April 2019
capture bench-scale plant (columns of 0.051 m ID and height of 1.067 m) at temperatures of 348e368 K.
Accepted 24 April 2019
Available online 7 May 2019
The results showed that the conversion increases relative to no catalyst for MEA were 55 and 74% while
for MEA-MDEA, they were 65 and 85.2% with g-Al2O3 and H-ZSM-5, respectively. A comprehensive
mechanistic LHHW rate model was developed for the catalytic CO2 desorption process.
Keywords:
CO2 desorption
© 2019 Elsevier Ltd. All rights reserved.
Solid acid catalyst
Kinetics
Mechanistic model
Activation energy
Acid sites

1. Introduction operational temperature introduces several drawbacks such as high


parasitic heat duty, solvent loss, and degradation, as well as stress
Carbon dioxide (CO2) is known to be a major contributor to cracking corrosion, thus making amine PCCC a cost-prohibitive
global warming due to its ever-increasing atmospheric concentra- process [5]. Meanwhile, about 70% of the operating cost is attrib-
tion resulting mainly from human activities such as fossils fuel uted to the energy requirement for amine regeneration [6], which
combustion and deforestation. In the fight to reduce the industrial indirectly reduces the overall efficiency of the power plant to
emission of CO2, a landmark agreement aimed at keeping the global approximately 30e40% if integrated with the CCS unit [7]. This
temperature rise below 2  C above the pre-industrial levels was generates a large interest in the heat duty for regeneration, making
reached at the 21st Conferences of Parties, COP 21, Paris summit [1]. it one of the most important parameters that should be carefully
This reduction in global temperature demands that more emphasis accounted for in the economic evaluation of the CCS technology [8].
is placed on the development and modification of new and existing The reaction kinetics is another essential parameter needed for
Carbon Capture Storage (CCS) technologies. Amine-based post- process design and optimization and highly depends on solvent
combustion CO2 capture (PCCC) has been reported to be one of the chemistry. Consequently, a robust and comprehensive model is
most promising and efficient mitigation technologies used in CO2 needed to account for the fundamental mechanism of the kinetic
capture due to its capability of handling large amounts of exhaust phenomena underlying CO2 absorption and desorption reactions.
gas streams in a cost-effective manner [2,3]. It is a well known fact Also, there is a need for a faster reaction rate because it allows for a
that the regeneration of CO2 from loaded amine occurs at an smaller absorber/desorber unit as well as reduced heat transfer
elevated temperature within the range of 120e140  C [4]. This high units.
Monoethanolamine (MEA), diethanolamine (DEA) and methyl-
diethanolamine (MDEA) are some commonly used solvents in
* Corresponding author. amine-based PCCC [9]. The reaction kinetics and mechanism of
E-mail address: Raphael.idem@uregina.ca (R. Idem).

https://doi.org/10.1016/j.energy.2019.04.174
0360-5442/© 2019 Elsevier Ltd. All rights reserved.
476 A. Akachuku et al. / Energy 179 (2019) 475e489

these amines have been summarized and reviewed by numerous mechanism, which was originally introduced by Caplow et al. [20]
researchers in the literature [10e12]. To increase the reaction ki- and later reintroduced by Danckwerts et al. [21], and the termo-
netics and reduce heat duty associated with CCS, researchers have lecular mechanism proposed by Crooks and Donnellan and further
proposed methods involving solvent enhancement and process modified by Ozutzurk et al. [22]. The two-step Zwitterion mecha-
optimization [8]. The addition of organic acid to CO2 loaded amine nism consists of the reaction of CO2 with the amine solution (RNH2 Þ
was reported to increase the CO2 desorption kinetics and lower to form Zwitterion intermediate (RNH þ 
2 COO ), equation (1) which
regeneration heat duty [13,14]. The overview of these technologies further reacts with any base (B) in the system to form carbamate
can be seen in the recent work of Li et al. [15]. Nevertheless, the ions and a protonated base (Bþ).
theoretical heat duty limit of 0.4356 MJ/kg CO2 [8] is far from reach Zwitterion formation:
and requires more work in order to obtain a reasonable energy
input. Since the major drawback associated with CO2 regeneration k1 K1
 RNHþ
CO2 þ RNH2 ! 2 COO

(1)
is the large heat duty - the ratio between the rate of heat addition k1
and the rate of CO2 desorption required for CO2 desorption, a sol-
Carbamate formation and breakdown:
vent blend of MEA-MDEA was used to reduce and compare the
regeneration energy with the base case scenario of single MEA in k2 K2
the absence and presence of solid acid catalyst. Subsequently, the RNHþ 
 Bþ þ RNHCOO
2 COO þ B ! (2)
k2
focus is on increasing the rate of CO2 desorption for a fixed rate of
heat addition at a fixed temperature. In the case where the base in step 2 is a primary amine, the
The addition of a solid acid catalyst was initially reported by intermediate reacts to give a stable carbamate ion. The overall re-
Idem et al. [16] and Shi et al. [17] who conducted their experiments action is then represented as:
using two solid acid catalysts: (i) a Brønsted-proton donor acid
k3 K3
catalyst (HZSM-5) and (ii) a Lewis-electron acceptor acid catalyst  RNH þ
CO2 þ 2RNH2 ! 3 þ RNHCOO

(3)
(g-Al2O3), and compared their effects on CO2-desorption from CO2 k3

loaded MEA, MEA-MDEA and MEA-DEAB, in terms of reaction rate On the other hand, tertiary amines (R1 R2 R3 NÞ react directly with
and heat duty. Optimum results were seen for all scenarios with the CO2 in an hydrolysis reaction to produce bicarbonate (HCO 3 Þ and
best scenario obtained at 5:1.25 M MEA-DEAB molar ratio using protonated amine (R1 R2 R3 NH þ Þ
HZSM-5 catalyst. A CO2 production rate of 0.167 CO2/mol amine was
obtained which was higher than the base case of no catalyst at k4 K4
 HCO
R1 R2 R3 N þ H2 O þ CO2 ! 3 þ R1 R2 R3 NH
þ
0.123 mol CO2/mol amine, indicating a tremendous increase in CO2 k4
production rate in the presence of a catalyst. This approach has (4)
been seen to achieve a significant reduction in heat duty by directly
removing the heat of vaporization of water by operating at a tem- where RNH2 and R1 R2 R3 N represents MEA and MEA-MDEA,
perature lower than the boiling point of water [17]. Several authors respectively.
have confirmed this to be true [17,18] by performing similar ex-
periments using the same solid acid catalyst recommended by
2.1. Non-catalytic regeneration mechanism
Idem et al. [16]. The low heat duty can be explained by the presence
of alternative catalytic pathways with a lower activation energy, as
Regeneration of CO2 loaded amine solvent involves the appli-
well as the solvent effect, thus, directly reducing the required en-
cation of heat, during which CO2-bonded molecules such as
ergy input for the same amount of CO2 produced. The major limi-
carbamate and HCO 3 release free CO2 and free amine. These pro-
tation to their work is that the experiments were carried out in a
cesses can be characterized into three subsequent reactions:
batch reactor and so did not reflect the effects of other process
variables such as solvent flow rate, rich amine loading, tempera- k5 K5
tures, the weight of the catalyst, etc. that are typical of a continuous RNHþ
2 COO

þ Bþ !
 RNH2 þ CO2 þ B (5)
k5
flow full cycle CO2 capture process. This catalytic method is ex-
pected to grow rapidly in the coming years and implemented at the
k6 K6
pilot plant scale where they can be extensively studied with the HCO þ
3 þ B !
 CO2ðaqÞ þ H2 O þ B (6)
k6
final goal aimed at industrial scale operation. This would drastically
reduce the heat duty and plant vessel [19]. Nevertheless, in order to
k7 K7
commercially implement these technologies into an existing power HCO  CO2ðaqÞ þ OH 
3 ! (7)
plant, an understanding of the fundamental kinetic phenomena of k7

the acid gas reactions with single and mixed amines are essential
for the effective design of the absorber-desorber units. The present where B is any base present in the system and Bþ is the corre-
work studies the effect of two solid acid catalysts HZSM-5 and g- sponding conjugate acid. In this work, the possible bases and
Al2O3 in the desorption of CO2 from CO2-loaded MEA and MEA- conjugate acids are MEA; OH  ; H2 O; HCO 2
3 ; CO3 and MEAHþ ;
þ 
MDEA solutions using a continuous flow full cycle CO2 capture H2 O; H3 O ; H2 CO3 ; HCO3 , respectively.
process plant. The kinetic performance was further analyzed to The concentrations of H3 Oþ ; and H2 CO3 are relatively small and
show the individual effects of each variable on the reaction are therefore neglected. During thermal desorption, the reaction
mechanism and was categorized as (i) catalytic effect (ii) solvent (equations (5)e(7)) leans towards a spontaneous endothermic re-
effect and (iii) temperature effect. The results are expressed in action. In the absence of solid acid catalyst, this process requires
terms of the rate constant, frequency factor and activation energy. proton transfer and bond cleavage which only occurs (spontane-
ously) at elevated temperatures, ranging from 120 to 140  C, thus
2. Reaction mechanism the high heat duty.
As shown in Fig. 1, a rich solvent containing carbamate and bi-
The reaction between CO2 and amine solutions is generally carbonate ions are heated up to a temperature in the range of
represented using two known mechanisms. The zwitterion 120e140  C via low-pressure steam. This heat weakens the
A. Akachuku et al. / Energy 179 (2019) 475e489 477

Fig. 1. Non-catalytic carbamate breakdown.

hydrogen bond of the hydronium ion which then attacks the ni-
trogen in carbamate ion, weakening the NeC bond. The NeC bond
then breaks forming free CO2, amine and water molecule. This high
heat input associated with regeneration brings about the energy
penalty.

2.2. Catalytic reaction mechanism

The focus is to make CO2 desorption possible at a temperature


less than 100  C by altering the reaction mechanism through the
addition of a solid acid catalyst to the desorber system. The acidic
properties of g-Al2O3 and HZSM-5 catalysts are precisely desirable
in the CO2 desorption process because of their ability to accelerate
chemical reactions and reduce the heat input needed to deproto-
nate and breakdown protonated molecule and carbamate ions, Fig. 2. Lewis acidity.
respectively. In order to attain a temperature below 100  C, the
experimental setup (Fig. 5) was modified to include a hot water
heating medium as the presence of a reboiler would be considered et al. [16], H-ZSM-5 catalyst also transfers its available proton to
redundant in our process due to its inability to produce adequate bicarbonate ions, thus increasing the amount of desorbed CO2 and
vapour needed for heat and mass transfer in the desorber column at reducing the need for higher external heat supply. The Lewis acid
a temperature below 100  C. The absence of the reboiler indirectly site attacks the free lone pair of electron density found in the N
reduces the heat duty by spending less energy on water vapor- atom of the carbamate ion. It is important to mention that the
ization which is a major constituent of heat duty. negative aluminosilicate ion sites present on the surface of the
catalyst inhibits the OH  group from approaching the interface,
thus eliminating any possible competition between carbamate ion
2.2.1. g-Al2O3 catalytic activity
and the hydroxyl group [26].
The partially uncoordinated metal cations and oxide anions that
appear at the surface of g-Al2O3 catalyst enables it to act as an acid
and base, respectively [23]. In the basic region, the lone pair elec- 2.3. Specific rate of desorption calculation
tron density found in the N atom of the carbamate is attacked by
the cus cation ion-Al3þ (Lewis acid). This weakens the NeC bond, The desorption kinetics of CO2 from loaded aqueous MEA and
resulting in the dissociative chemisorption of the carbamate ion. On MEA-MDEA solutions were studied over g-Al2O3 and HZSM-5
the other hand, the oxide anion-AlO þ
2 attacks the H of any pro- catalysts as a function amine flow rates, catalyst weight and reac-
tonated molecule. It has also been reported [24] that at tempera- tion temperature in a tubular packed-bed reactor. Power law and
tures < 423 K, the reaction between free CO2 and OH  groups in g- mechanistic models were used to describe the experimental kinetic
Al2O3 results in the formation of HCO 3 . This is an important feature data for CO2 desorption. Kinetic performances were evaluated in
that tends to be significant at lean CO2-loading region. Shi et al. [17] terms of conversion (i.e. %CO2 desorbed), activation energy and rate
and Liang et al. [18] further demonstrated via energy diagrams that constants.
much less energy is needed when protonated amine transfers its The rate of carbamate and bicarbonate breakdown is given
proton to a relatively strong base like HCO 
3 (21.9 kJ at 90 C) than to below:
water (78.22 kJ at 90  C). A typical Lewis acidity is shown in Fig. 2 Carbamate breakdown reaction
[25]. h i h i
rCO2MEA ¼ RNHCOO *k2 Bþ (8)
2.2.2. HZSM-5 catalytic activity
The presence of both Brønsted and Lewis acid sites on the active Bicarbonate breakdown reaction
sites of H-ZSM-5 increases its role in the desorption of CO2. A h i h i
typical Brønsted acidity reaction is shown in Fig. 3. The Brønsted rHCO3 ¼ HCO
3 *k6 H3 O
þ
þ k7 (9)
acid site donates a proton to the carbamate ion. This proton
donation converts the carbamate ion to a carbamic acid It is difficult to know the exact amount of HCO
3 that contributes
(RNHCOOH). This is then followed by chemisorption on the Al site to the deprotonation of MEAHþ and the amount that reacts with
which results in the weakening of the NeC bond. According to Idem H3 Oþ : Also, it is quite impossible to know how much MEAHþ will
478 A. Akachuku et al. / Energy 179 (2019) 475e489

Fig. 3. Brønsted acidity.

react with OH  and H2 O as well as the fraction that contributes to desorption rate, the bulk concentration of all the chemicals species
carbamate breakdown; therefore, the power law model is written present in the desorption reaction are needed. However, during
in terms of product formed which incorporates the formation of absorption and desorption reactions, a fraction of the amine con-
CO2 from HCO 
3 and MEACOO , (i.e. rCO2 ¼ rMEACOO þ rHCO3 Þ.
 verts to protonated amine (MEAHþ =MDEAHþ ) and the remaining
CO2 production rate is written as: fraction reacts to form carbamate ion. CO2 will either react with
MEA to produce carbamate ion or undergo hydrolysis reaction in
rCO2amine ¼  kCO2 ½CO2  amine (10) the presence of tertiary amines. Carbamate ion comes only from
reaction of CO2 with primary or secondary amines, with little
quantities of bicarbonate and carbonates ions which are only pre-
sent at higher loadings. These ions can be grouped into three cat-
2.4. Kinetic model
egories: bicarbonate/carbonate (HCO 2
3 =CO3 ), carbamate
 þ
ðMEACOO Þ and protonated/free amine (MEAH /MEA). The exact
Kinetic models which describe catalytic activities and considers
concentration of these ions is difficult to obtain independently as
all reacting species and their concentrations were developed and
the reactions occur simultaneously. There are fast proton exchange
used to fit the experimental data.
reactions between the ions HCO 3 =CO3
2
and MEAHþ /MEA. This
2.4.1. Power law model makes it very difficult to obtain the actual mole fractions of theses
An empirical power-law model in terms of CO2 desorption rate ions using common analytical techniques such as titration, high
was based on the overall reaction with respect to CO2 and is rep- performance liquid chromatography (HPLC) and gas chromatog-
resented in the form: raphy (GC). More accurate results can be achieved with NMR
analysis technique, and, from simulated ion speciation plot gener-
 
E
ated from ASPEN e-NRTL system model. These concentrations were
RTA
obtained using the mass balance equation and compared to that
rco2MEA ¼ ko exp ½CO2  aminen (11)
obtained from the ion speciation plot generated from ASPEN NRTL
Equation (11) shows the general dependence and sensitivity of system model. Both methods produced comparable results.
desorption rate on the concentration of reacting species expressed MEA balance:
in the Arrhenius form.
h i h i
2.4.2. Mechanistic model ½MEA þ MEAHþ þ MEA  COO ¼ ½MEAo (24)
To further understand the catalytic desorption pathway, mech-
MDEA balance:
anistic models were derived for both H-ZSM-5 and g-Al2O3 cata-
lysts. The developed kinetic models were based on the specific roles
of each catalyst and these involved all chemical species taking part
h i
½MDEA þ MDEAHþ ¼ ½MDEAo (25)
in the chemical reactions.
According to literature [27], certain assumptions are made to Total carbon balance:
allow for the simplicity of the model development. These as-
sumptions are as follows: h i h i h i
½CO2  þ HCO
3 þ CO3
2
þ MEACOO
 Reactants are present in a single fluid phase  
 All surface sites have the same adsorption energy þ ½MEAo þ ½MDEAo a (26)
 There is no interaction between adsorbed molecules
Charge Balance:
The developed rate models, their derivation and assumptions
h i h i h i
are given in Table S1.
MEAHþ þ MDEAH þ þ H3 Oþ
h i h i h i h i
2.5. Ion speciation of major and important ions ¼ HCO 2
3 þ CO3 þ MEACOO þ OH  (27)

To obtain the mechanistic rate model which describes the CO2


A. Akachuku et al. / Energy 179 (2019) 475e489 479

3. Materials and methods used to possibly just hot water instead of steam for stripping, as
well as the amount of heat spent on water vaporization which,
3.1. Chemicals conventionally, is a major constituent of heat duty.
At the beginning of each run, 100% CO2 and N2 were set to the
MEA and 1.0 M hydrochloric acid (HCL) were obtained from desired partial pressures and concentrations and verified using CO2
Sigma Aldrich with purity  98%. Ultra-high purity (99.9%) CO2 and infrared gas analyzer (Nova Analytical Systems Inc.) before being
N2, and 15% CO2 gas cylinders were supplied from Praxair lnc., introduced into the bottom of the absorber column through the gas
Canada. g-Al2 O3 catalyst was purchased from Zeochem, USA while flow meter which controlled the gas flows individually. Cooling
HZSM-5 zeolite catalyst was purchased from Zibo Yinghe Chemical water circulated through the condenser from the cold-water bath.
Co., Ltd, Zibo City, China. All supplied chemicals were used without The amine solution from the feed tank was let into the top of the
further purification. column with the help of a constant liquid flow pump. This ensured
that the column operated under contact counter-current flow,
allowing CO2 to be absorbed by providing maximum driving force
3.2. Sample analysis
for mass transfer. The treated gas escaped through the top of the
column while the rich solution exited at the bottom of the column,
An aqueous solution of MEA was prepared by mixing concen-
after which it passed through the lean/rich heat exchanger and
trated MEA with distilled water to the preferred concentrations,
then a hot water heater before flowing into the desorber column.
which were further verified by titration technique using a standard
Here the desorber is at elevated temperature, allowing desorption
known volume of 1.0 kmol/m3 of hydrochloric acid (HCl), with
of CO2 to occur. After desorption, the desorbed CO2 escapes through
methyl orange as indicator. Using the standard method presented
the top of the column and the lean amine is recirculated back into
by the Association of Official Analytical Chemists (AOAC), the
the absorber.
Chittick apparatus was used to calculate the CO2 loading of the
Each run was operated until steady state conditions were ach-
amine solution whereas the concentration of CO2 in the gas phase
ieved. Depending on the conditions, this normally took approxi-
was measured along the length of the column using IR gas analyzer.
mately 3e4 h. At steady state, the concentrations of CO2 in the gas
phase and temperature profile were measured along the length of
3.3. Experimental apparatus and procedure the column using IR gas analyzer and thermocouples, respectively.
To ensure that the CO2 absorption rates calculated from the CO2
The experiments were performed in complete absorber e concentration profile were certified, the CO2 rich solution at the
desorber CO2 capture bench-scale plant unit as shown in Fig. 4. bottom of the column was simultaneously sampled and further
Both the absorber and desorber stainless steel columns had an used for analyzing the amount of absorbed CO2 in the liquid phase.
internal diameter of 0.05 m and 1.50 m total height. Both columns Using the Chittick apparatus, the CO2 content in the sampled liquid
were filled with packing material to a height of 0.96 m. The sche- were used to calculate the CO2 loading of the amine solution. Each
matic of the orientation of the packing materials and operating experimental run was verified by calculating the mass balance error
parameters are given in Fig. 4 and Table 1, respectively. The using the formula:
absorber column was packed with a 0.047 m outside diameter of  
stainless steel LDX Sulzer structured packing, which had a surface absorbed CO2  removed CO2
Mass balance error ¼  100%
area of 900 m2 =m3 and void fraction of 0.775. The desorber column absorbed CO2
was divided into five segments, with the top and bottom sections (28)
filled with stainless steel LDX structured packing. Inert marbles of
10 mm diameter were placed at the second and fourth layers An error of or less than 10% was regarded as an acceptable test
occupying a height of 0.051 m. The middle section of 0.51 m was run.
randomly mixed with 6 mm diameter marble and varying weight of
catalyst. These structured packing and inert beads were used to 4. Results and discussion
increase the surface area for liquid distribution, allowing for more
wettability. The inert marble beads acted as a support for the The experimental analyses expressed the CO2 desorption char-
catalyst bed and ensured that plug flow conditions were met, acteristics of a catalytic packed bed reactor employing aqueous
whilst reducing pressure drop. MEA and MEA-MDEA blends. The catalytic desorption performance
Experimental kinetic data for CO2 desorption were obtained of CO2 was evaluated on the basis of the CO2 desorption rate, CO2
from the catalytic packed bed tubular reactor (i.e. desorber) at three conversion, activation energy, and frequency factor. Various process
temperatures (348 K, 358 K, and 368 K), with 5 mol/dm3 MEA and parameters such as reaction temperature (345e365 K), solvent
5/2 mol/dm3 MEA-MDEA concentrations, and CO2 loading ranging flow rate (60e80 ml), and catalyst weight (0e250 g) were consid-
from ~ 0.347e0.5 mol CO2/mol amine, for different ratios of catalyst ered for estimating the desorption characteristics of both catalysts.
weight/amine flow rate (W/FAo). Thermocouples and concentration The effect of all specified variable on desorption rate have been
measurements were located along the length of the absorber and discussed and are presented separately.
desorber column at equal intervals of 0.15 m and were used to
obtain gas phase CO2 concentration and temperature profiles. 4.1. Accuracy in the measurement of experimental data
Solvent losses were prevented by positioning a condenser at the
top of the absorber and desorber columns. Several other auxiliary A repeatability test was carried out to ensure that the repro-
units were used in this study, namely, CO2 cylinder, gas flow meter, ductivity of the experimental results were attainable. The experi-
feed tank, and liquid flow pump, rich-lean heat exchanger, and ments were carried out under the following conditions: 15 SLPM
amine storage tank. The experimental setup used in the current gas inlet, 150 g g-Al2O3 catalyst, 80 ml/min liquid amine flowrate
work does not use a reboiler. In its place, we have a rich amine and an average temperature of 358 K. The repeatability results in
heater. The aim of the heater is to provide a heating medium with Table 2 shows that repeatability is highly achievable.
temperature less than the boiling point of water (<100  C). The Furthermore, the uncertainty in the CO2 loading measurements
absence of the reboiler enables the reduction of the quality of heat as well as its propagation in the calculation of CO2 conversion and
480 A. Akachuku et al. / Energy 179 (2019) 475e489

Fig. 4. Packing orientation and schematic of the experimental Setup.


A. Akachuku et al. / Energy 179 (2019) 475e489 481

Fig. 5. Effect of MEA flowrate (50, 60, 70, 80 ml/min) on Conversion, X, at different catalyst weight (0e300 g) at 358 K for H-ZSM-5 and g-Al2O3.

Table 1 Table 3
Operating parameters of the desorber. Measurements for the lean loading, mol CO2/mol MEA.

Desorber operating condition Runs Lean loading, mol CO2/mol MEA Deviation from the mean

Inlet gas Composition 15% CO2, 85% N2 1 0.4216 0.0061


Inlet gas flowrate, kmol =m2 h 18.5 2 0.4294 0.0017
Amine circulation rate (mol/min) 60.00 3 0.4320 0.0043
MEA concentration (mol/dm3) 5 Average 0.4277 0.0040
MDEA concentration (mol/dm3) 2
Average Desorber temperature (K) 345e365
Desorber Pressure (atm) 1
measuring the concentration of the amine had an accuracy of
±0.05 mL. These values were used to obtain CO2 conversion of
16.000 and 17.694 respectively. Due to the propagated error in
desorption rate were calculated and quantified (Table 3). The CO2
loading and amine concentration, the uncertainty for each con-
loading measurements were carried out 3 times using a Chittick
version is given as ± 10%.
CO2 apparatus and the average deviations as well as their mean
deviation was obtained and averaged. Table 3 shows that the
loading measurement has a deviation of 0.004.
4.2. Catalyst characterization
The conversion was calculated from the lean and rich loadings
as shown in equation (29).
The BET surface area, pore volume and average pore size
measured for g-Al2O3 and H-ZSM-5 catalyst are presented in
CO2;in  CO2;out Table S2. It is well known in the literature that physical character-
XCO2 ¼ X 100% (29)
CO2;in istics alone cannot be used to evaluate catalyst performance. Other
critical chemical characteristics are needed which include the
where XCO2 is the conversion, CO2;in is the concentration of CO2 in Brønsted/Lewis acid ratio and the acid site strength. Fourier
the amine entering the desorber and CO2;out is the amount of CO2 in Transformed Infrared Spectroscopy (FTIR) experiments were car-
the amine leaving the desorber. The concentration of CO2 in the ried out to quantify the amount and ratio of Lewis (L) and Brønsted
amine was calculated from the CO2 loading. The result is shown in (B) acid sites in both H-ZSM-5 and g-Al2 O3 catalysts. The obtained
Table 4. Taking a base case loading of 0.5 mol CO2/mol amine and a results indicated the presence of Brønsted and Lewis acid sites in H-
lean loading of 0.42, the maximum and minimum deviations were ZSM-5, while g-Al2O3 had just Lewis acid sites. Similar trends were
obtained by increasing and decreasing the loading by 0.0040 to observed in the works of Pathak et al. [28] and other works re-
obtain 0.504 and 0.416 respectively. Also, the burette used in ported in the literature [28e33].

Table 2
Repeatability results for the experimental runs.

Runs Lean loading, mol CO2/mol MEA Rich loading, mol CO2/mol MEA CO2 desorbed, kg/h Mass balance error, %

1 0.43 0.49 0.0664 0.79


2 0.44 0.50 0.0672 1.80
3 0.43 0.50 0.0717 2.41
482 A. Akachuku et al. / Energy 179 (2019) 475e489

Table 4
Error Propagation in the calculation of CO2 conversion.

Runs Concentration, CO2 in, mol/dm3 Concentration CO2 out, mol/dm3 Conversion, XCO2

Base Case 2.500 2.100 16.000


At maximum deviation 2.515 2.070 17.694

4.3. Kinetic study


0
rA;obs pc R2c
4.3.1. Evaluation of possible heat and mass transport limitations Cwp;ipd ¼ (33)
Deff CAS
The possible effect of heat and mass limitations on the catalytic
reaction rates were examined at the highest reaction temperature Cwp;ipd is defined as the Weisz-Prater criterion for internal pore
368 K as reported in the literature [34].
diffusion, rc is the pellet density, Rc is the obtained values for g-
Al2O3 and HZSM-5 were far less than 1, indicating that the con-
centration at the catalysts surface is almost the same as that in the
catalyst pore. These results indicate the absence of internal mass
4.3.1.1. Heat transport effects. Internal pore heat transfer effect was
transfer diffusion [36]. The effect of film mass transfer limitations
estimated using the Prate analysis (equation (30))
was determined by estimating the ratio of observed rate to the rate
of film resistance. Equation illustrates this criterion:
Deff ðCAS  CAC ÞDHrxn
DTparticle;max ¼ (30)
leff 0
r A;obs db
observed rate
¼ (34)
rate if film transfer controls Kc CAs 6
where DTparticle;max is the upper limit to temperature variation be-
tween pellet centre and pellet surface, and DHrxn is the heat of The obtained value for the given ratio indicates that film transfer
reaction, CAS and CAC are the concentrations at the pellet surface rate is less than the observed rate. Therefore, the film resistance
and centre [35], and Deff is the effective mass diffusivity [36]. A should not influence the reaction rate [35]. A more rigorous set of
value of 3.4 K was obtained. The correlation adapted from Ibrahim criteria was applied to ensure the absence of mass transport limi-
and Idem [34] was used to account for the heat-transfer limitation tation in the film. This Mears’ criterion [36] considers the mass
across the gas film as shown in equation 31 transfer limitations during kinetic data collection and given as:

  0
Lc  rA;obs DHrxn r A;obs rb Rc n
DTfilm;max ¼ (31) < 0:15 (35)
h Kc CA

where DTfilm;max is the upper limit temperature difference between The summary of mass transfer limitations is given in Table S4.
the pellet surface and the gas phase, Lc is the characteristic length, The results show the absence of internal and external mass transfer
rA;obs is the observed reaction rate, and h is the heat-transfer co- limitations. Also, to ensure plug flow conditions, absence of back
  mixing and channeling certain relevant criterion [39,40] were
efficient estimated from the correlation JH ¼ JD ¼ Cphmr Npr 2=3 , adopted. These criteria include:
where JH is the heat-transfer J factor, Npr ¼ Cp m =l, NPr is Prandtl
Catalyst bed height L
number and l is the molecular thermal conductivity). The JD factor ð1Þ ¼  50 (36)
Catalyst Particle size dp
is given by the following correlations: JD ¼ (0.4548/εP)
d ur
NRe 0:4069 ð40Þ ¼ ðkc =vÞNsc
2=3
[37]. Where NRe ¼ mð1εÞ
p
: kc is the
5 Catalyst bed diameter d
mass-transfer coefficient obtained as 2.44  10 m/s. The heat ð2Þ ¼  10 (37)
Catalyst Particle size dp
transfer coefficient, h; was determined to be 1338:18 kJm2 s1 K 1 .
A value of 7:8 x 103 K was obtained for DTfilm;max. In this work, both requirements were met, indicating that the
A more rigorous criterion which determines the onset of the system operates under plug flow conditions as shown in Table S5.
heat-transport limitation during reaction [38] was also used to
further ascertain the insignificance of heat-transfer resistance in
the rate of the reaction, as shown in equation (32):
4.4. Effect of various parameters on the catalyst performance
rA;obs rb Rc EDHrxn
< 0:15 (32) In this work, the catalytic activity was evaluated in terms of
hT2 R
carbamate and bicarbonate conversion to CO2 and the rate at which
By substituting the values for the terms on the left-hand side the conversion occurred. Furthermore, to clearly understand the
(LHS) of equation 32, a value which is much less than 0.15 was catalytic role in the desorption mechanism, the kinetic perfor-
obtained for both catalysts, thereby proving the absence of any mance of the non-catalytic and catalytic reactions were further
heat-transport limitation. These values for DTfilm;max and Mears expressed in terms of activation energy, EA, and frequency factor A.
criterion are reported in Table S3. The conversion is expressed as:
In the absence of heat and mass limitations, intrinsic experi-
mental rates were calculated using the equal area differentiation
technique which was obtained from the plots of conversion, X (i.e.
4.3.1.2. Mass transfer effects. The internal pore mass transfer CO2 desorbed) vs. weight of catalyst per amine flow rate
resistance was calculated using WeiszePrater criterion as given: ðW =FAmine Þ:
A. Akachuku et al. / Energy 179 (2019) 475e489 483

anion-AlO 2 then deprotonates the protonated amines by attacking


dXCO2 the H þ found in MEAHþ =MDEAHþ . On the other hand, the lone pair
rA ¼ (38)
dðW=FAmine Þ electron density found in the N atom of the carbamate is attacked
by the cus cation ion-Al3þ (Lewis acid). This weakens the NeC bond,
resulting in the dissociative chemisorption of the carbamate ion.

4.4.1. Effect of feed flow rate


The liquid flowrate is a crucial factor to consider while calcu- 4.4.2.2. Effect of HZSM-5 catalyst. Fig. 9 shows the effect of H-ZSM-
lating specific desorption rate, as it plays a leading role in the 5 catalyst on conversion. The presence of Brønsted acid catalyst,
desorption process. It is also important that an optimum flow rate HZSM-5, expedites the rate of desorption by readily providing
be identified to obtain intrinsic kinetic which could only be ach- protons for carbamate breakdown. According to Shi et al. [17], the
ieved in the absence of heat and mass transfer limitations. Amine initial step in CO2 desorption is the deprotonation of pronated
flowrate was varied while keeping W =FAmine constant. The char- amine, MEAHþ. This simple but yet energy intensive step requires
acteristic effect of the liquid flowrate on conversion, X, is presented 73.4 kJ/mol at 25  C and 78.2 kJ/mol at 90  C free energy and
in Fig. 5. It is to be expected that conversion would increase with provides free protons for carbamate breakdown. As started earlier,
increase in amine flowrate, as there is more contact between the the aim of catalytic desorption is to reduce the parasitic energy
liquid and solid acid catalyst, i.e. increase in interfacial (wetted) associated with amine regeneration. Proton donating catalyst,
area via higher interactions, thus increasing the desorption rate and HZSM-5, greatly increases desorption rate by directly providing
mass transfer coefficient. However, such trend was not seen as the protons for carbamate breakdown. This implies that carbamate
flowrate was varied from 50 to 80 ml/min (Fig. 5). breakdown occurs without waiting to receive protons from
This can be explained by the absorber height limitation at MEAHþ. A 17.2% jump is seen when 50 g catalyst is added relative
increased flowrate. Equilibrium loading for 5 mol/dm3 MEA is to the non-catalytic reaction and rises to 74.4% at 250 g catalyst. The
achieved at 0.5 mol amine/mol CO2. However, with an absorber donated proton from HZSM-5 active site transforms the carbamate
height of 1.067 m, a loading range of 0.476e0.489 mol amine/mol ion into carbamic acid, this is followed by chemisorption of car-
CO2 were obtained at 70 and 80 mol/min, indicating that equilib- bamic acid via bonding of the O and Al atoms, which results in the
rium loading was not achieved due to the limited time spent in the weakening of the NeC bond. According to Idem et al. [16], HZSM-5
absorber. At 50 ml/min only a fraction of the catalyst surface was catalyst also transfers its available proton to bicarbonate ions, thus
utilized. The amine flowrate was too low and introduced mass increasing the amount of desorbed CO2. The Lewis acid site attacks
transfer limitation. Optimum flowrate which exploits most of the the free lone pair of electron density found in the N atom of the
catalyst surface area was achieved at 60 ml/min. This could be seen carbamate ion and the H atoms attached to the O atom moves to the
as maximum conversion was achieved for both catalysts at this rate. adjacent N atom. This transforms the carbamic acid into Zwitterion
ion which further dissociates into free MEA and CO2. Nevertheless,
4.4.2. Effect of catalyst type to accelerate the industrial deployment of this thermodynamically
Conversion of CO2-bonded molecules into free CO2 comprises of efficient catalytic post combustion CO2 capture process, the recy-
the combined effect of the catalyst acid strength, solvent flowrate clability of these catalyst materials and their direct effect in the
and regeneration temperature. The alternative pathway introduced absorber needs to be clearly demonstrated. Recent works [41,42]
by the presence of g-Al2O3 and HZSM-5 catalyst in the desorber showed that the absorber height limitations encountered at vary-
system, makes desorption feasible at faster reaction rates at tem- ing amine flowrates where overcome by addition of base catalyst.
perature as low as 75  C. The base catalyst increased the absorption rate and conversion and
also allowed for equilibrium loading in the absorber.
4.4.2.1. Effect of g-Al2O3 catalyst. The effect of g-Al2O3 catalyst on
conversion can be seen in Fig. 8 for both solvents. Three scenarios 4.4.3. Effect of the ðW =F Amine Þ ratio and temperature
were examined with the average desorber temperature fixed at Catalyst weight and temperature are vital process variables
345, 355 and 365 K and catalyst weight varied from 0 to 250 g with needed for CO2 desorption in a catalytic packed bed reactor. Figs. 6
an increment of 50 g per interval. As shown in Fig. 8, the conversion and 7 respectively for MEA and MEA-MDEA show the influence of
of CO2-bonded molecules to free CO2 increases as the catalyst temperature and weight of catalyst on conversion. Keeping the
weight is increased for both MEA and MEA-MDEA solvents. flowrate fixed at 60 ml/min, the weight of catalyst was varied from
Comparative to the non-catalytic process a 55.8% increase in con- 0 g to 250 g, for all three temperatures (348, 358 and 368 K) and
version is seen at 250 g catalyst. This percentage increase can be solvent types (MEA and MEA-MDEA).
explained by both the physical and chemical properties of the The temperature effect can be seen at 0 g catalyst weight. At
catalyst. Since catalytic desorption occurs at the catalyst active site, 75  C (348 K), the influence of the temperature effect on the reac-
it is important that the pore volume and pore size have enough tion is minimal. Desorption reaction is not kinetically favoured as
space to easily accommodate the reactant molecules, thus elimi- the average number of reactant molecules with kinetic energy
nating limitations in the mass transfer and diffusion steps. As given higher than the energy barrier is small. The catalyst activity dom-
in Table S2, the pore volume and pore size of g-Al2O3, allows the inates at this temperature (348 K) and we see a slight increase in
ease of reactant diffusion within the catalyst. This helps to accel- conversion as we increase catalyst weight. This can be explained by
erate the desorption process in the absence of mass transfer limi- the presence of an alternate catalytic pathway with lower energy
tation. Also, the BET surface area of the catalyst introduces an barrier, which increases the number of reactant molecules with
additional interfacial area which increases the mass transfer coef- kinetic energy able to exceed the barrier and transition into free
ficient and in turn the rate of desorption. The chemical property can CO2, thus increasing the rate of CO2 desorption. The chemical ac-
be explained by the amphoteric nature of g-Al2O3 which allows it to tivity which is of uttermost important at unfavorable kinetic tem-
react in both basic and acidic environment. The CO2-loaded amine perature is the acid strength. The temperature programmed
leaving the bottom of the absorber has a pH of 7e8. This creates a desorption (TPD) result (Fig. S1) showed that relative to HZSM-5, g-
basic environment where the surface of g-Al2O3 is negatively Al2O3 catalyst has high intensity strong acid site peak at 430  C
charged, resulting in the formation of oxide anion, AlO 2 : The oxide which contributes to its efficiency. This explains the closeness in
484 A. Akachuku et al. / Energy 179 (2019) 475e489

Fig. 6. Variation of catalyst weight (0e200 g) and temperature (348 K, 358 K, and 368 K) at constant MEA flowrate 60 ml/min.

Fig. 7. Variation of catalyst weight (0e200 g) and temperature (348 K, 358 K, and 368 K) at constant MEA-MDEA flow rate 60 ml/min.

performance of both catalyst at this temperature even though increase in temperature also increases the frequency of collision of
HZSM-5 clearly has higher chemical activity. reactant molecule which is reflected in the pre-exponential factor
The reaction becomes more spontaneous and kinetically with a value of 3:4  106 dm3:sec 1: g: catalyst  1 and 2:2  105
favourable with temperature increase from 75  C (348 K) to 85  C dm3:sec 1: g: catalyst  1 obtained for g-Al2O3 and H-ZSM-5
(358 K). At 0 g catalyst the percentage conversions for g-Al2O3 and respectively (Table S6).
HZSM-5 were calculated to be 81% and 83% respectively. The effect At 368 K (95  C), the temperature effect is more noticeable. This
of both catalyst and temperature is seen across the slope with a 50 g indirectly reduces the influence of the catalyst, making it to be less
catalyst increment up to a maximum of 250 g. As described by the effective and minimal on the reaction rate. This is so because at
Maxwell Boltzmann theory, increase in temperature and catalyst higher temperature, the fraction of kinetically active reactant
weight increases the average number of molecules with higher molecules increases. For MEA system, we see an increase of 32% at
kinetic energy, thus allowing the reaction to occur at a faster rate 0 g catalyst and a 36% for MEA-MDEA. This difference in conversion
and lower activation energy. Taking the base case at 85  C 0 g which is due solvent effect is also expected to occur in a large scale
catalyst, the combined effect for MEA system shows a 55% and 74% CO2 capture plant. From Fig. 8, it is seen that the conversions for
increase in conversion for g-Al2O3 and for HZSM-5 respectively and both g-Al2O3 and HZSM-5 are comparable. This can be explained by
a 65% (g-Al2O3) and 85.2% (HZSM-5) for MEA-MDEA system. An the temperature effect. At 365 K, most of the protons attached to
A. Akachuku et al. / Energy 179 (2019) 475e489 485

Fig. 8. Comparison of percentage increase in conversion-MEA and MEA-MDEA system g-Al2O3. Catalyst weight (0e200 g) and temperature (348 K, 358 K, and 368 K) at constant
amine flow rate (60 ml/min).

Fig. 9. Comparison of percentage increase in conversion for MEA and MEA-MDEA system H-ZSM-5 Catalyst weight (0e200 g) and temperature (348 K, 358 K, and 368 K) at constant
amine flow rate (60 ml/min).

MDEAHþ are easily deprotonated and available to assist in the seen at the point of 0 g catalyst for all temperature. Figs. 8 and 9
deprotonation of CO2, thus reducing the effect of both catalysts. show that the CO2 conversion in MEA-MDEA system is higher
than that of MEA for all cases. At 0 g catalyst, an average increase of
4.5. Effect of solvent 20.6% in conversion for MEA-MDEA relative to MEA was obtained
for all temperatures. The combined effect of solvent and catalyst is
Solvent type has a significant impact on CO2 desorption rate. seen in the slope as catalyst weight increases. At 200 g catalyst the
Figs. 8 and 9 show the comparison in percentage increase in con- combined effect is 20.9, 22.3 and 22.9% increase in conversion for g-
version for MEA and MEA-MDEA solvents. Solvent effect can be Al2O3 at 348, 358 and 368 K and 28.6, 30.4 and 31.2% increase in
486 A. Akachuku et al. / Energy 179 (2019) 475e489

conversion for H-ZSM-5 catalyst, respectively. According to Shi


et al. [17], tertiary amines such as MDEA helps to split the energy,
DGo , of one difficult step (equation (39)) into two less difficult steps,
DG1 þ DG2 (40) and (41). MDEA being a stronger base than water
easily accepts proton from MDEAHþ than H2O. This pathway
significantly accelerates the overall reaction rate and reduces the
energy penalty, since less energy is wasted on vaporization [43].

MEAHþ þ H2 O4MEA þ H3 Oþ DGo (39)

MEAHþ þ MDEA4MEA þ MDEAHþ DG1 (40)

MDEAHþ þ H2 O4MDEA þ H3 Oþ DG2 (41)


In MEA system at a loading of 0.25 mol/mol MEA bicarbonate
ions, HCO3 is exhausted, increasing the energy barrier from 21.9 to
78.2 kJ/mol [17]. The presence of tertiary amine MDEA allows the
provision of bicarbonate ions, HCO 3.
HCO 3 has two key roles, it either accepts protons from
MEAHþ =MDEAH þ equations (42) and (43) or further splits into H2 O
and CO2 equation (44) Fig. 11. Arrhenius plot (CO2 þ MEA þ MDEA þ H2O system).

MEAHþ þ HCO 
3 4MEA þ H2 CO3 (42)
g-Al2O3 and HZSM-5 catalyst increases the rate of desorption of
CO2 either by increasing the frequency of collision between the
MDEAH þ þ
HCO
3 4MDEA þ H2 CO
3 (43)
reacting molecules which is represented by the pre-exponential
factor ko, or by providing an alternative pathway with lower acti-
H2 CO
3 4 CO2ðaqÞ þ H2 O (44) vation energy, EA . The order of reaction, n, which shows to what
extent the rate of reaction depends on reactant concentration was 1
HCO3 being a stronger base than water easily accept protons from overall for both catalysts and solvent type. The pre-exponential
MEAHþ =MDEAH þ . This proton acceptor role reduces the free en- factor, ko , of g-Al2O3 was seen to be higher than that of HZSM-5.
ergy of regeneration from 73.4 kJ/mol to 21.9 kJ/mol [17]. This can be explained by physical property of the catalyst. g-
Al2O3 has larger pore size and volume than HZSM-5 (Table S2). This
4.6. Estimation of the values of the parameters of the rate models allows molecules to collide more frequently thus increasing the
and validation pre-exponential factor. On the other hand, the activation of HZSM-5
was seen to be lower than that of g-Al2O3 for both solvents. This can
The Power law and mechanistic model parameters were based be explained by the chemical activity of the catalyst. The Brønsted
on the minimization of the sum of the residual squares of the re- to Lewis acid site ratio of HZSM-5 was seen to be higher than that of
action rates by the Gauss-Newton and Levenberg-Marquardt al- g-Al2O3, with a ratio of 1.587 for HZSM-5 and 0.667 for g-Al2O3
gorithms by applying the nonlinear regression software (NLREG). [44,45]. The Brønsted acid sites provided an alternative pathway
The activation energies were obtained from Arrhenius equation by with lower activation energy compared to the Lewis acid sites,
plotting ln k vs. (1/T) as is shown in Figs. 10 and 11. The obtained thereby increasing the fraction of molecules with kinetic energy
parameters are presented in Tables S6 and S7. higher than the activation energy. HZSM-5 achieves this potentially
by directly transferring its available proton to the carbamate and
bicarbonate ions, thereby decreasing the amount of energy
required to desorb CO2. Another contributing factor that greatly
affects the kinetic parameters is the solvent type. MDEA
ðDGo ¼ 7:2kJ=molÞ being a weaker base than MEA requires less
energy to deprotonate [17]. Consequently, alongside the catalyst
contribution, the solvent effect also plays a significant role in
increasing the conversion and reducing the activation energy.
These major findings from the experimental runs show that the
overall kinetics performance is greatly affected by the catalyst and
solvent type used in the desorption process.
All models were validated by determining the absolute average
deviation (AAD %) between the experimental rate and predicted
rate obtained from the proposed models. Also, a parity chart which
depicts how well the predicted rates fit the experimental rate data
was plotted. As shown in Fig. 12, the experimental and predicted
rates are in close correlation. Furthermore, a restriction that an
accepted model should have an AAD%  15% and the corresponding
activation energy should be closer in value to that of the power law
model was applied to eliminate unrealistic models.
Even though all kinetic parameters were positive and fitted the
Fig. 10. Arrhenius plot (CO2þMEAþH2 O system). data very well, they were further subjected to thermodynamic
A. Akachuku et al. / Energy 179 (2019) 475e489 487

Fig. 12. Parity chart for all developed models.

scrutiny to determine the true models. This was achieved by on the Lewis acidic (Al3 þ ) and basic (AlO 2 Þ sites and the acidic
ensuring that the rate constant, k and equilibrium constants, K of sites of H-ZSM-5. After carefully analysing the experimental data in
each proposed models were thermodynamically consistent. Ther- the absence of heat and mass transfer, single and dual site LHHW
modynamic equilibrium constants for each proposed model were models were proposed to describe the desorption of CO2 in the
regressed at each temperature (348, 358, and 368 K) and the ob- presence of g-Al2O3 and H-ZSM-5.
tained parameters were further used to obtain the enthalpy, en-
tropy and Gibbs free energy. The variation in equilibrium constants
with temperature can be easily explained by Le Chatelier's principle 4.7.1. Dual site mechanism involving Lewis acid (Al3 þ ) and basic
which shows that for an endothermic reaction the equilibrium ðAlO2 Þsite
constant increases with increase in temperature. The presence of both cation ðAl3 þ ) and oxide anion (AlO 2Þ
The Van't Hoff equation was used to calculate the enthalpy and active sites in g  Al2 O3 allows the AlO
2 ion to attack and withdraw
entropy of reaction as in Equation (45). the proton, Hþ found in MEAHþ :It assumed that the AlO 2 is a
stronger base and therefore reduces the energy required to
DH DS deprotonate MEAHþ . Also the Al3þ cation attacks MEACOO ,
lnKeq ¼  þ (45)
RT R weakening the NeC bond. This results in dual site mechanism. It is
The Van't Hoff plot, which is a plot of lnKeq Vs 1 =T gives a linear proposed that MEAHþ and MEACOO are initially chemisorbed on
the catalyst surface followed by the transfer of the deprotonated
relationship, with the slope given as DRH and the intercept as. DRS
proton from the catalyst active site to HCO 
3 and/or MEACOO . The
The Gibbs free energy at each temperature was obtained using
attached free amine (AlO2 MEA Þ then desorbs from the active.
the Gibbs Equation: DG ¼  RTlnKeq . The equilibrium constant,
Another implicit assumption was that the chemical reversibility of
Keq increases with increasing temperature. A negative slope in-
the reaction between adsorbed MEAHþ and MEACOO is negligible
dicates an endothermic reaction, making the enthalpy and entropy due to the large equilibrium constant (K2) obtained at the highest
of reaction positive (i.e. DH > 0 and DS > 0). The positive value DH operating temperature.
shows it to be an endothermic reaction and absorbs a lot of heat. Adsorption of carbamate ion on (Al3 þ ) active site
DS > 0 indicates an increase in the degree of randomness of the
system as the temperatures rises. On the other hand, the negative MEACOO þ ðSÞ4MEACOO ðSÞ (46)
values of DG indicates a spontaneity of the process at an elevated
temperature. It can be seen that DG becomes more negative as Adsorption of protonated MEA on ðAlO
2Þ active site
temperature increases, indicating that CO2 desorption is more
MEAHþ þ ðSÞ4MEAHþ ðSÞ (47)
spontaneous at higher temperature illustrating the strong tem-
perature effect of the reaction as compared to catalyst effect at Surface reaction between adsorbed protonated MEA and
higher temperatures. Thermodynamic consistency was only adsorbed carbamate ion.
observed for models M3 g-Al2O3 and M5 HZSM-5 for MEA system
and M11 HZSM-5 for MEA-MDEA system as presented in Table S8. MEAHþ ðsÞ þ MEA  COO
ðsÞ 4MEAðsÞ þ CO2 þ MEAðsÞ (48)

Desorption of MEA from the catalyst active site.


4.7. Mechanisms involved in the reaction
2 MEAðsÞ 4 2MEA þ 2ðSÞ (49)
The catalyst characteristics served as a guideline for developing
the mechanistic models. LHHW formulation was used to describe Each elementary reaction was assumed to be the rate deter-
the surface reactions of MEAH þ , MDEAH þ , MEACOO and HCO 3 mining step (RDS) and rate expressions were developed
488 A. Akachuku et al. / Energy 179 (2019) 475e489

accordingly. Notations

4.8. Industrial relevance of catalytic desorption ½A Free amine concentration, mol/dm3
½B Bicarbonate concentration, mol/dm3
The use of catalyst in the desorber decreases the desorber size ½C Carbamate concentration, mol/dm3
and the heat duty requirement of which are major issues in amine- ½Hþ  Protonated amine concentration, mol/dm3
based CO2 capture. The current study does not directly involve AAD average absolute deviation
solvent degradation/emissions and absorber size. However, the use Cwp;ipd WeiszePrater criterion for internal pore diffusion
of catalysts enables the reduction of the industrial temperature d internal diameter of reactor, m
range used in the regeneration of CO2 from 120 to 140  C to below dp diameter of particle, mm
100  C. Also, it is well known that the rate of degradation increases D diffusivity coefficient, m2 s1
with temperature. Therefore, a decrease in the working tempera- EA activation energy, J mol1
ture is expected to reduce the degradation rate. Thus, low opera- FAmine Molar flow rate of species mol min1
tional temperature allows for negligible thermal and oxidative kc mass transfer coefficient, m2 s1
degradation in the desorber thus mitigating another major chal- k0 pre-exponential or collision factor
lenge faced in post combustion CO2 capture. Ki adsorption constant for the species (i ¼ 1,2,3,4)
L length of catalyst bed, m
P pressure, atm
5. Conclusions
r radius of the catalyst bed, m
ri rate of reaction based on a particular species, mol
 It has been demonstrated in this work that the addition of solid
gcat1 min1 (i ¼ A, B, C, …)
acid catalysts HZSM-5 and g-Al2O3 to the desorber unit at
R universal gas constant, kJ kmol1 K1
temperatures below 100  C increases CO2 conversion and re-
Rc radius of catalyst particle, m
duces the energy demand associated with CO2 desorption.
T temperature, K
 The presence of a rich-lean heat exchanger was used to attain a
W weight of catalyst, g
temperature below the boiling point of water hence, removing
ðW=FAmine Þ Contact time, min
the heat of vaporization of water associated with solvent
Xi conversion of component i
regeneration.
 The parasitic heat duty associated with the regeneration of 5 M
MEA is as a result of the difficulty in the deprotonation of Greek Letters
MEAHþ of water and the lack of protons. The g-Al2O3 weakens DH Enthalpy of reaction, KJ kmol1
the NeC bond present in the carbamate ion, resulting in the DS Entropy of reaction, KJ kmol1
dissociative chemisorption of the carbamate ion, while the ox- DG Gibbs free energy of reaction, KJ kmol1
ide anion - AlO þ þ
2 deprotonates MEAH =MDEAH by attacking ε Porosity
þ þ þ
H found in MEAH =MDEAH . On the other hand, HZSM-5
l Thermal conductivity KJ m1 s1 K 1
having both Lewis and Brønsted acid sites acts as both a pro-
m Viscosity
ton donor and receiver, thus having the ability to assist in
rb; rc Density, kg m3
MEAHþ deprotonation and carbamate breakdown. Also, the
available proton in HZSM-5 is transferred to bicarbonate ions, t Toutorsity factor
thus increasing the amount of desorbed CO2
 Optimum flowrate was achieved at 60 ml/min, with a 32% in- Superscripts
crease in conversion when temperature increased from 358 K n reaction order with respect to CO2
(85  C) to 368 K (95  C) for MEA system and 36% increase for inlet entering the reactor
MEA-MDEA system with a percent error of ±10%. outlet exiting the reactor
 The reaction kinetics is accelerated in the presence of H-ZSM-5
catalyst in comparison to g-Al2O3 catalyst. It was seen that Superscripts
HZSM-5 catalyst increased the reaction kinetics by decreasing b Bulk
the activation energy whereas g-Al2O3 increased the kinetics by c Catalyst
increasing the frequency factor. p Pellet or particle
 The activation energy decreased in the order: MEA-MDEA with eff Effective
H-ZSM-5(5:1 x 104 J =mol) < MEA-MDEA with g- ipd Internal pore diffusion
Al2O3 < (5:6 x 104 J =mol) < MEA with H-ZSM-5 obs Observed
(7:1 x 104 J =mol) < MEA with g-Al2O3 (8:0x 104 J =mol ). C Carbamate
 Comprehensive mechanistic rate models involving single and H Protonated MEA
dual site mechanism based on LHHW model was found to best A Free MEA
describe the experimental rates with an AAD <15%, were
obtained Appendix A. Supplementary data

Acknowledgements Supplementary data to this article can be found online at


https://doi.org/10.1016/j.energy.2019.04.174.
This work was supported by the Natural Sciences and Engi-
neering Research Council of Canada (NSERC), Canada Foundation References
for Innovation (CFI), Clean Energy Technologies Research Institute
[1] Rhodes CJ. The 2015 Paris climate change conference: COP21. Sci Prog
(CETRI), and Faculty of Graduate Studies and Research (FGSR), 2016;99(1):97e104.
University of Regina. [2] David J, Herzog H. The cost of carbon capture. 2000. p. 13e6.
A. Akachuku et al. / Energy 179 (2019) 475e489 489

[3] Rao AB, Rubin ES. A technical, economic, and environmental assessment of gamma.-alumina at 100-300. deg. J Phys Chem 1975;79(13):1280e4.
amine-based CO2 capture technology for power plant greenhouse gas control. [25] Campbell IM. Biomass, catalysts and liquid fuels. 1983.
Environ Sci Technol 2002;36(20):4467e75. [26] Trueba M, Trasatti SP. g-Alumina as a support for catalysts: a review of
[4] Finotello A, Bara JE, Camper D, Noble RD. Room-temperature ionic liquids: fundamental aspects. Eur J Inorg Chem 2005;(17):3393e403.
temperature dependence of gas solubility selectivity. Ind Eng Chem Res [27] Davis ME, Davis RJ. Fundamentals of chemical reaction engineering. 2012.
2008;47(10):3453e9. [28] Pathak K, Reddy KM, Bakhshi N, Dalai A. Catalytic conversion of glycerol to
[5] Heldebrant DJ, Yonker CR, Jessop PG, Phan L. CO2-binding organic liquids (CO2 value added liquid products. Appl Catal A Gen 2010;372(2):224e38.
BOLs) for post-combustion CO2 capture. Energy Procedia 2009;1(1):1187e95. [29] Anand R, Maheswari R, Gore K, Khaire S, Chumbhale V. Isopropylation of
[6] Goel M, Sudhakar M, Shahi R. Carbon capture, storage and utilization: a naphthalene over modified faujasites: effect of steaming temperature on ac-
possible climate change solution for energy industry. 2019. tivity and selectivity. Appl Catal A Gen 2003;249(2):265e72.
[7] Metz B, Davidson O, De Coninck H, Loos M, Meyer L, IPCC. IPCC special report [30] Adjaye JD, Katikaneni SP, Bakhshi NN. Catalytic conversion of a biofuel to
on carbon dioxide capture and storage. Prepared by Working Group III of the hydrocarbons: effect of mixtures of HZSM-5 and silica-alumina catalysts on
Intergovernmental panel on climate change442; 2005. Cambridge, United product distribution. Fuel Process Technol 1996;48(2):115e43.
Kingdom and New York, NY, USA. [31] Plyuto YV, Babich IV, Sharanda LF, De Wit AM, Mol JC. Thermolysis of Ru (acac)
[8] Rochelle GT. Amine scrubbing for CO2 capture. Science 2009;325(5948): 3 supported on silica and alumina. Thermochim Acta 1999;335(1):87e91.
1652e4. Sep. 25. [32] Dabbagh HA, Yalfani M, Davis BH. An XRD and Fourier-transformed infrared
[9] Kierzkowska-Pawlak H, Siemieniec M, Chacuk A. Investigation of CO2 and spectroscopy investigation of single and mixed g-alumina and thorium oxide.
ethylethanolamine reaction kinetics in aqueous solutions using the stopped- J Mol Catal A Chem 2005;238(1):72e7.
flow technique. Chem Paper 2013;67(9):1123e9. [33] Lewandowski M, Sarbak Z. The effect of boron addition on hydro-
[10] Li T, Keener TC. A review: desorption of CO2 from rich solutions in chemical desulfurization and hydrodenitrogenation activity of NiMo/Al2O3 catalysts.
absorption processes. Int J Greenh Gas Contr 2016;51:290e304. Fuel 2000;79(5):487e95.
[11] Mahajani V, Joshi J. Kinetics of reactions between carbon dioxide and alka- [34] Ibrahim HH, Idem RO. Kinetic studies of the partial oxidation of isooctane for
nolamines. Gas Sep Purif 1988;2(2):50e64. hydrogen production over a nickelealumina catalyst. Chem Eng Sci
[12] Vaidya PD, Kenig EY. CO2-Alkanolamine reaction kinetics: a review of recent 2006;61(17):5912e8.
studies. Chem Eng Technol 2007;30(11):1467e74. [35] Levenspiel O. Chemical reaction engineering. Ind Eng Chem Res 1999;38(11):
[13] Feng B, Du M, Dennis TJ, Anthony K, Perumal MJ. Reduction of energy 4140e3.
requirement of CO2 desorption by adding acid into CO2-loaded solventy. En- [36] Fogler HS. Elements of chemical reaction engineering. 2010.
ergy Fuels 2009;24(1):213e9. [37] Geankoplis C. Transport processes and separation process principles. includes
[14] Du M, Feng B, An H, Liu W, Zhang L. Effect of addition of weak acids on CO2 unit operations; 2003.
desorption from rich amine solvents. Kor J Chem Eng 2012;29(3):362e8. [38] Mears DE. Tests for transport limitations in experimental catalytic reactors.
[15] Li T, Keener TC. A review: desorption of CO2 from rich solutions in chemical Ind Eng Chem Process Des Dev 1971;10(4):541e7.
absorption processes. Int J Greenh Gas Contr 2016;51:290e304. [39] Froment GF, Bischoff KB, De Wilde J. Chemical reactor analysis and design, vol.
[16] Idem R, Shi H, Gelowitz D, Tontiwachwuthikul P. Catalytic method and 2; 1990.
apparatus for separating a gaseous component from an incoming gas stream. [40] Rase HF, Holmes JR. Chemical reactor design for process plants, vol. 2; 1977.
2017. U.S. Patent 9,586,175. [41] Afari DB, Coker J, Narku-Tetteh J, Idem R. Comparative kinetic studies of solid
[17] Shi H, Naami A, Idem R, Tontiwachwuthikul P. Catalytic and non catalytic absorber catalyst (K/MgO) and solid desorber catalyst (HZSM-5)-Aided CO2
solvent regeneration during absorption-based CO2 capture with single and absorption and desorption from aqueous solutions of MEA and blended so-
blended reactive amine solvents. Int J Greenh Gas Contr 2014;26:39e50. lutions of BEA-AMP and MEA-MDEA. Ind Eng Chem Res 2018;57(46):
[18] Liang Z, Idem R, Tontiwachwuthikul P, Yu F, Liu H, Rongwong W. Experi- 15824e39.
mental study on the solvent regeneration of a CO2-loaded MEA solution using [42] Narku-Tetteh J, Afari DB, Coker J, Idem R. Evaluation of the roles of absorber
single and hybrid solid acid catalysts. AIChE J 2016;62(3):753e65. and desorber catalysts in the heat duty and heat of CO2 desorption from
[19] Idem R, Supap T, Shi H, Gelowitz D, Ball M, Campbell C, Tontiwachwuthikul P. butylethanolaminee2-amino-2-methyl-1-propanol and
Practical experience in post-combustion CO2 capture using reactive solvents monoethanolamineemethyldiethanolamine solvent blends in a bench-scale
in large pilot and demonstration plants. Int J Greenh Gas Contr 2015;40:6e25. CO2 capture pilot plant. Energy Fuels 2018;32(9):9711e26.
[20] Caplow M. Kinetics of carbamate formation and breakdown. JAmChemSoc [43] Sakwattanapong R, Aroonwilas A, Veawab A. Behavior of reboiler heat duty
1968;90(24):6795e803. for CO2 capture plants using regenerable single and blended alkanolamines.
[21] P.V. Danckwerts and A. Lannus, '"Gas-liquid reactions," JElectrochemSoc, vol. Ind Eng Chem Res 2005;44(12):4465e73.
117, no. 10, pp. 369C-370C. [44] Danuthai T, Jongpatiwut S, Rirksomboon T, Osuwan S, Resasco DE. Conversion
[22] Ozkutlu M, Orhan OY, Ersan HY, Alper E. Kinetics of CO2 capture by ionic of methylesters to hydrocarbons over an H-ZSM5 zeolite catalyst. Appl Catal A
liquiddCO2 binding organic liquid dual systems. Chem Eng Process: Process Gen 2009;361(1):99e105.
Intensification 2016;101:50e5. [45] Weingarten R, Tompsett GA, Conner WC, Huber GW. Design of solid acid
[23] Busca G. Spectroscopic characterization of the acid properties of metal oxide catalysts for aqueous-phase dehydration of carbohydrates: the role of Lewis
catalysts. Catal Today 1978;41(1):191e206. and Brønsted acid sites. J Catal 2011;279(1):174e82.
[24] Rosynek MP. Isotherms and energetics of carbon dioxide adsorption on.

You might also like