Accepted Manuscript

An optimization approach for CO2 capture using ionic liquids

Darinel Valencia-Marquez, Antonio Flores-Tlacuahuac, Ruben Vasquez-Medrano

PII: S0959-6526(16)31903-5
DOI: 10.1016/j.jclepro.2016.11.064
Reference: JCLP 8453

To appear in: Journal of Cleaner Production

Received Date: 30 June 2016

Revised Date: 29 September 2016
Accepted Date: 10 November 2016

Please cite this article as: Valencia-Marquez D, Flores-Tlacuahuac A, Vasquez-Medrano R, An

optimization approach for CO2 capture using ionic liquids, Journal of Cleaner Production (2016), doi:
10.1016/j.jclepro.2016.11.064.

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

An Optimization Approach for CO2 Capture

PT
1

2 Using Ionic Liquids

RI
SC
2 ∗
3 Darinel Valencia-Marquez1 , Antonio Flores-Tlacuahuac , Ruben Vasquez-Medrano1

U
1
Departamento de Ingenierı́a y Ciencias Quı́micas
AN
Universidad Iberoamericana
M

Prolongación Paseo de la Reforma 880

México D.F., 01219, México.

D
TE

2
Escuela de Ingenieria y Ciencias

Tecnologico de Monterrey, Campus Monterrey

EP

Ave. Eugenio Garza Sada 2501

C

Monterrey, N.L. 64849, Mexico

AC

∗
To whom correspondence should be addressed. E-mail: antonio.flores.t@itesm.mx

1
 ACCEPTED MANUSCRIPT

4 Abstract

5 We propose a processing alternative for replacing the traditional and environmentally danger-

6 ous solvents (as monoethanolamine, diethanolamine) in the process of CO2 capture. We use a

PT
7 computer aided molecular design methodology for exploring the use of ionic liquids as solvents

8 for CO2 capture. The optimization formulation is solved as a mixed-integer nonlinear program-

RI
9 ming problem, the objective is to obtain an optimal molecular structure of a particular class of

ionic liquids for CO2 capture from a post-combustion gas stream. We use a subset of functional

SC
10

11 groups and anions, whose combination forms only certain kind of ILs that have the potential

12 to solubilize CO2 and remove it from gas streams. The properties of ILs are estimated by

U
13 group contributions methods and the equilibrium behaviour by empirical correlations. More-
AN
14 over, recognizing the strong interactions between ionic liquids design and the processing system

15 deployed to carried out the separation, a simultaneous optimal product and process design is
M

16 also proposed. Moreover, we also propose a multiobjective optimization framework to deal

17 with conflicting design objectives related to process economics and and sustainability issues.
D

Keywords: ionic liquids, carbon capture, optimization

TE

18
C EP
AC

2
 ACCEPTED MANUSCRIPT

19

1 Introduction

PT
20

RI
21 Nowadays, there is an enormous interest in the reduction of CO2 emissions from power generation

SC
22 systems. Because a significative portion of power generation systems deploy fossil fuels for this

23 aim, such industrial systems are one of the major contributors to CO2 increase in atmosphere. The

U
common industrial way of coping with large CO2 concentrations has been to use amines to address
24
AN
25 CO2 capture issues (Greer et al., 2010; Kemper et al., 2011; Nittaya et al., 2014; Romeo et al., 2008;
M

26 Ziaii et al., 2009). However, sustainability and large costs issues have been mentioned as some of
D

27 the disadvantages of this CO2 recovery technology (Abu-Zahra et al., 2007a,b; Alie et al., 2005;
TE

28 Chapel et al., 1999; Davis and Rochelle, 2009). Although, other CO2 capture strategies have been
EP

29 proposed (Agarwal et al., 2010; Buhre et al., 2005; Chansomwong et al., 2014a,b; Ekstrm et al.,

30 2009; Gonzalez-Salazar et al., 2012; Harun et al., 2012; Li et al., 2009a; MacDowell et al., 2010;
C

Shimizu et al., 1999; Sun et al., 2012; Trapp et al., 2015; Wang et al., 2010; Zhou et al., 2014), there
AC

31

32 is still need to develop new sustainable and cost competitive solutions.

33 One of the aims of this work is to replace traditional non-environmental friendly solvents used in

34 the CO2 capture process, in favour of a new kind of tailor-designed solvents known as ionic liquids.

3
 ACCEPTED MANUSCRIPT

35 Room-temperature ionic liquids (RTILs or ILs) can be used in a variety of different applications such

36 as: entrainers in extractive distillation (Orchillés et al., 2007) for separation of high-purity ethanol

PT
37 (Chapeaux et al., 2008; Li et al., 2009b), desulfurization and denitrogenation of fuels (Zhang et al.,

38 2004), bio-process for pretreatments and conversion of sugar (Ma et al., 2012; Sun et al., 2009; Yu

RI
39 et al., 2012), materials synthesis as catalyst and supports for nano-materials (Ueno et al., 2008),

SC
40 electrochemical process (Heinze et al., 2010) and batteries (Bansal et al., 2005; Lee et al., 2010).

U
41 These “green” solvents are feasible alternatives for replacing hazardous solvents, due to features such
AN
42 as negligible vapor pressure and chemical and thermal stability. Also, they can be readily tuned to
M

43 meet specific process conditions by tailoring the structure of their cation and/or anion components.

44 Moreover, because ILs are practically nonvolatile components under normal processing conditions,
D
TE

45 they can be recovered with negligible loss of solvent. ILs have proven to be an excellent candidate

46 for replacing amines used as solvent in the traditional process of CO2 absorption (Bara et al., 2009;
EP

47 Palgunadi et al., 2009; Revelli et al., 2010; ur Rahman et al., 2010; Wappel et al., 2010; Zhang
C

48 et al., 2011a). However, reliable information of their physical and thermodynamic behaviour can be
AC

49 difficult to estimate.

50 Optimal molecular design techniques can be used for approaching the design of ILs (González et al.,

51 2007; Hada et al., 2015; Kenny, 2012). The idea is to formulate the product design task as an

4
 ACCEPTED MANUSCRIPT

52 optimization problem, subject to a set of constraints that are enforced to meet target or desired

53 product properties (Gani et al., 2003; Karunanithi et al., 2005). The computer aided molecular

PT
54 design (CAMD) approach has been used for selecting organic solvents and in the design of polymers

55 (Churi and Achenie, 1996; Karunanithi et al., 2005; Odele and Macchietto, 1993). Chavez-Islas et al.

RI
56 (2011a); Valencia-Marquez et al. (2012) designed ILs for the separation of ethanol-water azeotropic

SC
57 mixtures to obtain high-purity ethanol. McLeese et al. (2010) designed a IL for solubilizing a refrig-

U
58 erant. However, presently one of the main difficulties in using CAMD techniques for the design of
AN
59 ionic liquids, has to do with the difficulty of obtaining accurate and reliable physical and thermody-
M

60 namic properties estimation. This is so, because since such products are novel, common physical and

61 thermodynamic information will not be available. Of course, the alternative to handle this problem
D
TE

62 consists in using accurate computations based on ab initio calculations (Hehre, 1976). However,

63 because of the high computational load related to ab initio calculations, presently this alternative
EP

64 is infeasible. Nevertheless, recently some steps have been taken to use estimation methods based on
C

65 fundamental grounds (Chen et al., 2013; Siougkrou et al., 2013). On the other hand, an easy, but
AC

66 potentially inaccurate, manner to estimate physical and thermodynamic properties of novel com-

67 pounds consists in using groups contribution methods (Ge et al., 2008; Huo et al., 2009; Valderrama

68 et al., 2012; Valderrama and Rojas, 2009) . In addition, attempts to take into account fundamental

5
 ACCEPTED MANUSCRIPT

69 phenomenological information to estimate reliable physical and thermodynamic properties of novel

70 compounds have been taken (Chen et al., 2013; Siougkrou et al., 2013). Recently, Chong et al.

PT
71 (2015) have proposed a similar approach for the optimal design of ILs for CO2 capture purposes.

72 There are some differences in scope between that work and the results reported in the present work,

RI
73 being some of them related to the family of cations and anions chosen to become part of the su-

SC
74 perstructure of potential compounds from which ILs will be synthesized. According to information

U
75 provided in Chong et al. (2015) local optimization results were also sought. Moreover, no attempt
AN
76 to test the performance of the optimal ILs in a separation unit was done. In the way it is stated
M

77 in our manuscript for product design purposes CAMD is simple to use and computationally less

78 demand compared to the deployment of molecular simulation techniques. Rigorously speaking even
D
TE

79 molecular simulation is an approximation to ab-initio calculations which is the formal approach we

80 would like to use for molecular design. However, by the time being both approaches are infeasible
EP

81 since they require huge computer loads and they cannot be used for the design of complex molecules
C

82 within an optimization framework. Nowadays, molecular simulation is suitable for the design of
AC

83 simple compounds. Even in this case, the computational load is not negligible.

84

85 In this work, we have addressed the design of the chemical structure of ionic liquids deploying an

6
 ACCEPTED MANUSCRIPT

86 optimal design approach, using formal gradient-based optimization algorithms (Valencia-Marquez

87 et al., 2012), aimed to CO2 capture from power generation post-combustion processes. To repre-

PT
88 sent the family of ionic liquids to be synthesized, we propose a superstructure approach composed

89 of a selected set of cations and anions. Experience and knowledge were the main tools to choose

RI
90 a promising set of cations and anions. To enforce target properties to be met by the ionic liq-

SC
91 uid, we used a proper set of constraints and objective function. Because integer and continuous

U
92 decision variables were involved and some of the constraints gave rise to nonlinear relationships,
AN
93 the optimization problem was cast as a mixed-integer nonlinear programming (MINLP) problem.
M

94 Having synthesized a promising ionic liquid, the performance of such compound for efficient CO2

95 capture was checked in an industrial scale absorption equipment. The results indicated that the
D
TE

96 optimally designed ionic liquid could be used for CO2 capture making feasible this conceptual de-

97 signed absorption process. Moreover, there are some evidences which indicate that there are strong
EP

98 interactions between product and process designs (Giovanoglou et al., 2003; Valencia-Marquez et al.,
C

99 2012). Such product-process interaction could be exploited to improve process economics and per-
AC

100 formance. Therefore, in this work we also address the simultaneous ionic liquid and optimal process

101 design issues to improve process economics by taking advantage of the natural interactions between

102 these problems. Due to the complexity of the underlying MINLP problem only local optimal so-

7
 ACCEPTED MANUSCRIPT

103 lutions are sought. Moreover, the effect of model and process uncertainty on the quality of the

104 optimal solution was not considered. However, we recognize the influence of uncertainty on optimal

PT
105 solutions especially when it comes to consider the impact of physical and thermodynamic properties

106 on the synthesis of novel compounds for which complete basic information in the form of physical

RI
107 constants (i.e. critical constants) may be lacking. Finally, an additional contribution of the present

SC
108 work lies in recognizing that when it comes to consider sustainability as a design goal, this design

U
109 objective may be in conflict with process economics. To cope with this issue in this work the si-
AN
110 multaneous ionic liquid-CO2 recovery design problem is approached and solved as a multiobjective
M

111 MINLP problem. Addressing the problem in this way allows the computation of the Pareto solution

112 curve from which it is up to the designer to pick the best optimal solution which meets the design
D
TE

113 objectives in the best possible manner.

EP

114 2 Problems definition

C

Overall, the problems addressed in the present work are: (a) The optimal molecular design of ionic
AC

115

116 liquids for CO2 capture and (b) The simultaneous optimal molecular and process design problems

117 of ionic liquids for the same purpose.

118 The aim of the first part of this work is to carry out the optimal molecular design of ILs to capture

8
 ACCEPTED MANUSCRIPT

119 large amounts of carbon dioxide (> 90%) from post-combustion streams, leading to a potential

120 reduction of the greenhouse gases commonly released to atmosphere. We will assume that the

PT
121 typical composition of a post-combustion stream is given as follows: N2 78 %, H2 O 9%, CO2 13%

122 (Shiflett et al., 2010).

RI
To obtain the optimal IL, the design problem was divided into two design scenarios. In the first

SC
123

124 case, we will use a processing stream composed only of pure CO2 , whereas in the second case we

125

U
will assume that CO2 is embedded in a post-combustion gas stream. Using the optimal molecular
AN
126 design approach proposed by Churi and Achenie (1996), the cationic part of the IL was supposed
M

127 to be part of a set of functional groups. Similarly, the anionic part of the IL was selected from a set
D

128 of anions. Knowledge and experience are probably the best tools to choose a proper set of cations
TE

129 and anions to design the IL from scratch. If the set of cations and anions are not properly selected,

then the performance of the designed IL will be poor. On the other hand, including a large set of
EP

130

131 cations and anions will make the underlying optimization problem difficult to solve. Therefore, a
C

132 trade-off between the number of cations and anions and their potential for CO2 recovery needs to be
AC

133 established. Because the optimally designed IL can be a novel compound, for which no information

134 regarding the computation of the physical and thermodynamic compound can be available, we will

135 use groups contribution methods (Ge et al., 2008; Lazzús, 2012; Valderrama et al., 2012; Valderrama

9
 ACCEPTED MANUSCRIPT

136 and Rojas, 2009), and some empirical correlations, for filling this gap. We must recognize that, in

137 most of the cases, groups contribution methods are rough estimates of physical and thermodynamic

PT
138 properties. However, they are easy to use and can lead to proper estimates of the target properties

139 in absence of additional information.

RI
Therefore, the problem to be addressed in the first part of this work can be stated as follows: “Given a

SC
140

141 set of cations and anions suitable for CO2 capture and two processing streams to be treated, the first

142

U
one composed of pure CO2 and the second one of a post-combustion stream of known composition,
AN
143 the problem consists in selecting the best cation-anion coupling which forms the IL molecule, such
M

144 that maximum recovery of CO2 is achieved holding a set of target physical properties. The steady-
D

145 state equilibrium model of an absorption column will be deployed to demonstrate that the optimally
TE

146 designed IL performs well in industrial scale equipment”.

EP

147 We also address the simultaneous optimal molecular design and process design of ionic liquids for

148 CO2 capture. Because of the complex behavior and natural interactions of processing systems, when
C
AC

149 addressing the optimization of such systems it is common to split the original problem in several

150 parts such that the decomposed problem becomes easier to solve in comparison to the original non-

151 decomposed problem. Since normally decoupled problems tend to ignore the strong interactions

152 embedded in the original problem, the optimal solution found can be suboptimal. Therefore, we hy-

10
 ACCEPTED MANUSCRIPT

153 pothesize that by approaching the optimal IL design and processing system in a simultaneous, rather

154 than in a sequential manner, improved optimal solutions can be found because natural interactions

PT
155 between these problems are explicitly taken into account.

RI
156 The problem addressed in this part of the work can be stated as follows:

SC
157 “Given: (a) a post-combustion flue gas stream , (b) a set of anions and cations suitable for CO2

158 recovery and (c) a fixed process flowsheet for CO2 recovery, the aim consists in simultaneously

U
AN
159 designing the best chemical structure of the ionic liquid together with the optimal values of the design

160 parameters of the fixed-structure flowsheet such that either maximum CO2 recovery is achieved
M

161 or minimum energy for CO2 recovery is deployed. Moreover, using multiobjective optimization
D

162 techniques, a trade-off between both design objectives will be reached”.

TE

163 It should be stressed that by solving two apparently different design problems (i.e. chemical structure
EP

164 of the ionic liquid and processing conditions of the CO2 capture process) we aim to exploit the
C

165 natural interactions between the two problems. Hence, superior optimal results can be found when
AC

166 the natural interactions embedded in a given system are exploited. We base this affirmation on

167 previous results observed in similar highly interactive processing systems (Chavez-Islas et al., 2011b;

168 Giovanoglou et al., 2003).

11
 ACCEPTED MANUSCRIPT

169 3 Optimization Formulation

170 In this part of the work the two optimization formulations deployed for approaching the IL optimal

PT
171 molecular design problem for CO2 capture and simultaneous ionic liquid and process design are

RI
172 given. All variables are fully described in the Nomenclature section.

SC
173 3.1 MINLP formulation of the CAMD model

U
AN
174 Because the optimal molecular design problem involves both binary and continuous variables under

175 steady-state processing conditions, the optimization problem is cast as a Mixed-Integer NonLinear
M

176 Programming (MINLP) problem, whose solution can be obtained by several well known methods
D

177 such as Bender’s decomposition, the outer-approximation algorithm and branch and bound methods
TE

178 (Floudas, 1995).

EP

179 Objective function

C

180 The objective function states our target design goal: the optimal IL should be able to recover
AC

181 as much CO2 as possible from a processing gas stream. For doing so, we will set up two design

182 scenarios. In the first case a pure CO2 gas stream will be assumed. According to this design goal

183 the objective function Ω1 reads as follows:

12
 ACCEPTED MANUSCRIPT

min Ω1 = (xdCO2 − xCO2 )2 (1)

u

PT
184 where xCO2 stands for the CO2 mole fraction in the product stream, whereas xdCO2 is the target

RI
185 CO2 mole fraction captured by the IL, and u stands for the vector of decision variables. The value

SC
186 used for xdCO2 was 0.7. The choice is based in the maximum mol fraction reported in literature for

187 anions deployed in this work (see references in Table 2). Similarly, in the second design scenario we

U
AN
188 will assume that the CO2 to be captured is an embedded component within a post-combustion gas

189 processing stream. For the purpose of designing the IL we will assume that the separation will take
M

190 place in a single separation stage. Our goal is also to recover the maximum amount of CO2 . To
D

191 meet this target, we propose the following objective function Ω2 :

TE

Lout xout
CO2
max Ω2 = in
(2)
EP

u Gin yCO 2

where x and y are the mole fractions in liquid and gas phases, respectively, L and G are the flows
C

192
AC

193 of liquid and gas streams, the subscripts in and out mean inlet and outlet streams. In both cases,

194 the vector of continuous decision variables (u) are: the mol fraction of the liquid phase (xi ), the

195 liquid phase flowrate (L), the gas phase flowrate (G), the boiling point temperature (Teb ) and the

196 melting point temperature (Tmp ). In the second design scenario, the mol fraction of the gas phase

13
 ACCEPTED MANUSCRIPT

197 (yi ) is also considered to be a decision variable. As clearly seen from Equation 2, Ω2 stands for

198 the CO2 recovery ratio. Therefore, when Ω2 =1 maximum recovery of CO2 in the IL stream takes

PT
199 place. On the other hand, when Ω2 =0, no CO2 is captured at all. Following, we will present the set

200 of constraints of the optimization formulation to select the IL which represents the best option for

RI
201 CO2 capture. Such a formulation contains binary decision variables used for obtaining the optimal

SC
202 IL. Of course, as previously stated, the performance of the IL depends upon the family of cations

U
203 and anions, from which the IL is designed. However, we should note that the performance of the
AN
204 IL can also depend upon both the processing conditions and the design of the equipment where
M

205 the process of CO2 capture will take place. Such interaction between the design of compounds and

206 the process synthesis of the equipment has been known for some time and it has been exploited for
D
TE

207 improving the economics and performance of chemical processes (Valencia-Marquez et al., 2015).

Molecular design constraints

EP

208

209 To approach the optimal molecular design of the IL, the set of constraints dealing with the chemical
C
AC

210 structure of the cations and anions set was divided into the following two parts: a) design of the

211 cation, and b) anion selection.

212 Design of the cation constraints

14
 ACCEPTED MANUSCRIPT

213 For addressing the cation design issue, the optimal molecular design methodology develop by Achenie

214 Churi and Achenie (1996) was used. The following constraints describe the structural feasibility of

PT
215 the molecule and their purpose is to generate molecules that do not violate basic feasibility criterion

216 such as the octet rule. The molecules should not have any unattached sites or multiple bonds

RI
217 attached to the same site. Moreover, in Table 1 the set of functional groups which were considered

SC
218 suitable for CO2 capture using ILs is shown.

219

U
Equation 3 sets up an upper bound on the number of structural groups that can constitute a
AN
220 molecule:
M
D

nX m
max X

uik ≤ nmax (3)

TE

i=1 k=1

Table 1: Functional groups used for defining the cationic part of the ILs.
EP

Functional Number of
group bonds
[mim]a 1
-CH3 1
C

-CH2 - 2
AC

>CH- 3
>C< 4
-OH 1
>C=O 2
-CHO 1
-COOH 1
-COO- 2
a
methylimidazolium

15
 ACCEPTED MANUSCRIPT

221 Equation 4 is an implementation of the octet rule:

nX
max S
Xmax m
X
zijp = uik υk , i = 1...nmax (4)

PT
p=1 j=1 k=1

RI
222 Equations 5-8 ensure that only one molecule is formed:

SC
i−1 SX
X max

zijp ≥ −ωi , i = 2...nmax (5)

p=1 j=1
nX m nX

U
max X max

uik + ωi = nmax (6)

i=1 k=1 i=1
AN
ω1 = 0 (7)
M

ωi ≤ ωi+1 , i = 1...(nmax − 1) (8)

D
TE

223 Equation 9 has to be introduced to account for different valencies of the groups:

SX
max nmax nX
max
EP

X
zijp − ziυk p + M uik ≤ M, i = 1...nmax , k = 1...m (9)
j=υk p=1 p=1
C

Equation 10 sets the symmetry constraint:

AC

224

SX
max SX
max

zijp = zpji , i = 1...(nmax − 1), p = (i + 1)...nmax (10)

j=1 j=1

225 This equation means that if a given group is attached to a second group, the second group is

16
 ACCEPTED MANUSCRIPT

226 automatically attached to the first one. Otherwise the second group could be attached to other

227 different group, giving rise to a non-feasible molecule.

PT
228 Equation 11 ensures that a site in a given group can be attached at most once to some other group:

RI
nX
max

zijp ≤ 1, i = ...nmax , j = 1...Smax (11)

p=1

SC
Equation 12 is applied to force existence of the ith group if the (i − 1)th group is present:

U
229
AN
m
X m
X
uik − ui−1,k ≤ 0, i = 2...nmax (12)
k=1 k=1
M

Equation 13 is added to forbid multiple bonds:

D

230
TE

SX
max

zijp ≤ 1, i = 1...(nmax − 1), p = (i + 1)...nmax (13)

j=1
EP

Equation 14 establishes that each functional group contributes with only half bond in a linear
C

molecule:
AC

nX
max

(ui,k υk ) = (n − 1) (14)
i,k

231 Figure 1 shows possible ways in which a molecule can be described. In general, there are several

17
 ACCEPTED MANUSCRIPT

232 multiplicities when specifying a molecule. To eliminate multiple solutions for the same molecule,

233 the solution found from the search space can be removed if its objective function value is the same

PT
234 than other previous solutions (Samudra and Sahinidis, 2013). A more rigorous approach would be

235 to introduce constraints for imposing order in the molecule (Churi and Achenie, 1996). This means

RI
236 that the molecule will have an orderly form in its graphical representation, and the functional groups

SC
237 always be counted consecutively beginning with one extreme (Figure 1a), this eliminate the other

U
238 solutions (figure 1b and 1c).
AN
M
D
TE
C EP

Figure 1: Multiplicity. One molecule with different solutions

AC

The following set of constraints take care of the order in the molecule (i.e. the molecule grows up

18
 ACCEPTED MANUSCRIPT

starting from the imidazolium ring if U1,k = [mim]):

u1,k = 1 k = [mim] (15)

PT
υk
X
zi,1,p = zp,j,i i ∈ 2, 3, ..., nmax (16)
j

RI
zi,j,p ≥ zi,j+1,p+1 i ∈ 1, 2, ..., nmax , (17)

SC
p ∈ 1, 2, ..., nmax , j ∈ 2, ..., (smax − 1)

239

U
These equations indicate that each functional group is attached to an existing and previous group.
AN
240 Design of the anion constraints
M

Table 2: Anions used for defining the ILs.

D

Anion ID Anion name Reference

TE

DMP Dimethylphosphate Palgunadi et al. (2009)

TFA Trifluoroacetate Muldoon et al. (2007); Shiflett and Yokozeki (2009); Yokozeki et al. (2008)
BF4 Tetrafluoroborate Bara et al. (2009); Blath et al. (2011); Revelli et al. (2010); Tokuda et al. (2006a)
Cl Cloride Carvalho et al. (2010)
TfO Trifluoromethanesulfonate Soriano et al. (2009)
EP

Ac Acetate Shiflett et al. (2010); Shiflett and Yokozeki (2009)

PF6 Hexafluorophospate Bara et al. (2009); Tokuda et al. (2006a); ur Rahman et al. (2010); Zhang et al. (2011b)
FEP Tris(pentafluoroethyl)trifluorophosphate Blath et al. (2011); Zhang et al. (2008)
Tf2 N Bis(trifluoromethylsulfonyl)imide Bara et al. (2009); Blath et al. (2011); Carvalho et al. (2010); ur Rahman et al. (2010)
C
AC

241 The set of anions which were considered suitable for CO2 capture using ILs are shown in Table 2.

242 For defining the proper anion one only needs to set-up the following constraint which indicates that

19
 ACCEPTED MANUSCRIPT

243 each molecule of IL should contain only one anion:

m
X
yak = 1 (18)

PT
k

RI
244

245

SC
246 Boiling point temperature

U
AN
The boiling point temperature is calculated using the group contribution methods proposed by

Valderrama et al. (2012):

M
D

X
Tb = 198.2 + uik ∆Tbk (19)
i,k
TE

247 Melting point temperature

EP

248 The melting point temperature is evaluated using the method developed by Lazzús (2012):
C
AC

! !
P P P P
(uik ∆Tmp,ik ) + (yak ∆Tmp,k ) + ZH (uik ∆Timp,ik ) + (yak ∆Timp,k )
i,k k i,k k
Tm = (20)
AH + BH σ + CH τ + DH δ

20
 ACCEPTED MANUSCRIPT

249

250

PT
251 Boiling and Melting point temperature bounds

RI
252 To enforce that the IL remains in liquid phase (negligible vapor pressure) under the processing

SC
253 conditions, the melting point temperature should be lower than the temperature of operation (set at

254 T = 303.15 K), whereas the boiling temperature should be greater than the processing temperature:

U
AN
Tm < T < T b (21)
M

255 where Tm = 283 K, Tb = 480 K.

D
TE

256 Vapor and liquid equilibrium

The thermodynamic physical vapor and liquid equilibrium model is based in the following correlation
EP

developed by Mortazavi-Manesh et al. (2011):

C
AC

P yi φi = xi Hi γi∗ (22)

257 where P is the total pressure, yi is fraction mol in gas phase, Hi,IL is the Henry’s constant and γi

258 is the asymmetric activity coefficient.

21
 ACCEPTED MANUSCRIPT

259 Vapor phase fugacity coefficient

260 The vapor phase fugacity coefficient is evaluated trough the Peng-Robinson equation of state

PT
RI
 
A Z + 2.414B
ln φi = Z − 1 − ln(Z − B) − √ ln

SC
2 2B Z − 2.414B

U
261 Henry’s law constant
AN
262 As it is well known, low Henry’s constant values mean greater solubility. Hence, we chose to
M

263 measure the solubility of CO2 in the IL in terms of Henry’s law because of the relationship between
D

264 the operating pressure and the amount of solubilized CO2 in the IL:
TE
EP

β P
ln H12 = ln Hi,IL = α + + (23)
T T
C

fi v∞P
ln Hi = ln = ln H12 + ln γi∗ + i (24)
xi RT
AC

X = X1 + X 2 θ (25)

22
 ACCEPTED MANUSCRIPT

265 where X can be α, β or . The parameters αi , βi and i are constants and θ is the molecular

266 weight. In Equation 24, Hi is the Henry’s law constant at the operating conditions, while H12 is

PT
267 Henry’s law constant at reference conditions, this is commonly obtained using experimental data.

268 To calculate Henry’s law constant, the methodology proposed Mortazavi-Manesh et al. (2011) uses

RI
269 an empirical relation similar to the Krichevsky-Kasarnovsky expression (Prausnitz, 1969), where

SC
270 activity coefficients are evaluated using the COSMO-RS software (Klamt and Schuurmann, 1993).

U
271 We assume that the term vi∞ is negligible.
AN
272 Activity coefficient
M

The Krichevsky-Kasarnovsky equation represents one of the best ways of modelling the solubility of
D

gases in liquids over wide pressure ranges (Prausnitz, 1969). Hence, in this work we have deployed
TE

this equation together with a correlation for estimating the asymmetric activity coefficient We also
EP

assume that the vapor pressure of the solvent is negligible. To calculate the activity coefficient,

experimental data were adjusted by least squares to obtain the empirical parameters embedded in
C
AC

the following correlation (equations (26),(27)), which are deployed to estimate the activity coefficient

23
 ACCEPTED MANUSCRIPT

in the solubility model used in this work :

Λ(x2 − 1)
γ∗ = (26)
RT

PT
Λ = λ1 + λ2 T + λ3 M W/P (27)

RI
273 A total of 394 experimental points were obtained from literature. This set of data includes only the

SC
274 anions selected for this work and shown in Table 2. Figure 2 shows the deviation of the solubility

U
275 model with respect to experimental data. The box represents the operating pressure region within
AN
276 which the search for the optimal IL will be conducted.
M

277 3.2 MINLP formulation for simultaneous ionic liquid and process de-
D

278 sign
TE

279 In this part we propose a Mixed-Integer Non-Linear Programming (MINLP) formulation to address
EP

280 the simultaneous optimal design of the chemical structure of an ionic liquid and the processing
C

281 operating conditions of a fixed flowsheet for CO2 recovery. The optimal ionic liquid should recover
AC

282 at least 90% of CO2 from a post-combustion gas stream by physical absorption.

24
 ACCEPTED MANUSCRIPT

200

180

PT
160

140

RI
120
Pcal [bar]

100

SC
80

60

U
40
AN
20

0
0 20 40 60 80 100 120 140 160 180 200
M

Pexp [bar]

Figure 2: Experimental total pressure of the CO2 -IL system vs. calculated pressure using the model
D

deployed in this work for the ILs shown in Table 2. The meaning of the solid line is the perfect fit
of the model to the experimental data.
TE

Single objective Function

EP

283
C

284 To quantify the performance of the CO2 recovery process three objective functions (Ω3 , Ω4 , Ω5 ) were
AC

285 used. The aim of the first objective function Ω3 is to remove the maximum amount of CO2 from

286 the post-combustion gas stream. This objective function reads as follows.

25
 ACCEPTED MANUSCRIPT

max Ω3 = RCO2 (28)

PT
287 where the CO2 removal term (RCO2 ) is defined as follows:

RI
out
yCO Gout
RCO2 = 1 − in
2
(29)
yCO 2
Gin

SC
288 where yCO2 is the CO2 mol fraction, G is gas molar flow, and the indexes in and out stand for

U
inlet and outlet streams of the absorber column, respectively. It should be remarked that RCO2 is
289
AN
290 referred to the inlet and outlet gas streams of the absorber. Therefore, when the amount of CO2 in
M

291 the stripped gas stream is minimum, RCO2 will be maximum.

D

292 The aim of the second objective function Ω4 consists in minimizing the flash unit thermal duty:
TE

min Ω4 = Q (30)
C EP

293 Since operating costs in separation systems are normally dominated by thermal loads, minimizing
AC

294 this term has a close relationship to the minimization of the operating costs. Of course, a formal

295 cost analysis should be conducted to estimate the profitability of the addressed system. The full

296 economic analysis of a related CO2 capture process using ionic liquids has been previously reported

297 elsewhere (Valencia-Marquez et al., 2015). It should be stressed that when Ω4 was deployed as the

26
 ACCEPTED MANUSCRIPT

298 objective function, Ω3 was enforced as an additional constraint to ensure minimum CO2 removal in

299 the absorber.

PT
300 Finally, since ionic liquids tend to be expensive compounds, a significant part of the operating costs

RI
301 has to do with the flow rate of ionic liquid. Hence, the aim of the third objective function Ω5 consists

in the minimization of the flow rate of ionic liquid:

SC
302

U
min Ω5 = FIL
AN (31)

303 where FIL stands for the flow rate of ionic liquid. In conclusion, the minimization of Ω5 should lead
M

304 to improve process economics.

D

Multiobjective Function
TE

305
EP

306 During processes operation we normally face a practical situation where more than one objective

307 function should be optimized. However, those objective functions can be contradictory or in conflict.
C
AC

308 In our case, we would like to meet the following two objectives simultaneously in the best possible

309 way: maximize the CO2 recovery and minimize the flash thermal duty. Unfortunately, maximizing

310 the amount of CO2 recovered will demand larger thermal duties. On the other hand, minimizing

311 the flash thermal duty will reduce the amount of CO2 recovered. Hence, the two objective func-

27
 ACCEPTED MANUSCRIPT

Table 3: Functional groups used in this work

Functional groups number
ID Name of bonds ref
-CH3 Methyl 1 —
-CH2 - Ethyl 2 —

PT
H2 O Water - —
N2 Nitrogen - —
CO2 Carbon dioxide - —

RI
BF4 Tetrafluoroborate 1 Bara et al. (2009); Blath et al. (2011); Revelli
Tf2 N Bis(trifluoromethylsulfonyl)imide 1 Bara et al. (2009); Blath et al. (2011); Carvalh
PF6 Hexafluorophospate 1 Bara et al. (2009); Hasibur Rahman et al. (201

SC
TfO Trifluoromethanesulfonate 1 Soriano et al. (2009)
MeSO4 Methyl sulfate 1 Jacek Kumelan and Maurer (2006)
EtSO4 Ethyl sulfate 1 Soriano et al. (2009)

U
Cl Cloride 1 Carvalho et al. (2010)
DEPO4 Diethyl phosphate 1 Palgunadi et al. (2009)
AN
DMPO4 Dimethyl phosphate 1 Palgunadi et al. (2009)
MDEGSO4 2-(2-methoxyethoxy)ethyl sulfate 1 Muldoon et al. (2007)
NO3 Nitrate 1 Lynnette A. Blanchard and Brennecke (2001)
SCN Thiocyanate 1 Revelli et al. (2010)
M

TFA Trifluoroacetate 1 Muldoon et al. (2007); Shiflett and Yokozeki (2

D

312 tions are contradictory and they cannot be met simultaneously. Under those situations a trade-off
TE

313 between these two objective functions should be reached by computing a Pareto curve solving the

314 optimization problem as a multiobjective optimization problem. In this work we have deployed
EP

315 the -constrained method for approaching the numerical solution of the multiobjective optimization
C

316 problem (Flores-Tlacuahuac et al., 2012).

AC

317 CAMD Formulation

318 The optimal molecular design formulation is the same as previously stated in equations 1-27.

28
 ACCEPTED MANUSCRIPT

319 Process description

320 In this part we describe the chemical process deployed for simultaneous ionic liquid and process

PT
321 design issue. We assume that CO2 will be removed from a post-combustion gas stream; these streams

RI
322 are part of power thermal combustion plants. We also assume that the polluted processing stream

SC
323 has been dehydrated to avoid water handling problems. The mol fraction of the post-combustion

324 stream used in this work is 0.14 and 0.86 of CO2 and N2 , respectively.

U
AN
325 Conceptually, two feedback-connected separation systems are required for CO2 capture using ionic

326 liquids. As shown in Figure 3, the proposed flowsheet consists of two separation equipments. In the
M

327 first separation equipment, an absorber column will be deployed such that a counter-current ionic
D

328 liquid stream will be used to physically absorb the CO2 present in the post-combustion gas stream.
TE

329 Since the vapor pressure of typical ionic liquids is low, they do not evaporate (at the processing
EP

330 conditions of the absorption column). Hence, the absorber top gas stream will be composed of N2
C

331 and small amounts of CO2 . In addition, the absorber liquid bottoms stream will be composed of a
AC

332 CO2 -ionic liquid mixture. The CO2 -ionic liquid mixture from the absorber is fed to a non-adiabatic

333 flash unit where almost pure CO2 and ionic liquid product streams will be obtained. The bottoms

334 warm liquid stream is first cooled and then recycled to the absorber. We should note that in Figure

335 3 normally a make-up ionic-liquid stream will be required to account for the small fractions of

29
 ACCEPTED MANUSCRIPT

336 ionic liquid that are unavoidably evaporated in the top gas streams of the absorber and flash units.

337 However, since ionic liquids feature negligible vapor pressure, the amount of evaporated ionic liquid

PT
338 is assumed to be enough small such that for simulation and design purposes a make-up stream is

339 not required. The model used for the absorber column is equilibrium-based, has 25 stages where

RI
340 stage 1 is at bottom and stage 25 is on the top; the column operates at 7 bar, whereas the flash

SC
341 unit is a single non-adiabatic equilibrium stage at atmospheric pressure. The operating pressure

U
342 in the absorber was selected on the basis that 7 bar was the minimum pressure to meet the CO2
AN
343 recovery specifications. In some of the published works about CO2 capture using ionic liquids (Aki
M

344 et al., 2004; Lynnette A. Blanchard and Brennecke, 2001; Manic et al., 2012b; Shin and Lee, 2008),

345 it has been stated that large pressures (i.e. over 20 bar) are required to hold product specifications.
D
TE

346 However, since high pressures demand compression systems, it is important to find design conditions

347 where the pressure is as small as possible to improve process economics. The atmospheric operating
EP

348 pressure in the flash drum was chosen to reduce power consumption since a compressor would not
C

349 be needed. However, we find that the adiabatic operation of the flash drum was not enough to meet
AC

350 separation specifications. For this reason, the flash drum was operated in a non-adiabatic form.

30
 ACCEPTED MANUSCRIPT

Vent gas Lean 2

Product

PT
Flash

RI
Absorber

SC
P1 Lean 1

Flue gas Rich

U
AN
Figure 3: Flowsheet of the CO2 capture system proposed in this work. An isothermal absorption
column and a non-adiabatic flash unit were deployed.
M

351 Steady-State MESH equations

D
TE

352 As shown in Figure 3, the separation process between CO2 and the components of the combustion

353 flue gas will take place in an absorption tower where the flue gas and ionic liquid are brought in
EP

354 counter-current contact.

C
AC

355 Absorption column

Isothermal operation of the absorption column has been assumed. Hence, only mass balance, equi-

librium relationships and sum of the mol fractions are formulated through the following equations

31
 ACCEPTED MANUSCRIPT

(see notation in the nomenclature section) :

Gs−1 ys−1,c + Ls+1 xs+1,c = Gs ysc + Ls xsc ∀s = 1, . . . , Nt (32)

PT
∀ c = 1, . . . , Nc

RI
Nc
X
ysc = 1 ∀ s = 1, . . . , Nt (33)
c

SC
N
X c

xsc = 1 ∀ s = 1, . . . , Nt (34)
c

U
P ysc φsc = γsc Pcsat xsc
AN ∀ s = 1, . . . , Nt (35)

∀ c = 1, . . . , Nc
M

356 Notice that in Equation 32 for s = 0 the terms G0 and yo,i refer to the flowrate and mol fraction of
D

the post-combustion gas stream fed to the absorber, respectively. Similarly, for s = Nt the terms
TE

357

358 LNt +1 and xNt +1,i refer to the flowrate of the recycle stream mainly composed of ionic liquid and its
EP

359 mol fraction, respectively.

C

360 Flash unit

AC

361 The non-adiabatic flash unit operates under non-isothermal operating conditions.

32
 ACCEPTED MANUSCRIPT

F z c = V yi + L x i ∀ c = 1, . . . , Nc (36)

PT
Nc
X
yc = 1 (37)
c

RI
Nc
X
xc = 1 (38)
c

SC
P yc φc = γc Pcsat xc ∀ c = 1, . . . , Nc (39)

U
F hF = V hV + LhL AN (40)

362 Under phase-equilibrium conditions the vapor phase fugacity coefficient was estimated using the
M

363 Peng-Robinson equation of state, the liquid phase activity coefficient was calculated by the UNIFAC
D

364 method and the vapor pressure by the Antoine equation as proposed by Shiflett and Yokozeki (2006).
TE

Performance constraints
EP

365
C

366 From a performance point of view the CO2 capture process using ionic liquids has to be competitive,
AC

367 particularly when compared to the traditional MEA-based process. To hold this objective, the

368 following recovery constraints are enforced :

33
 ACCEPTED MANUSCRIPT

369 CO2 purity

CO2 purity ≥ 0.95 (41)

PT
where the CO2 purity is the mol fraction in the gas stream leaving the flash unit.

RI
370

SC
371 CO2 removal

U
CO2 removal ≥ 0.90 (42)
AN
372 As previously mentioned, this constraint is only enforced when the Ω4 objective function is deployed.
M
D

373 4 Results
TE

374 Optimal ionic liquid design

EP

375 In this part of the work, the optimal molecular design of an IL for CO2 capture from a post-
C

376 combustion gas stream is obtained. The optimal molecular design problem is cast in terms of the
AC

377 MINLP formulation described in section 3.1. For doing so, two design scenarios were deployed. In

378 the first case, an IL is designed to capture CO2 from a processing stream containing pure CO2 . These

379 results are interesting by themselves. Moreover, they are used to provide initial values of the decision

34
 ACCEPTED MANUSCRIPT

380 variables of the design problem addressed in the second scenario. In the second case, another IL will

381 be designed to separate CO2 from a typical processing post-combustion gas stream. An interesting

PT
382 question here consists in checking whether the chemical structure of the IL has changed between

383 the two design scenarios due to different processing conditions. In these two design scenarios, the

RI
384 optimal design of the IL will be carried out assuming that the separation takes place in a single

SC
385 equilibrium stage. Next, the performance of the optimally designed IL will be checked using the

U
386 steady-state equilibrium model of an absorption column, which typically will be used for industrial
AN
387 scale applications.
M

388 The overall optimization problem is formulated as a MINLP. Only optimal local solutions were
D

389 sought using the SBB algorithm available in GAMS (Brooke et al., 1998). No attempt for computing
TE

390 global optimal solutions was done. Of course, global optimal solutions cannot be claimed. There is

always a chance that better solvents can be found using global MINLP algorithms. However, the
EP

391

392 large computational load can be a major problems when trying to obtain global MINLP solutions.
C

393 The problem statistics for the synthesis of the ILs are as follows: (a) number of constraints : 696,
AC

394 (b) Number of continuos variables : 663, (c) Number of integer variables : 623, (d) CPU time :

395 18690 seg. The computing platform was a laptop computer with an intel core i7 @ 2.67 Ghz, 4 Gb

396 RAM.

35
 ACCEPTED MANUSCRIPT

397 Case I: Pure CO2

398 Before considering the design of an IL for CO2 capture from a post-combustion gas stream, we

PT
399 decided to carry out the design of an IL for CO2 capture in a system with only CO2 in the gas

RI
400 phase. In this case we only focus in the thermodynamic equilibrium where the amounts of CO2

SC
401 and IL are not important. We decided to do so to recreate the experimental conditions reported

402 in several works (Palgunadi et al., 2009; Revelli et al., 2010), where the amount of CO2 is always

U
higher than the corresponding amount of IL (Mattedi et al., 2011; Zhang et al., 2011a). Moreover,
AN
403

404 most of the thermodynamic models proposed in the literature (Babarao et al., 2011; Li et al.,
M

405 2012; Manic et al., 2012a; Zhang et al., 2008) assume that the gas phase is composed of pure CO2 .
D

406 The model deployed in this work, regardless of the composition of the gas, must able to predict
TE

407 the thermodynamic equilibrium of the CO2 -IL system. In addition, there were other reasons to
EP

408 approach the optimal design of the IL using a pure CO2 gas stream: to analyse the dependence

409 of the optimal chemical structure of the IL with respect to the gas stream composition (in which
C
AC

410 CO2 is embedded), to provide initial values of the major decision variables for the second case (CO2

411 embedded in a post-combustion gas stream). Hence, to analyse the dependence of the IL optimal

412 design on the operating pressure, five different pressure values are considered: 1, 5, 10, 20 and 30

413 Bar. Table 4 shows the ILs with best performance. For each operating pressure different molecular

36
 ACCEPTED MANUSCRIPT

414 structures of the IL are obtained, which indicates a clear dependency of the IL chemical structure

415 upon the operating pressure. This partial conclusion is important since it stresses the fact that the

PT
416 IL optimal design and the operating conditions are related. This interaction can be exploited for

417 addressing the economical design of the CO2 capture process.

RI
It should be stressed that for the first three operating pressures the anion was always the same

SC
418

419 (tetrafluoroborate [BF4 ]). The ILs designed for 1, 5 and 10 Bar have a lower performance than

420

U
other ILs reported in the literature (Bara et al., 2009; Revelli et al., 2010; Zhang et al., 2008). This
AN
421 is so because the chain in the structure of the cation has many ramifications, leaving few space for
M

422 the CO2 staying inside the molecule, making also difficult the mobility of CO2 inside the IL (Cadena
D

423 et al., 2004; Sedláková and Wagner, 2012). Of course, the proportion of CO2 which can be captured
TE

424 by the IL also depends upon the operating pressure, and not only on the chemical structure of

the IL. On the other hand, the set of ILs designed for 20 and 30 Bar have a better performance
EP

425

426 than the ones reported in literature (Blath et al., 2011; Carvalho et al., 2009c). For the 20 Bar
C

427 case, the designed IL has a structure with few ramifications and its anion (trifluoromethylsulfonate
AC

428 [TfO4 ]) is bigger and more asymmetric than the [BF4 ] anion. At 30 Bar, the designed IL has a

429 larger chain and its anion (hexafluorophosphate [PF6 ]) is bigger than the anion for low pressures

430 ([BF4 ]), providing more free space for the CO2 , but it has more resistance to free movement. We

37
 ACCEPTED MANUSCRIPT

431 also observe that at higher operating pressures, the proportion of CO2 which can be captured by

432 the IL increases substantially. In Table 5 a comparison of the performance of the ILs designed

PT
433 in the present work against other ILs proposed in the research literature is done. As seen from

434 this table, the performance of the ILs designed in our work is better. We would like to highlight

RI
435 that, to make the comparison as fair as possible, the criterium used for choosing the ILs from the

SC
436 research literature was that they feature the same anion as the ILs designed in the present work.

U
437 In summary, the results shown in Table 5 highlight the fact that optimal recovery of CO2 using
AN
438 ILs depend upon the chemical structure of the IL, but also on the operating conditions. Therefore,
M

439 as previously stated, taking advantage of the existing interactions between both problems should

440 result in efficient profitable processes in comparison when no such interaction is exploited.
D
TE

441 Case II: Post-combustion gas stream

EP

442 The composition of the post-combustion gas stream is taken as follows: N2 78 %, H2 O 9%, CO2
C

443 13%, whereas the corresponding flowrate is taken as 8.7 kg/s (1.044 mol/h) (Shiflett et al., 2010).
AC

444 Moreover, an arbitrary flowrate of 48.2 kg/s (0.4658 mol/h) of IL is also used giving rise to a 5.5

445 kg IL/kg combustion gases ratio (0.4462 mol of IL/ mol of combustion gas). To carry out the IL

446 optimal molecular design, we assume that the separation takes place in a single absorption stage.

38
 ACCEPTED MANUSCRIPT

Table 4: Solubility of the ILs designed for a pure CO2 gas stream at 303.15 K.

Pressure Molality Ionic ID

[Bar] CO2 [molCO2 /KgIL ] Liquid
CH3 CH3

PT
CH3
CH2 CH2 C C COOH
-
N N [BF4]
+ CH3 CH3
1 0.059 IL-1
CH3

RI
COO CH3
CH3
C CH CH3
N N [BF4]
-
+ C
CH3

SC
5 0.316 CH3 CH3 IL-5
CH3

C COO CH3
CH3
CH2

U
N N CH2 CH3 -
+ [BF4]
C
AN
10 0.685 CH3 CH3 IL-10
CH3
CH3 C CH2 OH
M

N N [TfO4]
-
+ CH2 CH3
20 1.618 IL-20
CH3
D

N N COOH
+ [PF6]
-

30 4.502 IL-30
TE

When dealing with a typical post-combustion gas stream, the optimal chemical structure of the IL
EP

447

448 turns out to be the same for all pressures: 1-decyl-3-methylimidazolium trifluoromethylsulfonate
C

449 ([C10 mim][TfO]). This is in contrast with the ILs that were designed taking into account a pure
AC

450 CO2 gas stream, where, for the considered pressure range, the IL chemical structure was different.

451 In Table 6 the results concerning the amount of CO2 that was captured by the IL stream are shown.

452 It should be noted that the absorbed amount of CO2 is strong function of the operating pressure,

39
 ACCEPTED MANUSCRIPT

Table 5: Comparison of the performance of the ILs designed in this work vs ILs proposed in the
research literature for a pure CO2 gas stream at 303.15 K.

Pressure Mol fraction Ionic Mol fraction Ionic Reference

[Bar] of CO2 Liquid of CO2 Liquid

PT
1 0.019 IL-1 0.0162 [bmmim][BF4 ] a Cadena et al. (2004)
5 0.097 IL-5 0.0729 [bmmim][BF4 ] Cadena et al. (2004)
10 0.189 IL-10 0.134 [bmmim][BF4 ] Cadena et al. (2004)

RI
20 0.340 IL-20 0.343 [bmim][TfO]b Aki et al. (2004)
30 0.642 IL-30 0.483 [C9 mim][PF6 ]c Kim et al. (2011)

SC
a
1-butyl-2,3-dimethylimidazolium tetrafluoroborate
b
1-butyl-3-methylimidazolium trifluoromethanesulfonate
c
1-nonyl-3-methylimidazolium hexafluorophosphate

U
but otherwise the relationship is linear. Hence, simple extrapolation of these results, ignoring
AN
453

454 additional high-pressure effects, indicates that large operating pressures are required for acceptable
M

455 CO2 recovery. It must be highlighted that better CO2 absorption results have been obtained in
D

456 an absorption column (Janiczek et al., 2012). Janiczek et al. (2012) used [emim][Ac] as IL in an
TE

457 absorption column operating at 50 bar and 293.15 K to recover up to 60 % of CO2 . In this work
EP

458 the best result in a single recovery stage are close to 30% at 30 bar and 303.15 K. However, it is

459 clear that the results obtained in this work for a single stage can be improved by using multiple
C
AC

460 contacting stages. In fact, using multiple contact stages would allow to rise the amount of CO2

461 which was absorbed by the IL. This problem will be approached in the next section.

40
 ACCEPTED MANUSCRIPT

Table 6: Performance of the IL ([C10 mim][TfO]) designed in this work for CO2 capture from a PC
steam at 303.15 K

Pressure Mol fraction recovery Ionic

[Bar] of CO2 in IL [%] Liquid

PT
1 0.0029 0.9 [C10 mim][TfO]
5 0.0139 4.7 [C10 mim][TfO]
10 0.0277 9.5 [C10 mim][TfO]

RI
20 0.0546 19.3 [C10 mim][TfO]
30 0.0805 29.3 [C10 mim][TfO]

SC
462 CO2 capture in an absorption column

U
AN
463 After having designed an optimal IL featuring the potential for high recovery of CO2 , we checked

464 that the IL could realize such a separation task by using an absorption column. Feasible operation of
M

465 the absorption column for the capture of CO2 , embedded in a post-combustion gas stream, requires
D

466 to capture more than 90% of the CO2 . To check the feasibility of large CO2 absorption we use the
TE

467 IL previously designed in case II, and simulate the separation using the steady-state equilibrium
EP

468 model of a single absorption column. The column has 10 separation stages numbered from bottom

to upwards and operate at 30 Bar. The stages number was calculated with Kremser method (Seader
C

469
AC

470 et al., 2011). Detailed internal liquid and gas profiles and composition profiles are shown in Table

471 7, whereas in Figure 4 the molar fraction profiles inside the column are shown. In the gas phase

472 the molar fraction of the three components has a smooth slope, especially for water. In the liquid

473 phase the molar fraction of CO2 increases from the top to the bottom, and the IL concentrations

41
 ACCEPTED MANUSCRIPT

474 increases similarly. The slopes of the concentrations in the gas phase are smother because there

475 are more components, and removing CO2 has a small effect in the molar fractions. In Figure 5 the

PT
476 internal molar flowrates profiles are shown. The changes in the flows are due to the amount of CO2

477 which was captured by the IL. This means that CO2 will be present in larger proportion in the

RI
478 liquid phase. The major changes in the flowrates are located around the initial stages, because this

SC
479 part is closer to the post-combustion gas feedstream. Similarly, in Figure 6 the internal temperature

U
480 profile is shown. The results shown in Figure 6 correspond to the tabulated values of the last column
AN
481 in Table 7. The shape of the Temperature profile shown in Figure 6 displays two ”jumps” taking
M

482 place at the top and bottom parts of the absorption column. Such ”jumps” are nothing more than

483 the corresponding feed stream temperatures of the PC gas stream (tray 0) and ionic liquid stream
D
TE

484 (tray 10). Inside the column (trays 2-9) the temperature profile is rather smooth featuring 8 K

485 temperature difference between trays 2 and 9. So, the operating conditions within the column are
EP

486 almost isothermal. However, wide variations in the vapor and liquid streams traveling along the
C

487 column takes place, although such variations are smooth and this explains the internal shape of the
AC

488 temperature profile depicted in Figure 6. It should be remarked that, according to the results shown

489 in Table 7, the absorption column captures 97% of the CO2 contained in the post-combustion gas

490 stream. Also, none of the remaining components of the post-combustion gas stream is absorbed by

42
 ACCEPTED MANUSCRIPT

491 the IL. From the results previously discussed, we know that the amount absorbed of CO2 turns out

492 to be highly pressure dependent: as the operating pressure is increased, also does the amount of

PT
493 CO2 captured by the IL. This the reason why we decided to select 30 bar as the operating pressure

494 of the column, and because it is the higher pressure for which a IL was optimally designed. This

RI
495 pressure is similar to the one used for Basha et al. (2013) for developing a conceptual process for

SC
496 CO2 capture from a Shifted gas stream. Moreover, in the open literature there are some works that

U
497 use higher pressures up to 74 MPa (Carvalho et al., 2009a,b). Regarding the flow rate of IL in the
AN
498 feed stream, this was selected on a trial and error basis until the target amount of CO2 in the IL
M

499 bottoms stream was obtained.

Table 7: Flows and Composition Profiles of the absorption column

D

Molar fraction Molar flow

Temperature
TE

Stage Liquid phase Gas phase Liquid phase Gas phase

CO2 IL CO2 H2 O N2 [kmol/h] [kmol/h] [K]
Feed PC-gas — — 0.1300 0.0900 0.7800 — 1072.8 350.00
1 0.3287 0.6713 0.1009 0.0940 0.8151 414.37 1026.7 337.89
EP

2 0.2445 0.7555 0.0663 0.0966 0.8371 368.22 999.7 336.89

3 0.1847 0.8153 0.0493 0.0984 0.8524 341.18 981.8 335.89
4 0.1395 0.8605 0.0367 0.0997 0.8637 323.27 968.9 334.89
5 0.1039 0.8961 0.0269 0.1007 0.8724 310.43 959.2 333.89
C

6 0.0750 0.9250 0.0192 0.1015 0.8793 300.74 951.6 332.89

7 0.0511 0.9489 0.0129 0.1021 0.8850 293.16 945.6 331.89
AC

8 0.0311 0.9689 0.0078 0.1026 0.8896 287.10 940.7 330.89

9 0.0142 0.9858 0.0035 0.1031 0.8934 282.19 936.7 329.89
Feed IL 0.0000 1.0000 — — — 277.20 — 312.00

500 It should be highlighted that the structure of the ILs is not explicitly taken into account when

43
 ACCEPTED MANUSCRIPT

0.9

0.8

PT
0.7 COl2
Molar fraction

IL
0.6
COg2

RI
0.5
H2O
0.4 N2

SC
0.3

0.2

0.1

U
0
0 2 4 6 8 10
AN
Stage number

Figure 4: Molar fraction profiles

M

501 selecting an optimal molecule. Of course, we recognize that the physical and thermodynamic prop-
D

502 erties of the IL depend upon the chemical structure. However, our optimal design algorithm tends
TE

503 to select molecules featuring larger molecular weights without paying too much attention to the
EP

504 ramification structure. From a performance point of view, large molecular weights look more im-
C

505 portant than the ramification structure. Ideally, ILs featuring large molecular weights and reduced
AC

506 branching structure should result in better CO2 capture characteristics. However, as stated before,

507 the branching structure of the IL is not explicitly taken into account. Therefore, to obtain ILs

508 featuring larger molecular weights , the optimal design algorithm selects branched ILs. Finally, the

509 methods deployed in this work for computing required physical and thermodynamics properties of

44
 ACCEPTED MANUSCRIPT

1100

1000

900

PT
Molar Flow [kmol/h]

800
Gas flow
700 Liquid flow

RI
600

500

SC
400

300

U
200
0 2 4 6 8 10
AN
Stage number

Figure 5: Molar flows profiles

M

510 ILs were selected because they have the best accuracy for ILs based in the imidazolium cation. Also
D

511 a key feature was the reliability of such estimation methods.

TE

Simultaneous ionic liquid and process design

EP

512
C

513 In this section single and multiobjective optimization results are shown and discussed. Our aim is
AC

514 to remark how the chemical structure of the ionic liquid and the processing conditions are highly

515 dependent upon the type of objective function. This stresses the importance of choosing the proper

516 form of objective function and of solving in a simultaneous, rather than in a sequential manner, the

517 design of the ionic liquid and processing conditions. Moreover, recognizing that process economics

45
 ACCEPTED MANUSCRIPT

350

345
Temperature profile
340

PT
Temperature

335

RI
330

325

SC
320

315

U
310
0 2 4 6 8 10
AN
Stage number

Figure 6: Temperature profiles

M

518 is in contradiction with improved process performance, a multiobjective optimization problem, in-
D

519 troduced in section 3.2, is solved to account for trade-offs between these conflicting design and
TE

520 operation target objectives.

EP

521 Single objective function optimization

C
AC

522 The solution of the problem dealing with the simultaneous optimal design of the ionic liquid and

523 processing conditions of the proposed separation flowsheet was addressed using the fixed process

524 flowsheet shown in Figure 3 and the three different objective functions Ω3 , Ω4 and Ω5 and related

525 constraints introduced in section 3.2. For the first two objective functions two operating scenarios

46
 ACCEPTED MANUSCRIPT

526 were defined. In the first scenario no liquid ionic recycle stream was considered, whereas in the

527 second scenario the full liquid ionic recycle stream was taken into account. It should be stressed

PT
528 that the reason why these two objective functions were formulated has to do with the fact that

529 they represent target objectives. In Table 8 the set of optimal ionic liquids as a function of the

RI
530 type of objective function is shown, whereas Tables 9-13 contain the optimal values of pressure,

SC
531 temperature, flowrate and composition for each one of the processing streams shown in Figure 3

U
532 and for the 5 cases defined in Table 8. The results shown in tables 8-13 demonstrate the validity
AN
533 of our hypothesis: the chemical structure of the ionic liquid clearly depends upon the processing
M

534 conditions reflected trough the deployment of different objective functions and operation scenarios.

535 It should be stressed that in Case 5 the Ω5 objective function with full ionic liquid recycle was
D
TE

536 deployed.
EP

Table 8: Ionic liquids designed for 5 cases. Cases 1-2 deploys the Ω3 objective function with no
recycle and recycle of ionic liquid. Similarly, Cases 3-4 deploys the Ω4 objective function with no
C

recycle and recycle of ionic liquid. Finally, in Case 5 the Ω5 objective function with ionic liquid
recycle was deployed.
AC

Chemical Objective CO2 Heat duty IL flow

Case Ionic liquid name structure function removal [kJ/h] [kg/s]
1 1-decyl-3-methyl imidazolium ethyl sulfonate [C10 mim][EtSO4 ] 1 99.999 23.78 34.30
2 1-butyl-3-methyl imidazolium chloride [C4 mim][Cl] 1 90 1.90 17.34
3 1-decyl-3-methyl imidazolium bis(trifluoromethylsulfonyl)imide [C10 mim][Tf2 N] 2 95.572 7.31 50.03
4 1-hexyl-3-methyl imidazolium chloride [C6 mim][Cl] 2 90.665 2.41 19.70
5 1-decyl-3-methyl imidazolium bis(trifluoromethylsulfonyl)imide [C10 mim][Tf2 N] 3 95.677 7.15 48.90
537

47
 ACCEPTED MANUSCRIPT

538 The results shown in Table 8 for cases 1 and 2 indicate that maximum CO2 removal demands

539 maximum thermal load in the flash unit. However, the amount of ionic liquid needed to achieve

PT
540 high CO2 removal decreases, a result that at first sight seems contradictory. Table 8 also shows

541 that for case 1 CO2 removal is the highest. However, deploying this ionic liquid as solvent for CO2

RI
542 capture demands the highest energy requirement. Moreover, the results of case 4 are quite similar

SC
543 to case 2 results. Case 3 is the scenario which demands the largest amount of the corresponding

U
544 ionic liquid to achieve CO2 capture. It should be remarked that the molecular structure of the ionic
AN
545 liquid is different for each case, where the ionic liquid with the longest length of alkyl group in the
M

546 cation corresponds to case 1 which can capture more CO2 because it has more free space in the

547 chemical structure of the IL and lot of groups to interact with the CO2 . However, this becomes a
D
TE

548 disadvantage for ionic liquid recovery because more energy must be supplied for the CO2 and ionic

549 liquid separation. For case 2 the molecular structure is the smallest and has few free space into
EP

550 the ionic liquid and few interaction with CO2 . This explains why the amount of CO2 captured is
C

551 lower when compare to similar results from case 1. Results for case 5 shown in Table 8 also indicate
AC

552 that minimizing just the flowrate of ionic liquid turns out to be a potential expensive solution since,

553 although CO2 removal is high, it also leads to large energy and ionic liquid consumption. Globally

554 speaking, from the results shown in Table 8 we might conclude that Case 2 represents the best trade-

48
 ACCEPTED MANUSCRIPT

Table 9: Data regarding to streams in the CO2 capture system for case 1.
Stream Flow Temperature Pressure composition (mol fraction)
(kmol/h) (K) (Bar) CO2 N2 IL
Flue gas 118.81 298.15 7 0.1429 0.8571 0
Vent gas 100.30 298.15 7 0 1 0

PT
Rich 372.83 298.15 7 0.0455 0.0041 0.9503
Lean 1 354.31 407.50 1 0.0009 0.0043 0.9948
Lean 2 356.17 298.15 7 0 0 1

RI
Product 16.67 407.50 1 1 0 0

SC
555 off between the flash thermal duty and the amount of recovered CO2 . In summary, from information

556 shown in Tables 9-13, we conclude that no ionic liquid is evaporated in the outlet streams displayed

U
in Figure 3. Moreover, deploying the optimal chemical structure of the ionic liquid and optimal
557
AN
558 parameters of the separation process obtained in this work, most of the CO2 in flue gas stream can
M

559 be efficiently captured.

D

Table 10: Data regarding to streams in the CO2 capture system for case 2.
Stream Flow Temperature Pressure composition (mol fraction)
TE

(kmol/h) (K) (Bar) CO2 N2 IL

Flue gas 118.81 298.15 7 0.1429 0.8571 0
Vent gas 102.53 298.15 7 0.0166 0.9834 0
EP

Rich 373.61 298.15 7 0.0409 0.0027 0.9564

Lean 1 357.33 323.15 1 0.0041 0.0028 0.9931
Lean 2 359.82 298.15 7 0 0 1
Product 13.79 323.15 1 1 0 0
C
AC

560 Multiobjective function optimization

561 In CO2 recovery systems one of the main issues consists in designing a process for maximum CO2

562 recovery at minimum cost. It turns out that process economics and CO2 recovery are contradictory

49
 ACCEPTED MANUSCRIPT

Table 11: Data regarding to streams in the CO2 capture system for case 3.
Stream Flow Temperature Pressure composition (mol fraction)
(kmol/h) (K) (Bar) CO2 N2 IL
Flue gas 118.81 298.15 7 0.1429 0.8571 0
Vent gas 101.87 298.15 7 0.0074 0.9926 0

PT
Rich 376.94 298.15 7 0.0465 0.0047 0.9489
Lean 1 360.09 407.50 1 0.0036 0.0032 0.9933
Lean 2 360.09 298.15 7 0.0036 0.0032 0.9933

RI
Product 16.85 407.50 1 0.9630 0.0370 0

SC
Table 12: Data regarding to streams in the CO2 capture system for case 4.
Stream Flow Temperature Pressure composition (mol fraction)
(kmol/h) (K) (Bar) CO2 N2 IL
Flue gas 118.81 298.15 7 0.1429 0.8571 0

U
Vent gas 103.34 298.15 7 0.0153 0.9847 0
Rich 367.51 298.15 7 0.0453 0.0030 0.9517
AN
Lean 1 352.04 323.15 1 0.0036 0.0014 0.9950
Lean 2 352.04 298.15 7 0.0036 0.0014 0.9950
Product 16.01 323.15 1 0.9617 0.0383 0
M

563 objectives. Therefore, a trade-off between both target objectives should be reached. During the
D

564 operation of the CO2 recovery process the operating costs will be dominated by the amount of energy
TE

565 deployed in the flash. Of course, operating costs are also incurred in the absorber column since it
EP

566 will be operated at non-atmospheric conditions. However, because the absorber column pressure has

been fixed, in principle its contribution to the total operating cost will remain constant. Therefore,
C

567
AC

568 in this work we will use the flash heat duty as an indicator of process economics because such a

569 variable is considered a decision variable.

570 Instead of just combining the two objective functions, given by the Ω3 and Ω4 objective functions

50
 ACCEPTED MANUSCRIPT

Table 13: Data regarding to streams in the CO2 capture system for case 5.
Stream Flow Temperature Pressure composition (mol fraction)
(kmol/h) (K) (Bar) CO2 N2 IL
Flue gas 118.81 298.15 7 0.1429 0.8571 0.0000
Vent gas 101.33 298.15 7 0.0074 0.9926 0.0000

PT
Rich 368.85 298.15 7 0.0474 0.0047 0.9479
Lean 1 351.57 323.15 1.01325 0.0025 0.0029 0.9945
Lean 2 355.57 298.15 7 0.0025 0.0029 0.9945

RI
Product 17.28 323.15 1.01325 0.9603 0.0397 0.0000

SC
571 defined in section 3.2, in this work we will approach the simultaneous optimization of such objective

572 functions as a multiobjective optimization problem (Flores-Tlacuahuac et al., 2012). The solution of

U
the multiobjective optimization problem was approached deploying the -constrained method giving
573
AN
574 rise to the Pareto curve. No solution on the Pareto curve is worse or better than any other solution
M

575 on this curve and it is up to the designer to pick the point along this curve that represents the best
D

576 trade-off between both objective functions.

TE

577 In Figure 7 the Pareto curve is shown, whereas in Table 14 tabular information regarding the Pareto
EP

578 curve and additional information is provided. One of the first interesting results shown in Table

14 concerns the chemical structure of the ionic liquid. Except for the last two points, the chemical
C

579
AC

580 structure of the ionic liquid turns out to be different for each one of the points on the Pareto

581 curve. Also, from the information displayed in Table 14, we note that as the % of CO2 recovered

582 is increased, so it does the molecular weight of the corresponding ionic liquid. This result indicates

583 that to increase the amount of CO2 captured would require ionic liquids featuring larger chains.

51
 ACCEPTED MANUSCRIPT

584 However, increasing the molecular weight of the ionic liquid could also increase its viscosity making

585 harder its physical processing. From Table 14 it is also clear that increasing the CO2 recovery would

PT
586 also demand larger amounts of the corresponding ionic liquid flowrate.

RI
Table 14: Pareto curve information for assessing the trade-off between the flash thermal duty and

SC
the % CO2 removed.
% CO2 Rem heat duty IL mass flow CO2 Molecular IL
[kJ/s] [kg/s] purity weight
90.67 2.4058 18.355 0.9617 202.76 [C6mim][Cl]

U
92 2.2329 20.408 0.9597 174.71 [C4mim][Cl]
93 2.4966 21.568 0.95 188.73 [C5mim][Cl]
AN
94 2.9550 22.465 0.95 230.81 [C8mim][Cl]
95 4.7714 33.885 0.95 348.43 [C8mim][DEPO4]
96 5.5306 35.543 0.95 422.59 [C10mim][MDEGSO4]
97 6.3765 41.11 0.95 422.59 [C10mim][MDEGSO4]
M

587

6.5
D

6
TE

5.5

5
Heat duty [kJ/h]

EP

4.5

4
C

3.5
AC

2.5

2
90 91 92 93 94 95 96 97 98
% CO 2 Removed

Figure 7: Pareto curve deploying the flash heat duty and the % CO2 recovery as conflicting objec-
tives.

52
 ACCEPTED MANUSCRIPT

588 5 Conclusions

589 Carbon capture from power plants has gained interest, not only for reducing the greenhouse emis-

PT
590 sions, but also to recycle the CO2 into other materials and processes. In this work, we have proposed

RI
591 a computer-aided molecular design optimization formulation for approaching the optimal design of a

SC
592 new kind of solvents tailored to recover CO2 from post-combustion gas-streams. We have formulated

593 the CAMD problem as an MINLP problem, whose solution provides both the best cation-anion cou-

U
AN
594 pling for CO2 capture and meets target physical and thermodynamic properties. Accordingly, some

595 ILs have been designed assuming that the separation will take place in a single separation stage,
M

596 and using also a pure CO2 stream and a post-combustion stream. The results demonstrate that the
D

597 chemical nature of the IL depends upon the operating pressure, mainly for the pure CO2 stream.
TE

598 When considering a post-combustion gas stream, the results indicate that the chemical nature of
EP

599 the IL remains the same independently of the operating pressure. In fact, using 1-decile-3-methyl

imidazolium trifluoromethanesulfonate [C10 mim][TfO] as the IL, the carbon recovery is close to 30%
C

600
AC

601 at 30 bar in a single stage. Furthermore, as a way to increase carbon capture, the steady-state

602 operation of an absorption column was analyzed. Even when the design of the absorption column

603 was not optimized, by simple trial and error we were able to obtain operating conditions to recover

604 97.65% of the CO2 embedded in the post-combustion gas stream. Considering mixtures of ILs and

53
 ACCEPTED MANUSCRIPT

605 using improved property estimation methods (Lei et al., 2014) can be a interesting question worth

606 to analyze.

PT
607 We have also addressed the simultaneous optimal design of the chemical structure of a ionic liquid,

RI
608 suitable for CO2 capture from post-combustion gas streams, and the optimal value of the processing

parameters in a fixed-structure chemical process deployed for this purpose. Our results indicate

SC
609

610 that both problems are strongly related and that, to obtain better optimal solutions, the interaction

611

U
between both problems should be taken into account. Moreover, it should be stressed that taking
AN
612 into account the interaction between both problems allowed us to set values of the operating pressure
M

613 of the equipments shown in Figure 3 which are relatively low compared to larger operating pressure
D

614 values reported in other works (Ali et al., 2013; Manic et al., 2012b; Sedláková and Wagner, 2012;
TE

615 Valencia-Marquez et al., 2015; Yim et al., 2011). This result is particularly interesting since high

pressure values imply also large electricity costs. Therefore, reducing the operating pressure should
EP

616

617 also contribute to reduce operating costs. Moreover, since commonly sustainability and economics
C

618 goals are contradictory objectives, we have shown that by solving this problem as a multiobjective
AC

619 optimization problem allows us to compute the Pareto solution curve along which the best trade-off

620 solutions between these design objectives can be chosen.

621 Finally, we should stress that the best way of checking the accuracy of our results would be to

54
 ACCEPTED MANUSCRIPT

622 compare them against similar experimental data. Since no experimental work was done for this

623 purpose we cannot claim experimental accuracy. However, from a computational point of view

PT
624 the results seems to indicate that CO2 capture deploying the silicon synthesized ionic liquids look

625 feasible. In our opinion the most uncertain part in our approach has to do with the groups con-

RI
626 tribution methods for the estimation of physical and thermodynamic properties. In future work

SC
627 we will address the impact of process uncertainty on the quality of the results. Another way to

U
628 improve accuracy consists in using molecular simulation-like approaches coupled to free-derivative
AN
629 optimization strategies. Both approaches will be pursued in future work.
M
D
TE
C EP
AC

55
 ACCEPTED MANUSCRIPT

630 Nomenclature

Index

PT
c components (CO2 , H2 O, N2 , IL)
i,p position of the functional group in the molecular structure
j number of valence

RI
k type of functional group or anion (see tables 1,2)
m maximum number for type of functional group
in inlet to absorber column

SC
out outlet from absorber column
L in liquid phase
V in gas phase

U
s stages in absorber column
AN
631
M

632 Binary variables

uik indicate the type (k ) and position (i ) of one

D

functional group in the molecular structure

ωi auxiliary variables in MINLP model
TE

ya binary variable for selection of the anion

z indicate the bond between functional groups (i )
and (p) in the valence (j ) of the group functional (i )
C EP
AC

56
 ACCEPTED MANUSCRIPT

Variables

A coefficient in PR-EOS
B coefficient in PR-EOS
AH coefficient in the model equation for the melting temperature

PT
BH coefficient in the model equation for the melting temperature
CH coefficient in the model equation for the melting temperature
DH coefficient in the model equation for the melting temperature
hF feed enthalpy in flash unit [kJ/kmol]

RI
hV vapor enthalpy in flash unit [kJ/kmol]
hL liquid enthalpy in flash unit [kJ/kmol]
Hi Henry’s law constant [Bar]

SC
Hc enthalpy of component (c) [kJ/kmol]
H12 empirical reference Henry constant [Bar]
x molar fraction in liquid phase

U
y molar fraction in gas or vapor phase
F Molar flow feed to flash unit [kmol/h]
AN
FIL Ionic liquid molar flow [kmol/h]
G Gas molar flow [kmol/h]
L Liquid molar flow [kmol/h]
M

V Vapor molar flow [kmol/h]

MW Molecular weight
nmax maximum number of the functional groups in the molecular structure
D

Nc Total number of components

Nt Total stage number
Smax maximum number of valence of the functional groups
TE

T Temperature [K]
Tb Boiling temperature [K]
∆Tbk contribution of functional group type k to the boiling temperature
EP

Tm Melting temperature [K]

∆Tmp contribution of functional group to the melting temperature
∆Timp ionic contribution of functional group to the melting temperature
C

P Pressure [Bar]
Pcsat vapor pressure of component c [Bar]
AC

Q Heat duty in flash tank [kJ/h]

R Gas constant
RCO2 Removal of CO2 from postcombustion stream
u decision variables
vi∞ infinite dilution volume
z molar fraction fed to flash unit
Z compressibility factor from PR-EOS
ZH coefficient in the model equation for the melting temperature

57
 ACCEPTED MANUSCRIPT

633 Greek variables

PT
α coefficient in the empiric model equation for Henry’s constant
β coefficient in the empiric model equation for Henry’s constant
 coefficient in the empiric model equation for Henry’s constant

RI
γ activity coefficient
γ∗ asymmetric activity coefficient
Λ constants in empirical activity coefficient model

SC
λi constants in empirical activity coefficient model
φc fugacity coefficient of the component (c)
σ number of the same groups that connect to >NH and =N- (ring of imidazolium)
τ number of ring groups in the molecule

U
δ number of C substitutional groups in the molecule
υk valence number of the functional groups type k
AN
φ Fugacity coefficient
Ω Objective function
M
D
TE
C EP
AC

58
 ACCEPTED MANUSCRIPT

634 References

635 Abu-Zahra, M. R., Niederer, J. P., Feron, P. H., and Versteeg, G. F. (2007a). CO2 capture

PT
636 from power plants: Part II. A parametric study of the economical performance based on mono-

RI
637 ethanolamine. Int. J. Greenh. Gas Control., 1(2):135 – 142.

SC
638 Abu-Zahra, M. R., Schneiders, L. H., Niederer, J. P., Feron, P. H., and Versteeg, G. F. (2007b).

639 CO2 capture from power plants: Part I. A parametric study of the technical performance based

U
AN
640 on monoethanolamine. Int. J. Greenh. Gas Control., 1(1):37 – 46.

Agarwal, A., Biegler, L. T., and Zitney, S. E. (2010). Superstructure-Based Optimal Synthesis
M

641

642 of Pressure Swing Adsorption Cycles for Precombustion CO2 Capture. Ind. Eng. Chem. Res.,
D

49(11):5066–5079.
TE

643

644 Aki, S. N. V. K., Mellein, B. R., Saurer, E. M., and Brennecke, J. F. (2004). High-pressure phase be-
EP

645 havior of carbon dioxide with imidazolium-based ionic liquids. J. Phys. Chem. B., 108(52):20355–
C

646 20365.
AC

647 Ali, E., Alnashef, I., Ajbar, A., Mulyono, S., Hizaddin, H., and Hadj-Kali, M. (2013). Determina-

648 tion of cost-effective operating condition for CO2 capturing using 1-butyl-3-methylimidazolium

649 tetrafluoroborate ionic liquid. Korean J. Chem. Eng., 30(11):2068–2077.

59
 ACCEPTED MANUSCRIPT

650 Alie, C., Backham, L., Croiset, E., and Douglas, P. (2005). Simulation of CO2 capture using MEA

651 scrubbing: a flowsheet decomposition method. Energy Convers. Manage., 46(3):475 – 487.

PT
652 Babarao, R., Dai, S., and Jiang, D. (2011). Understanding the High Solubility of CO2 in an Ionic

RI
653 Liquid with the Tetracyanoborate Anion. J. Phys. Chem. B., 115(32):9789–9794.

SC
654 Bansal, D., Cassel, F., Croce, F., Hendrickson, M., Plichta, E., and Salomon, M. (2005). Conduc-

655 tivities and Transport Properties of Gelled Electrolytes with and without an Ionic Liquid for Li

U
AN
656 and Li-Ion Batteries. J. Phys. Chem. B., 109(10):4492–4496.

Bara, J. E., Carlisle, T. K., Gabriel, C. J., Camper, D., Finotello, A., Gin, D. L., and Noble, R. D.
M

657

658 (2009). Guide to CO2 Separations in Imidazolium-Based Room-Temperature Ionic Liquids. Ind.
D

Eng. Chem. Res., 48(6):2739–2751.

TE

659

660 Basha, O. M., Keller, M. J., Luebke, D. R., Resnik, K. P., and Morsi, B. I. (2013). Development
EP

661 of a Conceptual Process for Selective CO2 Capture from Fuel Gas Streams Using [hmim][Tf2 N]
C

662 Ionic Liquid as a Physical Solvent. Energy Fuels., 27(7):3905–3917.

AC

663 Blath, J., Christ, M., Deubler, N., Hirth, T., and Schiestel, T. (2011). Gas solubilities in room

664 temperature ionic liquids - Correlation between RTiL-molar mass and Henry’s law constant.

665 Chem. Eng. J., 172:167–176.

60
 ACCEPTED MANUSCRIPT

666 Brooke, A., Kendrick, D., Meeraus, A., and Raman, R. (1998). GAMS: A User’s Guide. GAMS

667 Development Corporation. , http://www.gams.com.

PT
668 Buhre, B., Elliott, L., Sheng, C., Gupta, R., and Wall, T. (2005). Oxy-fuel combustion technology

RI
669 for coal-fired power generation. Prog. Energy Combust. Sci., 31(4):283 – 307.

SC
670 Cadena, C., Anthony, J. L., Shah, J. K., Morrow, T. I., Brennecke, J. F., and Maginn, E. J. (2004).

671 Why Is CO2 So Soluble in Imidazolium-Based Ionic Liquids?. J. Am. Chem. Soc., 126:5300–5308.

U
AN
672 Carvalho, P. J., Álvarez, V. H., Machado, J. J., Pauly, J., Daridon, J.-L., Marrucho, I. M., Aznar,

M., and Coutinho, J. A. P. (2009a). High pressure phase behavior of carbon dioxide in 1-alkyl-3-
M

673

674 methylimidazolium bis(trifluoromethylsulfonyl)imide ionic liquids. J. Supercrit. Fluids,, 48(2):99–

D

107.
TE

675

676 Carvalho, P. J., Álvarez, V. H., Marrucho, I. M., Aznar, M., and Coutinho, J. A. P.
EP

677 (2009b). High pressure phase behavior of carbon dioxide in 1-butyl-3-methylimidazolium

C

678 bis(trifluoromethylsulfonyl)imide and 1-butyl-3-methylimidazolium dicyanamide ionic liquids. J.

AC

679 Supercrit. Fluids., 50(2):105–111.

680 Carvalho, P. J., Álvarez, V. H., Marrucho, I. M., Aznar, M., and Coutinho, J. A. P. (2010). High

61
 ACCEPTED MANUSCRIPT

681 carbon dioxide solubilities in trihexyltetradecylphosphonium-based ionic liquids. J. Supercrit.

682 Fluids., 52(3):258–265.

PT
683 Carvalho, P. J., Álvarez, V. H., Schröder, B., Gil, A. M., Marrucho, I. M., Aznar, M., Santos, L.

RI
684 M. N. B. F., and Coutinho, J. A. P. (2009c). Specific Solvation Interactions of CO2 on Acetate

and Trifluoroacetate Imidazolium Based Ionic Liquids at High Pressures. J. Phys. Chem. B.,

SC
685

686 113(19):6803–6812.

U
AN
687 Chansomwong, A., Douglas, P. L., Croiset, E., Zanganeh, K., Shafeen, A., and Ricardez-Sandoval,

688 L. (2014a). Control of An Oxy-fuel Capture and Purification Unit For Coal-Based Power Plants.
M

689 Energy Procedia, 63(0):476 – 483.

D

Chansomwong, A., Zanganeh, K., Shafeen, A., Douglas, P., Croiset, E., and Ricardez-Sandoval, L.
TE

690

691 (2014b). Dynamic modelling of a CO2 capture and purification unit for an oxy-coal-fired power
EP

692 plant. Int. J. Greenh. Gas Control., 22(0):111 – 122.

C

693 Chapeaux, A., Simoni, L. D., Ronan, T. S., Stadtherr, M. A., and Brennecke, J. F. (2008). Extrac-
AC

694 tion of alcohols from water with 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide.

695 Green Chem., 10:1301–1306.

62
 ACCEPTED MANUSCRIPT

696 Chapel, D. G., Mariz, C. L., and Ernest, J. (1999). Recovery of CO2 from flue gases: commercial

697 trends. In Canadian Society of Chemical Engineers Annual Meeting, pages 4–6.

PT
698 Chavez-Islas, L. M., Vasquez-Medrano, R., , and Flores-Tlacuahuac, A. (2011a). Optimal Molecular

RI
699 Design of Ionic Liquids for High-Purity Bioethanol Production. Ind. Eng. Chem. Res., 50(9):5153–

5168.

SC
700

701 Chavez-Islas, L. M., Vasquez-Medrano, R., and Flores-Tlacuahuac, A. (2011b). Optimal synthesis of

U
AN
702 a high purity bioethanol distillation column using ionic liquids. Ind. Eng. Chem. Res., 50(9):5175–

703 5190.
M

704 Chen, Q., Aaron, M., and Camarda, K. (2013). Application of quantum-based descriptors to the
D

molecular design of ionic liquids. In 13 AICHE Annual Meeting, San Francisco, CA, USA.
TE

705

706 Chong, F. K., Foo, D. C., Eljack, F. T., Atilhan, M., and Chemmangattuvalappil, N. G. (2015). Ionic
EP

707 liquid design for enhanced carbon dioxide capture by computer-aided molecular design approach.
C

708 Clean Techn. Environ. Policy, 17:1301–1312.

AC

709 Churi, N. and Achenie, L. E. K. (1996). Novel Mathematical Programming Model for Computer

710 Aided Molecular Design. Ind. Eng. Chem. Res., 35(10):3788–3794.

63
 ACCEPTED MANUSCRIPT

711 Davis, J. and Rochelle, G. (2009). Thermal degradation of monoethanolamine at stripper conditions.

712 Energy Procedia, 1(1):327 – 333.

PT
713 Ekstrm, C., Schwendig, F., Biede, O., Franco, F., Haupt, G., de Koeijer, G., Papapavlou, C., and

RI
714 Rkke, P. E. (2009). Techno-Economic Evaluations and Benchmarking of Pre-combustion CO2

Capture and Oxy-fuel Processes Developed in the European ENCAP Project. Energy Procedia,

SC
715

716 1(1):4233 – 4240.

U
AN
717 Flores-Tlacuahuac, A., Morales, P., and Rivera-Toledo, M. (2012). Multiobjective nonlinear model

718 predictive control of a class of chemical reactors. Ind. Eng. Chem. Res., 51:5891–5899.
M

719 Floudas, C. A. (1995). Nonlinear and Mixed-Integer Optimization Fundaments and Applications.
D

Oxford University Press.

TE

720

721 Gani, R., Sinha, M., and Achenie, L. E. K. (2003). Blanket Wash Solvent Blend Design Using
EP

722 Interval Analysis. Ind. Eng. Chem. Res., 42:516–527.

C

Ge, R., Hardacre, C., Jacquemin, J., Nancarrow, P., and Rooney, D. W. (2008). Heat Capacities of
AC

723

724 Ionic Liquids as a Function of Temperature at 0.1 MPa. Measurement and Prediction. J. Chem.

725 Eng. Data., 53:2148–2153.

64
 ACCEPTED MANUSCRIPT

726 Giovanoglou, A., Barlatier, J., Adjiman, C. S., and Pistikopoulos, E. (2003). Optimal solvent design

727 for batch separation based on economic performance. AIChE J., 49:3095–3109.

PT
728 González, H. E., Abildskov, J., and Gani, R. (2007). A computer aided framework for prediction of

RI
729 properties of organic systems. Comput. Aided Chem. Eng., 24:141–146.

SC
730 Gonzalez-Salazar, M. A., Perry, R. J., Vipperla, R.-K., Hernandez-Nogales, A., Nord, L. O.,

731 Michelassi, V., Shisler, R., and Lissianski, V. (2012). Comparison of Current and Advanced

U
AN
732 Post-Combustion CO2 Capture Technologies for Power Plant Applications. Energy Procedia,

733 23(0):3 – 14.

M

734 Greer, T., Bedelbayev, A., Igreja, J., Gomes, J., and Lie, B. (2010). A simulation study on the abate-
D

ment of CO2 emissions by de-absorption with monoethanolamine. Environ. Technol., 31(1):107–

TE

735

736 115.
EP

737 Hada, S., Herring, R., Davis, S., and Eden, M. (2015). Multivariate characterization, modeling, and
C

738 design of ionic liquid molecules. Comput. Chem. Eng., 81:310–322.

AC

739 Harun, N., Nittaya, T., Douglas, P. L., Croiset, E., and Ricardez-Sandoval, L. A. (2012). Dynamic

740 simulation of MEA absorption process for CO2 capture from power plants. Int. J. Greenh. Gas

741 Control., 10(0):295 – 309.

65
 ACCEPTED MANUSCRIPT

742 Hasibur Rahman, M., Siaj, M., and Larachi, F. (2010). Ionic liquids for CO2 capture - Development

743 and progress. Chem. Eng. Proc., 49:313–322.

PT
744 Hehre, W. J. (1976). Ab initio molecular orbital theory. Acc. Chem. Res., 9(11):399–406.

RI
745 Heinze, J., Frontana-Uribe, B. A., and Ludwigs, S. (2010). Electrochemistry of Conducting

SC
746 Polymerss-Persistent Models and New concepts. Chem. Rev., 110(8):4724–4771.

U
747 Huo, Y., Xia, S., and Ma, P. (2009). Group contribution method for predicting melting points of
AN
748 imidazolium and benzimidazolium ionic liquids. Ind. Eng. Chem. Res., 48(4):2212 – 2217.
M

749 Jacek Kumelan, Álvaro Pérez-Salado Kamps, D. T. and Maurer, G. (2006). Solubility of co2 in the

ionic liquids [bmim][ch3so4] and [bmim][pf6]. J. Chem. Eng. Data, 51(5):1802–1807.

D

750
TE

751 Janiczek, P., Kalb, R. S., Thonhauser, G., and Gamse, T. (2012). Carbon dioxide absorption in a

technical-scale-plant utilizing an imidazolium based ionic liquid. Sep. Purif. Technol., 97:20–25.
EP

752

Karunanithi, A. T., Achenie, L. E. K., and Gani, R. (2005). A New Decomposition-Based Computer-
C

753
AC

754 Aided Molecular/Mixture Design Methodology for the Design of Optimal Solvents and Solvent

755 Mixtures. Ind. Eng. Chem. Res., 44(13):4785–4797.

756 Kemper, J., Ewert, G., and Grnewald, M. (2011). Absorption and regeneration performance of

757 novel reactive amine solvents for post-combustion CO2 capture. Energy Procedia, 4(0):232 – 239.

66
 ACCEPTED MANUSCRIPT

758 Kenny, P. W. (2012). Computation, experiment and molecular design. J. Comput. Aided. Mol.

759 Des., 26:69–72.

PT
760 Kim, J. E., Lim, J. S., and Kang, J. W. (2011). Measurement and correlation of solubility of carbon

RI
761 dioxide in 1-alkyl-3-methylimidazolium hexafluorophosphate ionic liquids. Fluid Phase Equilib.,

306(2):251 – 255.

SC
762

763 Klamt, A. and Schuurmann, G. (1993). COSMO: a new approach to dielectric screening in solvents

U
AN
764 with explicit expressions for the screening energy and its gradient. J. Chem. Soc., Perkin Trans.

765 2., 5:799–805.

M

766 Lazzús, J. A. (2012). A group contribution method to predict the melting point of ionic liquids.
D

Fluid Phase Equilib., 313(1):1–6.

TE

767

768 Lee, S.-Y., Ogawa, A., Kanno, M., Nakamoto, H., Yasuda, T., and Watanabe, M. (2010). Nonhu-
EP

769 midified Intermediate Temperature Fuel Cells Using Protic Ionic Liquids. J. Am. Chem. Soc.,
C

770 132(28):9764–9773.
AC

771 Lei, Z., Dai, C., Wang, W., and Chen, B. (2014). UNIFAC model for ionic liquid-CO2 systems.

772 AIChE J., 60(2):716–729.

67
 ACCEPTED MANUSCRIPT

773 Li, H., Yan, J., Yan, J., and Anheden, M. (2009a). Impurity impacts on the purification process in

774 oxy-fuel combustion based CO2 capture and storage system. Appl. Energy, 86(2):202 – 213.

PT
775 Li, Q., Zhang, J., Lei, Z., Zhu, J., Zhu, J., and Huang, X. (2009b). Selection of Ionic Liquids as

RI
776 Entrainers for the Separation of Ethyl Acetate and Ethanol. Ind. Eng. Chem. Res., 48:9006–9012.

SC
777 Li, X., Schatz, G. C., and Nesbitt, D. J. (2012). Anion Effects in the Scattering of CO2 from

778 the Room Temperature Ionic Liquids [bmim][BF4] and [bmim][Tf2N]: Insights from Quantum

U
AN
779 Mechanics/Molecular Mechanics Trajectories. J. Phys. Chem. B., 116(11):3587–3602.

Lynnette A. Blanchard, Z. G. and Brennecke, J. F. (2001). High-pressure phase behavior of ionic

M

780

781 liquid/co2 systems. J. Phys. Chem. B, 105(12):2437–2444.

D
TE

782 Ma, S., Xue, X., Yu, S., and Wang, Z. (2012). High-intensity ultrasound irradiated modification of

783 sugarcane bagasse cellulose in an ionic liquid. Ind. Crops Prod., 35(1):135 – 139.
EP

784 MacDowell, N., Florin, N., Buchard, A., Hallett, J., Galindo, A., Jackson, G., Adjiman, C. S.,
C

Williams, C. K., Shah, N., and Fennell, P. (2010). An overview of CO2 capture technologies.
AC

785

786 Energy Environ. Sci., 3:1645–1669.

787 Manic, M. S., Macedo, E. A., and Najdanovic-Visak, V. (2012a). Trihexyl(tetradecyl)phosponium

68
 ACCEPTED MANUSCRIPT

788 bromide: Liquid density, surface tension and solubility of carbondioxide. Fluid Phase Equilib.,

789 324(25):8–12.

PT
790 Manic, M. S., Queimada, A. J., Macedo, E. A., and Najdanovic-Visak, V. (2012b). High-pressure

RI
791 solubilities of carbon dioxide in ionic liquids based on bis(trifluoromethylsulfonyl)imide and chlo-

ride. J. Supercrit. Fluids, 65(0):1 – 10.

SC
792

793 Mattedi, S., Carvalho, P. J., Coutinho, J. A. P., Álvarez, V. H., and Iglesias, M. (2011). High

U
AN
794 pressure CO2 solubility in N-methyl-2-hydroxyethylammonium protic ionic liquids. J. Supercrit.

795 Fluids., 56(3):224 – 230.

M

796 McLeese, S. E., Eslick, J. C., Hoffmann, N. J., Scurto, A. M., and Camarda, K. V. (2010). Design
D

of ionic liquids via computational molecular design. Comput. Chem. Eng., 34(9):1476–1480.
TE

797

798 Mortazavi-Manesh, S., Satyro, M., and Marriott, R. A. (2011). A semi-empirical Henry’s law
EP

799 expression for carbon dioxide dissolution in ionic liquids. Fluid Phase Equilib., 307:208–215.
C

Muldoon, M. J., Aki, S. N. V. K., Anderson, J. L., Dixon, J. K., and Brennecke, J. F. (2007).
AC

800

801 Improving carbon dioxide solubility in ionic liquids. J. Phys. Chem. B., 111(30):9001–9009.

802 Nittaya, T., Douglas, P. L., Croiset, E., and Ricardez-Sandoval, L. A. (2014). Dynamic Modeling

69
 ACCEPTED MANUSCRIPT

803 and Evaluation of an Industrial-Scale CO2 Capture Plant Using Monoethanolamine Absorption

804 Processes. Ind. Eng. Chem. Res., 53(28):11411–11426.

PT
805 Odele, O. and Macchietto, S. (1993). Computer Aided Molecular Design: A Novel Method for

RI
806 Optimal Solvent Selection. Fluid Phase Equilib., 82:47–54.

SC
807 Orchillés, A. V., Miguel, P. J., Vercher, E., and Martı́nez-Andreu, A. (2007). Ionic Liquids as

808 Entrainers in Extractive Distillation: Isobaric Vapor-Liquid Equilibria for Acetone + Methanol

U
AN
809 + 1-Ethyl-3-methylimidazolium Trifluoromethanesulfonate. J. Chem. Eng. Data., 52:141–147.

Palgunadi, J., Kang, J. E., Nguyen, D. Q., Kim, J. H., Min, B. K., Lee, S. D., Kim, H., and Kim,
M

810

811 H. S. (2009). Solubility of CO2 in dialkylimidazolium dialkylphosphate ionic liquids. Thermochimi.

D

Acta., 494:94–98.
TE

812

813 Prausnitz, J. (1969). Molecular Thermodinamics of Fluid-Phase Equilibria. Prantice-Hall, Inc.

EP

814 Revelli, A.-L., Mutelet, F., and Jaubert, J.-N. (2010). High Carbon Dioxide Solubilities in
C

Imidazolium-Based Ionic Liquids and in Poly(ethyleneglycol) Dimethyl Ether. J. Phys. Chem.

AC

815

816 B., 114(40):12908–12913.

817 Romeo, L. M., Bolea, I., and Escosa, J. M. (2008). Integration of power plant and amine scrubbing

818 to reduce CO2 capture costs. Appl. Therm. Eng., 28(89):1039 – 1046.

70
 ACCEPTED MANUSCRIPT

819 Samudra, A. P. and Sahinidis, N. V. (2013). Optimization-based framework for computer-aided

820 molecular design. AIChE J., 59(10):3686–3701.

PT
821 Seader, J., Henley, E., and Roper, D. (2011). Separation Process Principles. John Wiley & Sons,

RI
822 Inc.

SC
823 Sedláková, Z. and Wagner, Z. (2012). High-Pressure Phase Equilibria in Systems Containing CO2

824 and Ionic Liquid of the [Cn mim][Tf2 N] Type. Chem. Biochem. Eng. Q., 26(1):55–60.

U
AN
825 Shiflett, M. B., Drew, D. W., Cantini, R. A., and Yokozeki, A. (2010). Carbon Dioxide Capture

Using Ionic Liquid 1-Butyl-3-methylimidazolium Acetate. Energy Fuels., 24:5781–5789.

M

826

Shiflett, M. B. and Yokozeki, A. (2006). Solubility and diffusivity of hydrofluorocarbons in room-

D

827
TE

828 temperature ionic liquids. AIChE J., 52(3):1205–1219.

EP

829 Shiflett, M. B. and Yokozeki, A. (2009). Phase behavior of carbon dioxide in ionic liquids:

830 [emim][acetate], [emim][trifluoroacetate], and [emim][acetate] + [emim][trifluoroacetate] mixtures.

C

J. Chem. Eng. Data., 54(1):108–114.

AC

831

832 Shimizu, T., Hirama, T., Hosoda, H., Kitano, K., Inagaki, M., and Tejima, K. (1999). A Twin Fluid-

833 Bed Reactor for Removal of CO2 from Combustion Processes. Chem. Eng. Res. Des., 77(1):62 –

834 68.

71
 ACCEPTED MANUSCRIPT

835 Shin, E.-K. and Lee, B.-C. (2008). High-Pressure Phase Behavior of Carbon Dioxide with Ionic Liq-

836 uids: 1-Alkyl-3-methylimidazolium Trifluoromethanesulfonate. J. Chem. Eng. Data, 53(12):2728–

PT
837 2734.

RI
838 Siougkrou, E., Galindo, A., and Adjiman, C. (2013). Qm - camd: Advances in computer-aided

molecular design of solvents for reactions. In 13 AICHE Annual Meeting, San Francisco, CA,

SC
839

840 USA.

U
AN
841 Soriano, A. N., Doma, B. T. J., and Li, M. H. (2009,). Carbon dioxide solubility in some ionic

842 liquids at moderate pressures. J. Taiwan Inst. Chem. Eng., 40:3873–3893.

M

843 Sun, N., Rahman, M., Qin, Y., Maxim, M. L., Rodrı́guez, H., and Rogers, R. D. (2009). Complete
D

dissolution and partial delignification of wood in the ionic liquid 1-ethyl-3-methylimidazolium

TE

844

845 acetate. Green Chem., 11:646–655.

EP

846 Sun, R., Li, Y., Liu, H., Wu, S., and Lu, C. (2012). CO2 capture performance of calcium-based
C

847 sorbent doped with manganese salts during calcium looping cycle. Appl. Energy, 89(1):368 – 373.
AC

848 Tokuda, H., Ishii, K., Susan, M. A. B. H., Tsuzuki, S., Hayamizu, K., and Watanabe, M. (2006a).

849 Physicochemical Properties and Structures of Room Temperature Ionic Liquids. 3. Variation of

850 Cationic Structures. J. Phys. Chem. B., 110(6):2833–2839.

72
 ACCEPTED MANUSCRIPT

851 Tokuda, H., Tsuzuki, S., Susan, M. A. B. H., Hayamizu, K., and Watanabe, M. (2006b). How Ionic

852 Are Room-Temperature Ionic Liquids? An Indicator of the Physicochemical Properties. J. Phys.

PT
853 Chem. B., 110(39):19593–19600.

RI
854 Trapp, C., de Servi, C., Casella, F., Bardow, A., and Colonna, P. (2015). Dynamic modelling and

validation of pre-combustion CO2 absorption based on a pilot plant at the buggenum IGCC power

SC
855

856 station. Int. J. Greenh. Gas Control., 36(0):13 – 26.

U
AN
857 Ueno, K., Hata, K., Katakabe, T., Kondoh, M., and Watanabe, M. (2008). Nanocomposite Ion Gels

858 Based on Silica Nanoparticles and an Ionic Liquid: Ionic Transport, Viscoelastic Properties, and
M

859 Microstructure. J. Phys. Chem. B., 112(30):9013–9019.

D

ur Rahman, M. H., Siaj, M., and Larachi, F. (2010). Ionic liquids for CO2 capture-Development
TE

860

861 and progress. Chem. Eng. Process., 49(4):313 – 322.

EP

862 Valderrama, J. O., Forero, L. A., and Rojas, R. E. (2012). Critical Properties and Normal Boiling
C

863 Temperature of Ionic Liquids. Update and a New Consistency Test. Ind. Eng. Chem. Res.,
AC

864 51(22):7838–7844.

865 Valderrama, J. O. and Rojas, R. E. (2009). Critical Properties of Ionic Liquids. Revisited. Ind.

866 Eng. Chem. Res., 48(14):6890–6900.

73
 ACCEPTED MANUSCRIPT

867 Valencia-Marquez, D., Flores-Tlacuahuac, A., , and Vasquez-Medrano, R. (2012). Simultaneous

868 Optimal Design of an Extractive Column and Ionic Liquid for the Separation of Bioethanol

PT
869 Water Mixtures. Ind. Eng. Chem. Res., 51(17):5866–5880.

RI
870 Valencia-Marquez, D., Flores-Tlacuahuac, A., and Ricardez-Sandoval, L. (2015). Technoeconomic

and dynamical analysis of a co2 capture pilot-scale plant using ionic liquids. Ind. Eng. Chem.

SC
871

872 Res., 54(45):11360–11370.

U
AN
873 Wang, J., Manovic, V., Wu, Y., and Anthony, E. J. (2010). A study on the activity of cao-based

874 sorbents for capturing CO2 in clean energy processes. Appl. Energy, 87(4):1453 – 1458.
M

875 Wappel, D., Gronald, G., Kalb, R., and Draxler, J. (2010). Ionic liquids for post-combustion CO2
D

absorption. Int. J. Greenh. Gas Control., 4(3):486–494.

TE

876

877 Yim, J.-H., Song, H. N., Lee, B.-C., and Lim, J. S. (2011). High-pressure phase behavior of binary
EP

878 mixtures containing ionic liquid [HMP][Tf2N], [OMP][Tf2N] and carbon dioxide. Fluid Phase
C

879 Equilib., 308(12):147 – 152.

AC

880 Yokozeki, A., Shiflett, M. B., Junk, C. P., Grieco, L. M., and Foo, T. (2008). Physical and

881 chemical absorptions of carbon dioxide in room-temperature ionic liquids. J. Phys. Chem. B.,

882 112(51):16654–16663.

74
 ACCEPTED MANUSCRIPT

883 Yu, H., Hu, J., Fan, J., and Chang, J. (2012). One-Pot Conversion of Sugars and Lignin in Ionic

884 Liquid and Recycling of Ionic Liquid. Ind. Eng. Chem. Res., 51:3452–3457.

PT
885 Zhang, J., Sun, J., Zhang, X., Zhao, Y., and Zhang, S. (2011a). The recent development of CO2

RI
886 fixation and conversion by ionic liquid. Greenhouse Gas Sci. Technol., 1(2):142–159.

SC
887 Zhang, L. L., Wang, J. X., Xiang, Y., Zeng, X. F., and Chen, J. F. (2011b). Absorption of Carbon

888 Dioxide with Ionic Liquid in a Rotating Packed Bed Contactor: Mass Transfer Study. Ind. Eng.

U
AN
889 Chem. Res., 50(11):6957–6964.

Zhang, S., Zhang, Q., and Zhang, Z. C. (2004). Extractive Desulfurization and Denitrogenation of
M

890

891 Fuels Using Ionic Liquids. Ind. Eng. Chem. Res., 43:614–622.
D
TE

892 Zhang, X., Liu, Z., , and Wang, W. (2008). Screening of Ionic Liquids to Capture CO2 by COSMO-

893 RS and Experiments. AIChE J., 54(10):2717–2728.

EP

894 Zhou, Q., Koiwanit, J., Piewkhaow, L., Manuilova, A., Chan, C., Wilson, M., and Tontiwach-
C

wuthikul, P. (2014). A Comparative of Life Cycle Assessment of Post-combustion, Pre-combustion

AC

895

896 and Oxy-fuel CO2 Capture. Energy Procedia, 63(0):7452 – 7458.

897 Ziaii, S., Rochelle, G. T., and Edgar, T. F. (2009). Dynamic Modeling to Minimize Energy

75
 ACCEPTED MANUSCRIPT

898 Use for CO2 Capture in Power Plants by Aqueous Monoethanolamine. Ind. Eng. Chem. Res.,

899 48(13):6105–6111.

PT
RI
U SC
AN
M
D
TE
C EP
AC

76