Separation and Purification Technology 233 (2020) 116050

Contents lists available at ScienceDirect

Separation and Purification Technology

journal homepage: www.elsevier.com/locate/seppur

Assessment of ionic liquids as H2S physical absorbents by thermodynamic T

and kinetic analysis based on process simulation
Rubén Santiago, Jesús Lemus , Ana Xiao Outomuro, Jorge Bedia, José Palomar
⁎

Chemical Engineering Department, Universidad Autónoma de Madrid, 28049 Madrid, Spain

ARTICLE INFO ABSTRACT

Keywords: A comprehensive evaluation of ionic liquids (ILs) as potential H2S absorbents was performed using both mo-
H2S lecular and process simulation. First, the Conductor-like-Screening MOdel for Real Solvents (COSMO-RS
Absorption method) was applied to select promising ILs absorbents and to understand the H2S gas solubility from a mo-
COSMO-RS lecular point of view. The ILs screening more than 700 ionic combinations determines that H2S physical ab-
Ionic liquids
sorption is mainly controlled by the hydrogen-bond acceptor capacity of the anion, due to the easily hydrogen
Aspen plus
bond formation when mixed with the acidic solute. Based on molecular simulation analysis, 6 ILs of different
nature were evaluated in a typical industrial packed absorption column using COSMO-based/Aspen Plus
methodology. Equilibrium based simulations demonstrated higher H2S separation efficiency (i.e. lower solvent
expenses and smaller equipment sizes) when increasing H2S absorption capacity of the IL solvent. In contrast,
rigorous process simulation analysis (including kinetic equations) reveals a strong mass transfer kinetic control
in the H2S absorption in commercial packed column, which severely limits the maximum H2S absorption given
by thermodynamics. As a result, ILs that present the best performance in the thermodynamic aspect, become
worse for the operation. In fact, it was found that H2S recovery at given operating conditions increases when
decreasing the viscosity of IL, being 1-ethyl-3-methylimidazolium dicyanamide, the one that presents the best
absorbent performance, requiring the lowest operating temperatures and liquid volume flows. Lastly, the ab-
sorption operation was designed to achieve fixed H2S recovery using different liquid/gas feed ratios, resulting in
column heights and diameters inside the typical range marked by heuristic rules for usual industrial packed
columns. In sum, current prospective study based on COSMO-RS and Aspen Plus have been proved as a useful
tool to analyze the potential industrial application of ILs in the H2S capture and to select the most adequate ILs,
before starting with experimental tests, highly demanding in cost and time.

1. Introduction Some technologies are available for H2S treatment. Starting with the
adsorption operation, some adsorbents such as silica gel, active carbon
In recent years, natural gas has become one of the most promising and zeolites are traditionally used. Specially, high specific surface
alternatives to the traditional and pollutant energy sources due to its zeolites demonstrated a great efficiency with high H2S selectivity, and
cleanliness [1]. The active component in natural gas is methane (CH4), great stability until high temperatures [6]. However, the main dis-
but it usually contains undesired impurities such as carbon dioxide advantage of this technology is the high energy demanding in the re-
(CO2) and hydrogen sulfide (H2S) [2]. H2S is an extremely toxic and generation operation that hinders the application of this technology
corrosive gas that can cause several diseases and even the potential [7]. The technology most widely studied is H2S absorption. In fact, a
dead in enough concentrations [3]. H2S is one of the responsible huge number of solvent-based processes were analyzed to remove this
compounds of the well-known “acid rain” when it is oxidized to SO2, acid gas [8]. The separation has been accomplished by using chemical
causing extremely negative effects to the different ecosystems [4]. For and/or physical absorption processes. Starting with the former, the
all these reasons, the complete removal of H2S from natural gas is re- most typical absorbents used are the aqueous organic amine solutions
quired not only for a safety transport, but also for its correct utilization. [9,10]. They present high absorption capacities at low partial pressures,
Furthermore, the removal of this acid gas increases the calorific value of but they become less effective when the absorbent is reaching the sa-
the natural gas [5]. turation [8]. These absorbents show some disadvantages in their use,

⁎
Corresponding author.
E-mail address: [email protected] (J. Lemus).

https://doi.org/10.1016/j.seppur.2019.116050
Received 30 May 2019; Received in revised form 2 September 2019; Accepted 7 September 2019
Available online 09 September 2019
1383-5866/ © 2019 Elsevier B.V. All rights reserved.
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

such as the high energy consumption due to the heat requirements to process at different operating temperatures was analyzed by using
release the H2S from the solvent and their high losses because of their rigorous Rate-Based column model. Finally, the sizing of the absorption
high volatility [11]. In addition, the presence of CO2 in the streams, column was performed for the best IL in order to achieve 95% H2S
results in competitive absorption, being the H2S separation capacity recovery at different L/G ratios, obtaining in each case the required
reduced [11]. In the case of physical absorption, it presents some ad- height and diameter of the packed column and comparing the results
vantages when compared with the chemical one, such as the easy re- with the heuristic industrial parameters.
generation at lower temperatures. Some physical absorbents are me-
thanol, dimethylether, and morpholine derivatives, which are 2. Methodology
employed in commercial processes: Rectisol®, Selexol® and Morphy-
sorb®, respectively [12]. However, Rectisol® presents a high volatility 2.1. COSMO-RS calculations
and low H2S selectivity when CO2 is present on the stream to treat.
Selexol® presents lower vapor pressure than traditional solvents usually The first stage in COSMO-RS calculations is to create the *.cosmo
employed, but it presents high viscosity, which will result in low mass files of the ILs for their use in COSMOtherm software. For this, the
transfer kinetics. Morphysorb® presents high volatility, which will make optimization of the compounds was performed at BP86/TZVP compu-
more difficult the regeneration stage. [13]. These advantageous prop- tational level by using Turbomole 7.0 software. ILs are defined using
erties are the high and tunable solvent capacity, negligible volatility, the molecular model of independent counter ions (C + A) in which the
non-flammability, low corrosivity and relatively high chemical and cation and the anion are optimized separately. These optimizations
thermal stability [13,14]. Furthermore, the possibility of customizing were carried out taking the compounds to the structure of minimum
the cation and/or the anion for specific tasks made them considered as energy, with the solvent effect through the COSMO continuum solva-
“designer solvents” [15]. The introduction of the ILs to physically ab- tion method. Once the optimization was finished, the polarized charge
sorb H2S has attracted a great scientific community interest in recent distribution of the compounds is obtained by a COSMO single point
years. Several works studying the H2S solubility in different ILs physical calculation, saving it into *.cosmo files. Then, the COSMO-RS calcula-
absorbents were reported [16–20]. In that way, a pioneer work by tions were performed using the COSMOtherm program package (ver-
Pomelli et al. [21] reported an experimental and theoretical study on sion C30_1704) and its parametrization BP_TZVP_C30_1701. The charge
solubility using different nature based ILs, concluding the importance of distributions of both H2S, as pure solute, and ILs, as absorbents, can be
hydrogen bond interactions between the anion and the acid gas, easily visualized in σ-profile and σ-potential. In addition, vapor pres-
showing less importance the cation selection. This finding was con- sure calculation of H2S at different temperatures was performed. In the
firmed later by Aparicio et al. [22] and Sakhaeinia et al. [23]. In par- case of the solute-solvent mixtures, Henry’s law constants of H2S in ILs
allel, several studies using COSMO-RS method performed computa- and detailed contributions to excess enthalpies in equimolar mixtures
tional screenings to select ILs with high H2S gas solubility, in order to were calculated as well. First, Henry’s law constants (KH) are calculated
minimize the high cost and time demanding experimental tests as follows:
[24–26].
KH = ·p0vap [1]
In recent years, we have followed a multiscale research strategy, i

that integrates molecular and process simulations, to design new gas where KH is calculated as the product of i , which is the activity
separation process based on ILs [27–29], successfully applied to the coefficient of the solute at infinite dilution in the IL (calculated by
capture of CO2 [30–33], NH3 [34], toluene [35], acetylene [27] and COSMOtherm) and p0vap , which is the vapor pressure of the pure gas,
other volatiles organic compounds [36–38]. This computational meth- estimated by Antoine equation (experimental data from reference
odology can be applied to preliminary evaluate the technical viability [41]).
of novel gas absorption process using ILs prior to experimental tests. For Second, the excess enthalpy of an equimolar mixture (HE) is cal-
this purpose, the free ILUAM database [28] was recently developed to culated by COSMOtherm following the next expression:
include 100 common ILs as compounds in Aspen Plus/Hysys software,
HE = H E (MF ) + HE (HB ) + HE (vdW ) [2]
in order to easily model individual operation units and complete se-
paration processes based on ILs. This COSMO-based/Aspen Plus ap- where HE (MF ) is the energy contribution associated to Misfit forces;
proach allows considering, simultaneously, thermodynamic, kinetic, HE (MF ) is that attributed to the hydrogen bonding and HE (vdW ) is the
technical, energetic and economic criteria in the selection of the most one ascribed to Van der Waals forces.
adequate IL absorbent. In addition, operating conditions, equipment
design and process configuration can be optimized in order to promote 2.2. Process simulation details
the technical and economic viability of the new separation process
based on ILs [30,31,39,40]. ILs were directly used from ILUAM database [28], which includes
The main aim of this work is to evaluate and select adequate ILs for information and parameters of 100 commercial ILs for its application in
the H2S absorption by means of the COSMO-based/Aspen Plus metho- Aspen Process simulators. COSMOSAC property package was selected
dology. First, molecular simulation was used to select ILs with favorable for the thermodynamic calculations. In the case of the conventional
thermodynamic properties for H2S capture. For this purpose, a gaseous compounds used in this work, which are N2, CH4, H2S, and CO2,
COSMO–RS screening of Henry’s law constants of H2S in more than 700 they were loaded from Aspen Plus Database including two parameters
ILs was carried out to determine the more favorable cation-combination for the complete definition of COSMOSAC property package. In this
for H2S absorption, selecting a sample of ILs with favorable H2S ab- sense, the σ-profile and COSMO volume of the gaseous compounds were
sorbent properties. Then, the excess enthalpies of H2S-IL binary mix- first calculated by using COSMOtherm software and then introduced to
tures were analyzed in terms of the solute-solvent intermolecular in- Aspen Plus process simulator. The use of COSMOSAC property package
teractions, in order to obtain deeper insight of H2S physical absorption lets the user select the model that better fits to the system. Therefore,
phenomena in IL. Finally, process simulation with Aspen Plus process the user can select between: (i) the original COSMO-SAC model pro-
simulator was accomplished to describe the H2S capture by IL in posed by Lin and Sandler [42] (denoted as COSMOSAC 1); (ii) the
packing column at industrial scale. An equilibrium-based model was original COSMO-RS model developed by Klamt [43] (COSMOSAC 2);
preliminary applied to analyze the role of thermodynamics in H2S ab- and (iii) a modification of the Lin and Sandler model performed by P.
sorption operation, including the influence of the operating pressure, Mathias et al. [44] (COSMOSAC 3). In the present work, the three
gas/liquid ratio (L/G) and number of column stages (N) on the H2S models were used for selecting the one that best fits to the experimental
recovery. Then, the potential kinetic control on the H2S absorption data. Mean relative error were calculated by Eq. (3) using the

2
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

Table 1 polarized charge distribution (σ). This charge distribution can be

Inlet gas feed general characteristics. summarized in a histogram named σ-profile that permits a simple
Stream characteristics Value analysis of the chemical nature of the compounds involved. Next,
P, atm 0.98 COSMO-RS introduces an additional tool called σ-potential in which the
T, °C 25 charge density concept is introduced. This is defined as the chemical
Total Molar Flow, kmol/h 1301 potential of a surface unit with a polarized charge distribution in a
Total Mass Flow, ton/h 22.5
solvent, and allows analyzing the intermolecular interactions between
Component MW, kg/kmol yi Flow, kmol/h the compound and its environment [46]. In this sense, Fig. 1 shows the
CH4 16 0.940 1222.9
σ-profile and σ-potential results of the solute studied (H2S). It is im-
CO2 44 0.025 32.5
N2 14 0.015 19.5 portant to remark the three different regions in which the histograms
H2S 34 0.020 26.1 are divided. The first one that includes the range of polarities
σ < −0.0082 e/Å2 is referred to the hydrogen bond donor groups, the
second one (−0.0082 < σ > +0.0082 e/Å2) includes the non-polar
experimental and calculated gas–liquid equilibria (GLE) data. region and the last one σ > +0.0082 e/Å2 associated to hydrogen
1 |xcalculated x experimental | bond acceptor groups. The σ-profile of H2S presents a peak at the ne-
Mean relative error (%) = ·100 gative region (hydrogen bond donor group), in agreement with the
Npoint x calculated [3]
weak H2S acidity. On the other hand, attending to the σ–potential
The RADFRAC model implemented as default in Aspen Plus v9.0 (Fig. 1B), it is observed that H2S would present attractive interactions
was used to model the H2S absorption column. To accomplish this, two with non-polar compounds (minimum energy potential with −0.13 kJ/
operation modes of RADFRAC model were used: Equilibrium mode, in mol Å2 value located at 0.001 e/Å2) and polar compounds presenting
which only the thermodynamic equilibrium is considered, and Rate- weak hydrogen bond acceptor groups (minimum energy potential with
Based mode, in which also the mass transfer kinetics is taken into ac- −0.08 kJ/mol Å2 value located at 0.018 e/Å2), as seen on the exo-
count. For Rate-Based calculations, a fractional capacity of 65% was thermic behavior near the cut-off region.
fixed to ensure a similar fluid dynamic behavior. A packing type Raschig The next stage of molecular simulation is to select ILs among a huge
Rings with a diameter of 76.2 mm was selected following the heuristic database. A widely used key parameter for the selection of ILs as ab-
rules for maintaining a D/d ratio around 20 [45]. sorbent -attending to thermodynamic criteria- is the Henry’s law con-
The inlet gas feed presents the composition summarized in Table 1, stant of the gas solute in the IL. Following the equation (1), COSMO-RS
accordingly to Faramawy et al. work, for typical acid gas natural method calculates KH by means of the activity coefficient of the solute
streams of industrial plants [2]. at infinite dilution in the mixture and the vapor pressure of the pure
Different simulation cases were performed for studying both the solute. In current COSMO-RS calculations, the experimental vapor
thermodynamics and the kinetics of the H2S absorption operation using pressure was included in the software by means of Andrade equation
ILs: Case 1): The Equilibrium mode of RADFRAC column was used for parameters (A = 4.52887; B = 958.587 and C = -0.539 units of bar).
studying the influence of the L/G ratio at two given pressures in the H2S For validation purposes, Fig. 2 shows the experimental KH vs the cal-
recovery in different ILs. Case 2): Once again using the equilibrium culated KH for several different solutes in ILs, using the [C + A] model
column, the number of stages was varied for obtaining 99% of recovery to describe the IL component in COSMO-RS simulations, following the
at a fixed L/G ratio. Case 3): The H2S recoveries as a function of the methodology of previous works [27,32,36]. It is important to remark
temperature are estimated for the selected ILs using Equilibrium and that the experimental KH values were collected from Gonzalez-Miquel
Rate-Based mode in RADFRAC column (1 m of height). Case 4): et al. work [36] (more information in Table S1 of Supporting
Comparison of L/G ratio for reaching a 95% H2S recovery at different Information). As can be seen, the statistical parameters (slope = 0.99
temperatures between Equilibrium and Rate-Based mode of different ILs. and square correlation coefficient R2 = 0.98) show the good predict-
All the used conditions for each case are summarized in Table 2. Last, ability of COSMO-RS approach used in this work, including the avail-
the column design (height and diameter) was accomplished for the best able data for H2S-IL systems (Table S1 of Supporting Information col-
case taking into account heuristic rules. In this sense, the enough height lects all the data and references presented in Fig. 2).
for reaching a 95% H2S recovery, and the diameter needed for a frac- Then, a complete Henry’s law constant screening among more than
tional approach to maximum capacity of 65% as a function of the L/G 20 combinations cation-anion (> 700 ILs) was performed for selecting
ratio. ILs with potential properties for H2S absorption. Remarkably different
nature cations and anions were used for the screening (see Table S2 in
3. Results the Supplementary Material for the nomenclature of the ILs used).
Fig. 3 presents the results of predicted KH of H2S in the different cation-
3.1. Molecular simulation anion combinations. It is important to remark that, following the Eq.
(1), low KH values are traduced into high solubility values in molar
COSMO-RS methodology allows predicting the thermodynamic terms. Therefore, we search combinations cation-anion that present the
properties of solute-IL systems and analyzing the intermolecular inter- lowest KH values. Analyzing the Fig. 3, it is observed that the KH values
actions governing the mixture behavior. This is due to the quantifica- for several ILs are lower than 2 bar, indicating a higher H2S solubility
tion of the interaction energy between the surfaces by means of the than that of CO2 in ILs [33] and in the order of that of similar to NH3 in
ILs [47]; i.e, we found favorable cation-anion combinations for the H2S
Table 2
absorption in comparison to other solutes [36]. Attending to COSMO-
Conditions of the different cases of study. RS predictions depicted in Fig. 3, the selection of both anion and cation
has significant influence on the H2S absorption capacity of the IL [40],
Case 1 2 3 4
obtaining higher values with those ILs composed by anion with basic
Mode EQ EQ EQ/RB EQ/RB character ([MePO3]−, [(Me)2PO4]−, [MeSO3]− …) and aprotic cations
Pressure, atm 3–30 30 30 30 with non-polar character (phosphonium, ammonium …).
T, °C 25 25 0–200 0–100 To analyze the possible relationship between the H2S solubility in
L/G, kg/kg 0–200 – 30 –
ILs and the excess enthalpy of H2S-IL mixtures, COSMO-RS method was
H2S recov, % – 99 – 95
Number of Stages 2 2–10 5 5
applied to calculate the HE values of the H2S-IL mixtures at H2S con-
centrations corresponding to the solubility data in ILs. Fig. 4 relates the

3
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

Fig. 1. (A) σ-profile and (B) σ-potential of pure H2S.

Fig. 4. Molar Excess Enthalpies (HE) vs Henry’s law constants (KH) of equimolar
H2S-ILs mixtures at 25 °C.
Fig. 2. Henry’s law constant prediction validation using [C + A] model and
experimental vapor pressure of solutes using COSMO-RS calculations at near
room temperature.

[bpyr]
[hxpyr]
[Choline]
[N111,10] 1
[nmim]
[npy] 2
[omim]
[hxmim] 3
[hxpy]
[mpy]
Cations

4
[N1114]
[bmim] Fig. 5. Excess Enthalpies (HE) contributions (MF, HB and vdW) and Henry’s law
10
[bpy] constants (KH) of H2S with different anions nature with the same cation
[ompyr]
[ompip] [emim]+.
[hxmpip]
[N6666]
[P5555]
[bmpip] In order to obtain more insights in the intermolecular contributions
[P222,10]
[empyr] to the H2S absorption, different COSMO-RS analysis can be performed.
[N2222] In this sense, we first analyze the effect of the anion nature on H2S-IL
[P2222]
mixture behavior. It is going to be analyzed different chemical nature
[BETI]

[PF6]
[FEP]
[Cl]

[EtPO3]

[(Et)2PO4]

[EtSO4]

[TCM]
[ClO4]

[B(CN)4]
[MePO3]

[(Me)2PO4]

[BuPO3]
[MeSO3]
[TFA]
[NO3]

[MeSO4]
[OcSO4]
[BF4]
[TfO]
[DCN]
[SCN]
[NTf2]

effects of ILs. First, Fig. 5 presents the calculated HE and KH for ILs series
of common [emim]+ cation and different anions. As can be seen, re-
Anions
markably enhanced H2S gas solubility can be obtained by the proper
Fig. 3. Henry’s law constants screening (bar) of H2S in 529 ILs at 25 °C. selection of the anion. Thus, ILs with strong hydrogen-bond acceptor
groups promote favorable solute-solvent intermolecular interactions
predicted KH values of H2S in IL to the excess enthalpy of the equimolar that increase the gas solubility. On the contrary, anion with delocalized
solute-solvent mixtures at 25 °C. As can be seen, the gas solubility in- charge, such as [PF6]−, is not able to form local HB interactions with
creases (lower KH) in more exothermic mixtures, indicating that the the solute, and as result, low solubility of H2S in the ILs formed by this
favorable intermolecular interactions between H2S and IL determine anion is expected. It was demonstrated that the anion contributes ma-
the absorption process, in good agreement with the previously observed jority to the H2S physical absorption and the importance of selecting
for CO2, NH3, VOCs, etc. capture [27,30,32,34,36,40,47]. As can be anions that present basic character.
seen, from molecular simulations, we can select ILs that present higher It is also analyzed the effect of different alkyl chains with the same
exothermic mixtures and lower KH values than common ILs widely used anion ([(Me)2PO4]−). Fig. 6 presents the contributions to excess en-
in bibliography [16–18,48,49]. thalpy and KH as a function of the alkyl chain of the methylimidazolium

4
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

Table 3
Mean relative errors of calculated solubilities using the different COSMOSAC
property models, at temperatures between 30 and 80 °C.
IL [omim][PF6] [emim][NTf2] [hxmim] [omim][NTf2]
[NTf2]
Refs. [51] [49] [48] [48]
P range (MPa) 0.12–1.92 0.16–1.61 0.11–2.02 0.13–1.69

COSMOSAC Relative Error (%)

model
1 15.3 16.0 14.3 8.5
2 22.5 27.2 27.7 15.3
Fig. 6. Excess Enthalpies (HE) contributions (MF, HB and vdW) and Henry’s law
3 16.1 18.7 16.4 9.1
constants (KH) of H2S in different imidazolium based cations with the same
anion [(Me)2PO4]−.

comparing to available experimental data. Table 3 collects the mean

based cation. As can be seen, the cation nature also affects the H2S relative errors of the gas-liquid equilibria (Eq. (3)) of four different ILs
solubility in the IL, but with less influence than that of the anion. In all using the three COSMOSAC models commented in the methodology
studied cases, the hydrogen bonding interactions present the higher section. The experimental and calculated gas–liquid equilibria (GLE)
contribution, being almost 68% of the total. The exothermic behavior of data used to calculate these errors are collected in Fig. S1 of the
mixture is related to favorable solute-solvent interactions. ILs formed by Supplementary Material. All the models describe reasonably well the
methylimidazolium based cations with short alkyl chains present GLE behavior of the mixture, presenting relative error < 30%. COS-
slightly better performance in the H2S absorption, since the attractive MOSAC 1 property model shows the best fit to the experimental data,
electrostatic interactions are more effective when using shorter alkyl with errors lower than 16%. This low dispersion can be considered an
chains. admissible error on initial stages of conceptual engineering of new
The last effect is shown in Fig. 7 presenting the results of all the processes based on ILs [28]. Therefore, COSMOSAC 1 is selected as the
contributions to excess enthalpies and KH of H2S in different ILs pre- property model for next stages of the work.
senting the same anion [MePO3]−, but varying the head group of the Once COSMO-based/Aspen approach was validated, different pro-
cation. As in previous analysis, the main contribution to the H2S-IL cess simulations were carried out to systematically evaluate the per-
mixing behavior is the hydrogen bond interactions (HB), representing a formance of the selected ILs in H2S absorption from a typical acid gas
70% of the total excess enthalpy value. ILs with cations presenting natural stream (Table 1), attending to the four cases detailed in Table 2.
hydrogen-bond donor groups (as imidazolium or pyrrolidinium) pre- Case 1: The performance of the selected ILs was first evaluated by
sent lower H2S gas solubility, due to the competition between hy- modeling the absorption column in Equilibrium mode at a given number
drogen-bond interactions of cation and anion and H2S–anion interac- of stages and operating temperature. Fig. 8 presents the achieved H2S
tions, as it was observed with other solutes [36,40,50]. In sum, current recovery when using the 6 different ILs with increasing liquid mass flow
COSMO-RS results reveal that both anion and cation present great in- at two operating pressures (3 and 30 atm).
fluence in the behavior of H2S-IL mixtures. For the following process It is clearly shown the advantage of using [emim][MePO3] from the
simulation analysis in this work, we selected 3 ILs that present favor- thermodynamic point of view, due to the lower L/G ratios required to
able thermodynamic properties ([emim][MePO3], [bmim][TFA] and reach high H2S recoveries when compared to the other ILs. Comparing
[emim][OcSO4]) from COSMO-RS analysis. In addition, we selected the two operation pressures (3 to 30 atm), it can be seen that a higher
three ILs ([emim][DCN], [emim][SCN] and [emim][NTf2]) that present operating pressure allows using lower L/G ratios to reach the same H2S
worse thermodynamic behavior but lower viscosities and thus better recovery. [emim][MePO3] presents the best thermodynamic behavior
expected transport properties [33]. at both pressures. The overall process analysis suggests that H2S ab-
sorption by ILs is not impeded from the thermodynamic point of view,
3.2. Process simulation analysis reaching at 30 atm high solute recoveries with L/G ratios in the in-
dustrial range [52].
As first stage, COSMO-based/Aspen methodology was validated for Case 2: The aim of this study is to analyze the effect of the number of
the prediction gas-liquid equilibria data of H2S-IL mixtures by equilibrium stages of the absorption column in H2S separation. Fig. 9

Fig. 7. Excess Enthalpies (HE) contributions (MF, HB and vdW) and Henry’s law constants (KH) of H2S with different cations nature with the same anion [MePO3]−.

5
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

Fig. 8. H2S recoveries as a function of the L/G ratio at (A) 3 atm and (B) 30 atm (Case 1 in Table 2).

stages.
Case 3: This case of study introduces the mass transfer kinetics in the
description of the absorption operation by using RADFRAC column in
Rate-based mode. Thus, a more realistic scenario is studied, considering
the transport properties of the mixture to estimate the H2S separation
efficiency by ILs in commercial packed columns. Fig. 10 compares the
H2S recovery as a function of the absorption temperature using RAD-
FRAC column in Equilibrium and Rate-Based modes with [emim]
[MePO3] and [emim][DCN] ILs as absorbers. Simulations based on
equilibrium show curves that follow the expected trends from ex-
clusively thermodynamic consideration, i.e. the H2S recovery increases
when temperature decreases for both ILs, due to the higher gas solu-
Fig. 9. Stages number vs L/G ratio needed to reach a 99% H2S recovery, using
bility. In contrast, when absorption rate is considered in calculations,
different ILs at 30 atm of pressure (Case 2 in Table 2).
very low H2S separation recovery was obtained for [emim][MePO3] at
near room temperature, as it was already observed when analyzing CO2
presents the calculated number of stages required for reaching a H2S capture by physical absorption with ILs [31,33]. The results observed in
recovery of 99% when using different L/G ratio for the selected ILs. As Fig. 10A reveal a strong kinetic control in the H2S absorption phe-
can be seen, from equilibrium-based simulations, it is possible to obtain nomena in [emim][MePO3] when using a commercial packed column at
high H2S recoveries in a wide range of L/G and Nstages design variables. relatively low temperatures, assignable to the high viscosity of IL
As expected, increasing the number of stages in the absorption column [31,33]. Then, when increasing the operating temperature, the diffu-
decreases the IL mass flow required to achieve 99% of recovery. sivity of H2S in IL is progressively enhanced, as consequence of the
However, when increasing more than 5 stages, the required L/G ratio decreased IL viscosity [54], obtaining a maximum of H2S recovery
decreases in less proportion, converging into the same value. The curve at 120 °C, then thermodynamics start to control the absorption
shaded zone of the graph is referred to the work range (3 – 5 stages) process. Similar results are obtained for [emim][DCN] (Fig. 10B), but
more usually employed. This range was selected taking into account the with less remarkable differences between Equilibrium and Rate-Based
heuristic value of 1 – 1.5 of L/G minimum for the separation [53]. For mode due to the much lower viscosity of [emim][DCN] (16.83 cP @
next stages, it was decided to work with an absorption column of 5

Fig. 10. Comparative of the different H2S recoveries as a function of the absorption temperature in both Equilibrium and Rate-Based modes of (A) [emim][MePO3]
and (B) [emim][DCN] (Case 3 in Table 2).

6
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

Fig. 13. Absorption column design using [emim][DCN] as absorbent.

Fig. 11. H2S recoveries of the different ILs as a function of the absorption
temperature (Case 3 in Table 2).
favorable thermodynamic behavior is [emim][MePO3]. In contrast,
25 °C [55]) when compared to [emim][MePO3] (149.12 cP @ 25 °C Fig. 12B shows opposite trends of required L/G when increasing the
[56]). operating absorption temperature. In this case, the IL solvent expenses
Fig. 11 completes current analysis, presenting the obtained H2S are higher at low temperatures due to the increasing IL viscosities [33].
recovery for all selected ILs when rigorous Rate-Based absorption In fact, the curves in Fig. 12B follow an exponential decrease in good
column model was used in simulations at different operating tem- agreement with the dependence of viscosity values with temperature
peratures (25–200 °C). The curves pass through a maximum in all cases, [54]. Again, it is observed that the IL [emim][DCN] is the less viscous
but at different temperature depending on the IL. Thus, [emim][DCN], solvent. It is also important to remark the much higher L/G ratio needed
[emim][SCN] and [emim][NTf2] allow obtaining the maximum H2S (1000 time higher) when the mass transfer kinetics is taken into ac-
recovery at lower operating temperatures than those of [emim] count in separation simulations at near room temperatures.
[MePO3], [emim][OcSO4] and [emim][TFA]. This may be explained by
the lower viscosities of the first IL group compared to those of the 3.3. Absorption column design
second one [54]. In fact, at near room temperatures, the H2S absorbent
performance follows the trend of the IL viscosity values, which confirms Based on above results, the last purpose of the analysis is to design
that the mass transfer rate is controlling the H2S absorption process in an H2S absorption operation using [emim][DCN] at 50 °C and 30 atm,
the column. On the contrary, at higher operating temperatures (than to obtain a H2S recovery of 95%. Fig. 13 presents the required height
that corresponding to curve maximum) the absorption separation is (H) and diameter (D) of the packed column as a function of the L/G
determined by the thermodynamics of the H2S-IL mixture (as in Fig. 9). ratio. As can be seen, the use of a higher L/G ratio entails smaller
Based on current results, it is concluded that [emim][DCN], operating column heights. On the other hand, an increase in the liquid mass flow
at an absorption temperature of 50 °C, presents the best performance as demands larger column diameters for maintaining the fixed fractional
H2S absorbent. capacity of 65%. The grey region corresponds to the design zone in
Case 4. Finally, simulations to estimate the solvent consumption which all the heuristic parameters are satisfied. In this sense, the
required to reach 95% of H2S recovery at fixed operating conditions height/diameter ratio in industrial absorption columns is between 3
(see Table 2) were carried out for the selected ILs. Fig. 12 presents the and 15 [45] so different absorption columns with [emim][DCN] are
calculated L/G ratios needed to achieve this separation at different possible. Thus, this study confirms the viability of [emim][DCN] IL to
temperatures by using Equilibrium (only thermodynamic equilibrium is be used as absorbent of H2S in gas streams in real applications. Future
considered) and Rate-Based mode (mass transfer kinetic is also taken works with detailed cost estimations will be needed in order to optimize
into account). the design variables of H2S absorption process based on ILs and com-
As can be seen, huge differences are obtained when using the two pare their behavior with that of conventional absorbents.
different simulation approaches (Equilibrium or Rate-Based mode).
Attending to Fig. 12A, all the ILs present the same trend with absorption 4. Conclusions
temperature, increasing IL mass flows by increasing temperatures, due
to the lower H2S gas solubility. In this case, the IL that presents the most In the present work, the modeling of H2S absorption operation using
ILs was accomplished. For this, a COSMO-based/Aspen Plus multiscale

Fig. 12. L/G mass ratio needed to reach a 95% of H2S recovery at different temperatures for each IL using the RADFRAC model in (A) Equilibrium and (B) Rate-Based
mode (Case 4 in Table 2).

7
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

methodology was used. This methodology allows the a priori evaluation Springer, Netherlands: Dordrecht, 2012, pp. 155–167.
of the thermodynamic properties of a huge number of H2S-IL mixtures. [12] Burr, B. and L. Lyddon, A Comparison of Physical Solvents for Acid Gas Removal.
Vol. 1. 2008.
First, COSMOtherm molecular simulator was successfully applied to [13] J.S. Wilkes, P. Wasserscheid, T. Welton, Ionic Liquids in Synthesis, Wiley-VCH
select different ILs among more than 700 attending to thermodynamic Verlag GmbH & Co. KGaA, 2008, pp. 1–6.
criteria (Henry’s law constants). The results showed that it is expected a [14] T. Welton, Ionic liquids: a brief history, Biophys. Rev. 10 (3) (2018) 691–706.
[15] D. Shang, et al., Ionic liquids in gas separation processing, Curr. Opin. Green
high H2S solubility in some ILs due to the low KH obtained. In addition, Sustain. Chem. 5 (2017) 74–81.
it was clear that the selection of the anion would have a great influence [16] F.-Y. Jou, A.E. Mather, Solubility of hydrogen sulfide in [bmim][PF6], Int. J.
in the solubility, being favored with ions of hydrogen-bond acceptor Thermophys. 28 (2) (2007) 490.
[17] A.H. Jalili, et al., Solubility of H2S in ionic liquids [bmim][PF6], [bmim][BF4], and
character. The intermolecular interactions analysis remarks the high [bmim][Tf2N], J. Chem. Eng. Data 54 (6) (2009) 1844–1849.
contribution on hydrogen bonding between H2S solute and IL anion, [18] A.H. Jalili, et al., Solubility and diffusion of CO2 and H2S in the ionic liquid 1-ethyl-
traducing this into more exothermic mixtures. Second, Aspen Plus 3-methylimidazolium ethylsulfate, J. Chem. Thermodyn. 42 (10) (2010)
1298–1303.
process simulator was used to test the H2S solvent performance of se-
[19] M.B. Shiflett, A.M.S. Niehaus, A. Yokozeki, Separation of CO2 and H2S using room-
lected ILs in absorption column. Despite the fact that no thermo- temperature ionic liquid [bmim][MeSO4], J. Chem. Eng. Data 55 (11) (2010)
dynamic restrictions to H2S separation were found (equilibrium-based 4785–4793.
recoveries of more than 99%), the strong kinetic control of the opera- [20] Y. Zhao, et al., Hydrogen sulfide solubility in ionic liquids (ILs): an extensive da-
tabase and a new ELM model mainly established by imidazolium-based ILs, J.
tion was determinant for the selection of both the operation conditions Chem. Eng. Data 61 (12) (2016) 3970–3978.
and the IL used. In this sense, the IL that shows the best thermodynamic [21] C.S. Pomelli, et al., Influence of the interaction between hydrogen sulfide and ionic
behavior presents severe limited transport properties, implying very liquids on solubility: experimental and theoretical investigation, J. Phys. Chem. B
111 (45) (2007) 13014–13019.
low H2S recovery at near room temperatures. The IL that presents the [22] S. Aparicio, M. Atilhan, Computational study of hexamethylguanidinium lactate
best performance as H2S absorbent is [emim][DCN], mainly due to its ionic liquid: a candidate for natural gas sweetening, Energy Fuels 24 (9) (2010)
low viscosity, and thus improved transport properties. The operation 4989–5001.
[23] H. Sakhaeinia, et al., Solubility of H2S in 1-(2-hydroxyethyl)-3-methylimidazolium
temperature is the key parameter of the column design; being 50 °C the ionic liquids with different anions, Fluid Phase Equilib. 298 (2) (2010) 303–309.
one selected in which the recovery curve reaches the maximum ab- [24] Y. Zhao, et al., Predicting H2S solubility in ionic liquids by the quantitative struc-
sorption capacity. Lastly, the column design was accomplished with ture-property relationship method using S sigma-profile molecular descriptors, RSC
Adv. 6 (74) (2016) 70405–70413.
different design proposals, confirming the technical viability of the [25] A. Kamgar, F. Esmaeilzadeh, Prediction of H2S solubility in hmim Pf(6), hmim Bf(4)
operation. Therefore, integrated molecular (COSMO-RS) and process and hmim Tf2N using UNIQUAC, NRTL and COSMO-RS, J. Mol. Liq. 220 (2016)
(Aspen Plus) simulations analysis have been proved as a useful tool to 631–634.
[26] S. Mortazavi-Manesh, M.A. Satyro, R.A. Marriott, Screening ionic liquids as can-
achieve a prospective analysis of the potential application of ILs in the
didates for separation of acid gases: solubility of hydrogen sulfide, methane, and
H2S capture, providing a key comprehension on the IL selection and its ethane, AIChE J. 59 (8) (2013) 2993–3005.
application in a commercial absorption column before required ex- [27] R. Santiago, et al., Acetylene absorption by ionic liquids: a multiscale analysis based
perimental tests. on molecular and process simulation, Sep. Purif. Technol. 204 (2018) 38–48.
[28] V.R. Ferro, et al., Enterprise ionic liquids database (ILUAM) for use in aspen ONE
programs suite with COSMO-based property methods, Ind. Eng. Chem. Res. 57 (3)
Acknowledgments (2018) 980–989.
[29] V.R. Ferro, et al., Introducing process simulation in ionic liquids design/selection
for separation processes based on operational and economic criteria through the
The authors thank Comunidad Autónoma de Madrid (Project example of their regeneration, Sep. Purif. Technol. 97 (2012) 195–204.
P2018/EMT4348) and Ministerio de Economía y Competitividad [30] J. De Riva, et al., Aspen plus supported analysis of the post-combustion CO2 capture
(MINECO) of Spain (Project CTQ2017-89441-R) for financial support. by chemical absorption using the [P2228][CNPyr] and [P66614][CNPyr] AHA ionic
liquids, Int. J. Greenhouse Gas Contr. 78 (2018) 94–102.
We are very grateful to Centro de Computación Científica de la [31] J. de Riva, et al., Ionic liquids for post-combustion CO2 capture by physical ab-
Universidad Autónoma de Madrid for computational facilities. sorption: thermodynamic, kinetic and process analysis, Int. J. Greenhouse Gas
Control 61 (2017) 61–70.
[32] J. Palomar, et al., Understanding the physical absorption of CO2 in ionic liquids
Appendix A. Supplementary material
using the COSMO-RS method, Ind. Eng. Chem. Res. 50 (6) (2011) 3452–3463.
[33] J. Palomar, et al., Demonstrating the key role of kinetics over thermodynamics in
Supplementary data to this article can be found online at https:// the selection of ionic liquids for CO2 physical absorption, Sep. Purif. Technol. 213
(2019) 578–586.
doi.org/10.1016/j.seppur.2019.116050.
[34] J. Palomar, et al., Task-specific ionic liquids for efficient ammonia absorption, Sep.
Purif. Technol. 82 (2011) 43–52.
References [35] J. Bedia, et al., Optimized ionic liquids for toluene absorption, AIChE J. 59 (5)
(2013) 1648–1656.
[36] M. Gonzalez-Miquel, J. Palomar, F. Rodriguez, Selection of ionic liquids for en-
[1] British Petroleum, Statistical review of world energy. 2018. hancing the gas solubility of volatile organic compounds, J. Phys. Chem. B 117 (1)
[2] S. Faramawy, T. Zaki, A.A.E. Sakr, Natural gas origin, composition, and processing: (2013) 296–306.
a review, J. Nat. Gas Sci. Eng. 34 (2016) 34–54. [37] M. Fallanza, et al., Screening of RTILs for propane/propylene separation using
[3] Astaria, G., D.W. Savage, and A. Bisio, Gas treating with chemical solvents. 1983: J. COSMO-RS methodology, Chem. Eng. J. 220 (2013) 284–293.
Wiley and Sons,New York, NY; None. Medium: X; Size: Pages: 493. [38] X. Liu, et al., High solubilities for methane, ethane, ethylene, and propane in tri-
[4] M. Ozekmekci, G. Salkic, M.F. Fellah, Use of zeolites for the removal of H2S: a mini- methyloctylphosphonium bis(2,4,4-trimethylpentyl) phosphinate ([P8111][TMPP]),
review, Fuel Process. Technol. 139 (2015) 49–60. Ind. Eng. Chem. Res. 53 (1) (2014) 363–368.
[5] L. Peters, et al., CO2 removal from natural gas by employing amine absorption and [39] J. de Riva, et al., Aspen plus supported conceptual design of the aromatic–aliphatic
membrane technology—a technical and economical analysis, Chem. Eng. J. 172 (2) separation from low aromatic content naphtha using 4-methyl-N-butylpyridinium
(2011) 952–960. tetrafluoroborate ionic liquid, Fuel Process. Technol. 146 (2016) 29–38.
[6] L. Sigot, G. Ducom, P. Germain, Adsorption of hydrogen sulfide (H2S) on zeolite (Z): [40] D. Moreno, et al., Absorption refrigeration cycles based on ionic liquids: re-
retention mechanism, Chem. Eng. J. 287 (2016) 47–53. frigerant/absorbent selection by thermodynamic and process analysis, Appl. Energy
[7] L. Zhou, et al., Feasibility study on pressure swing sorption for removing H2S from 213 (2018) 179–194.
natural gas, Chem. Eng. Sci. 59 (2004) 2401–2406. [41] D.R. Stull, Vapor pressure of pure substances. Organic and inorganic compounds,
[8] A.J. Kidnay, W.R. Parrish, Fundamentals of Natural Gas Processing, CRC Press, Ind. Eng. Chem. 39 (4) (1947) 517–540.
2006. [42] S.-T. Lin, S.I. Sandler, A priori phase equilibrium prediction from a segment con-
[9] B.P. Mandal, A.K. Biswas, S.S. Bandyopadhyay, Selective absorption of H2S from gas tribution solvation model, Ind. Eng. Chem. Res. 41 (5) (2002) 899–913.
streams containing H2S and CO2 into aqueous solutions of N-methyldiethanolamine [43] A. Klamt, Conductor-like screening model for real solvents: a new approach to the
and 2-amino-2-methyl-1-propanol, Sep. Purif. Technol. 35 (3) (2004) 191–202. quantitative calculation of solvation phenomena, J. Phys. Chem. 99 (7) (1995)
[10] J.-G. Lu, Y.-F. Zheng, D.-L. He, Selective absorption of H2S from gas mixtures into 2224–2235.
aqueous solutions of blended amines of methyldiethanolamine and 2-tertiar- [44] S.-T. Lin, P.M.M., Y. Song, C.-C. Chen, and S. I. Sandler, Improvements of Phase-
ybutylamino-2-ethoxyethanol in a packed column, Sep. Purif. Technol. 52 (2) Equilibrium Predictions for Hydrogen-Bonding Systems from a New Expression for
(2006) 209–217. COSMO Solvation Models, in AIChE Annual Meeting. 2002: Indianapolis.
[11] J.E. Bara, Potential for Hydrogen Sulfide Removal Using Ionic Liquid Solvents, [45] G.D. Ulrich, Chemical Engineering Process Design and Economics: A Practical

8
R. Santiago, et al. Separation and Purification Technology 233 (2020) 116050

Guide, Process Publishing, 2004. liquid 1-octyl-3-methylimidazolium hexafluorophosphate: experimental and mod-
[46] A. Klamt, F. Eckert, W. Arlt, COSMO-RS: an alternative to simulation for calculating elling, J. Chem. Thermodyn. 65 (2013) 220–232.
thermodynamic properties of liquid mixtures, Ann. Rev. Chem. Biomole. Eng. 1 (1) [52] M. Azizi, et al., Simulation of hydrogen sulphide absorption in alkaline solution
(2010) 101–122. using a packed column, Environ. Technol. 35 (2014) 3105–3115.
[47] E. Ruiz, et al., Evaluation of ionic liquids as absorbents for ammonia absorption [53] Sinnott, R.K., Chapter 11 - Separation Columns (Distillation and Absorption), in
refrigeration cycles using COSMO-based process simulations, Appl. Energy 123 Coulson and Richardson's Chemical Engineering (Second Edition), R.K. Sinnott,
(2014) 281–291. Editor. 1993, Pergamon: Amsterdam. p. 439–564.
[48] A.H. Jalili, et al., Solubility of CO2, H2S, and their mixture in the ionic liquid 1- [54] J. de Riva, et al., Statistical refinement and fitting of experimental viscosity-to-
Octyl-3-methylimidazolium bis(trifluoromethyl)sulfonylimide, J. Phys. Chem. B temperature data in ionic liquids, Ind. Eng. Chem. Res. 53 (25) (2014)
116 (9) (2012) 2758–2774. 10475–10484.
[49] H. Sakhaeinia, et al., Solubility of H2S in ionic liquids 1-ethyl-3-methylimidazolium [55] C.M.S.S. Neves, et al., Systematic study of the thermophysical properties of imi-
hexafluorophosphate ([emim][PF6]) and 1-ethyl-3-methylimidazolium Bis(tri- dazolium-based ionic liquids with cyano-functionalized anions, J. Phys. Chem. B
fluoromethyl)sulfonylimide ([emim][Tf2N]), J. Chem. Eng. Data 55 (12) (2010) 117 (35) (2013) 10271–10283.
5839–5845. [56] Y. Hiraga, et al., Densities at pressures up to 200 MPa and atmospheric pressure
[50] E. Ruiz, et al., Interactions of ionic liquids and acetone: thermodynamic properties, viscosities of ionic liquids 1-ethyl-3-methylimidazolium methylphosphate, 1-ethyl-
quantum-chemical calculations, and NMR analysis, J. Phys. Chem. B 117 (24) 3-methylimidazolium diethylphosphate, 1-butyl-3-methylimidazolium acetate, and
(2013) 7388–7398. 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide, J. Chem. Eng.
[51] M. Safavi, et al., Study of the solubility of CO2, H2S and their mixture in the ionic Data 60 (3) (2015) 876–885.