Deep Eutectic Solvent Formed by Imidazolium Cyanopyrrolide Andethylene Glycol For Reactive CO2Separations

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

This is an open access article published under a Creative Commons Non-Commercial No

Derivative Works (CC-BY-NC-ND) Attribution License, which permits copying and


redistribution of the article, and creation of adaptations, all for non-commercial purposes.

pubs.acs.org/journal/ascecg Letter

Deep Eutectic Solvent Formed by Imidazolium Cyanopyrrolide and


Ethylene Glycol for Reactive CO2 Separations
Yun-Yang Lee, Drace Penley, Aidan Klemm, William Dean, and Burcu Gurkan*
Cite This: ACS Sustainable Chem. Eng. 2021, 9, 1090−1098 Read Online

ACCESS Metrics & More Article Recommendations *


sı Supporting Information
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

ABSTRACT: Solvents made from a reactive ionic liquid, with an


imidazolium cation and pyrrolide anion, and ethylene glycol at a wide
compositional range were studied for separations of CO2 at low
partial pressures (≪0.1 bar up to 1 bar). Thermal analysis and
measurements of viscosity and density show compacting of the liquid
Downloaded via 76.34.42.37 on February 18, 2024 at 03:38:31 (UTC).

upon mixing with enhanced stability achieved by hydrogen bonding.


A detailed mechanistic study was performed by IR, quantitative NMR,
and ab initio calculations that show significant CO2 absorption
capacity below 5000 ppm of CO2 in N2. Three reversible routes are
found that yield carbonate (major product), carboxylate (moderate),
and carbamate (minor) species. With CO2 at 100% RH, bicarbonate
along with carbonate species form. The CO2-ethlyene glycol reaction
complex, the carbonate anion, is stabilized by the hydrogen bonding
and Coulombic interactions, thus preventing evaporation of the solvent during regeneration. This study demonstrates a promising
approach to designer green solvents for CO2 separations in open systems such as direct air capture.
KEYWORDS: Direct Air Capture, DAC, carbon capture, Negative emissions science, Ionic liquid, Carbon dioxide, Gas separations

■ INTRODUCTION
As CO2 emissions continue to rise due to human activity,1 so
upon repeated absorption−desorption cycles. The capsules
demonstrated superior stability in the presence of water
does the need for novel carbon capture materials.2 Ionic liquids compared to zeolites, although with more sluggish break-
(ILs), particularly CO2-reactive ionic liquids (rILs), have been through curves under dry conditions. Here, we report dilution
a steady target of research as relatively benign solvents for CO2 of the rIL with ethylene glycol (EG) to improve the CO2
capture, owing to their negligible volatility.3,4 ILs also have absorption rate while also changing the reaction mechanism, as
high CO2 solubilities, relatively high thermal stabilities revealed by NMR, IR, and computational studies.
EG is a common hydrogen bond donor (HBD) in deep
compared to volatile organic solvents, and versatile chemical
eutectic solvents (DESs).11 DESs have been investigated for
structures for tailorable physical properties.5 rILs with aprotic
CO2 capture, as they can absorb CO2 through the entropic
heterocyclic anions have shown distinct promise in CO2
voids within the liquid, similar to nonreactive ILs.12,13 A DES
capture due to their favorable reaction energetics (about −45
is a mixture for which the melting point is significantly lower
kJ/mol of CO2 at 25 °C), achieving regeneration easily at
than either of the parent compounds: a hydrogen bond
moderate temperatures (i.e., 60 °C) without a significant
acceptor (HBA), usually a halide salt, and a hydrogen bond
viscosity increase upon absorption.6 Still, initially high
donor (HBD), such as EG. Recent studies showed that DESs
viscosities, even prior to CO2 absorption, and low gravimetric
and DES-like mixtures with superbase additives,14 superbase
capacities due to the large molecular weight of rILs have
HBAs,15 or superbases as the tertiary component16 have CO2
limited their widespread application. It has been reported that
capacities ranging from 3.6 to 17.34 wt % at 1 bar of CO2 with
the gravimetric capacity and viscosity can be further improved
typical absorption temperatures of 45−60 °C. The highest
by the choice of the cation.7,8 We have previously shown that
gravimetric CO2 capacity at 33.7 wt % is reported for a mixture
the challenge of CO2 transport in viscous rILs can be mitigated
by encapsulation, where the gas−liquid surface area is
enhanced dramatically for improved mass transport.9 More Received: September 30, 2020
recently, we have reported CO2 capacities of 1.5 mol CO2 per Revised: December 30, 2020
kg of sorbent for capsules of 1-ethyl-3-methylimidazolium 2- Published: January 14, 2021
cyanopyrrolide, [EMIM][2-CNpyr] (60 wt % of capsule), at
25 °C and low partial pressures of CO2 (<0.1 bar).10
Regeneration was achieved at 40 °C with no loss of capacity

© 2021 American Chemical Society https://dx.doi.org/10.1021/acssuschemeng.0c07217


1090 ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

Figure 1. (a) DSC curves of [EMIM][2-CNpyr]:EG under a heating rate of 2 °C/min. Ratios in parentheses are the molar ratios of rIL to EG. (b)
Melting points of rIL:EG with respect to the mole fraction of rIL (lower x-axis) and molar ratio (upper x-axis). Samples 1:2, 1:3, and 1:5 did not
present a phase transformation within the accessible temperature range of the DSC. The gray dashed line is the hypothetical trend of the melting
temperature, while the solid lines are connecting the data points to guide the eye. (c) Excess molar volume as a function of the mole fraction of rIL
(lower x-axis) and molar ratio (upper x-axis). Excess molar volumes were calculated from the measured densities reported in Table S1. The dashed
line is the Redlich−Kister fit.31 (d) Temperature-dependent viscosity of rIL:EG (1:2) in comparison to the neat rIL10 (1:0) and EG30 (0:1). The
dashed line is the expected viscosity from the Grunberg−Nissan ideal mixture. Solid lines are Vogel−Tamman−Fulcher fits to the viscosity data;
fitting parameters are in Table S2.

of monoethanolammonium chloride, [MEA.Cl], and ethlyene- combination with ATR-FTIR show the imidazolium cation,
diamine at 1 bar of CO2 and 30 °C.17 It should be noted that pyrrolide anion, and EG, all bind with CO2, resulting in three
these high gravimetric capacities were at the expense of very different reaction routes that yield carboxylate, carbamate, and
slow rates as the viscosity of the DESs made of alkanolamines carbonate species, respectively. Calculations of the transition
reached about 4000 cP and required sorbent regeneration at states suggest that the formation of carbonate from EG has the
100 °C.17 Cui, Lv, and Yang studied mixtures of ammonium/ smallest energy barrier, in agreement with the experimentally
phosphonium azolide ILs with EG and reported capacities of determined majority product. All of the reactions are reversible
0.118 g CO2 per gram of solvent (10.5 wt %) at 25 °C and 1 and solvent regeneration was achieved at as low as 40 °C. This
atm of CO2 with regeneration at 70 °C under N2.18 While study demonstrates a route to achieving CO2 reactive sorbents
these reports are promising in terms of the development of that are water-lean, thermally stable, minimally volatile, and
regenerable, which is promising for open systems such as DAC.


alternative solvents for CO2 capture, DESs have not been
explored for direct air capture (DAC) or indoor air where the
partial pressure of CO2 is extremely low (410−5000 ppm). EXPERIMENTAL SECTION
In this study, we report a new reactive DES formed between Preparation of rIL:EG Mixtures and Characterization. Details
the rIL, [EMIM][2-CNpyr], and EG. Various rIL:EG to the materials, synthesis of [EMIM][2-CNpyr] and [EMMIM][2-
compositions are studied by differential scanning calorimetry CNpyr] (methyl substitution at α carbon of imidazolium), and
characterization techniques are described in the Supporting
(DSC) with the following molar ratios: 1:0.5, 1:1, 1:2, 1:3, 1:4,
Information (see Figures S1 and S2 for confirmation of synthesis by
1:5, and 1:10. The rIL:EG (1:2) forms a thermally stable and NMR). Mixtures of [EMIM][2-CNpyr] with EG at molar ratios of
low viscosity DES with an absorption capacity of 10.3 and 11.4 1:0.5, 1:1. 1:2, 1:3, 1:4, 1:5, and 1:10 were prepared using a vortex
wt % at 0.005 bar (0.5% CO2 in N2) and 1 bar of CO2 (100% mixer in an argon-purged glovebox (VTI, H2O < 0.1 ppm, O2 < 0.1
CO2), respectively. Quantitative analysis of 13C NMR in ppm). The water contents of samples were measured by a Karl

1091 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

Figure 2. (a) CO2 absorption capacities measured by TGA using 2.5 mg of rIL:EG sorbent under 1 bar of CO2. (b) Gravimetric CO2 capacities at
25 °C. Red circles are for neat [EMIM][2-CNpyr] (1:0) measured gravimetrically from a mixed CO2/N2 feed at 1 bar. Blue triangles and green
diamonds are for rIL:EG mixtures with a molar ratio of 1:1 and 1:2, respectively. CO2 capacity from TGA data in panel a (hexagons) and capacities
measured under 100% RH (stars) are also shown. The partial pressure of CO2 was adjusted for each set point by the mass flow controllers. The
capacity previously reported for [EMIM][2-CNpyr] under pure CO2 is included for comparison (gray squares).10 Uncertainties of the gravimetric
capacities are estimated to be 10%. (c) CO2 capacities in panel b replotted for the low pressure region (left; 0−6 mbar) and the higher pressure
region (right; 900−1100 mbar). (d) Three consecutive absorption (1 bar CO2, 25 °C) and desorption (1 bar N2, 40 °C) cycles of rIL:EG (1:2).

Fischer titrator (Metrohm Coulometric KF 889D) and were in the Micrometrics TriStar II sorption analyzer, following our previous
range of 700−900 ppm. Viscosity and density were measured method.9
following procedures reported previously.19 Thermal analysis of the CO2 desorption and solvent regeneration following an absorption
samples was performed by DSC (Mettler Toledo DSC3). In DSC, the at 25 °C was studied by the gravimetric method described above.
samples were initially chilled from 25 to −84 °C, held isothermally for Briefly, the feeding gas was switched from N2 to CO2 for absorption at
10 min, and then heated back to 25 °C with a rate of 2 °C/min under 25 °C for 40 min and then switched back to N2 during desorption at
N2 atmosphere. A three-step heating−cooling−heating cycle was 40 °C for 30 min. The sample gas flow rate was 200 mL/min at 1 bar.
performed, and it was confirmed for [EMIM][2-CNpyr] that there Identification of Reaction Routes and Distribution of
are no significant differences between the heating cycles. Therefore, Products. The chemical composition of the samples before and
only the first heating scan is reported for all of the solvents. after an absorption experiment was characterized by ATR-FTIR using
CO2 Absorption and Desorption Experiments. The CO2 a Nicolet iS50 FTIR (Thermo Scientific) with 32 scans, at a
capacity of the rIL:EG was measured gravimetrically by weighing resolution of 4 cm−1, on a diamond crystal and quantitatively by 13C
the sample before and after CO2 absorption using an analytical NMR (125.75 MHz, Bruker Ascend HD NMR). The relaxation time
balance (accuracy of ±0.001 g). In brief, 0.25 g of liquid was placed in of the C nucleus was drastically reduced by a paramagnetic
a 9 mL glass scintillator vial with a stir bar, and a CO2/N2 gas mixture compound, chromium acetylacetonate in deuterated dimethyl
was introduced at a flow rate of 200 mL/min and 1 bar at 25 °C until sulfoxide (0.1 M Cr(ACAC)3 in DMSO). The T1 relaxation times
the weight of the sample reached equilibrium as established by the of the different CO2 complexes were measured by the inversion
maintenance of constant weight for 1 h. All CO2 absorption recovery method (see Figure S3 and SI for details). The energetics
experiments were done under anhydrous feed, unless noted otherwise. and the transition states associated with the reactions were computed
The partial pressure of CO2 in the feed was varied by mass flow at the level of density functional theory (DFT) using the
controllers on N2 and CO2 lines (Brooks 5850I) using LabVIEW. The Turbomole20 software. The calculations were performed with the
CO2 concentration was monitored by a CO2 analyzer (SBA-5, def2-SVP21 basis set using the Karlsruhe split valence polarization,
PPSystems Inc.). This gravimetric method is confirmed to yield Tao−Perdew−Staroverov−Scuseria (TPSS)22 functional with D3
capacities within 10% of the CO2 isotherm reported previously for dispersion corrections,23 Becke and Johnson (BJ) damping,24 and
[EMIM][2-CNpyr].9 This comparison also verifies that the CO2 resolution of identity (RI) approximation.25 All calculations were run
capacity of the rIL is virtually unaffected by the presence of N2 during at a m4 grid size. For simplicity, the transition state search for each
absorption within our uncertainties. The CO2 isotherm of neat reaction route is performed separately. A bond was created between
[EMMIM][2-CNpyr] was measured barometrically at 25 °C using a the CO2 and the investigated reaction site according to the scheme

1092 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

Table 1. Measured CO2 Capacities in This Study in Comparison to Previously Reported DESs with Specified Absorption and
Desorption Conditionsa
Solvent Absorption conditions Regeneration under N2 CO2 capacity (mol CO2/kg DES) ref
[Ch][Cl]+MEA+Piperazine (1:7:1) 1.57 bar CO2, 25 °C N/A 0.96 32

DBN+EU (2:1) 1 bar CO2, 45 °C 90 °C 5.23 15


1 bar CO2, 45 °C 90 °C 4.54 (14% CO2 in N2) 15

[N2222][Im]+EG (1:2) 1 bar CO2, 25 °C 70 °C 2.68 18

[N2222][Triz]+EG (1:2) 1 bar CO2, 25 °C 70 °C 2.68 18

[P2222][Im]+EG (1:2) 1 bar CO2, 25 °C 70 °C 2.84 18

[P2222][Triz]+EG (1:2) 1 bar CO2, 25 °C 70 °C 2.93 18

DBN (9 v%) in [Ch][Cl]+EG (1:2) 1 bar CO2, 25 °C 100 °Cb 0.86 14

DBU (9 v%) in [Ch][Cl]+EG (1:2) 1 bar CO2, 25 °C 100 °Cb 0.83 14

DBN (9 v%) in [Ch][Cl]+MEA (1:4) 1 bar CO2, 25 °C 100 °Cb 5.11 14

[MEA.Cl]+EDA (1:3) 1 bar CO2, 30 °C 100 °C 7.16 17

[Ch][Cl]+Glycerol+DBN (1:3:10) 1 bar CO2, 25 °C 60 °Cc 2.36 16

[TETA.Cl]+EG (1:3) 1 bar CO2, 40 °C 100 °C 3.98 33

[HDBU][Im]+EG (7:3) 1 bar CO2, 40 °C 70 °Cc 3.20 27

[HDBU][Ind]+EG (7:3) 1 bar CO2, 40 °C 70 °Cc 2.66 27

[HDBU][Triz]+EG (7:3) 1 bar CO2, 40 °C 70 °Cc 2.45 27

[EMIM][2-CNpyr]+EG(1:2) 1 bar (CO2:N2), 25°C 40 °C 2.59 (100% CO2) This study


1 bar (CO2:N2), 25°C 40 °C 2.35 (0.5% CO2) This study
1 bar (CO2:N2), 25°C 40 °C 2.08 (2000 ppm of CO2) This study
1 bar (CO2:N2), 25°C 40 °C 0.95 (410 ppm of CO2) This study
a
Compositions in parentheses are molar ratios. All of the reported CO2 capacities are converted to units of mol CO2 per kg of sorbent for
comparison purposes. For this study: 2.59 mol CO2 per kg DES corresponds to 11.4 wt % (g CO2 per g DES). b35% regeneration. cSignificant mass
loss; EU = ethylene urea; MEA = monoethanolamine; [MEA.Cl] = monoethanolamine and HCl mixture; DBN and DBU are superbases; [Im] =
imidazolide; [Triz] = triazolide; [TETA.Cl] = triethylenetetramine and HCl mixture; [Ch][Cl] = choline chloride; [HDBU] = protonated DBU;
[Ind]=indolide.

suggested by experimental spectroscopic analysis. These bonds were heating step of the 1:10 mixture, indicating that an exothermic
scanned via a potential energy surface scan to obtain a bonding structural reordering process occurred before melting. Figure
distance that resulted in the highest optimized energy that 1b shows the dependence of the melting temperature with
corresponds to the transition state. The obtained transition states
were confirmed via frequency analysis where a transition state yields respect to the mole fraction of the rIL (lower axis) and also in
one imaginary frequency corresponding to the CO2 bonding with the terms of the molar ratio of rIL:EG (upper axis). While it is not
molecule of interest.26−28 certain how deep the melting point depression is between 1:2


and 1:5 mixtures compared to the rIL and EG parent
RESULTS AND DISCUSSION compounds, excess molar volumes calculated from measured
densities are negative, indicating that there is compacting of
The [EMIM][2-CNpyr]:EG mixtures at compositions ranging
from 1:0.5 to 1:10 were observed to be liquids at room the liquid, as shown in Figure 1c. The negative deviation may
temperature, and their DSC curves are shown in Figure 1a. stem from a hydrogen bond network formation, a characteristic
While the 1:10 mixture has a melting point of −22.6 °C, trait of DESs, further supporting the DES formation between
between that of EG (−11.5 °C) and rIL (−60.8 °C), it is seen [EMIM][2-CNpyr] and EG. The measured viscosities of
that 1:0.5 and 1:1 mixtures form eutectics. The other rIL:EG (1:2) also show a closer interaction between rIL and
compositions, 1:2, 1:3, and 1:5 did not present a phase EG than would be expected from an ideal mixture: see
transition within the temperature range studied possibly due to comparisons of viscosities in Figure 1d with rIL, EG, and the
the melting occurring below the temperature limit of the estimated viscosity from the classic Grunberg−Nissan29 mixing
instrument, −80 °C. A cold crystallization is observed in the rule (dashed line).30 These results suggest tight hydrogen
1093 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

Figure 3. (a) Proposed reaction network with NMR labeling of the reactants (1, a, b, c, ...) and CO2 bound products (α, 2, 3, 4, m, l, a′, b′, ...). (b)
Quantitative 13C NMR of [EMIM][2-CNpyr]:EG (1:2) prior to (black) and after CO2 absorption at 2000 ppm (red) and 5000 ppm (blue) of CO2
in N2 (1 bar, 25 °C). For comparison, measurements with dry (0% RH, green) and moist (100% RH, purple) CO2 at 1 bar are included. The
integration of peaks was done by setting the d or d′ position carbon of the anion to 1. The carbamate peak (l) is difficult to see due to the strong H-
bonds between rIL and EG; Figure S8 is the magnified NMR where it is better seen.

bonding alongside Coulombic interactions, similar to the case shown in Figure 2b with hexagonal symbols. When the feeding
of ionic DESs. gas was saturated with water by bubbling through a water
Figure 2a shows the TGA curves of rIL:EG mixtures with reservoir (100% RH), the resulting CO2 capacities (star
1:0, 1:0.5, 1:1, and 1:2 as they absorb CO2 under atmospheric symbols in Figure 2b) measured for 1:0 and 1:2 mixtures show
pressure at 25 °C (initial masses were tared). Despite having no significant difference from the dry feed experiments. At
about the same initial mass of liquid (2.5 mg), the time to CO2 concentrations of 2000 and 5000 ppm, the capacity of
reach equilibrium varies slightly between samples because of rIL:EG (1:1 and 1:2) exceeds that of the neat rIL which is
their different viscosities. For the neat rIL, equilibrium is still better seen in Figure 2c. This suggests EG is participating in
not reached after 40 min due to its relatively high viscosity, the reaction network as discussed later. The CO2 cyclability of
while it takes 10 min for the rIL:EG (1:2) to reach equilibrium. rIL:EG (1:2) was studied by performing three consecutive
The viscosities before and after CO2 for rIL and rIL:EG (1:2) absorption−desorption cycles. As shown in Figure 2d, the
are given in Table S3. The mixtures with less EG (1:0.5, 1:1, absorbed CO2 at 25 °C can easily be released under N2 at 40
and 1:2) show a stable equilibrium with no mass loss, whereas °C. While the CO2 capacity stays about the same in the second
rIL:EG (1:3), as shown in Figure 2a, loses mass following an and third cycles, small mass loss is observed during
initial mass gain after 10 min due to evaporation of EG. regeneration following the third cycle when the regeneration
CO2 capacities measured by the gravimetric method using is prolonged from 30 to 50 min. This is possibly due to the
the stirred absorption cell for rIL:EG (1:1 and 1:2) are shown evaporation of EG. However, the CO2 capacity is maintained
in Figure 2b. The CO2 capacity obtained under a CO2/N2 from the first cycle to the third, suggesting no evaporation until
mixture for the neat rIL (red circles) compares similarly to the CO2 is stripped. Therefore, we believe the CO2 complexation
isotherm of neat rIL (gray squares) from our previous report stabilizes the sorbent, preventing EG evaporation. Spectro-
based on a barometric measurement.10 The equilibrium scopic evidence suggests the stabilization of EG as discussed
capacities obtained by TGA in Figure 2a are very similar and later. Table 1 summarizes the measured CO2 capacities for the
1094 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

rIL+EG system in comparison to similar DESs investigated. in the quantitative 13C NMR (Table S5), the CO2 capacity can
The main advantage of the studied solvent in this work is the be broken down into the contributions from each reaction
improved CO2 absorption rates, milder regeneration temper- route, yielding the final product distributions represented in
atures, and the high CO2 capacity at extremely low partial Figure 4. Under dry CO2, the complexation between CO2 and
pressures of CO2 that have not been reported for these types of
sorbents.
IR analysis of the reaction products between dry CO2 and
rIL:EG (1:2) indicates multiple species for bound CO2, as
shown in Figure S4a. The reaction scheme in Figure 3a is
proposed to take place. Route 1 is the complexation of CO2
with the pyrrolide anion that forms carbamate (−N−COO).
Route 2 is the deprotonation of the imidazolium cation by the
anion that forms a carbene zwitterion, which then binds with
CO2 to form carboxylate (−C−COO). Route 3 is the
protonation of the anion by EG, resulting in deprotonated
EG that subsequently reacts with CO2 to form carbonate
(−O−COO). In the presence of water, bicarbonate (HO−
COO) also forms via Route 4. The carbamate and carboxylate
formations are consistent with CO2 absorption products by the
neat rIL (Figure S4b), similar to prior literature.7 Yan et al. also
reported the observation of both carbonate and carbamate
routes in a superbase and EG mixture upon CO2 absorption.27 Figure 4. Breakdown of CO2 capacities determined from the NMR
integrals (Table S5) with respect to the three reaction routes: from
To further probe the CO2 absorption mechanism, Route 2 is anion (carbamate, peak l at 146 ppm), cation (carboxylate, peak m at
eliminated by blocking the α carbon (k) on the imidazolium 154 ppm), and EG (carbonate, peak 4 at 158 ppm) as a function of
with a methyl group, synthesizing [EMMIM][2-CNpyr]. The CO2 percentage in the feed.
resulting CO2 capacity of [EMMIM][2-CNpyr] at 25 °C is
measured to be significantly smaller than [EMIM][2-CNpyr]
(isotherms compared in Figure S5). At 0.002 bar (2000 ppm, EG is the dominating route, as opposed to the case of CO2
1.5 Torr), the capacity of [EMMIM][2-CNpyr] is 0.16 mol absorption in [EMIM][2-CNpyr] (Figure S7b), where the
CO2 per mol IL, compared to 0.35 for [EMIM][2-CNpyr]. CO2-cation pathway dominates. Under moist CO2 (100%
Even under these very low pressures, [EMIM][2-CNpyr] has RH), the formation of carbonate (Route 3) is superseded by
significantly higher capacity due to the strong CO2 affinity of bicarbonate (Route 4). The mixture absorbs as much as 30 wt
the α carbon (k) on the imidazolium (Route 2). In the absence % water under 100% RH feed. New hydrogen bonds occur
of a carbene formation, there is still measurable CO2 capacity, under water-rich conditions, as indicated by the downshift of
suggesting that the carbamate formation in Route 1 takes place the bicarbonate peak from 158 ppm (α) to 160 ppm, as well as
with [EMMIM][2-CNpyr]. However, the IL solidifies at 0.013 the downshift of EG (1). However, the total CO2 capacity is
bar CO2 (10 Torr). In the presence of EG and carbene, Route not compromised by the presence of water. We hypothesize
1 is suppressed for [EMIM][2-CNpyr] at low partial pressures that the deprotonated EG is stabilized by the H-bonding
(≪1 bar). network and the presence of the [2-CNpyr]/2-CNpyrH
A more quantitative analysis of the extent of each reaction adduct (conjugate base and acid). Stabilization of the EG-
route was performed by 13C NMR as shown in Figure 3b. The CO2 reaction complex through H-bonds and Coulombic
relaxation times obtained for each reaction product are shown interactions (since the product is an anion) supports our
in Figure S4 and summarized in Table S4. Comparing the interpretation for the consistent CO2 capacities obtained in
NMR of rIL:EG (1:2) before and after exposure to CO2 in Figure 2d.
Figure 3b, it is shown that the cation peaks split into two sets, In order to understand the reaction mechanism and
similar to what is observed for the neat IL (Figure S7a). One energetics, calculations at the DFT level were performed.
set (a, b, c, g, i, and k, Figure 3a) at the original peak locations The calculated relative energies of the transition states and
is for the unreacted cation. The other set is the downfield shift final products are shown in Figure 5 for the proposed reaction
of the original set of peaks, indicated by (′), corresponding to network (see Table S6 for tabulated energies). DFT analysis
the cation-CO2 complex (carboxylate). As the CO2 concen- suggests the carbonate formation via Route 3 has the smallest
tration increases, the downshifted peaks of the cation increase activation energy barrier at 2.6 kJ/mol (TS3 in Figure 5), in
in intensity. The split of peaks in the cation is ascribed to the comparison to 6.2 for carbamate (TS1 for Route 1), 5.9 for
change in aromatic ring current after CO2 complexation.34 carbene intermediate (TS2*), and 10.7 (TS2**) kJ/mol for
Such a split is not seen for the anion, and instead, a downfield carboxylate via carbene (Route 2). These relatively low
shift is observed for the set of anion peaks (d′, e′, f′, h′, and j′, activation energies are consistent with the literature for gas
Figure 3a) possibly due to participation in hydrogen bonding phase simulations of CO2 absorption by azolide28 and
(see Figure S8 for zoomed in spectra). The HMBC spectra superbase27 ILs as well as imidazolium acetate.26 Route 2 is
confirm the correlation between cation and anion, and the absent in the previously investigated phosphonium ILs with
observed couplings (Figure S9a) indicate interaction through the [2-CNpyr] anion where the reaction energy for Route 1
H-bonding upon complexation via Routes 1 and 2. HMBC was reported as −64 kJ/mol.6 We calculated a similar reaction
spectra in Figure S9b also confirm the assignments of energy of −70.7 kJ/mol for CO2 binding directly to the
carbamate at 146 ppm (l), carboxylate at 154 ppm (m), and individual anion alone, with the difference attributed to the
carbonate at 158 ppm (4). Using the integral values obtained different basis sets used (see Table S7 for species energies). In
1095 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

Figure 5. Calculated relative energies (left) of the reaction systems shown for Routes 1−3 (right). Calculated transition states correspond to the
CO2 binding to the anion (TS1 in Route 1), CO2 binding to the cation via carbene (TS2* and TS2** Route 2), and CO2 binding to EG (TS3 in
Route 3). TS2* is the hydrogen transfer step from the [EMIM] cation to the [2-CNpyr] anion, in the presence of CO2, whereas TS2** is the
carboxylate formation from carbene. Color code: carbon (green), oxygen (red), nitrogen (blue), and hydrogen (gray). Dotted lines are H-bonds.

the presence of [EMIM] and the associated H-bonding that is why we do not experimentally find Route 2 products as the
observed in the calculated TS1 product in Figure 5, a much majority. Future ab initio molecular dynamics simulations
smaller reaction energy at −18.63 kJ/mol was obtained for investigating the solvation environment would be helpful in
Route 1. Similarly, for TS3 product, the presence of H-bonds determining the possible complexes for a transition state search
yields a reaction energy of −16 kJ/mol. This is much smaller in Route 2 and better compare with experimental results.
than the calculated −56.9 kJ/mol for CO2 binding directly to
EG without any interaction between the anion and EG. The
impact of H-bonding in ILs has been discussed in the literature
■ CONCLUSIONS
In summary, we present a new sorbent for reactive CO2
to result in lower estimated reaction energies.35 In a study by separations based on [EMIM][2-CNpyr] and EG. The 1:2
Low et al.,36 it is discussed that the H-bonds in ILs is usually molar mixture of rIL:EG forms a DES via hydrogen bonding. A
predicted to be shorter when calculated in the gas phase detailed mechanistic investigation of the CO2 absorption by
compared to the implicit solvent. This results in lower rIL:EG shows that carbonate formation by the deprotonated
estimated total energies due to the underestimated electro- EG, in the presence of the pyrrolide anion, is the most
statics. According to our calculations of the rIL pair, the H- significant reaction. Even at extremely low partial pressures of
bond lengths between the anion and cation are in the orders of CO2, carbonate formation is accompanied by the carboxylate
1.9 and 2.2 Å which indicate moderate H-bond formation with formation via the carbene intermediate. Unlike previous
induced electrostatics. The H-bond length between EG and studies on reactive ILs with aprotic heterocyclic anions,
the anion that is calculated in Route 3 is 1.5 Å in reactants, 1.3 carbamate formation in the rIL:EG sorbent is minimal. The
in TS3 (consistent with covalent bonding), and 1.6 in specific advantages of the rIL:EG solvent are the high capacity
carbonate, all of which are shorter than that of the ion pair at low CO2 concentration, improved absorption rates, and mild
of rIL. This can explain the reduced nucleophilicity of the solvent regeneration temperature compared to amines and
anion toward CO2 and Route 1 products being the minority. rILs. The stability of the sorbent under absorption and
Although the carboxylate formation in Route 2 seems to be desorption conditions studied are a result of the extended
thermodynamically most favorable, with an overall reaction hydrogen bonded network among all species. While the
energy of −69.2 kJ/mol, it involves the carbene intermediate presence of a significant amount of water (30 wt %) favored
by the cation deprotonation which requires a higher activation bicarbonate over carboxylate and carbamate products, the CO2
energy (TS2**). The determination of the transition state capacity was maintained. Further studies on solvent regener-
from [EMIM] to [EMIM]+-CO2− in a single step was not ation in the presence of water and thermal stability are needed
possible due to the complexity of simultaneous CO2 binding to assess the suitability of these new sorbents for direct air
and hydrogen transfer, as also discussed previously for CO2 capture of CO2.


binding to [EMIM][acetate] by Mao et al.26 In the absence of
solvation effects, we treated the [EMIM]-CO2 complex to ASSOCIATED CONTENT
occur by a two-step reaction. This approach does not account
*
sı Supporting Information
for the stabilization of the carboxylated imidazolium by
The Supporting Information is available free of charge at
another imidazolium via H-bond as reported by Mao et al.26
https://pubs.acs.org/doi/10.1021/acssuschemeng.0c07217.
After a certain CO2 loading, there would be no free
imidazoliums to stabilize the carboxylated imidazolium; Materials; synthesis; tabulated densities, viscosities, and
therefore, Route 2 would be terminated. This may explain water contents; details of NMR and IR analysis;
1096 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

calculated energies for reaction species; inverse recovery accelerating CO2 emissions. Proc. Natl. Acad. Sci. U. S. A. 2007, 104
method; and HMBC spectra (PDF) (24), 10288−10293.
(2) Lu, X.; Jin, D.; Wei, S.; Wang, Z.; An, C.; Guo, W. Strategies to

■ AUTHOR INFORMATION
Corresponding Author
enhance CO2 capture and separation based on engineering absorbent
materials. J. Mater. Chem. A 2015, 3 (23), 12118−12132.
(3) Ravula, S.; Larm, N. E.; Mottaleb, M. A.; Heitz, M. P.; Baker, G.
A. Vapor pressure mapping of ionic liquids and low-volatility fluids
Burcu Gurkan − Department of Chemical and Biomolecular using graded isothermal thermogravimetric analysis. ChemEngineering
Engineering, Case Western Reserve University, Cleveland, 2019, 3 (2), 42.
Ohio 44106, United States; orcid.org/0000-0003-4886- (4) Wang, C.; Luo, X.; Zhu, X.; Cui, G.; Jiang, D.-e.; Deng, D.; Li,
3350; Email: beg23@case.edu H.; Dai, S. The strategies for improving carbon dioxide chemisorption
by functionalized ionic liquids. RSC Adv. 2013, 3 (36), 15518−15527.
Authors (5) Hasib-ur-Rahman, M.; Siaj, M.; Larachi, F. Ionic liquids for CO2
Yun-Yang Lee − Department of Chemical and Biomolecular captureDevelopment and progress. Chem. Eng. Process. 2010, 49
Engineering, Case Western Reserve University, Cleveland, (4), 313−322.
Ohio 44106, United States; orcid.org/0000-0002-4165- (6) Gurkan, B.; Goodrich, B. F.; Mindrup, E. M.; Ficke, L. E.;
0857 Massel, M.; Seo, S.; Senftle, T. P.; Wu, H.; Glaser, M. F.; Shah, J. K.;
Drace Penley − Department of Chemical and Biomolecular Maginn, E. J.; Brennecke, J. F.; Schneider, W. F. Molecular Design of
High Capacity, Low Viscosity, Chemically Tunable Ionic Liquids for
Engineering, Case Western Reserve University, Cleveland,
CO2 Capture. J. Phys. Chem. Lett. 2010, 1 (24), 3494−3499.
Ohio 44106, United States (7) Seo, S.; DeSilva, M. A.; Brennecke, J. F. Physical Properties and
Aidan Klemm − Department of Chemical and Biomolecular CO2 Reaction Pathway of 1-Ethyl-3-Methylimidazolium Ionic
Engineering, Case Western Reserve University, Cleveland, Liquids with Aprotic Heterocyclic Anions. J. Phys. Chem. B 2014,
Ohio 44106, United States 118 (51), 14870−14879.
William Dean − Department of Chemical and Biomolecular (8) Brown, P.; Gurkan, B. E.; Hatton, T. A. Enhanced gravimetric
Engineering, Case Western Reserve University, Cleveland, CO2 capacity and viscosity for ionic liquids with cyanopyrrolide
Ohio 44106, United States anion. AIChE J. 2015, 61 (7), 2280−2285.
(9) Huang, Q. W.; Luo, Q. M.; Wang, Y. F.; Pentzer, E.; Gurkan, B.
Complete contact information is available at: Hybrid Ionic Liquid Capsules for Rapid CO2 Capture. Ind. Eng.
https://pubs.acs.org/10.1021/acssuschemeng.0c07217 Chem. Res. 2019, 58 (24), 10503−10509.
(10) Lee, Y.-Y.; Edgehouse, K.; Klemm, A.; Mao, H.; Pentzer, E.;
Author Contributions Gurkan, B. Capsules of Reactive Ionic Liquids for Selective Capture of
Y-Y.L. synthesized the solvent, performed CO2 absorption/ Carbon Dioxide at Low Concentrations. ACS Appl. Mater. Interfaces
desorption measurements, and analysis of NMR and IR 2020, 12 (16), 19184−19193.
spectroscopy. D.P. performed DFT calculations. A.K. meas- (11) Ibrahim, R. K.; Hayyan, M.; AlSaadi, M. A.; Ibrahim, S.;
ured densities and viscosities and performed excess molar Hayyan, A.; Hashim, M. A. Physical properties of ethylene glycol-
volume analysis. W.D. measured DSC. B.G. oversaw the based deep eutectic solvents. J. Mol. Liq. 2019, 276, 794−800.
(12) Sarmad, S.; Mikkola, J.-P.; Ji, X. Carbon Dioxide Capture with
experiments, computations, and analysis. All authors con- Ionic Liquids and Deep Eutectic Solvents: A New Generation of
tributed to the writing of the manuscript. Sorbents. ChemSusChem 2017, 10 (2), 324−352.
Notes (13) García, G.; Aparicio, S.; Ullah, R.; Atilhan, M. Deep Eutectic
The authors declare no competing financial interest. Solvents: Physicochemical Properties and Gas Separation Applica-


tions. Energy Fuels 2015, 29 (4), 2616−2644.
ACKNOWLEDGMENTS (14) Bhawna; Pandey, A.; Pandey, S. Superbase-Added Choline
Chloride-Based Deep Eutectic Solvents for CO2 Capture and
The authors thank Dr. Nalinda P. Wickramasinghe at the Sequestration. ChemistrySelect 2017, 2 (35), 11422−11430.
Instrumentation Facility of the Department of Chemistry at (15) Jiang, B.; Ma, J.; Yang, N.; Huang, Z.; Zhang, N.; Tantai, X.;
CWRU for feedback on the inverse recovery measurements by Sun, Y.; Zhang, L. Superbase/Acylamido-Based Deep Eutectic
NMR and Prof. Shane Parker for discussions on transition state Solvents for Multiple-Site Efficient CO2 Absorption. Energy Fuels
calculations. The authors acknowledge the Northeast Ohio 2019, 33 (8), 7569−7577.
High Field NMR Facility and the Soft Material Character- (16) Sze, L. L.; Pandey, S.; Ravula, S.; Pandey, S.; Zhao, H.; Baker,
ization Laboratories for the access to TGA. The ionic liquid G. A.; Baker, S. N. Ternary Deep Eutectic Solvents Tasked for Carbon
Dioxide Capture. ACS Sustainable Chem. Eng. 2014, 2 (9), 2117−
synthesis and CO2 measurements were supported by an Early
2123.
Career Faculty grant from NASA’s Space Technology Research (17) Trivedi, T. J.; Lee, J. H.; Lee, H. J.; Jeong, Y. K.; Choi, J. W.
Grants Program under Award No. 80NSSC18K1505. The Deep eutectic solvents as attractive media for CO2 capture. Green
analysis of the absorption mechanism was in part supported by Chem. 2016, 18 (9), 2834−2842.
the American Chemical Society Petroleum Research Fund (18) Cui, G.; Lv, M.; Yang, D. Efficient CO2 absorption by azolide-
59520-DNI4. The phase behavior and the physical property based deep eutectic solvents. Chem. Commun. 2019, 55 (10), 1426−
characterization of the solvents were supported by the 1429.
Breakthrough Electrolytes for Energy Storage (BEES), an (19) Gurkan, B.; Squire, H.; Pentzer, E. Metal-Free Deep Eutectic
Energy Frontier Research Center funded by the U.S. Solvents: Preparation, Physical Properties, and Significance. J. Phys.
Department of Energy, Office of Science, Basic Energy Chem. Lett. 2019, 10 (24), 7956−7964.
Sciences under Award No. DE-SC0019409. (20) Ahlrichs, R.; Bär, M.; Häser, M.; Horn, H.; Kölmel, C.


Electronic structure calculations on workstation computers: The
program system turbomole. Chem. Phys. Lett. 1989, 162 (3), 165−
REFERENCES 169.
(1) Raupach, M. R.; Marland, G.; Ciais, P.; Le Quéré, C.; Canadell, (21) Zheng, J.; Xu, X.; Truhlar, D. G. Minimally augmented
J. G.; Klepper, G.; Field, C. B. Global and regional drivers of Karlsruhe basis sets. Theor. Chem. Acc. 2011, 128 (3), 295−305.

1097 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098
ACS Sustainable Chemistry & Engineering pubs.acs.org/journal/ascecg Letter

(22) Tao, J.; Perdew, J. P.; Staroverov, V. N.; Scuseria, G. E.


Climbing the Density Functional Ladder: Nonempirical Meta–
Generalized Gradient Approximation Designed for Molecules and
Solids. Phys. Rev. Lett. 2003, 91 (14), 146401.
(23) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and
accurate ab initio parametrization of density functional dispersion
correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010,
132 (15), 154104.
(24) Johnson, E. R.; Becke, A. D. A post-Hartree-Fock model of
intermolecular interactions: Inclusion of higher-order corrections. J.
Chem. Phys. 2006, 124 (17), 174104.
(25) Hernández Vera, M.; Jagau, T.-C. Resolution-of-the-identity
approximation for complex-scaled basis functions. J. Chem. Phys.
2019, 151 (11), 111101.
(26) Mao, J. X.; Steckel, J. A.; Yan, F.; Dhumal, N.; Kim, H.;
Damodaran, K. Understanding the mechanism of CO2 capture by 1,3
di-substituted imidazolium acetate based ionic liquids. Phys. Chem.
Chem. Phys. 2016, 18 (3), 1911−1917.
(27) Yan, H.; Zhao, L.; Bai, Y.; Li, F.; Dong, H.; Wang, H.; Zhang,
X.; Zeng, S. Superbase Ionic Liquid-Based Deep Eutectic Solvents for
Improving CO2 Absorption. ACS Sustainable Chem. Eng. 2020, 8 (6),
2523−2530.
(28) Izadyar, M.; Rezaeian, M.; Victorov, A. Theoretical study on
the absorption of carbon dioxide by DBU-based ionic liquids. Phys.
Chem. Chem. Phys. 2020, 22 (35), 20050−20060.
(29) Grunberg, L.; Nissan, A. H. Mixture Law for Viscosity. Nature
1949, 164 (4175), 799−800.
(30) Zhang, Y.; Poe, D.; Heroux, L.; Squire, H.; Doherty, B. W.;
Long, Z.; Dadmun, M.; Gurkan, B.; Tuckerman, M. E.; Maginn, E. J.
Liquid Structure and Transport Properties of the Deep Eutectic
Solvent Ethaline. J. Phys. Chem. B 2020, 124 (25), 5251−5264.
(31) Redlich, O.; Kister, A. T. Algebraic Representation of
Thermodynamic Properties and the Classification of Solutions. Ind.
Eng. Chem. 1948, 40 (2), 345−348.
(32) Sarmad, S.; Nikjoo, D.; Mikkola, J.-P. Amine functionalized
deep eutectic solvent for CO2 capture: Measurements and modeling.
J. Mol. Liq. 2020, 309, 113159.
(33) Zhang, K.; Hou, Y.; Wang, Y.; Wang, K.; Ren, S.; Wu, W.
Efficient and Reversible Absorption of CO2 by Functional Deep
Eutectic Solvents. Energy Fuels 2018, 32 (7), 7727−7733.
(34) Matthews, R. P.; Welton, T.; Hunt, P. A. Competitive pi
interactions and hydrogen bonding within imidazolium ionic liquids.
Phys. Chem. Chem. Phys. 2014, 16 (7), 3238−3253.
(35) Dong, K.; Zhang, S.; Wang, J. Understanding the hydrogen
bonds in ionic liquids and their roles in properties and reactions.
Chem. Commun. 2016, 52 (41), 6744−6764.
(36) Low, K.; Tan, S. Y. S.; Izgorodina, E. I. An ab initio Study of the
Structure and Energetics of Hydrogen Bonding in Ionic Liquids.
Front. Chem. 2019, 7 (208), 208.

1098 https://dx.doi.org/10.1021/acssuschemeng.0c07217
ACS Sustainable Chem. Eng. 2021, 9, 1090−1098

You might also like