ARTICLE

Received 11 Jul 2014 | Accepted 5 Sep 2014 | Published 9 Oct 2014 DOI: 10.1038/ncomms6147 OPEN

A hybrid absorption–adsorption method to

efficiently capture carbon
Huang Liu1,*, Bei Liu1,*, Li-Chiang Lin2,*, Guangjin Chen1, Yuqing Wu1, Jin Wang1, Xueteng Gao1, Yining Lv1,
Yong Pan1, Xiaoxin Zhang1, Xianren Zhang3, Lanying Yang1, Changyu Sun1, Berend Smit2,4 & Wenchuan Wang3

Removal of carbon dioxide is an essential step in many energy-related processes. Here we

report a novel slurry concept that combines specific advantages of metal-organic frameworks,
ion liquids, amines and membranes by suspending zeolitic imidazolate framework-8 in
glycol-2-methylimidazole solution. We show that this approach may give a more efficient
technology to capture carbon dioxide compared to conventional technologies. The carbon
dioxide sorption capacity of our slurry reaches 1.25 mol l  1 at 1 bar and the selectivity of
carbon dioxide/hydrogen, carbon dioxide/nitrogen and carbon dioxide/methane achieves
951, 394 and 144, respectively. We demonstrate that the slurry can efficiently remove
carbon dioxide from gas mixtures at normal pressure/temperature through breakthrough
experiments. Most importantly, the sorption enthalpy is only  29 kJ mol  1, indicating that
significantly less energy is required for sorbent regeneration. In addition, from a technological
point of view, unlike solid adsorbents slurries can flow and be pumped. This allows us to use a
continuous separation process with heat integration.

1 State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Beijing 102249, P. R. China. 2 Department of Chemical and Biomolecular

Engineering, University of California, Berkeley, California 94720, USA. 3 State Key Laboratory of Organic-Inorganic Composites, Beijing University of Chemical
Technology, Beijing 100029, P. R. China. 4 Laboratory of Molecular Simulation, Institut des Sciences et Ingénierie Chimiques, Ecole Polytechnique Fédérale de
Lausanne (EPFL), CH-1015 Lausanne, Switzerland. * These authors contributed equally to this work. Correspondence and requests for materials should be
addressed to G.C. (email: gjchen@cup.edu.cn) or to C.S. (email: cysun@cup.edu.cn) or to B.S. (email: berend-smit@berkeley.edu).

NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6147

R
eduction of CO2 emission is directly influenced by the out in a continuous process, whereas the solid adsorbent is
efficiency with which we are able to capture carbon from typically used in a less efficient batch process. To take full
flue gas and other gas mixtures related to energy advantage of the exciting developments in the field of MOFs,
generation, such as biogas, integrated gasification combined cycle it is essential to remove these intrinsic difficulties related to a
gas, syngas, shift gas produced from steam reforming of natural solid-adsorbent process.
gas or coal and natural gas1–4. Motivated by the staggering energy The approach we develop in this work is based on the idea
costs associated with these gas separations, finding the optimal that we can make slurries by suspending solid adsorbents in a
material for a given separation has become a very active area of liquid absorbent. From a process engineering point of view,
modern chemistry5–7. In particular, metal-organic frameworks slurries are very similar to liquids. Hence, with slurries we can
(MOFs)8, a new class of crystalline nanoporous materials, are develop a continuous process and use heat integration. For
regarded as promising candidates for CO2 separations. MOFs are conventional adsorbents, slurries would be a terrible idea, as the
three-dimensional networks of metal clusters that are connected liquid would fill the pores. The beauty of MOFs is, however, that
with organic linkers; by changing the metal and/or linker we can we can use their tunability to select a material with pores that are
synthesize millions of different materials. Moreover, one can use sufficiently small to prevent liquid-absorbent molecules from
this tunability to synthesize a material that has exactly the right entering the materials, but sufficiently large for the gas
pore volume, surface area and selectivity to efficiently separate molecules to be adsorbed. A carefully chosen absorbent/adsorbent
CO2 (refs 9–11).The enthusiasm of the scientific community combination may lead to a significantly enhanced separation
about MOFs and other nanoporous materials as solid adsorbents, performance, which is denoted as the absorption–adsorption
however, does not yet resonate in the process engineering system.
community12. To understand why not, consider a simple solid-
adsorption separation process for carbon capture from flue gas. Results
The first step involves an adsorber, containing the nanoporous
Separation with ZIF-8/glycol slurry. As a proof of concept,
material, which selectively adsorbs CO2 from the flue gas. Once we first studied slurries of ZIF-8 (zeolitic imidazolate
the adsorber is saturated, regeneration is required, which is
framework-8) suspended in liquid glycol (see Fig. 1). ZIFs are a
typically done by supplying heat (temperature swing adsorption) sub-class of MOFs14, and generally possess excellent solution,
or applying vacuum (pressure swing adsorption) to the adsorbent.
chemical and thermal stability15–17. ZIF-8 has narrow six-
This process needs at least two columns, which alternate between membered ring pore windows (3.4 Å), which are smaller than
the adsorption and the regeneration mode. Liquid absorption
the kinetic diameter of glycol molecules (4.5 Å)18, and hence
uses a similar process, replacing the nanoporous material with, glycol is unlikely to influence the adsorption capacity of CO2 in
for example, an amine solution. If we now compare one of the
suspended ZIF-8. To demonstrate that our ZIF-8/glycol slurry
most promising MOFs with commercially available amine can separate different CO2 mixtures, we performed adsorption
solutions, the energy required to regenerate the amine solutions
measurements on different CO2 gas mixtures (that is, CO2 with
is about one order of magnitude larger than the energy required CH4, N2 or H2, see Supplementary Tables 1–5). Figure 2 shows
to regenerate this MOF4,13. This is because in amine solutions,
the selectivity as a function of pressure for these three mixtures
the CO2 is so strongly bound that one needs to boil the amine (that is, up to B745 for CO2/H2, B286 for CO2/N2 and B37 for
solution to reverse the chemical bonding; as the amine solution
CO2/CH4), which are sufficiently large for an effective separation.
contains 70% water, most of this energy is actually used for
No remarkable loss of separation ability was observed after
boiling water. Nevertheless, liquid absorption is the current state- several times of cycling use of the slurry (Supplementary Table 2),
of-the-art process for carbon capture and, surprisingly, solid
where the slurry was regenerated by applying vacuum. Further
adsorption is not considered as such a promising alternative. The characterization of the solid ZIF-8, recovered from the slurry after
fact is due to two significant advantages of the use of liquids. First,
these cycles showed that the ZIF-8 structure remained intact (see
the liquid phase allows us to use advanced heat integrations to scanning electron microscopy images, X-ray diffraction (XRD)
recover a large fraction of the heat. In solid adsorption by
patterns, Fourier transform infrared spectra, Fourier transform
contrast, efficient heat integration is very difficult. Without heat Raman spectra and energy dispersive X-ray spectroscopy patterns
recovery, the energy efficiency of a solid-adsorption process will
in Supplementary Figs 1–5). In addition to glycol, other liquids
be low. Second, pumping allows liquid absorption to be carried such as ethanol, cyclohexane, normal hexane, methylbenzene,
tetrachloromethane and triethylene glycol have been tested to
form slurries with ZIF-8 to separate a CO2/N2 gas mixture
(Supplementary Table 6). It was found that only triethylene glycol
is effective with a selectivity higher than 50. Compared with
triethylene glycol, the other studied liquids consist of smaller
molecules that can enter into ZIF-8 frameworks. Note that, to
provide a reference for showing the superiority of the slurry
approach proposed in this work, absorption separation using pure
glycol or water for the CO2/N2 and CO2/CH4 gas mixtures
(Supplementary Table 7) and adsorption separation using solid
ZIF-8 for the CO2/N2, CO2/H2 and CO2/CH4 gas mixtures
(Fig. 2, Supplementary Tables 8–10) were also performed.

Separation with ZIF-8/glycol-2-methylimidazole slurry.

Though we have obtained promising CO2 selectivities, Fig. 3a
indicates that, as in most application the partial CO2 pressure is
Figure 1 | Formation of ZIF-8/glycol slurries. Photographs of low, the solubility coefficient of ZIF-8/glycol slurry at these
experimentally prepared slurries with solid ZIF-8 suspended in liquid glycol: conditions is too low for practical applications. We further tuned
(a) before mixing, (b) after mixing. the absorbent by adding 2-methylimidazole (mIm) to the glycol.

2 NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6147 ARTICLE

3.5
Water
3.0 [p5mim][bFAP]

CO2 solubility coefficient

2.5 [bmim][pF6]

(mol l–1 per bar)

MEA
2.0
102 MDEA
Selectivity

1.5 TEA
CO2/N2 mixture 1.0 DEAE
ZIF-8/glycol
Solid ZIF-8
0.5
Solvent glycol-mIm
ZIF-8/glycol-mIm
ZIF-8/glycol-mIm 0.10
Glycol-mIm
101 0.05 ZIF-8/glycol
0.00
0 1 2 3 4 5 6 7 8
5 10 15 20 25
Pressure (bar)
Pressure (bar)

–26
Glycol-mIm

CO2 ab(d)sorption enthalpy

102 –28 ZIF-8/glycol-mIm

–30

(kJ mol–1)
Selectivity

–32

–34
101
–36
CO2/CH4 mixture
ZIF-8/glycol –38
Solid ZIF-8
ZIF-8/glycol-mIm –40
100 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
0 10 20 30 40 50 60 CO2 ab(d)sorption amount (mol l–1)
Pressure (bar)
3.0
Ab(d)sorption amount (mol l–1)
CO2
CH4
103 2.5
N2
CO2/H2 mixture H2
ZIF-8/glycol
2.0
Selectivity

Solid ZIF-8
ZIF-8/glycol-mIm 1.5
102
1.0

0.5

101 0.0
0 10 20 30 40 50 60 70
5 10 15 20 25 30 35 Pressure (bar)
Pressure (bar) Figure 3 | Ab(d)sorption solubility and enthalpies. (a) The comparison of
Figure 2 | Separation performance of different media. This figure shows CO2 solubility coefficients in ZIF-8/glycol–mIm slurry at 303.15 K with that
the selectivity of CO2 over (a) N2, (b) CH4 and (c) H2 as a function of in ZIF-8/glycol at 293.15 K and that reported in the literature ((MEA (mass
pressure in the solid ZIF-8, ZIF-8/glycol slurry at 293.15 K and in liquid fraction 30%), MDEA (mass fraction 30%), TEA (mass fraction 30%),
glycol–mIm, ZIF-8/glycol–mIm slurry at 303.15 K. DEAE (mass fraction 30%)) at 313.15 K, ([p5mim][bFAP], [bmim][PF6])
at 298.15 K)4,20–23, (b) sorption enthalpies of CO2 in glycol–mIm liquid and
ZIF-8/glycol–mIm slurry at 303.15 K (c) isotherms of CO2, CH4, N2 and H2
The solubility of CO2 in glycol–mIm (3:2) solution is
at 303.15 K in ZIF-8/glycol–mIm slurry.
0.64 mol l  1 at 303.15 K and 1 bar (see Fig. 3a) and the selectivity
of CO2 over N2 is higher than 200 (Supplementary Table 11). The
CO2 absorption enthalpy in glycol–mIm solution is only about bar at 298.15 K)4 and about 15 times higher than that in ion
 34 kJ mol  1 at 303.15 K (Fig. 3b), which can lead to a much liquid [bmim][PF6] (B0.108 mol l  1 per bar at 298.15 K)22.
lower regeneration cost compared with many of the aqueous We anticipate that the solubility might further increase with
alkanolamines (around  100 kJ mol  1)12 and ion liquids19. higher ZIF-8 fraction (indicated by Supplementary Table 13).
For the ZIF-8/glycol–mIm slurry system, we now obtain a Furthermore, the CO2 absorption–adsorption enthalpy in the
solubility coefficient, Sc, in (ZIF-8 15 wt% þ mIm 34 wt% þ glycol slurry is only about  29 kJ mol  1 (see Fig. 3b, which was
51 wt%) of 1.63 mol l  1 per bar (Fig. 3a, Supplementary Fig. 6 obtained from Supplementary Fig. 7). Compared with the pure
and Supplementary Table 12). Such solubility is sufficiently high glycol–mIm liquid, the prepared slurry has an enhanced CO2
at low CO2 partial pressures to ensure a good sorption capacity at solubility with a reduced energy requirement for CO2 desorption.
these conditions of practical interest. Although this solubility is One can expect that an efficient CO2 capture process at lower
still slightly smaller than that in MEA20 and MDEA21 solutions energy cost can be achieved by using this advanced slurry in this
that are used in the current technologies, it is much higher than regard.
some promising ion liquids for CO2 capture: about 33 times Another important result is that the selectivity of the ZIF-8/
higher than that in ion liquid [p5mim][bFAP] (0.048 mol l  1 per glycol–mIm slurry is significantly higher than the selectivities

NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications 3

& 2014 Macmillan Publishers Limited. All rights reserved.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6147

100 100

Concentration in the outlet

Concentration in the outlet

80 CO2 80 CO2 (slurry)
N2 CH4 (slurry)

(molar %)

(molar %)
60 60 CO2 (water)
CH4 (water)

40 40
CO2 (feed gas)
CO2 (feed gas)
20 20

0 0
0.00
0.02
1 2 3 4 5 6 7 0.0 0.4 0.8 1.2 1.6 4 6 8 10
Time (h) Time (h)

Figure 4 | Breakthrough experiment. (a) Bubble column breakthrough experiment for a CO2/N2 gas mixture (z1 ¼ 0.2065) (303.15 K, 1 bar) carried
out in ZIF-8/glycol–mIm slurry, (b) Column breakthrough experiments for a CO2/CH4 gas mixture (z1 ¼ 0.276) (303.15 K, 1 bar) carried out in both
ZIF-8/glycol–mIm slurry and pure water. The x axis is the time of the breakthrough experiment, the left y axis represents the concentration of gas
components in outlet gas.

observed in both ZIF-8/glycol slurry and glycol–mIm solution short. That suggests that this simple bubbling operation gives a
(see Fig. 2, Supplementary Tables 12–15). The highest selectivity high ab(d)sorption rate as most of CO2 in the inlet gas can be
for CO2/H2, CO2/N2 and CO2/CH4 reaches 951, 394 and 144, removed despite the short contacting time.
respectively, which are sufficiently large for an effective separa- At this point, it is instructive to compare our ZIF-8/
tion. To the best of our knowledge, these selectivities are better glycol–mIm slurry approach with the water-based technology
than those reported in MOFs23,24. Moreover, as shown in Fig. 2, that is currently used to separate CO2 from biogas. Biogas is a gas
such a high selectivity was obtained for a large pressure range, mixture of CO2 and CH4 coming from the breakdown of organic
which demonstrates that this kind of slurry is suitable for the matter in the absence of oxygen25. To illustrate the superior
CO2 capture from different kinds of feed gases such as flue gas performance of the absorption–adsorption approach compared
(CO2/N2, B1 bar), biogas (CO2/CH4, B1 bar), integrated with water, CO2/CH4 (zCO2 ¼ 0.276) mixture breakthrough
gasification combined cycle gas (CO2/H2, 3B5 MPa) or natural experiment using the slurry was compared with the
gas (CO2/CH4, 45 MPa). To further quantify the performance of conventional process based on water. Figure 4b (and
the ZIF-8/glycol–mIm slurry in a separation, we measured the Supplementary Table 18) shows that CO2 breakthrough
absorption–adsorption isotherms for single gas components occurred in the water-based system within 1 min, much faster
(CO2, CH4, N2 and H2). As seen from Fig. 3c, the uptake of than that in ZIF-8/glycol–mIm slurry (10 min). In addition, after
CO2 in the slurry is much higher than those of other components, about 1 h, the concentration of CO2 in the outlet gas nearly
especially in lower pressure range. It should be noted that all equalled to that in the inlet gas (27.6 mol%) for the water-based
measurements for both solubility and selectivity in the ZIF-8/ system, while for our slurry the CO2 concentration was only
glycol–mIm slurry shown in Figs 2 and 3 were performed using 3.04 mol%, and it increased to only 12.73 mol% after 9.4 h.
the same slurry through cycles of sorption/desorption (that is,
apply vacuum for desorption). We observed no remarkable loss of Discussion
the mass and separation ability of the slurry during these A better understanding of the slurry system can play an
measurements. Similar to the ZIF-8/glycol system, further important role in further improving the proposed slurry
characterization of the solid ZIF-8, which were recovered from approach. It is interesting to compare selectivities in the slurry
the slurry that has been reused by 33 times in 25 days, showed with those in the pure solid. Let us first, take a purely
that the ZIF-8 structure remained intact (Supplementary Figs 1e thermodynamic look and assume that gases in the three phases
and 2c). Supplementary Movie 1 illustrates the CO2 capture are in equilibrium. At equilibrium, there is equal temperature and
process (fresh slurry, equilibrium slurry, gas desorption process equal chemical potentials of each of the components in the gas,
and recovered slurry). liquid and solid phases. If we further assume that the liquid
absorbent and solid adsorbent in the slurry do not influence each
Column breakthrough tests. To mimic an actual separation other, the expected selectivity of a particular mixture will be the
process for capturing carbon in ZIF-8/glycol–mIm slurry, column weighted average of the selectivities in the (pure) liquid absorbent
breakthrough tests using two binary mixtures, CO2/N2 or (pure) solid adsorbent26. Surprisingly, Fig. 5a shows that the
(zCO2 ¼ 0.2065) and CO2/CH4 (zCO2 ¼ 0.276), at 303.15 K were selectivity of the slurry phase for all mixtures are significantly
performed in a stainless bubbling column (Supplementary Fig. 8). higher than the selectivity observed in the pure solid ZIF-8 phase,
For the CO2/N2 (zCO2 ¼ 0.2065) mixture, N2 breakthrough the pure liquid phase as well as the weighted average value
occurred within 1 min, whereas CO2 breakthrough occurred after of them.
about 17 min for CO2/N2 (Fig. 4a and Supplementary Table 16). To explain these observations, we have plotted in Fig. 5b the
After the breakthrough of CO2, notably, the concentration of CO2 pure component isotherms inside the pure solid ZIF-8 and the
in the outlet gas still kept low for a long time; even after 6.7 h, the ZIF-8 suspended in the slurry phase. These isotherms show that
concentration of CO2 in the outlet gas was still less than half of the adsorption of CO2 in the solid ZIF-8 suspended in a slurry is
the CO2 concentration in the feed gas. Similar results were similar to adsorption observed in pure solid ZIF-8, but the
obtained for CO2/CH4 (Fig. 4b and Supplementary Table 17). It corresponding adsorption of CH4, N2 and H2 is significantly
should be noted that the slurry used for CO2/N2 was regenerated lower. From the Henry constants of these gases in glycol at
by purging with Helium at 318.15 K and atmospheric pressure, 293.15 K (Supplementary Table 19), we see that, compared with
and then used for CO2/CH4. As the height of slurry in the column CO2, the solubility of the other gases is much lower. In the slurry,
is only 1.31 m, the contact time of gas bubbles with slurry is very all gases that enter the solid phase need to first pass through the

4 NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6147 ARTICLE

103
CO2/N2 mixture CO2
10

Adsorption amount (mol kg–1)

(IAST) solvent glycol CH4
(IAST) solid ZIF-8 N2
8 H2
(Cal.) slurry at equilibrium

Selectivity
(Exp.) ZIF-8/glycol
6
102
4

101 0
5 10 15 20 25 0 10 20 30 40 50
Pressure (bar) Pressure (bar)

Figure 5 | Enhanced selectivity in ZIF-8/glycol slurries. (a) The experimentally measured CO2/N2 selectivity at 293.15 K in ZIF-8/glycol slurries
(solid circles). For comparison, we have calculated the expected selectivities in the pure solid ZIF-8 phase (black lines), the pure liquid glycol phase
(red lines) and the corresponding weighted average (blue lines). In these calculations, we used the ideal adsorbed solution theory (IAST)26 to predict
the mixture adsorption isotherm from the experimental single-component adsorption isotherms. For these calculations, we used 15% CO2 concentration
in gas phase and a mass fraction of ZIF-8 equals to 0.152 in liquid glycol for the weighted average, (b) adsorption isotherms of CO2, CH4, N2 and H2
at 293.15 K in pure solid ZIF-8 (closed symbols) and the ZIF-8 suspended in ZIF-8/glycol slurry (open symbols).

(ref. 27). However, as we have seen from this study, it has strong
positive effect when ZIF-8 is used in slurry state. The lessened
demand of product purity makes the production of ZIF-8 much
Gas phase less expensive, because the removal of surplus mIm by
vapourizing the product requires high cost of energy.
To interpret the high CO2 absorbing ability of the glycol–mIm
solution, we propose a mechanism as illustrated in
Supplementary Fig. 9. We suggest that there is a quasi-
ZIF-8 chemisorptive formation of rather unstable N–C bonded
carbamate species between CO2 and mIm. The rather unstable
carbamate leads to a smaller CO2 absorption enthalpy in glycol–
Slurry phase mIm solution, as it is usually much larger in typical chemical
absorption12 or chemical adsorption28 processes. In aqueous
CO2 Other gas molecule amine, carbamate can further react with water and subsequently
lead to quite stable carbonate species29. In the glycol–mIm
Figure 6 | Hybrid absorption–adsorption process. Schematic of the hybrid solution, however, the carbamate species cannot further react
absorption–adsorption separation process for CO2 gas mixtures in the with glycol. The mechanism is indirectly supported by
slurry formed by ZIF-8 suspended in glycol solution. experimental observations that the presence of CO2 can
obviously increase the quantity of mIm dissolved in glycol. It
should be noted that other researchers also proposed a similar
liquid film formed by glycol molecules surrounding solid particle. mechanism for the absorption of CO2 in ion liquids that have
Therefore the low solubilities of CH4, N2 and H2 create significant function groups of  NH or  NH2 (refs 30,31). In addition,
mass-transfer limitations. The flux of these molecules is so low there might be a structural interplay between glycol and mIm,
that on the relevant timescale the adsorption of these gases in which effectively lower the absorption enthalpy. A more detailed
ZIF-8 is far less than one can expect from equilibrium investigation, however, is required to validate the proposed
consideration and hence the selectivity in the slurry at low mechanism and to better understand the fundamental
pressures is enhanced. Or stated differently, the liquid absorbent physics/chemistry behind the system. In particular, as a future
functions as a semipermeable membrane for which the work, experimental characterizations, start-of-the-art quantum
permeability of those components other than CO2 is so low that, mechanical calculations and molecular simulations will be carried
on the timescale of the separation, these components never reach out to study the pure solvent and slurry systems in detail.
the equilibration loading in the ZIF-8 suspended in the absorbent. Three important insights emerge from this work. The unique
A schematic diagram of the system can be seen in Fig. 6. If we properties of MOFs allow us to develop a slurry-based process to
increase the pressure (Fig. 2), we observe a decrease in the mitigate the main difficulties in the conventional solid-adsorption
selectivity. These observations are consistent with an increase of processes. As the slurry can be pumped, we can carry out a
the gas concentration in the liquid phase. On the other hand, process very similar to liquid absorption. Second, the absorbent–
if we decrease the temperature we see a further increase in adsorbent combination gives a surprisingly extra dimension for
selectivity (Supplementary Table 3). When we used glycol–mIm optimizing the design of a separation. The absorbent can act as a
solution instead of pure glycol as solvent, we observed higher semipermeable membrane, preventing equilibration of one of the
selectivity (Fig. 2a–c) as the solubility of CO2 in the glycol–mIm components and hence enhancing the selectivity of the separa-
solution is much higher than that in pure glycol. At this point, it tion. Finally, the single ligand mIm used in ZIF-8 synthesis (that
is instructive to point out that mIm is the single ligand used for is, also the main impurity of ZIF-8) could substantially increase
synthesizing ZIF-8 and usually becomes the primary impurity of the carbon capture capacity of the MOF/solvent-ligand slurry.
the product. Such an impurity has been shown to impose a It suggests a lower requirement on the MOF purities as well as a
seriously negative effect on the adsorption ability of solid ZIF-8 reduced cost in their production while using the slurry approach.

NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications 5

& 2014 Macmillan Publishers Limited. All rights reserved.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6147

In fact, in our example, we have chosen ZIF-8 that has a very in the equilibrium gas phase of the sapphire cell after absorption and adsorption
equilibrium is determined by:
modest selectivity towards CO2, yet in the slurry shows excellent
separation performance because of this enhancement of the PE Vg
nE ¼ ð2Þ
selectivity caused by the absorbent and mIm that is a general ZE RT
ligand in the synthesis of MOFs and ion liquids, yet in the slurry where PE is the equilibrium pressure of the sapphire cell and ZE is the
significantly increases the slurry’s carbon capture and separation compressibility factor corresponding to T, PE and gas composition. Vg is the
volume of equilibrium gas phase in the sapphire cell at the end of each
efficiency. All of the above suggests that this work opens up many experimental run. The total uptake of CO2 (n1) and that of N2 (CH4 or H2) (n2) in
exciting possibilities to further optimize CO2 removal process by slurry are calculated as follows:
combining the specific advantages of MOFs, ion liquids, amines n1 ¼ nt z1  nE y1 ð3Þ
and membranes.
n2 ¼ nt z2  nE y2 ð4Þ
Methods where z1 and y1 are the mole fraction of CO2 in the synthetic gas and equilibrium
Materials. Materials used in this work include ZIF-8, mIm, glycol, ethanol, gas phase, respectively; z2 and y2 are the mole fraction of N2 (CH4 or H2) in the
n-hexane, cyclohexane, methybenzene, tetrachloromethane, triethylene glycol, synthetic gas and equilibrium gas phase, respectively.
water and feed gases. Among them, both ZIF-8 and mIm were purchased from Accordingly, the apparent mole fractions of CO2 (x1) and N2 (CH4 or H2) (x2)
Sigma-Aldrich. The scanning electron microscopy images (Supplementary Fig. 1a) in the equilibrium slurry phase can be obtained by the following formulas:
and XRD patterns of the purchased ZIF-8 sample (Supplementary Fig. 2a) are n1
in good agreement with the ones reported in the literature32. Glycol, ethanol, x1 ¼ ð5Þ
n 1 þ n2
n-hexane, cyclohexane, methylbenzene, tetrachloromethane and triethylene
glycol were purchased from Beijing Chemical Reagents Company, China. n2
Analytical grade carbon dioxide (99.99%), nitrogen (99.99%), methane (99.99%) x2 ¼ ð6Þ
n 1 þ n2
and hydrogen (99.999%) were purchased from Beijing AP Beifen Gas Industry
Company, China. The synthetic gases CO2/N2, CO2/H2 and CO2/CH4 were In an equilibrium-based separation process, a good indication of the CO2
prepared in our own laboratory. A Hewlett-Packard gas chromatograph (HP 7890) separation efficiency is the sorption selectivity34. In this study, the apparent
was used to analyze the composition of the prepared gases. selectivity of CO2 over other component in slurry (b) is calculated as:
x1 =y1
b¼ ð7Þ
Ab(d)sorption measurements. All the ab(d)sorption measurement experiments x2 =y2
were performed using the experimental apparatus as schematically illustrated in To provide a reference for showing the superiority of the proposed0 absorption–
Supplementary Fig. 10. A detailed description of the setup can be found in our adsorption hybrid method, the selectivity of CO2 in solid ZIF-8 (b ) is also
previous report33. The key parts of the apparatus are a transparent sapphire cell provided, which is calculated as:
and a steel-made blind cell, which are both installed in an air bath. The effective
volume of the sapphire cell is 60 cm3 and that of the blind cell plus connected tubes x10 =y1
b0 ¼ ð8Þ
is 112 cm3. The maximum working pressures of these two cells are designed to be x20 =y2
20 MPa. To directly observe samples in the cell, a lamp with luminescence source
where x10 and x20 are the mole fractions of CO2 and N2 (CH4 or H2) in adsorbed
(type LG100H) is mounted on the outside of the cell. A secondary platinum
phase (ZIF-8), respectively.
resistance thermometer (type-pt100) is used as the temperature sensor. A
The solubility coefficient (Sc) of CO2 in sorbents, an important indication of the
calibrated Heise pressure gauge and differential pressure transducers are used to
separation capacity, is calculated as follows:
measure the system pressure. The uncertainties of pressure and temperature
measurements are ±0.01 MPa and±0.1 K, respectively. Real-time readings of the n1
Sc ¼ ð9Þ
system temperature and pressure are recorded. Vs PE y1
Before the experiments, the sapphire cell was dismounted from the apparatus, where Vs is the volume of sorbents (liquid or slurry).
washed with distilled water and dried, then loaded with a known quantity of dry The apparent volumetric solubility of CO2 in slurry (Sv) is defined as:
porous material. After that, a known amount of solvent was immersed into the
sapphire cell slowly and evenly. Both the used dry porous material and solvent were n1
Sv ¼ ð10Þ
weighed by an electrical balance with a precision of ±0.1 mg. The mixture of Vs
porous material and liquid solvent was stirred to form a suspension mixture (that The initial gas–slurry volume ratio (F) and gas–solid adsorbent volume ratio (F0 )
is, slurry). Subsequently, the cell was installed back into the apparatus. The system are defined as:
(sapphire cell þ blind cell þ tubes connecting two cells) was then purged through
vacuuming. Enough amount of synthetic gas was injected into the blind cell, f ¼ nt RTSTP /PSTP
ð11Þ
then the desired value of temperature was set through the air bath. Once both Vs
temperature and pressure of the blind cell were kept constant, the pressure of gas
mixture in the blind cell was recorded as the initial pressure P0. The top valve of the f0 ¼ nt RTSTP /PSTP
ð12Þ
sapphire cell was opened slowly then, letting the desired amount of synthetic gas Vs0
flow into the sapphire cell from the blind cell. Afterwards, this valve was closed
and the magnetic stirrer was turned on. The pressure of the residual gas mixture where Vs0 is the volume of solid adsorbent. TSTP and PSTP are standard temperature
in the blind cell was recorded as P1. With the sorption of gas mixture by the slurry, and pressure, respectively.
the system pressure in the sapphire cell decreased gradually. During each The adsorption capacity of solid ZIF-8 suspended in liquid absorbents for gas
measurement, the pressure in the sapphire cell as a function time was recorded. components is calculated as follows:
When the system pressure remained as a constant for at least 2 h, we considered p1 ms
n10 ¼ n1  ð13Þ
the equilibrium of system was achieved. The equilibrium pressure of the sapphire H1
cell was recorded as PE. Gas mixture in the equilibrium gas phase of the sapphire
cell was sampled under constant pressure by pushing the connected hand pump p2 ms
n20 ¼ n2  ð14Þ
and analysed by a HP 7890 gas chromatograph. The volume of the slurry in the H2
sapphire cell can be obtained by measuring the height of the equilibrium liquid n10 n20
where and are the total uptake of CO2 and that of N2 (CH4 or H2) in ZIF-8
phase. The inner radius of the sapphire cell is known to be 1.27 cm. In this work, that is suspended in slurry, H1 and H2 are their Henry constants in liquid
the amount of each gas species absorbed and adsorbed in the measured sample was absorbents, and p1 and p2 are the corresponding equilibrium partial pressure,
determined through mass balance as described below. respectively. ms is the mass of absorbents. Henry constants of CO2, CH4, N2 and H2
The total mole number of gas mixtures (nt) that was injected into the sapphire in glycol at 293.15 K are experimentally determined and reported in Supplementary
cell is calculated by the following formula: Table 19.
The mass balance method stated above has been used in our previous work33,35.
P0 Vt P1 Vt
nt ¼  ð1Þ To validate our measurement approach, the adsorption isotherm of pure CO2 and
Z0 RT Z1 RT the gas selectivity obtained for CO2/N2 (z1 ¼ 0.2286) mixture in solid ZIF-8 at
303.15 K were measured using this method and compared with the literature data
where T is the system temperature, P0 is the initial pressure of the blind cell, P1 is (Supplementary Figs 11 and 12).
the equilibrium pressure of the blind cell after injecting gases into the sapphire cell,
Vt is the total volume of the blind cell plus tubes connecting to it and R is the
gas constant. Compressibility factors Z0 and Z1 were calculated using the Characterization. The adsorbents (solid ZIF-8) were characterized by XRD
Benedict–Webb–Rubin–Starling equation of state. The total gas amount (nE) (SIMADU XRD 6000) with Cu Ka radiation (0.1542 nm, 40 kV and 400 mA) at a

6 NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6147 ARTICLE

scanning rate of 2 °C per minute. The morphologies and energy dispersive X-ray 23. Nugent, P. et al. Porous materials with optimal adsorption thermodynamics
spectroscopy measurements were obtained using a FEI Quanta 200F scanning and kinetics for CO2 separation. Nature 495, 80–84 (2013).
electron microscope. Fourier transform infrared spectra were obtained using a 24. Babarao, R. & Jiang, J. Unprecedentedly high selective adsorption of gas
Bruker 80v spectrometer. Fourier transform Raman spectra were obtained using mixtures in rho zeolite-like metal-organic framework: a molecular simulation
a HORIBA XploRA spectrometer. The thermogravimetric measurements were study. J. Am. Chem. Soc. 131, 11417–11425 (2009).
carried out at a NETZSCH STA 409 PC/PG instrument. 25. Abatzoglou, N. & Boivin, S. A review of biogas purification processes. Biofuel,
Bioprod. Biorefin. 3, 42–71 (2009).
Column breakthrough tests. The experimental setup used for dynamic break- 26. Myers, A. L. & Prausnitz, J. M. Thermodynamics of mixed-gas adsorption.
through measurements is shown in Supplementary Fig. 8. The gas manifold AIChE J. 11, 121–127 (1965).
consists of two lines, A and B, equipped with mass flow controllers. From line ‘A’, 27. Kida, K., Okita, M., Fujita, K., Tanaka, S. & Miyake, Y. Formation of
an inert gas (that is, helium) is injected to purge air out of the column and pipelines high crystalline ZIF-8 in an aqueous solution. CrystEngComm 15, 1794–1801
before experiment and to regenerate the slurry after each experiment. From line ‘B’, (2013).
a gas mixture containing CO2 is injected with a constant flow rate for the break- 28. McDonald, T. M., D’Alessandro, D. M., Krishna, R. & Long, J. R. Enhanced
through experiment. Both lines ‘A’ and ‘B’ are connected to the column using a carbon dioxide capture upon incorporation of N,N0 -dimethylethylenediamine
three-way valve. The stainless column is 165 cm high with a jacket that is connected in the metal-organic framework CuBTTri. Chem. Sci. 2, 2022–2028 (2011).
to a water batch for the temperature control. The inner diameter of the column is 29. Vaidya, P. D. & Kening, E. Y. CO2-alkanolamine reaction kinetics: a review of
2 cm. About 400 ml absorbent such as ZIF-8/glycol–mIm slurry is loaded in the recent studies. Chem. Eng. Technol. 30, 1467–1474 (2007).
column. A gas distributor was fixed at the bottom of the column. The designed 30. Li, X. et al. Absorption of CO2 by ionic liquid/polyethylene glycol mixture and
maximum working pressure of this apparatus is 5 MPa. The outlet gas from the top the thermodynamic parameters. Green Chem. 10, 879–884 (2008).
of the column is sampled and analysed by a HP 7890 gas chromatograph. 31. Gutowski, K. E. & Maginn, E. J. Amine-functionalized task-specific ionic
liquids: A mechanistic explanation for the dramatic increase in viscosity upon
References complexation with CO2 from molecular simulation. J. Am. Chem. Soc. 130,
1. Chu, S. & Majumdar, A. Opportunities and challenges for a sustainable energy 14690–14704 (2008).
future. Nature 488, 294–303 (2012). 32. Pan, Y., Liu, Y., Zeng, G., Zhao, L. & Lai, Z. Rapid synthesis of zeolitic
2. Chu, S. Carbon capture and sequestration. Science 325, 1599 (2009). imidazolate framework-8 (ZIF-8) nanocrystals in an aqueous system. Chem.
3. Metz, B., Davidson, O., de Coninck, H., Loos, M. & Meyer, L. IPCC special Commun. 47, 2071–2073 (2011).
report on Carbon Dioxide Capture and Storage (Intergovernmental Panel on 33. Liu, H. et al. Separation of ethylene from refinery dry gas via forming hydrate in
Climate Change (IPCC), 2005). w/o dispersion system. Sep. Purif. Technol. 116, 342–350 (2013).
4. Wilcox, J. Carbon Capture (Springer, 2012). 34. Keskin, S. Atomistic simulations for adsorption, diffusion, and separation of gas
5. D’Alessandro, D. M., Smit, B. & Long, J. R. Carbon dioxide capture: prospects mixtures in zeolite imidazolate frameworks. J. Phys. Chem. C 115, 800–807
for new materials. Angew. Chem. Int. Ed. 49, 6058–6082 (2010). (2011).
6. Férey, G. Hybrid porous solids: past, present, future. Chem. Soc. Rev. 37, 35. Mu, L. et al. A novel method to improve the gas storage capacity of ZIF-8.
191–214 (2008). J. Mater. Chem. 22, 12246–12252 (2012).
7. Dawson, R., Cooper, A. I. & Adams, D. J. Chemical functionalization strategies
for carbon dioxide capture in microporous organic polymers. Polym. Int. 62, Acknowledgements
345–352 (2013). The financial supports received from the National Natural Science Foundation of China
8. Yaghi, O. M. et al. Reticular synthesis and the design of new materials. Nature (U1162205, 21276272), the National 973 Project of China (No. 2012CB215005), the
423, 708–714 (2003). Research Fund of China University of Petroleum, Beijing (2462013YXBS005) and the
9. Li, J.-R., Sculley, J. & Zhou, H.-C. Metal–organic frameworks for separations. Program for New Century Excellent Talents from Ministry of Education (NCET-12-
Chem. Rev. 112, 869–932 (2012). 0968) are gratefully acknowledged. L.-C.L. and B.S. were supported as part of the Center
10. Furukawa, H., Cordova, K. E., O’Keeffe, M. & Yaghi, O. M. The chemistry and for Gas Separations Relevant to Clean Energy Technologies, an Energy Frontier Research
applications of metal-organic frameworks. Science 341, 974 (2013). Center funded by the U.S. Department of Energy, Office of Science, Office of Basic
11. Sumida, K. et al. Carbon dioxide capture in metal–organic frameworks. Chem. Energy Sciences under Award Number DE-SC0001015.
Rev. 112, 724–781 (2012).
12. Rochelle, G. T. Amine scrubbing for CO2 capture. Science 325, 1652–1654
(2009). Author contributions
13. Lin, L.-C. et al. In silico screening of carbon-capture materials. Nat. Mater. 11, G.C., C.S. and B.S. designed the experiments; B.S. and G.C. led data analysis. H.L., Y.W.,
633–641 (2012). J.W., X.G., Y.L., Y.P., X.X.Z. and L.Y. performed laboratory experiments. B.L., L.-C.L.,
14. Banerjee, R. et al. High-throughput synthesis of zeolitic imidazolate frameworks H.L., G.C., B.S., C.S., X.R.Z. and W.W. performed data analysis. H.L. and L.-C.L. pre-
and application to CO2 capture. Science 319, 939–943 (2008). pared figures and tables. G.C., B.S., B.L., L.-C.L. and C.S. developed the project concept.
15. Küsgens, P. et al. Characterization of metal-organic frameworks by water L.-C.L., B.L., H.L., G.C., B.S., C.S. and W.W. wrote the manuscript.
adsorption. Micropor. Mesopor. Mat. 120, 325–330 (2009).
16. Park, K. S. et al. Exceptional chemical and thermal stability of zeolitic Additional information
imidazolate frameworks. Proc. Natl Acad. Sci. USA 103, 10186–10191 (2006). Supplementary Information accompanies this paper at http://www.nature.com/
17. Cychosz, K. A. & Matzger, A. J. Water stability of microporous coordination naturecommunications
polymers and the adsorption of pharmaceuticals from water. Langmuir 26,
17198–17202 (2010). Competing financial interests: The authors declare no competing financial interests.
18. Hu, C. et al. Development of novel mordenite-filled chitosan-poly(acrylic acid)
polyelectrolyte complex membranes for pervaporation dehydration of ethylene Reprints and permission information is available online at http://www.nature.com/
glycol aqueous solution. J. Membr. Sci. 293, 142–150 (2007). reprints/index.html.
19. Wang, C. et al. Turning the basicity of ion liquids for equimolar CO2 capture. How to cite this article: Liu, H. et al. A hybrid absorption–adsorption method to
Angew. Chem. Int. Ed. 50, 4918–4922 (2011). efficiently capture carbon. Nat. Commun. 5:5147 doi: 10.1038/ncomms6147 (2014).
20. Jou, F. Y., Mather, A. E. & Otto, F. D. The solubility of CO2 in a 30 mass
percent monoethanolamine solution. Can. J. Chem. Eng. 73, 140–147 (1995). This work is licensed under a Creative Commons Attribution 4.0
21. Chowdhury, F. A., Yamada, H., Higashii, T., Goto, K. & Onda, M. CO2 capture International License. The images or other third party material in this
by tertiary amine absorbents: A performance comparison study. Ind. Eng.
article are included in the article’s Creative Commons license, unless indicated otherwise
Chem. Res. 52, 8323–8331 (2013).
in the credit line; if the material is not included under the Creative Commons license,
22. Anthony, J. L., Maginn, E. J. & Brennecke, J. F. Solubilities and thermodynamic
users will need to obtain permission from the license holder to reproduce the material.
properties of gases in the ionic liquid 1-n-butyl-3-methylimidazolium
To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
hexafluorophosphate. J. Phys. Chem. B 106, 7315–7320 (2002).

NATURE COMMUNICATIONS | 5:5147 | DOI: 10.1038/ncomms6147 | www.nature.com/naturecommunications 7

& 2014 Macmillan Publishers Limited. All rights reserved.