Capture of carbon dioxide from air and flue gas in the alkylamine-appended metal-organic framework mmen-Mg2(dobpdc)

Article abstract of DOI:10.1021/ja300034j

Two new metal-organic frameworks, M₂(dobpdc) (M = Zn (1), Mg (2); dobpdc^4- = 4,4′-dioxido-3,3′-biphenyldicarboxylate), adopting an expanded MOF-74 structure type, were synthesized via solvothermal and microwave methods. Coordinatively unsaturated Mg²⁺ cations lining the 18.4-A-diameter channels of 2 were functionalized with N,N′-dimethylethylenediamine (mmen) to afford Mg₂(dobpdc)(mmen) _1.6(H₂O)_0.4 (mmen-Mg₂(dobpdc)). This compound displays an exceptional capacity for CO₂ adsorption at low pressures, taking up 2.0 mmol/g (8.1 wt %) at 0.39 mbar and 25 °C, conditions relevant to removal of CO₂ from air, and 3.14 mmol/g (12.1 wt %) at 0.15 bar and 40 °C, conditions relevant to CO₂ capture from flue gas. Dynamic gas adsorption/desorption cycling experiments demonstrate that mmen-Mg₂(dobpdc) can be regenerated upon repeated exposures to simulated air and flue gas mixtures, with cycling capacities of 1.05 mmol/g (4.4 wt %) after 1 h of exposure to flowing 390 ppm CO₂ in simulated air at 25 °C and 2.52 mmol/g (9.9 wt %) after 15 min of exposure to flowing 15% CO₂ in N₂ at 40 °C. The purity of the CO₂ removed from dry air and flue gas in these processes was estimated to be 96% and 98%, respectively. As a flue gas adsorbent, the regeneration energy was estimated through differential scanning calorimetry experiments to be 2.34 MJ/kg CO₂ adsorbed. Overall, the performance characteristics of mmen-Mg₂(dobpdc) indicate it to be an exceptional new adsorbent for CO₂ capture, comparing favorably with both amine-grafted silicas and aqueous amine solutions.

Full text of DOI:10.1021/ja300034j

Article
pubs.acs.org/JACS
Capture of Carbon Dioxide from Air and Flue Gas in the Alkylamine-
Appended Metal−Organic Framework mmen-Mg2(dobpdc)
Thomas M. McDonald,† Woo Ram Lee,§ Jarad A. Mason,† Brian M. Wiers,† Chang Seop Hong,*,§
and Jeffrey R. Long*,†
†Department of Chemistry, University of California, Berkeley, California 94720, United States
§Department of Chemistry, Research Institute for Natural Sciences, Korea University, Seoul 136-713, Republic of Korea
S
* Supporting Information
ABSTRACT: Two new metal−organic frameworks, M2(dobpdc) (M =
Zn (1), Mg (2); dobpdc4− = 4,4′-dioxido-3,3′-biphenyldicarboxylate),
adopting an expanded MOF-74 structure type, were synthesized via
solvothermal and microwave methods. Coordinatively unsaturated Mg2+
cations lining the 18.4-Å-diameter channels of 2 were functionalized with
N,N′-dimethylethylenediamine (mmen) to afford Mg2(dobpdc)-
(mmen)1.6(H2O)0.4 (mmen-Mg2(dobpdc)). This compound displays
an exceptional capacity for CO2 adsorption at low pressures, taking up
2.0 mmol/g (8.1 wt %) at 0.39 mbar and 25 °C, conditions relevant to
removal of CO2 from air, and 3.14 mmol/g (12.1 wt %) at 0.15 bar and 40 °C, conditions relevant to CO2 capture from flue gas.
Dynamic gas adsorption/desorption cycling experiments demonstrate that mmen-Mg2(dobpdc) can be regenerated upon
repeated exposures to simulated air and flue gas mixtures, with cycling capacities of 1.05 mmol/g (4.4 wt %) after 1 h of exposure
to flowing 390 ppm CO2 in simulated air at 25 °C and 2.52 mmol/g (9.9 wt %) after 15 min of exposure to flowing 15% CO2 in
N2 at 40 °C. The purity of the CO2 removed from dry air and flue gas in these processes was estimated to be 96% and 98%,
respectively. As a flue gas adsorbent, the regeneration energy was estimated through differential scanning calorimetry experiments
to be 2.34 MJ/kg CO2 adsorbed. Overall, the performance characteristics of mmen-Mg2(dobpdc) indicate it to be an exceptional
new adsorbent for CO2 capture, comparing favorably with both amine-grafted silicas and aqueous amine solutions.
variety of advanced amines are available, 30% monoethanol-
amine (MEA) in water is the benchmark solvent against which
competing technologies are generally compared. The low
solvent cost and proven effectiveness make MEA an attractive
absorbent for many applications. However, if MEA were to be
utilized for CCS, electricity prices are projected to increase by
86%.8 The U.S. Department of Energy has targeted a maximum
35% increase for the cost of electricity produced from a coal
power plant that captures 90% of the CO2 it generates. The
diversion of steam from the electricity generation cycle to the
solvent regeneration cycle sharply reduces the net electricity
output of the plant, drastically increasing electricity costs.
Previous work has demonstrated that plant efficiency is highly
dependent on the solvent regeneration energy.9
Presently, there is significant interest in the development of
solid adsorbents that selectively adsorb CO2 at partial pressures
applicable to CCS.10 Solid adsorbents are promising candidates
because the significantly smaller heat capacities of solids may
reduce the sensible heat required for regeneration. In addition,
solvent loss and corrosion issues resulting from the use of
aqueous amines would be minimized if solids adsorbents were
instead utilized.9
1. INTRODUCTION
The concentration of CO2 in the Earth’s atmosphere is
presently 390 ppm,1 an increase of approximately 110 ppm
since the start of the Industrial Revolution.2 The combustion of
fossil fuels is largely responsible for this increase,3 yet fossil
fuels will continue to be heavily utilized for energy production
during the 21st century. Currently, there is significant interest
in the development and implementation of technologies that
slow CO2 emissions and thus forestall the most severe
consequences of global warming. For limiting future CO2
emissions from large, stationary sources like coal-fired power
plants, carbon capture and sequestration (CCS) has been
proposed.4 The CCS process involves the selective removal of
CO2 from gas mixtures, the compression of pure CO2 to a
supercritical fluid, transportation to an injection site, and finally
permanent subterranean or submarine storage.5 For the retrofit
of existing power plants, post-combustion CO2 capture is a
likely configuration. In this design, fuel is burned in air and CO2
is removed from the effluent. For coal-fired power plants, the
largest flue gas components by volume are N2 (70−75%), CO2
(15−16%), H2O (5−7%), and O2 (3−4%), with total pressures
near 1 bar and temperatures between 40 and 60 °C.6
Aqueous amine solutions are currently the most viable
absorbents for carbon capture under the aforementioned
conditions, and they are presently used for the removal of
CO2 from industrial commodities like natural gas.7 While a
Received: January 2, 2012
Published: April 4, 2012
© 2012 American Chemical Society
7056
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
While CCS is perhaps unlikely to be widely implemented
within the next decade,5 a number of current industrial
processes utilize liquid or solid adsorbents to remove CO2
from gas mixtures. These processes could benefit greatly from
the next generation of adsorbents that are currently being
proposed for CCS applications. Currently, aqueous amines are
used industrially to separate CO2 from gas mixtures with high
CO2 partial pressures like natural gas, while solid adsorbents are
used to remove CO2 from mixtures with very low CO2 partial
pressures.
Among the most challenging CO2 separations is the removal
of CO2 directly from air. Prior to the cryogenic distillation of air
for N2, O2, and Ar production, CO2 is removed from the air to
minimize solid CO2 formation on heat exchangers.11 To
maximize the capacity of zeolite 13X for CO2 adsorption, the
air stream is dried over alumina, cooled to 5 °C, and
pressurized to 5−7 bar. The effective capacity of zeolite 13X,
the most widely used adsorbent for this process, under these
conditions is ca. 0.35 mmol/g, which corresponds to a
crystallographic volumetric capacity of approximately 0.5
mmol/cm3.11c,12 In this “pre-purification” step, the CO2
concentration in the air is reduced to less than 1 ppm. New
higher-capacity adsorbents could potentially eliminate the costs
associated with pre-cooling feed air and reduce the adsorbent
regeneration energy.
Carbon dioxide scrubbers are critical life support systems in
confined spaces with limited air exchange, such as spacecraft,
submarines, and breathing suits.13 Because of the very low
capacity of solid adsorbents for 390 ppm CO2 in unpressurized
gas streams, weight and volume limitations prevent the
implementation of systems capable of maintaining CO2
concentrations at atmospheric levels. Thus, the average CO2
concentration aboard the International Space Station ranges
from 3000 to 7000 ppm,14 which approaches the currently
established safe limit for chronic CO2 exposure.14,15 Improved
adsorbents could potentially reduce CO2 concentrations within
confined spaces to significantly lower and potentially safer
levels, while simultaneously reducing the adsorbent mass and
volume.
In addition to current processes that remove CO2 from air,
proposed technologies may benefit greatly from improved air
capture adsorbents. For example, alkaline fuel cells (AFCs)16
and iron-air batteries17 require CO2-free O2 sources to avoid
electrolyte side reactions. It has been previously suggested that
improved CO2 adsorbents could solve many issues associated
with the operation of AFCs because air purification imposes
considerable engineering and financial burdens on the system.18
Lastly, directly adsorbing CO2 from the atmosphere
combined with geologic sequestration has been proposed as a
potential solution for offsetting CO2 emissions from mobile or
diffuse generators.19 Direct air capture, if widely implemented,
could also theoretically reduce atmospheric CO2 concentrations
by capturing historic emissions rather than simply abating
future emissions. Significant obstacles remain for direct air
capture including its substantially higher cost compared to
traditional CCS.20 Yet many estimates rely on the use of
traditional inorganic bases, which require the construction of
very large physical structures to ensure sufficient surface area
for air to contact the adsorbent.21 Porous solid adsorbents with
high surface areas could potentially drastically reduce the large
capital costs associated with air capture.
inorganic bases and zeolites. For example, amine-modified
porous polymers have been tested for use aboard spacecraft.22
Silica,17,23 alumina,24 and carbon25 adsorbents functionalized
with amines have also been analyzed for their efficacy as air
capture adsorbents, as have ammonium ion-exchange mem-
branes.26 The primary advantage of amine-functionalized
adsorbents is their high capacity for 390 ppm CO2, in some
cases in excess of 2 mmol/g. However, existing materials of this
type frequently require hours to reach saturation because of
slow adsorption kinetics.
Metal−organic frameworks are a class of porous, crystalline
adsorbents that have recently attracted much attention for use
in gas separations.10,27 The high tunability of their design may
enable greater functionality with reduced adsorbent mass and
volume compared to traditional solid adsorbents. Among the
most interesting features of some metal−organic frameworks is
the presence of coordinatively unsaturated metal centers (open
metal sites) along the pore surfaces.28,29 These five-coordinate
metal cations, known to behave as Lewis acids that strongly
polarize gas adsorbents, are further amenable to post-synthetic
functionalization.30
In chemically robust metal−organic frameworks with well-
separated open metal sites, one amine of a diamine molecule
can bind to a metal cation as a Lewis base, while the second
amine remains available as a chemically reactive adsorption site.
The modification of open metal sites within the metal−organic
framework Cu-BTTri with the secondary amine N,N′-
dimethylethylenediamine (mmen) was recently reported.31
Upon grafting mmen onto the exposed Cu2+ sites of the
framework, a 3.5 times enhancement in CO2 capacity at 0.15
bar and 25 °C was realized. We believe that the incorporation
of very basic alkylamines into framework pores will be a
promising strategy for increasing the capacity of metal−organic
frameworks for CO2 uptake at low partial pressures.
The high concentration of open metal sites within the
M2(dobdc) (M = Mg, Mn, Fe, Co, Ni, Zn; dobdc4− = 2,5-
dioxido-1,4-benzenedicarboxylate; M-MOF-74 or CPO-27-M)
series of metal−organic frameworks makes them attractive
candidates for diamine functionalization.29 To date, however,
we have been unable to synthesize promising amine
functionalized derivatives of the M2(dobdc) series. We
hypothesized that the relatively narrow, one-dimensional
channels (∼11 Å diameter) may be hindering effective diffusion
of the diamines into the framework. Thus, we sought to
synthesize expanded analogues of the M2(dobdc) structure via
a ligand extension. Larger pores should enable more facile
functionalization, enhance gas diffusion, and potentially unlock
unrealized functionality within this interesting structural
topology replete with open metal sites.
Herein, we report the first expanded analogues of the
M2(dobdc) structure type, featuring 18.4-Å-wide channels and
exhibiting exceptional CO2 adsorption properties upon
functionalization with mmen.
2. EXPERIMENTAL SECTION
General. All reagents were obtained from commercial vendors at
reagent grade purity or higher and used without further purification.
4,4′-Dihydroxy-(1,1′-biphenyl)-3,3′-dicarboxylic Acid (H4dobpdc).
The compounds 4,4′-dihydroxybiphenyl (1.16 g, 6.24 mmol), KHCO3
(2.00 g, 20.0 mmol), dry ice (4 g), and 1,2,4-trichlorobenzene (3 mL)
were added to a PTFE insert within a steel acid digestion bomb (23
mL) and heated at 255 °C for 17 h.32 After cooling to room
temperature, the mixture was collected via vacuum filtration and
washed with diethyl ether. The solid was suspended in 300 mL of
Recently, amine-functionalized porous solids have been
proposed as superior adsorbents for air capture compared to
7057
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
distilled water and filtered again. To the filtrate, neat HCl was slowly
added until a pH between 1 and 2 was reached. The resulting crude
product was collected via filtration. Recrystallization using 50 mL of
acetone and 50 mL of water per gram of crude material afforded 0.68 g
(40%) of pure product as a white powder. Anal. Calcd for C14H10O6:
C, 61.32; H, 3.68. Found: C, 61.34; H, 3.60. IR (KBr) 1659 (vs), 1612
(s), 1481 (s), 1450 (vs), 1319 (s), 1290 (s), 1240 (s), 1099 (w), 1047
(w), 894 (w), 825 (m), 792 (m), 723 (w), 686 (m), 573 (w), 530 (w)
cm−1. 1H NMR (300 MHz, dmso-d6): δ = 10.75−11.95 (br, 2H), 7.93
(s, 2H), 7.74 (d, 2H, J = 11.2 Hz), 7.00 (d, 2H, J = 8.6 Hz).
Zn2(dobpdc)(DEF)2·DEF·H2O (DEF-1). To a 2-mL Pyrex tube,
H4dobpdc (4.0 mg, 0.015 mmol), ZnBr2·2H2O (8.9 mg, 0.034 mmol),
and 0.5 mL of mixed solvent (1:1 DEF:EtOH; DEF = N,N-
diethylformamide) were added. The tube was sealed and placed in a
pre-heated oven at 100 °C. After 72 h, needle-shaped, colorless crystals
had formed. The crystals were isolated by filtration and washed with
hot DEF to afford 3.7 mg (35%) of product. Anal. Calcd for
C29H41N3O10Zn2: C, 47.84; H, 5.64; N, 5.79. Found: C, 48.21; H,
5.72; N, 5.82. IR (KBr): 1655 (vs), 1610 (s), 1544 (s) 1462 (s), 1412
(vs), 1286 (s), 1234 (s), 1149 (m), 1103 (m), 1047 (w), 881 (m), 825
(m), 760 (w), 690 (m), 586 (m), 505 (w) cm−1.
Mg2(dobpdc)(DEF)2·DEF1.5·H2O (DEF-2). Into a 10-mL Pyrex cell,
H4dobpdc (24 mg, 0.088 mmol), MgBr2·6H2O (60 mg, 0.21 mmol),
and 3 mL of solvent (1:1 DEF:EtOH) were loaded and sealed with a
PTFE cap. The mixture was irradiated in a microwave reactor (CEM
Discover) for 30 min at 120 °C. After 30 min, the solution was cooled,
and the resulting solid was collected via filtration and washed with hot
DEF. The solid was dried under vacuum to yield 57.5 mg (95%) of
product as a white powder. Anal. Calcd for C31.5H46.5Mg2N3.5O10.5: C,
54.77; H, 6.78; N, 7.10. Found: C, 54.85; H, 7.07; N, 6.86. IR (KBr):
1661 (vs), 1612 (s), 1570 (s), 1468 (vs), 1419 (s), 1298 (s), 1242 (s),
1149 (m), 1111 (m), 1047 (w), 945 (w), 885 (m), 842 (m), 825 (m),
725 (w), 692 (m), 660 (w), 590 (m), 501 (w), 447 (w) cm−1. Heating
at 420 °C for 65 min in vacuo yielded the fully activated adsorbent
Mg2(dobpdc) (2).
diffractometer using Cu Kα radiation (λ = 1.5406 Å). The unit cell
dimensions of DEF-2 and mmen-2 were determined by performing a
full-pattern decomposition using the Le Bail method, as implemented
in TOPAS-Academic.35 Owing to the isomorphism with Zn2(dobpdc),
the trigonal space group P3221 was used for the refinements. Crystal
data for DEF-2: a = 21.761(2) Å, c = 6.9721(7) Å, V = 2859.1(5) Å3
(Rwp = 0.093, Rp = 0.067). Crystal data for mmen-2: a = 21.500(2) Å, c
= 6.8275(9) Å, V = 2733.2(6) Å3 (Rwp = 0.042, Rp = 0.033).
Gas Adsorption Measurements. Gas adsorption data for
pressures in the range 0−1.1 bar were obtained by volumetric
methods using a Micromeritics ASAP2020 instrument. All gases were
99.998% purity or higher. Isotherms at 77 K were measured in liquid
nitrogen baths. Isotherms at 25, 35, 45, 50, and 75 °C were measured
using liquid circulators to maintain a constant temperature. Isotherms
at 100 and 120 °C were measured using a heated sand bath controlled
by a programmable temperature controller. BET surface areas were
calculated from N2 adsorption at 77 K. DFT pore size distributions
and pore sizes were calculated from N2 adsorption at 77 K with the
Micromeritics DFT Plus Models Kit (Ver. 2.02) software suite with
cylinder pore geometries for an oxide surface. The compound mmen-2
was regenerated at 100 °C under dynamic vacuum for 4 h after
measurement of each isotherm.
Isosteric Heats of Adsorption Calculations. A dual-site
Langmuir−Freundlich equation (eq 1) was employed to model the
CO2 adsorption at 25, 50, and 75 °C for mmen-2 in the region before
the step in the isotherms.
qsat,AbApα
qsat,BbBpα
A
B
q =
+
1 + bApα
1 + bBpα
A
B
(1)
Here, q is the amount of CO2 adsorbed (mmol/g), p is the pressure
(bar), qsat is the saturation capacity (mmol/g), b is the Langmuir−
Freundlich parameter (bar−α), and α is the Langmuir−Freundlich
exponent (dimensionless) for two adsorption sites A and B. In order
to model the CO2 adsorption in the region after the step, a modified
Langmuir−Freundlich equation (eq 2) was employed.
Mg2(dobpdc)(mmen)1.6(H2O)0.4 (mmen-Mg2(dobpdc) or mmen-
2). A sample of fully activated 2 (77 mg, 0.24 mmol) was immersed in
anhydrous hexane, and 20 equiv of N,N′-dimethylethylenediamine
(mmen, 0.53 mL, 4.8 mmol) was added. The suspension was stirred
for one day, filtered, and rinsed copiously with hexanes. The solid was
then evacuated of residual solvents at 100 °C for 24 h to afford 87 mg
(77%) of product as a gray-white powder. Anal. Calcd for
C20.4H26Mg2N3.2O9.6: C, 52.46; H, 5.62; N, 9.60. Found: C, 52.15,
H, 5.41; N, 9.52. IR (ATR, neat): 3320 (w), 2952 (w), 2910 (w), 2862
(w), 2806 (w), 1616 (s), 1575 (s), 1538 (w), 1468 (vs), 1421 (vs),
1295 (m), 1244 (s), 1152 (m), 1104 (m), 1053 (m), 1000 (w), 887
(m), 844 (m), 828 (m), 727 (w), 692 (s), 618 (m), 589 (s) cm−1.
X-ray Structure Determination. A crystal of DEF-1 was
mounted on a cryoloop under a cooling stream of dinitrogen.
Diffraction data were collected with synchrotron radiation using a 6B
MX-I ADSC Quantum-210 detector with a silicon (111) double-
crystal monochromator at the Pohang Accelerator Laboratory. The
ADSC Quantum-210 ADX program (Ver. 1.92) was used for data
collection and HKL2000 (Ver. 0.98.699) was used for cell refinement,
data reduction, and absorption corrections. The structure was solved
by direct methods and refined by full-matrix least-squares analysis
using anisotropic thermal parameters for non-hydrogen atoms with the
SHELXTL program.33 The C2 and C5 atoms were isotropically
refined due to poor thermal behavior. Guest molecules in the pores
were highly disordered and unable to be modeled. To account for this
electron density, the program SQUEEZE34 was employed. All
hydrogen atoms were calculated at idealized positions and refined
using a riding model. Crystal data for DEF-1: empirical formula =
C12H14NO4Zn, Mr = 301.61, T = 100(2) K, space group = P3221, a =
21.698(3) Å, c = 6.8690(14) Å, α = 90°, β = 90°, γ = 120°, V =
2800.7(8) Å3, Z = 6, Dcalc = 1.073 g/cm3, μ = 1.319 mm−1, 13927
reflections collected, 2941 unique (Rint = 0.0418), R1 = 0.0466, wR2 =
0.1204 (I > 2σ(I)).
qsat,AbA(p − p*)α
qsat,BbB(p − p*)α
A
B
q =
+
1 + bA(p − p*)αA
1 + bB(p − p*)α
B
qsat,CbC(p − p*)α
C
+
1 + bC(p − p*)α
C
(2)
Here, adsorption is considered at three sites, A, B, and C, and the extra
parameter p* is used to account for the pressure at which the step in
the isotherm occurs and the strongest adsorption sites are first
populated. After carefully refining the parameters in eqs 1 and 2,
excellent agreement was achieved between the experimental isotherm
data and the corresponding isotherm fits (see Figures S10−S13).
Using the appropriate isotherm fits, Mathematica software was used to
solve for the exact pressures, p, corresponding to constant amounts of
CO2 adsorbed, q, at 25, 50, and 75 °C. The Clausius−Clapeyron
equation (eq 3) was then used to calculate the isosteric heats of
adsorption, Qst, by determining the slope of the best-fit line for ln p
versus 1/T at each loading.
⎛Q st
⎞
⎛
⎜
⎞
⎟
1
T
(ln p) =
+ C
⎜
⎟
⎠
q
⎝
⎠
⎝
R
(3)
As indicated by the residual sum of square values, R2, of close to 1
(see Figure S14), the isotherm data were consistent with the
Clausius−Clapeyron equation across the entire loading range
considered, even with the changes in the location of the step in the
isotherms.
The isosteric heats of adsorption for CO2 in the unmodified
Mg2(dobpdc) were determined by fitting the adsorption isotherms at
25, 35, and 45 °C with a dual-site Langmuir−Freundlich equation (eq
1). Each temperature was fit independently, and the Clausius−
Clapeyron equation was used to determine Qst as a function of loading.
All fit parameters for Mg2(dobpdc) and mmen-Mg2(dobpdc) are
specified in Tables S2−S4.
Powder X-ray Diffraction. Powder X-ray diffraction data were
collected with either a Rigaku Ultima III or a Bruker D8 Advance
7058
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
Figure 1. Synthesis of mmen-Mg2(dobpdc) (mmen = N,N′-dimethylethylenediamine; dobpdc4− = 4,4′-dioxido-3,3′-biphenyldicarboxylate). From the
microwave reaction of MgBr2·6H2O and H4dobpdc (left), Mg2(dobpdc) (2) is obtained following evacuation of the as-synthesized solid at high
temperatures (middle). The framework structure depicted is that obtained from single crystal X-ray analysis of the isostructural zinc compound DEF-
1. Green, red, and gray spheres represent Mg, O, and C atoms respectively; H atoms are omitted for clarity. Addition of an excess of mmen to the
evacuated framework yields the amine-appended CO2 adsorbent Mg2(dobpdc)(mmen)1.6(H2O)0.4 (right).
Model CO2 Isotherms. Isotherms for CO2 uptake in mmen-2 at
other temperatures were predicted from the experimental isotherm at
75 °C via application of a Clausius−Clapeyron relation (eq 4).
Differential Scanning Calorimetry. Thermal analysis was
performed on a TA Instruments Q200 differential scanning
calorimeter equipped with a refrigerated cooling system (RCS40).
Through the sample cell, 15% CO2 in N2 or N2 was flowed over T-
Zero aluminum pans that were not hermetically sealed. An empty,
aluminum T-Zero pan provided the reference sample for thermal
analysis. The calorimeter was calibrated with the TA Instruments
software package; the melting point of indium (156.60 °C) was
utilized for the temperature calibration. Sample and pan masses were
determined after activation at 150 °C for 60 min on a TA Instruments
TGA Q5000 via the readout from the internal balance under flowing
N2. Identical aliquots of mmen-2 were utilized for TGA and DSC
measurements to approximate heats of adsorption. Integrated heats
were calculated with TA Instruments Universal Analysis software suite.
Other Physical Measurements. Elemental analyses for C, H, and
N were performed at the Elemental Analysis Service Center of Sogang
University or the Microanalytical Laboratory of the University of
California, Berkeley. 1H spectra were obtained using a 300 MHz
Varian instrument. Infrared spectra were obtained from KBr pellets
with a Bomen MB-104 spectrometer or on a Perkin-Elmer Spectrum
400 FTIR spectrometer equipped with an attenuated total reflectance
(ATR) accessory. For diffuse reflectance infrared Fourier transform
spectroscopy (DRIFTS) spectra, the Perkin-Elmer spectrometer was
equipped a Harrick Praying Mantis Diffuse Reflectance accessory and a
temperature-controlled high-pressure gas cell with Swagelok valves
connecting 5% CO2 in He (Praxair certified standard HE CD5C-K)
and an oil-free vacuum.
⎛
⎜
⎜
⎝
⎞
⎟
⎟
⎠
RTT
1
2
Q
+
ln p
(
2
)
)
st
(
(
1
T − T
2
1
(ln p2 ) =
RTT
1
⎜
⎜
⎟
⎟
)
T − T
2
1
(4)
Calculated pressures (p2) at constant loadings (q) were determined
assuming an isosteric heat of adsorption (Qst) of 71 kJ/mol for all
loadings. Here, p1 is a pressure experimentally measured at 75 °C, R is
the universal gas constant, T1 is 75 °C, and T2 is the temperature for
which a model isotherm is desired.
CO2 Selectivity Calculations. Here, the adsorption capacities of
component n (qn) are defined to be molar excess adsorption capacities
determined experimentally without correction for absolute adsorption,
and pn is defined to be the pressure of component n as experimentally
measured. Selectivity (S) is defined according to eq 5, while purity is
defined according to eq 6.
qCO /q2
2
S =
p
/p2
CO2
(5)
q
CO2
purity =
× 100%
q
+ q2
CO2
(6)
3. RESULTS AND DISCUSSION
Thermogravimetric Analysis and Gas Cycling Measure-
ments. Thermogravimetric analyses (TGA) were carried out at
ramp rates between 5 and 10 °C/min under a nitrogen flow with a TA
Instruments TGA Q5000 (Ver. 3.1 Build 246) or a Scinco TGA N-
1000.
Synthesis, Structure, and Activation. Reaction of
H4dobpdc with ZnBr2·2H2O or MgBr2·6H2O in 1:1
DEF:EtOH afforded DEF-1 and DEF-2, respectively (see
Figure 1). The coordination environment of the divalent metal
cations within 1 and 2 are analogous to those in the M2(dobdc)
series.29 In the crystal structure of DEF-1, four different
dobpdc4− ligands and one DEF molecule are bonded to each
Zn2+ ion in a distorted octahedral geometry. There are three
unique O donor types from the dobpdc4− ligand: bridging (μ2)
aryloxide O atoms (O1), bridging (μ2) carboxylate O atoms
(O2), and nonbridging carboxylate O atoms (O3). The
equatorial plane of each Zn2+ is composed of two trans-
disposed O1 ligands from different linkers, one O3 donor atom,
and one O2 donor atom. An O2 donor atom occupies one axial
coordination site, while the other axial site is occupied by an O
donor atom from DEF, the reaction solvent. This coordination
Carbon dioxide cycling experiments were performed on the
aforementioned TA Instruments analyzer using 15% CO2 in N2
(Praxair NI-CD15C-K), 390 ppm CO2 in air (Praxair AI-CD-390C-
K; 390 ppm CO2, 21% O2, balance N2), CO2 (Praxair 99.998%), and
N2 (Praxair, 99.9%). A flow rate of 25 mL/min was employed for all
gases. Prior to cycling, the sample was activated by heating at 150 °C
for 1 h. For Figure 6, sample mass was normalized to be 0% at the
adsorption temperature (25 °C for 390 ppm CO2; 40 °C for 15%
CO2) under flowing N2. For Figure 7, sample mass was normalized to
be 0% at 150 °C under flowing 100% CO2. Masses were uncorrected
for buoyancy effects. In Figure 6, the difference between the quantity
of N2 adsorbed at the tare temperature and the regeneration
temperature likely accounts for the negative apparent mass of
mmen-2 at high temperatures.
7059
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
mode results in the formation of helical chains of Zn2+ atoms
running along the c axis of the crystal. The resulting framework
consists of a honeycomb lattice of hexagonal, one-dimensional
channels approximately 18.4 Å in width. Bound DEF molecules
occupy the Zn2+ coordination sites along the corners of
hexagonal channel walls. As shown in Figure S1, powder X-ray
diffraction (PXRD) data indicate DEF-2 to be isostructural with
DEF-1.
Heating DEF-2 at 420 °C for 65 min under dynamic vacuum,
removed the DEF molecules bound to the metal atoms,
completely activating the material and generating open Mg2+
coordination sites. Such extreme thermal treatment was
necessary because soaking in methanol at 100 °C for 20 h
did not lead to exchange of the bound DEF molecules. The
porosity of activated 2 was confirmed via N2 adsorption at 77 K
(see Figure S4), resulting in a BET surface area of 3270 m2/g.
Note that, in line with the expanded structure, this is
significantly greater than the BET surface area of 1495 m2/g
reported for Mg2(dobdc).36
The synthesis and structure of mmen-2 is depicted
schematically at the right of Figure 1. An activated sample of
2 was suspended in hexanes and an excess of mmen was added.
As shown by powder X-ray diffraction (see Figure S3),
framework crystallinity was not significantly affected by
activation or subsequent amine functionalization. A much
reduced BET surface area of 70 m2/g was calculated for mmen-
2, while DFT pore size distributions indicated a reduction in
average pore size (see Figures S5−S7).
Upon exposure of DEF-2 to atmospheric air, the white
powder turns blue and a loss of crystallinity occurs, as observed
via powder X-ray diffraction measurements. Amine functional-
ization, however, appears to enhance framework stability,
because no similar degradation was observed for mmen-2 upon
exposure to air for one week.
CO2 Adsorption Isotherms. The CO2 adsorption data
obtained for 2 are depicted in Figure S8. Isosteric heats of
adsorption were calculated to approach −44 kJ/mol at low
coverage, as shown in Figure S9. This value is similar to those
previously reported for the analogous Mg2(dobdc) frame-
work.37 The adsorption capacity of 2 at 25 °C is 4.85 mmol/g
(17.6 wt %) and 6.42 mmol/g (22.0 wt %) at 0.15 and 1 bar,
respectively. The gravimetric capacity of 2 for CO2 at 0.15 bar
exceeds the capacity of all metal−organic frameworks except for
Mg2(dobdc),10c which adsorbs 6.1 mmol/g (21.2 wt %) at 25
°C and 0.15 bar.36
Figure 2. Adsorption of CO2 in mmen-2 at 25 °C (blue squares), 50
°C (green triangles), and 75 °C (red circles). Inset: The isotherms at
very low pressures exhibit a step that shifts to higher pressures at
higher temperatures. The dashed, vertical line marks the current partial
pressure of CO2 in air (390 ppm).
surface area accessible to CO2, and the pore size distribution
calculated from N2 adsorption at cryogenic temperatures likely
does not accurately represent the true pore size distribution
within mmen-2. On a per mass basis, the amine-functionalized
framework adsorbed less CO2 than 2 at pressures higher than
20 mbar. The large density difference between the two
frameworks is primarily responsible for the lower gravimetric
capacity of mmen-2. Crystallographic densities of 0.58 and 0.86
g/cm3 were calculated for 2 and mmen-2, respectively. At 0.15
bar, 2 and mmen-2 adsorb 2.8 and 2.7 mmol/cm3, respectively.
Based upon the calculated number of dangling amine groups
in mmen-2, a loading of 3.4 mmol/g would correspond to one
CO2 per amine, yet uptake of only ca. 2.2 mmol/g was
observed. Here, pore blockages, hydrogen bonded amines, or
cooperative binding mechanisms between two amines and one
CO2 may be limiting the accessible stoichiometry of mmen-2.
Thus, significant additional capacity improvements might be
realized in the material if conditions can be identified for
appending one mmen per magnesium and binding one CO2
molecule per dangling amine.
Isosteric heat of adsorption calculations were hindered by the
presence of a prominent step in the isotherms at low pressures
and convex to the pressure axis. Generally, continuous
mathematical functions are used to model experimental
isotherms, which then become the input parameters for the
Clausius−Clapeyron relation. Since we were unable to
mathematically model the CO2 isotherms of mmen-2 with
continuous equations over the entire pressure range, each
isotherm was modeled with two Langmuir−Freundlich
equations. Data sets corresponding to the adsorption regions
before and after the steps were compiled and then modeled
individually.
The alkylamine-functionalized metal−organic framework
mmen-2 displayed an extremely high affinity for CO2 at
extraordinarily low pressures. The CO2 adsorption isotherms
obtained at 25, 50, and 75 °C are presented in Figure 2. At 25
°C and 0.39 mbar, near the current partial pressure of CO2 in
Earth’s atmosphere, the compound adsorbed 2.0 mmol/g (8.1
wt %), which is 15 times the capacity of 2. At the much higher
pressure of 5 mbar, the median partial pressure of CO2 within
the International Space Station, the framework adsorbed 2.6
mmol/g (10.3 wt %). For comparison, zeolite 5A, which is
currently used aboard the station to adsorb CO2, adsorbs 0.85
mmol/g (3.6 wt %, crystallographic volumetric capacity 1.3
mmol/cm3) at 5 mbar.12,38
As shown in Figure 3, isosteric heats of adsorption for
mmen-2 were calculated from the 25, 50, and 75 °C isotherm
models. At low loadings, heats significantly lower than those
expected for chemical adsorption of CO2 onto an amine were
calculated. However, calculated heats quickly approached and
maintained a value of −71 kJ/mol, which likely corresponds to
the chemical adsorption of CO2 onto the free amine of mmen.
Here, a carbamate with a weak C−N bond is probably formed
At 25 °C, the CO2 adsorption in mmen-2 reaches 3.13
mmol/g (12.1 wt %) at 0.15 bar and 3.86 mmol/g (14.5 wt %)
at 1 bar. Remarkably, its CO2 uptake at 1 bar and 25 °C exceeds
the amount of N2 adsorbed at 77 K. Thus, the low surface area
measured at 77 K does not appear to accurately reflect the
7060
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
mmen-2 is unlikely due to a lack of surface hydroxyl groups,
and the broad NH stretches expected for ammonium cations
are not definitively resolvable from the DRIFTS difference
spectrum. Furthermore, the slow reversibility of mmen-2 at
room temperature (see Figure S15) appears to preclude the
formation of ammonium carbamates, which have been reported
to desorb CO2 from primary amines readily at room
temperature.39 Additional experiments are necessary to
ascertain whether adjacent functional groups are necessary for
CO2 adsorption within such alkylamine-functionalized metal−
organic frameworks.
The step in each isotherm marks the pressure at which CO2
adsorption becomes dominated by chemisorption. Interest-
ingly, the step moves to significantly higher pressures as the
adsorption temperature increases. As shown in Figure S16, the
location of the step at 75 °C was reproduced four times with
little variation. Thus, the step cannot be attributed to slow
kinetics, but appears to be an inherent characteristic of mmen-2
CO2 adsorption isotherms. All batches of mmen-2 have
exhibited steps at approximately the same pressures, even
when functionalized with different amounts of amines. Figure
S17 presents the 75 °C CO2 adsorption isotherm of
Figure 3. Isosteric heats of CO2 adsorption onto mmen-2, as
calculated using the Clausius−Clapeyron relation. The calculated
values indicate that chemical adsorption of CO2 onto the amines did
not occur at very low coverage.
through interaction the lone pair of the free amine of mmen
and the electrophilic carbon of CO2.
Mg2(dobpdc)(mmen)1.75(H2O)0.25
.
In situ diffuse reflectance infrared Fourier transform spec-
troscopy (DRIFTS) was employed to probe the chemical
nature of adsorbed CO2. At 25 °C, 1 bar of 5% CO2 in He was
introduced into an airtight gas cell containing a sample of
activated mmen-2. The difference spectrum between mmen-2
under a 1 bar atmosphere and the activated framework under
vacuum is shown in Figure 4. The prominent loss peak at 3316
The location of the step is modeled well by a simple
Clausius−Clapeyron relation (see eq 4), which predicts how
isotherms move as a function of temperature. In Figure 5,
Figure 5. Adsorption of CO2 in mmen-2 at 25, 50, 75, 100, and 120
°C, represented as blue, green, red, purple, and orange squares,
respectively. The locations of the step for the 25, 50, 100, and 120 °C
isotherms were modeled with a Clausius−Clapeyron derived equation
constructed from the 75 °C data (shown as black lines for each
temperature). The heat of adsorption was assumed to be a constant
−71 kJ/mol. The onset pressure of each isotherm step was predicted
well by the simple model.
Figure 4. DRIFTS spectra of CO2 adsorption in mmen-2. The loss
band at 3316 cm−1 in the difference spectrum (black line) between
evacuated mmen-2 (blue line) and mmen-2 under a 5% CO2 in He
atmosphere (green line) indicates CO2 adsorption occurs via a
chemical adsorption mechanism. The blue line is offset by −0.5
absorbance units for visual clarity.
experimental isotherms at 25, 50, 75, 100, and 120 °C are
shown as blue, green, red, purple, and orange squares,
respectively. The black lines through the 25, 50, 100, and 120
°C data are predicted isotherms for those temperatures
modeled from the experimental data collected at 75 °C by
assuming that the heat of adsorption is −71 kJ/mol for all CO2
loadings. The black lines are not continuous functions, but
connect the calculated points listed in Table S5. Utilizing the
approximate heat of adsorption calculated from the 25, 50, and
75 °C isotherms, the model was able to predict the general
shapes of the 100 and 120 °C isotherms, as well as the
cm−1, assigned to the NH stretch of free mmen-2, is indicative
of chemical adsorption of CO2 onto amines.
Recent work on alkylamine-grafted silica surfaces have
suggested that chemical adsorption of CO2 onto alkylamines
is not possible without neighboring amines or surface hydroxyl
groups to stabilize the resulting carbamates; ammonium
carbamates or surface-bonded carbamates are formed,
respectively.39 The formation of surface-bonded carbamates in
7061
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
Table 1. Approximate CO2 Selectivities of mmen-2 for Selected Gas Mixtures at 25 °C
pressure (mbar)
uptake (mmol/g)
mixture
comp. 2
CO2
comp. 2
CO2
comp. 2
molar selectivity
est. purity (%)
air
N2
O2
N2
0.4
0.4
800
200
750
2.05
2.05
3.13
0.083
0.038
0.079
49 000
27 000
200
96
98
98
air
flue gas
150
pressures at which the steps arise. Thus, it seems that
thermodynamics alone account for the movement of the step
to higher pressures at higher temperatures.
calculated in Table S6 indicate that the while the purity of
CO2 decreases within increasing void space, any adsorbent bed
with less than 80% void space should be capable of achieving
CO2 purity in excess of 90%. From these calculations, we
believe mmen-2 to be a promising, high-capacity adsorbent for
removing high-purity CO2 from dry gas mixtures containing N2
and O2.
CO2 Capture from Air. To evaluate the performance of
mmen-2 as a regenerable adsorbent, thermogravimetric analysis
(TGA) was utilized to monitor sample mass under dynamic
environments. The top panel of Figure 6 plots changes in
The existence of the step, however, is unexpected in a
strongly adsorbing material with large pores. The amines, as the
strongest adsorption sites, should preferentially adsorb CO2
before other, weaker adsorption sites adsorb significant
quantities of CO2. To explain the nonclassical adsorption
behavior of mmen-2, we presently hypothesize that adsorption
of CO2 onto mmen is disfavored at low adsorptive
concentrations (the density of gas phase CO2 in the pores)
because of the large positive entropy associated with
reorganization of the amines, as required to form a chemical
bond with CO2. Before the chemical potential necessary for
amine reorganization afforded to the system by the adsorptive
is achieved, CO2 adsorption is dominated by non-amine or
weak amine-CO2 adsorption sites. The minimum chemical
potential required for amine reorganization increases as thermal
motion increases, in line with the shift of the step to higher
pressures at higher temperatures. Additional experiments and
modeling will be required to test this hypothesis and to
understand the relationship between pore filling and adsorption
in mmen-2. However, the shift of the step to higher pressures at
higher temperatures could afford the opportunity for unique
regeneration conditions, whereby a weak vacuum could nearly
fully regenerate the material at moderate temperatures.
CO2 Selectivity. Adsorption isotherms for N2 and O2 in
mmen-2 at 25 °C are shown in Figure S18. The uptake
capacities for these gases relative to CO2 suggest that it would
be a highly selective adsorbent. Since we were unable to
mathematically model the CO2 isotherms of mmen-2 with a
meaningful, single equation40 over the entire pressure range of
interest, ideal adsorbed solution theory (IAST)41 selectivities
could not be calculated. Table 1 therefore instead presents the
approximate molar selectivities of mmen-2 for relevant gas
mixtures, as calculated according to eq 5 above using the excess
adsorption capacities. This method for determining selectivities
was recently shown to be less accurate than IAST, in part
because it generally underestimates the values.37
The selectivity of mmen-2 for CO2 over N2 in air is thus
estimated to be at least 49 000. However, the purity of gas
adsorbed is perhaps a more physically meaningful value than
selectivity, and is an important criterion for evaluating
adsorbents if captured CO2 is to be transported or sequestered.
The purity of CO2 adsorbed onto mmen-2, calculated
according to eq 6, would likely be more than 96% or 98%
for capture from air or flue gas, respectively.
Figure 6. Adsorption−desorption cycling for mmen-2, demonstrating
reversible uptake of 1.05 mmol CO2/g from dry air after 1 h (upper)
and 2.52 mmol CO2/g from dry flue gas after 15 min (lower). Changes
in sample mass are plotted as green circles and sample temperatures
are plotted as blue lines; CO2 was introduced into the sample chamber
at the points marked by black squares. Samples were regenerated
under flowing N2.
sample mass (normalized to the mass of the framework under
N2 at 25 °C) while simulated air containing 390 ppm CO2 was
flowed over the sample. Despite the very low concentration of
CO2, a 4.6%mc (%mc = percent mass change;42 1.05 mmol/g;
4.4 wt %) was realized after 60 min. The adsorbent was then
regenerated under flowing N2 at 150 °C for 30 min and the
cycle repeated 10 times with no apparent loss of capacity.
To the best of our knowledge, no studies evaluating the
efficacy of air capture within metal−organic frameworks have
yet been reported. The equilibrium capacity (2.0 mmol/g, 1.72
mmol/cm3) of mmen-2 is similar to the capacities of the very
best amine-grafted silica and alumina adsorbents reported to
While the purity of gas adsorbed within individual mmen-2
crystallites is very high, as shown in Table 1, the purity of gas
adsorbed within a bed of adsorbent will be affected by the
composition of gas phase molecules that fill the pore and
intercrystalline spaces. For carbon capture applications, the high
concentration of N2 in these pore volumes will negatively affect
the selectivity and purity values of real adsorbent beds.
However, the approximate selectivity and purity values
7062
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
date. However, the kinetics of adsorption appear to be
significantly faster in mmen-2 than for amines deposited via
evaporation or polymerization methodologies. For example,
while the pseudo-equilibrium capacity of an outstanding
poly(ethyleneimine) impregnated silica gel was reported to
be ca. 2.4 mmol/g, it took nearly 200 min for the silica based
adsorbent to realize 4.6%mc, the capacity of mmen-2 for CO2
after only 60 min.23d The easily accessed amines within mmen-
2 appear to enhance adsorption rates greatly, enabling rapid
adsorption−desorption cycles to be utilized.
increased the amount of CO2 adsorbed (point D). The working
capacity of mmen-2 is the difference between the amounts
adsorbed at points A and C. The mass of the material at 150 °C
under 100% CO2 was normalized to be 0% mass in Figure 7.
The relationship between points A and C and isotherms is
illustrated in Figure S19.
Differential Scanning Calorimetry. To approximate the
adsorbent regeneration energy, heats of adsorption and
desorption were quantified for mmen-2 via differential scanning
calorimetry (see Figure 8).44 When 15% CO2 in N2 at 40 °C
CO2 Capture from Dry Flue Gas. The capabilities of
mmen-2 as an adsorbent for removing CO2 from the flue gas of
coal-fired power stations were also investigated. The bottom
panel of Figure 6 presents the dynamic cycling behavior of
mmen-2 under the relevant, dry conditions: 15% CO2 in N2
flowing over the solid at 40 °C. After adsorbing CO2 for 15
min, the sample was heated at 120 °C for 15 min under N2. A
capacity of 11.1%mc (2.52 mmol/g, 9.9 wt %) relative to the
sample mass of mmen-2 under N2 at 40 °C was realized. After
50 cycles, only a 0.2%mc capacity loss was observed. Longer
adsorption and desorption times did not significantly improve
the cycling capacity of the material, nor did higher desorption
temperatures. Note that the apparent capacity of mmen-2
greatly exceeds the ca. 2 wt % working capacity43 of aqueous
monoethanolamine (MEA) scrubbers, which would likely swing
between the same adsorption and desorption temperatures.
If captured CO2 is to be sequestered, high-purity CO2 is
essential. To desorb the ca. 98% pure CO2 adsorbed onto
mmen-2, a N2 or air purge cannot be utilized to strip the
adsorbent bed. Hence, to approximate the working capacity of
mmen-2 using a temperature swing without a N2 purge, a pure
CO2 purge was utilized instead. A 7.8%mc (1.8 mmol/g, 7.2 wt
%) was realized when 15% CO2 in N2 at 40 °C was desorbed
with 100% CO2 at 150 °C. In Figure 7, 15% CO2 was
Figure 8. Heat flows from mmen-2, as determined via differential
scanning calorimetry. First, 15% CO2 in N2 was adsorbed at 40 °C
(event A). After 15 min, mmen-2 was heated to 120 °C under N2
(event B). Lastly, the regenerated material was cooled under N2 to 40
°C (event C). The integrated heats calculated for each event are given
in the legend.
was introduced over a sample of mmen-2 activated at 150 °C
under N2, an exothermic heat flow of −153 J/g was observed
(event A). Note that, to enable rapid gas exchange, the pan was
not hermetically sealed and the true value of heats evolved are
likely slightly larger than those registered by the instrument due
to heat loss to the environment. When this heat flow was
correlated to the 2.5 mmol/g capacity calculated via TGA
adsorption under similar conditions, an approximate, average
heat of adsorption of −61 kJ/mol CO2 was calculated.
Following a 15 min adsorption period, 100% N2 was
introduced into the cell and the sample was heated to 120 °C
(event B); 260 J/g were required to heat the sample and
maintain the regeneration temperature for 15 min. Lastly, −107
J/g were evolved as the material was cooled under a N2
atmosphere (event C) from 120 to 40 °C. Based upon this
laboratory-scale experiment, approximately 2.34 MJ of energy
would be required to regenerate 1 kg of CO2 adsorbed onto
mmen-2. This value compares very favorably with the 3.6−4.5
MJ of energy necessary to release 1 kg of CO2 from an MEA
scrubber.5,43
However, the aforementioned regeneration energies were
measured under a N2 purge, a laboratory-scale analogue for a
temperature-vacuum swing regeneration cycle. While a direct
measurement of regeneration energy with a 100% CO2 purge
was not possible, calculations shown in the SI indicate that the
regeneration energy of mmen-2 under CO2 will be higher
primarily due to the reduced working capacity of the adsorbent.
The regeneration energy of mmen-2 under CO2 is still lower
than the regeneration energy of the most highly optimized
MEA solutions, and it is likely that optimization of adsorption
and desorption temperatures will further lower the heat of
regeneration of mmen-2.
Figure 7. A temperature swing processes, as simulated for CO2 in
mmen-2 using TGA and desorption with a CO2 purge. A 1.8 mmol/g
capacity difference was calculated from the quantity of CO2 adsorbed
at 150 °C under a 100% CO2 atmosphere (point A) and the quantity
of CO2 adsorbed at 40 °C under flowing 15% CO2 in N2 (point C).
Changes in sample mass are plotted as green circles and sample
temperatures are plotted as blue lines; points B and D arise from
introducing 15% CO2 at 150 °C and 100% CO2 at 40 °C respectively.
introduced over the sample at 150 °C (point A). The change in
atmosphere from 100% to 15% CO2 partially activated the
framework, resulting in a negative effective mass (point B). As
the sample cooled to 40 °C under the 15% atmosphere, CO2
was adsorbed. After 15 min at 40 °C (point C), 100% CO2 was
introduced and the sample was heated at 10 °C/min. The
increase in CO2 concentration at low temperatures temporarily
7063
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
Approximate heat capacities (CP) were calculated for mmen-
2. Over the 80 °C temperature range, the average heat required
to regenerate CO2-loaded mmen-2 was 3.3 J/g·K. The
regenerated framework released an average of 1.3 J/g·K upon
cooling, a reasonable approximation of the heat capacity of
mmen-2. For comparison, the heat capacity of a 30 wt % MEA
solution without CO2 is ∼3.8 J/g·K over the same 80 °C
range.43 The lower regeneration energy of mmen-2 is thus
attributable to its lower heat capacity. In addition, the
significantly greater working capacity of mmen-2 means that
a smaller mass is required to adsorb the same amount of CO2.
The heat required to desorb CO2 from the amines is similar for
mmen-2 and MEA solutions.43
Basic Energy Sciences, under Award No. DE-SC0001015. We
also thank NSF for providing graduate fellowship support
(J.A.M.). We thank Prof. T. Xu for use of the differential
scanning calorimetry instrument.
REFERENCES
■
(1) National Oceanic and Atmospheric Administration. Trends in
Atmospheric Carbon Dioxide. http://www.esrl.noaa.gov/gmd/ccgg/
trends/ (accessed Nov 15, 2011).
(2) Carbon Dioxide Information Analysis Center. Recent Green-
house Gas Concentrations. http://cdiac.ornl.gov/pns/ current_ghg.
html (accessed Dec 16, 2011). DOI: 10.3334/CDIAC/atg.032.
(3) Climate Change 2007: The Physical Science Basis; Solomon, S.,
Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K. B., Tignor,
M., Miller, H. L., Eds.; The Fourth Assessment Report of the
Intergovernmental Panel on Climate Change; Cambridge Univ. Press:
New York, 2007.
4. CONCLUSIONS AND OUTLOOK
With the new metal−organic frameworks M2(dobpdc) (M =
Zn (1), Mg (2)), an expanded variant of the M2(dobdc)
(MOF-74) structure type has been achieved, leading to
enlarged, 18.4-Å-wide channels lined with open metal
coordination sites. Functionalization of 2 with mmen, afforded
a remarkable new CO2 adsorbent, mmen-Mg2(dobpdc). The
large capacity, high selectivity, and fast kinetics of this material
for adsorbing CO2 from dry gas mixtures with N2 and O2 make
it an attractive new adsorbent for applications in which zeolites
and inorganic bases are currently used, including the removal of
CO2 from air.
Future work will focus on evaluating the efficacy of
alkylamine-appended metal−organic frameworks such as
mmen-2 under humid conditions. Whether the presence of
water in gas mixtures will significantly affect the stability,
capacity, selectivity, or regeneration energy of mmen-2 is
presently unknown. In addition, the peculiar CO2 isotherm
shapes warrant further investigation, since a detailed under-
standing of the adsorption behavior within mmen-2 may enable
the synthesis of even better materials for CO2 capture. Finally,
efforts are also underway to improve the separation perform-
ance of the material through use of various other polyamines in
place of mmen.
(4) Chu, S. Science 2009, 325, 1599.
(5) DOE/NETL Carbon Dioxide Capture and Storage RD&D
Roadmap; National Energy Technology Laboratory, 2010.
(6) Granite, E. J.; Pennline, H. W. Ind. Eng. Chem. Res. 2002, 41,
5470.
(7) Rochelle, G. T. Science 2009, 325, 1642.
(8) Fout, T.; Murphy, J. T. DOE/NETL’s Carbon Capture R&D
Program for Existing Coal-Fired Power Plants; DOE/NETL 2009/1356;
National Energy Technology Laboratory, 2009.
(9) Carbon Dioxide Capture from Existing Coal Fired Power Stations;
Revision Nov 2007; DOE/NETL 401/110907; National Energy
Technology Laboratory, 2007.
(10) (a) D’Alessandro, D. M.; Smit, B.; Long, J. R. Angew. Chem., Int.
Ed. 2010, 49, 6058. (b) Wang, Q.; Luo, J.; Zhong, Z.; Borgna, A.
Energy Environ. Sci. 2011, 4, 42. (c) Sumida, K.; Rogow, D. L.; Mason,
J. A.; McDonald, T. M.; Bloch, E. D.; Herm, Z. R.; Bae, T.-H.; Long, J.
R. Chem. Rev. 2012, 112, 724.
(11) (a) Acharya, D. R.; Jain, R. Sep. Sci. Technol. 1995, 30, 3489.
(b) Rege, S. U.; Yang, R. T.; Buzanowski, M. A. Chem. Eng. Sci. 2000,
55, 4827. (c) Kumar, R.; Huggahalli, M.; Deng, S.; Andrecovich, M.
Adsorption 2003, 9, 243. (c) Kerry, F. G. Industrial Gas Handbook: Gas
Separation and Purification. CRC: Boca Raton, 2007.
(12) International Zeolite Association. Database of Zeolite Structures.
http://www.iza-structure.org/databases/ (Accessed Dec 30, 2011).
(13) (a) Carhart, H. W.; Thompson, J. K. Removal of Trace
Contaminants from Air; ACS Symposium Series 17; American
Chemical Society: Washington, DC, 1975; p 1. (b) Wieland, P. O.
Controlling Carbon Dioxide. Living Together in Space: The Design and
Operation of the Life Support Systems on the International Space Station;
NASA/TM-98-206956/VOL1; NASA Marshall Space Flight Center:
Marshall Space Flight Center, AL, 1998; Chapter 3.3.1, p 121.
(14) Law, J.; Watkins, S.; Alexander, D. In Flight Carbon Dioxide
Exposures and Related Symptoms: Association, Susceptibility, and
Operational Implications; NASA/TP-2010-216126; NASA: Hanover,
MD, 2010.
(15) Carbon Dioxide. NIOSH Pocket Guide to Chemical Hazards;
NIOSH Publication 2005-149; Department of Health and Human
Services, Centers for Disease Control and Prevention, National
Institute for Occupational Safety and Health: Cincinnati, OH, 2007.
(16) Ahuja, V.; Green, R. Int. J. Hydrogen Energy 1998, 23, 131.
(17) Goeppert, A.; Czaun, M.; May, R. B.; Prakash, G. K. S.; Olah, G.
A.; Narayanan, S. R. J. Am. Chem. Soc. 2011, 133, 20164.
(18) McLean, G. F.; Niet, T.; Prince-Richard, S.; Djilali, N. Int. J.
Hydrogen Energy 2002, 27, 507.
ASSOCIATED CONTENT
* Supporting Information
■
S
Powder X-ray diffraction patterns and fits, additional CO2, N2,
and O2 isotherms and fits, FTIR spectra, TGA traces, and
additional purity and regeneration energy calculations. This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
cshong@korea.ac.kr; jrlong@berkeley.edu
■
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENTS
■
The synthesis and characterization of compounds 1 and 2 was
funded by the National Research Foundation of Korea (KRF-
2009-220-C00021 and NRF20110018396) and PAL (2010-
2063-04), while the remainder of this research, including the
synthesis and characterization of mmen-2, was funded through
the Center for Gas Separations Relevant to Clean Energy
Technologies, an Energy Frontier Research Center funded by
the U.S. Department of Energy, Office of Science, Office of
(19) Keith, D. Science 2009, 325, 1654.
(20) (a) Jones, C. W. Annu. Rev. Chem. Biomol. Eng. 2011, 2, 31.
(b) Ranjan, M.; Herzog, H. J. Energy Procedia 2011, 4, 2869.
(21) American Physical Society. Direct Air Capture of CO2 with
Chemicals. http://www.aps.org/policy/reports/assessments/upload/
dac2011.pdf (accessed Nov. 20, 2011).
7064
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065

Journal of the American Chemical Society
Article
(22) (a) Satyapal, S.; Filburn, T.; Trela, J.; Strange, J. Energy Fuels
2001, 15, 250. (b) Schaldt, M. J.; Filburn, T. P.; Helble, J. J. Ind. Eng.
Chem. Res. 2007, 46, 1590.
(42) In this work, weight percent (wt %) is defined as
mass of adsorbate
mass of adsorbate + mass of adsorbent
wt% =
× 100%
(8)
(23) (a) Belmabkhout, Y.; Serna-Guerrero, R.; Sayari, A. Chem. Eng.
Sci. 2010, 65, 3695. (b) Choi, S.; Drese, J. H.; Eisenberger, P. M.;
Jones, C. W. Environ. Sci. Technol. 2011, 45, 2420. (c) Chaikittisilp, W.;
Lunn, J. D.; Shantz, D. F.; Jones, C. W. Chem.Eur. J. 2011, 17,
10556. (d) Choi, S.; Gray, M. L.; Jones, C. W. ChemSusChem 2011, 4,
628. (e) Stuckert, N. R.; Yang, R. T. Environ. Sci. Technol. 2011, 45,
10257. (f) Wurzbacker, J. A.; Gebald, C.; Steinfeld, A. Energy Environ.
Sci. 2011, 4, 3584.
The mass change registered by the TGA does not account for
adsorbate mass in the denominator and thus is not strictly wt %. To
clearly differentiate this difference, the abbreviation %mc (percent mass
change) is used instead to describe the experimentally observed mass
changes. As an additional note, because TGA only records net mass
change, the mass of N2 in the activated framework likely displaced by
CO2 is not registered as part of the %mc.
(24) Chaikittisilp, W.; Kim, H.-J.; Jones, C. W. Energy Fuels 2011, 25,
5528.
(43) Song, H.-J.; Lee, S.; Park, K.; Lee, J.; Spah, D. C.; Park, J.-W.;
Filburn, T. P. Ind. Eng. Chem. Res. 2008, 47, 9925.
(25) Gebald, C.; Wurzbacher, J. A.; Tingaut, P.; Zimmermann, T.;
Steinfeld, A. Environ. Sci. Technol. 2011, 45, 9101.
(44) Mu, B.; Walton, K. S. J. Phys. Chem. C 2011, 115, 22748.
(26) Wang, T.; Lackner, K. S.; Wright, A. Environ. Sci. Technol. 2011,
45, 6670.
(27) (a) Li, J.-R.; Kuppler, R. J.; Zhou, H.-C. Chem. Soc. Rev. 2009,
38, 1477. (b) Keskin, S.; van Heest, T. M.; Sholl, D. S. ChemSusChem
2010, 3, 879.
(28) (a) Chen, B.; Ockwig, N. W.; Millward, A. R.; Contreras, D. S.;
̆
Yaghi, O. M. Angew. Chem., Int. Ed. 2005, 44, 4745. (b) Dinca, M.;
Dailly, A.; Liu, Y.; Brown, C. M.; Neumann, D. A.; Long, J. R. J. Am.
Chem. Soc. 2006, 128, 16876. (c) Bae, Y.-S.; Farha, O. K.; Spokoyny, A.
M.; Mirkin, C. A.; Hupp, J. T.; Snurr, R. Q. Chem. Commun. 2008,
4135. (d) Llewellyn, P. L.; Bourrelly, S.; Serre, C.; Vimont, A; Daturi,
M; Hamon, L; Weireld, G. D.; Chang, J.-S.; Hong, D.-Y.; Hwang, Y.
́
K.; Jhung, S. H.; Ferey, G. Langmuir 2008, 24, 7245. (e) Yazaydin, A.
̈
O.; Benin, A. I.; Faheem, S. A.; Jakubczak, P.; Low, J. J.; Willis, R. R.;
Snurr, R. Q. Chem. Mater. 2009, 21, 1425.
(29) (a) Rosi, N. L.; Kim, J.; Eddaoudi, M.; Chen, B. L.; O’Keeffe,
M.; Yaghi, O. M. J. Am. Chem. Soc. 2005, 127, 1504. (b) Dietzel, P. D.
C.; Morita, Y.; Blom, R.; Fjellvag, H. Angew. Chem., Int. Ed. 2005, 44,
6354. (c) Dietzel, P. D. C.; Panella, B.; Hirscher, M.; Blom, R.;
Fjellvag, H. Chem. Commun. 2006, 959. (d) Zhou, W.; Wu, H.;
Yildrim, T. J. Am. Chem. Soc. 2008, 130, 15268. (e) Bloch, E. D.;
Murray, L. M.; Queen, W. L.; Chavan, S.; Maximoff, S. N.; Bigi, J. P.;
Krishna, R.; Peterson, V. K.; Grandjean, F.; Long, G. J.; Smit, B.;
Bordiga, S.; Brown, C. M.; Long, J. R. J. Am. Chem. Soc. 2011, 133,
14814.
(30) (a) Hwang, Y. K; Hong, D.-Y.; Chang, J.-S.; Jhung, S. H.; Seo,
́
Y.-K.; Kim, J.; Vimont, V.; Daturi, M.; Serre, C.; Ferey, G. Angew.
Chem., Int. Ed. 2008, 47, 4144. (b) Demessence, A.; D’Alessandro, D.
M.; Foo, M. L.; Long, J. R. J. Am. Chem. Soc. 2009, 131, 8784.
(c) Wiers, B. M.; Foo, M.-L.; Balsara, N. P.; Long, J. R. J. Am. Chem.
Soc. 2011, 133, 14522.
(31) McDonald, T. M.; D’Alessandro, D. M.; Krishna, R.; Long, J. R.
Chem. Sci. 2011, 2, 2022.
(32) Tomic, E. A. J. Appl. Polym. Sci. 1965, 9, 3745.
(33) Sheldrick, G. M. SHELXTL, Version 6.12; Bruker Analytical X-
Ray Systems, Inc.: Madison, WI, 2000.
(34) Van der Sluis, P.; Spek, A. L. Acta Crystallogr., Sect. A 1990, 46,
194.
(35) Coelho, A. TOPAS-Academic, Version 4.1; Coelho Software:
Brisbane, 2007.
(36) Caskey, S. R.; Wong-Foy, A. G.; Matzger, A. J. J. Am. Chem. Soc.
2008, 130, 10870.
(37) Mason, J. A.; Sumida, K.; Herm, Z. R.; Krishna, R.; Long, J. R.
Energy Environ. Sci. 2011, 4, 3030.
(38) Mulloth, L. M; Finn, J. E. Carbon Dioxide Adsorption on a 5A
Zeolite Design for CO2 Removal in Spacecraft Cabins; NASA/TM-1998-
208752; NASA Ames Research Center: Moffett Field, CA, 1998.
(39) Danon, A.; Stair, P. C.; Weitz, E. J. Phys. Chem. C 2011, 115,
11540.
(40) While virial-type equations were capable of describing the CO2
isotherm data for mmen-2 at each temperature independently,
unsuitably large numbers of parameters were required for good fits
to be obtained.
(41) Myers, A. L.; Prausnitz, J. M. AIChE J. 1965, 11, 121.
7065
dx.doi.org/10.1021/ja300034j | J. Am. Chem. Soc. 2012, 134, 7056−7065