R ES E A RC H

MOLECULAR SORPTION tions confirmed that H2O and CO2 occupy distinct
adsorption sites.
We obtained fluorinated MOFs NiAlF5(H2O)(pyr)2·
Hydrolytically stable fluorinated 2(H2O) (KAUST-8, AlFFIVE-1-Ni; pyr, pyrazine) and
its isomorph NiFeF5(H2O)(pyr)2·4(H2O) (KAUST-9,

metal-organic frameworks for FeFFIVE-1-Ni), which we expected to be iso-

structural to SIFSIX materials, based on pillaring
the M′(II)-pyrazine square-grid layers with [SiF6]2–
energy-efficient dehydration anions (21–23) but replacing the original [SiF6]2–
building block with a suitable inorganic pillar: an
Amandine Cadiau,1* Youssef Belmabkhout,1* Karim Adil,1 Prashant M. Bhatt,1
octahedrally coordinated metal center that should
accommodate five fluoride anions (F–) and a H2O
Renjith S. Pillai,1 Aleksander Shkurenko,1 Charlotte Martineau-Corcos,2,3
molecule in its coordination sphere. We selected
Guillaume Maurin,4 Mohamed Eddaoudi1†

Downloaded from https://www.science.org at King Abdullah University of Science and Technology on November 08, 2022
trivalent metals M, Al(III) and Fe(III), for the in-
organic pillar [MF5(H2O)]2–, as these are observed
Natural gas must be dehydrated before it can be transported and used, but conventional drying
to adopt an octahedral fluorinated environment
agents such as activated alumina or inorganic molecular sieves require an energy-intensive
(24–26).
desiccant-regeneration step. We report a hydrolytically stable fluorinated metal-organic
Although AlFFIVE-1-Ni and FeFIVE-1-Ni are
framework, AlFFIVE-1-Ni (KAUST-8), with a periodic array of open metal coordination sites
isomorphic to the SIFSIX materials (21–23),
and fluorine moieties within the contracted square-shaped one-dimensional channel. This
the explicit replacement of Si(IV) by Al(III) or
material selectively removed water vapor from gas streams containing CO2, N2, CH4, and higher
Fe(III) is not synthetically straightforward be-
hydrocarbons typical of natural gas, as well as selectively removed both H2O and CO2 in
cause no direct source of MF5(H2O) is available.
N2-containing streams. The complete desorption of the adsorbed water molecules contained
Consequently, a hydrothermal synthesis that per-
by the AlFFIVE-1-Ni sorbent requires relatively moderate temperature (~105°C) and about
mitted the in situ formation of the requisite in-
half the energy input for commonly used desiccants.
organic pillar, [MF5(H2O)]2–, was performed using

H
pyrazine, Ni(NO3)2·6H2O, and Al(NO3)3·9H2O [or
ydrolytically stable adsorbent materials porous metal-organic frameworks (MOFs) (4–8) Fe(NO3)3·9H2O] under highly acidic conditions
that can efficiently remove traces amounts have been reported that have the requisite hy- (48% aqueous HF) that yielded to square-shaped
of H2O from natural gas (NG) are critical drolytic stability for H2O-related applications crystals insoluble in water and common organic
for NG transport via pipeline systems, where (9, 10). The hydrolytic stability of MOFs has solvents (see materials and methods). The single-
H2O can be corrosive or create catastrophic been intensively investigated, either under H2O crystal structures are similar to SIFSIX, adopting
blockages through methane ice formation. Cur- vapor or in liquid media (9–16), and some ap- tetragonal space groups (I4/mcm for AlFFIVE-1-
rently, NG is dried with absorbents such as glycol plications such as heat transfer (heat pumps and Ni and P4/nbm FeFFIVE-1-Ni) with primitive
derivatives (1, 2) or adsorbents (solid drying agents) chillers) have been successfully demonstrated cubic topology resulting from the pillaring of
(3). Solid desiccants such as zeolites 3A, 4A, and (16–18). However, the use of MOFs as desic- Ni(II)-pyrazine 2-periodic square-grid layers with
13X achieve lower H2O(g) concentrations com- cants at very low H2O concentrations remains [MF5(H2O)]2– (M = Al3+ or Fe3+) inorganic pillars
pared with liquid absorbents, but an energy- to be explored. We targeted the deliberate in- (Fig. 1A). Le Bail refinements (27) of the powder
intensive activation step (heating to ~200°C) troduction of coordinately unsaturated metal XRD (PXRD) patterns for the as-synthesized mate-
is needed to achieve sufficient working capac- sites, using reticular chemistry and appropri- rials were carried out for full profile fitting against
ity. However, this requirement lengthens each ate inorganic molecular building blocks, in fluo- the resultant structure from SCXRD, displaying a
adsorption-desorption cycle and instigates carbon rinated porous MOF materials that have periodic good agreement factor [unweighted-profile R fac-
buildup (coke formation) within the adsorbent arrays of fluorine moieties in contracted one- tor (Rp) and weighted-profile R factor (Rwp) for
pore system. Commercial solid desiccants show dimensional (1D) channels that have gas sepa- AlFFIVE-1-Ni and FeFFIVE-1-Ni are Rp = 0.086,
highly competitive adsorption of H2O and CO2; ration properties useful for various industrially Rwp = 0.108 and Rp = 0.110, Rwp = 0.085, re-
thus, an excess of one species can greatly lower relevant applications (19–22). These materials spectively] and confirming the phase purity of
the uptake of the other. In NG purification, CO2 adsorb CO2 at the very low partial pressures the as-synthesized fluorinated MOFs (fig. S1).
often must be captured as well, and in CO2 ca- needed for air capture requirements, because In the pillars, the coordinated H2O molecule
pture from flue gas for sequestration, H2O is also charge density is highly and uniformly localized and the three axial F atoms that point into the 1D
present. in a contracted (1D) channels (20–23). channel are disordered over four positions and
A more efficient dehydration process should We show that the inclusion of the Al3+ (or Fe3+) not crystallographically distinguishable. This dis-
have high H2O selectivity in the presence of cations introduces open metal coordination sites order is confirmed by the 19F solid-state magic-
other components, appreciable water uptake ca- within these MOFs and enhances their H2O angle spinning (MAS) NMR spectrum of the
pacity at low H2O partial pressures, and a water affinity without affecting CO2 adsorption. Single- as-synthesized AlFFIVE-1-Ni, which contains
desorption cycle operating at <150°C. Micro- crystal x-ray diffraction (SCXRD) measurements one single and broad isotropic resonance (figs. S2
revealed exposed open metal coordination sites and S3). The 27Al MAS NMR spectrum (Fig. 1J, †)
1
King Abdullah University of Science and Technology within precisely controlled square-shaped channels contains a single resonance in agreement with
(KAUST), Division of Physical Science and Engineering (PSE),
Advanced Membrane and Porous Materials (AMPM),
with periodically arrayed fluoride anions. Calori- the structural model. The large spinning side-
Functional Materials Design, Discovery and Development metric studies revealed a relatively low energy band manifold and unusually high chemical shift
(FMD3), Thuwal 23955-6900, Kingdom of Saudi Arabia. requirement for the MOF adsorbent regeneration- [~1808 parts per million (ppm)] are caused by
2
Institut Lavoisier de Versailles (ILV), UMR CNRS 8180, recycling compared with the solid desiccants the neighboring presence of the paramagnetic
Université de Versailles St–Quentin en Yvelines (UVSQ), 45
Avenue des Etats-Unis, 78035 Versailles Cedex, France.
such as zeolite 4A. Single- and multicomponent Ni ions. The available free volume within the
3
CEMHTI-CNRS, UPR 3079, 1D Avenue de la Recherche adsorption experiments showed that these mate- KAUST-8 channels is occupied by extra non-
Scientifique, 45071 Orléans Cedex 2, France. 4Institut Charles rials selectively adsorbed H2O and CO2 from coordinated water molecules, as evidenced by
Gerhardt Montpellier (UMR CNRS 5253), Université gas streams containing N2, CH4, and higher the noted electron density distribution along the
Montpellier, Place Eugène Bataillon, 34095 Montpellier Cedex
05, France.
hydrocarbons such as C 2+n. Multiprobe solid-state channels and supported by thermogravimetric
*These authors contributed equally to this work. †Corresponding nuclear magnetic resonance (NMR) and peri- analysis (fig. S4) and 1H MAS NMR results
author. Email: mohamed.eddaoudi@kaust.edu.sa odic density functional theory (DFT) calcula- (Fig. 1I, †). Crystallographic independent H2O

Cadiau et al., Science 356, 731–735 (2017) 19 May 2017 1 of 5

R ES E A RC H | R E PO R T

ment that is further sustained by additional H-

bonding interactions from the coordinated H2O
and the F moieties (fig. S5, C and D).
To evaluate the thermal and hydrolytic sta-
bilities, we performed variable humidity PXRD
(Fig. 2A and fig. S6) and variable temperature
PXRD measurements (fig. S7). The crystallin-
ity for both AlFFIVE-1-Ni and FeFFIVE-1-Ni
was maintained up to 400° and 250°C, respec-
tively, confirming their relative high thermal
stability. The water stabilities of AlFFIVE-1-Ni
and FeFFIVE-1-Ni were corroborated by sim-
ilarities between theoretical PXRD patterns

Downloaded from https://www.science.org at King Abdullah University of Science and Technology on November 08, 2022
derived from the crystal structure and the ex-
perimental PXRD patterns collected up to 95%
relative humidity (RH).
We evaluated the adsorption properties of
these fluorinated materials for H2O against probe
molecules—specifically, highly polarizable mole-
cules such as CO2, n-C4H10, and iso-C4H10, as
well as poorly polarizable molecules such CH4
and N2. Related materials with no open metal sites
but with the requisite confined channels hosting
a periodic array of fluorine moieties, such as
SIFSIX and NbOFFIVE-1-Ni, showed high affinity
for CO2 as compared with H2O (20–23). The water
adsorption isotherms for both materials displayed
a pronounced adsorption isotherm steepness at
very low partial pressures, indicative of high H2O
affinity (Fig. 2B). Both AlFFIVE-1-Ni and FeFFIVE-
1-Ni showed substantial water adsorption capaci-
ties with an associated saturation of 22 and 18
weight % (wt %), respectively, at P/P0 = 0.05 rel-
ative pressure (where P is the equilibrium pressure
and P0 is the saturation pressure) and 25°C. This
marked steepness in the water adsorption iso-
therms made it challenging to quantify the heat
of adsorption (Qst), so we used thermogravimetric–
differential scanning calorimetry (TG-DSC). The
H2O Qst values for both AlFFIVE-1-Ni and FeFFIVE-
1-Ni were 63 kJ/mol (Fig. 3F) and 64.7 kJ/mol (fig.
S9B), respectively.
For conventional dehydrating agents such as
zeolites 3A, 4A, and 5A, the H2O Qst at very low
loading is markedly higher (80 to 120 kJ/mol)
than in the MFFIVE-1-Ni (M = Al3+ or Fe3+) ad-
sorbents, which leads to high regeneration tem-
Fig. 1. Structure description and characterization of as-synthesized, dehydrated, and peratures (~250°C) (Fig. 2C). However, the requisite
rehydrated forms of AlFFIVE-1-Ni (KAUST-8). (A) Repetitive square motif in the Ni-pyrazine temperatures for AlFFIVE-1-Ni and FeFFIVE-
(4,4′) square-grid layer. (B) Projection along [001] of the crystal structure of AlFFIVE-1-Ni, with a 1-Ni are only ~105°C (Fig. 2C). Adsorption and
primitive cubic topology illustrating the connectivity of the hosted water guest molecules within the desorption cyclic properties and the activation
1D channel. (C) Projection along [001] of the crystal structure of the dehydrated form of AlFFIVE-1- and reactivation assessment of the AlFFIVE-
Ni, showing the trigonal bipyramid (AlF5)2– inorganic building block. (D) Projection along [001] of the 1-Ni adsorbent revealed full and rapid dehydra-
crystal structure of the rehydrated form AlFFIVE-1-Ni, revealing the water molecule within the tion, achievable after a simple heating at 105°C
coordination sphere of Al3+ cations. (E to H) DFT calculation by applying a progressive removal under vacuum or by a simple flushing with a
of H2O molecules (from 3 to 0 per unit cell) on AlFFIVE-1-Ni, confirming the crystal structure of less adsorbable gas, such as N2 (Fig. 2D). Silica gel
the as-synthesized form and the dehydration mechanism. (I) 1H and (J) 27Al MAS NMR spectra of desiccants exhibit a H2O Qst of 43 to 50 kJ/mol
AlFFIVE-1-Ni recorded on as-synthesized sample (†), sample previously heated at 105°C under (28) and can be recycled by heating at 90° to
vacuum for 8 hours (‡), and rehydrated sample (§), confirming the departure and reversion of water 100°C but have only nominal H2O adsorption
molecules within the materials. uptake at relatively low H2O vapor partial pres-
sures (Fig. 2C). Similar to zeolite molecular sieves,
MFFIVE-1-Ni exhibited an excellent rate of H2O
molecules are present within the cavities—two generated H2O molecules are interconnected by adsorption (Fig. 2B) and largely improved adsorp-
for AlFFIVE-1-Ni and four for FeFFIVE-1-Ni. Dis- H-bonding to each other and further connected tion kinetics (fig. S10) compared with other types
tinct H-bonding networks between F anions of to F atoms belonging to pillars (Fig. 1B and fig. of desiccants (2). In fact, AlFFIVE-1-Ni (KAUST-8)
pillars and H2O guest molecules were observed S5, A and B). In FeFFIVE-1-Ni, four water mol- showed the same water vapor adsorption per-
in both cases. In AlFFIVE-1-Ni, the symmetrically ecules constitute a tetrahedral-shaped arrange- formance after evacuation in vacuum at 105°C

Cadiau et al., Science 356, 731–735 (2017) 19 May 2017 2 of 5

R ES E A RC H | R E PO R T

much lower uptake at 0.05 P/P0 (19.8 wt % versus

5.4 wt % for 4A).
We evaluated adsorption competition between
water and CO2 in N2- and CH4-containing gas
streams, as would occur for separation involving
NG, flue gas, and syngas. The CO2 adsorption
isotherms at 25°C showed the saturation of the
adsorbent pore system at a very low relative pres-
sure (0.05 P/P0), with a marked steepness, from
which we derived apparent surface areas and pore
volumes, respectively, of 258 m2/g and 0.102 cm3/g
for AlFFIVE-1-Ni and 324 m2/g and 0.129 cm3/g for
FeFFIVE-1-Ni. The Qst of CO2, determined from

Downloaded from https://www.science.org at King Abdullah University of Science and Technology on November 08, 2022
variable temperature adsorption isotherms (figs.
S14A and S15A) for AlFFIVE-1-Ni and FeFFIVE-
1-Ni, was estimated to be 45 ± 2.2 kJ/mol (fig.
S14B) and 48 ± 2.5 kJ/mol, respectively (fig.
S15B); this value was also measured directly by
calorimetry for the aluminum (43 kJ/mol) and
iron (45.7 kJ/mol) analogs (Fig. 3F and figs. S8
and S9). By comparison with the Qst associated
with H2O adsorption (63 kJ/mol), the framework-
CO2 interactions (45 kJ/mol) are much weaker
than the framework-H2O interactions (63 kJ/
mol). The DFT calculations showed that CO2
preferentially adsorbs in a distinct site from
H2O (Fig. 3A). The electropositive C interacts
with F atoms of pillars (F···CCO2 = ~2.8 Å), and
the electronegative O is surrounded by pyrazine
Fig. 2. Stability and water sorption properties of MFFIVE-1-Ni. (A) Variable humidity PXRD H (H···OCO2 = ~2.6 Å) (Fig. 3A). This geometry
patterns or AlFFIVE-1-Ni. q, diffraction angle; a.u., arbitrary units. (B) Water adsorption is reminiscent of that previously reported for
isotherm at 35°C for AlFFIVE-1-Ni (violet) and FeFFIVE-1-Ni (pink). (C) Comparison and relation NbOFFIVE-1-Ni, although the presence of open
between water uptake at 0.05 P/P0, heat of sorption, and regeneration temperature of metal sites in AlFFIVE-1-Ni prevents CO2 to
AlFFIVE-1-Ni and FeFFIVE-1-Ni, along with other well-known dehydrating agents (see materials interact simultaneously with the four F cen-
and methods). (D) Adsorption-desorption cycles for AlFFIVE-1-Ni (adsorption at 25°C ters, thus leading to a slightly lower Qst than for
and desorption at 105°C by flushing N2). NbOFFIVE-1-Ni (~54 kJ/mol) (20).
Breakthrough adsorption column experiments
for 480, 60, and 30 min, demonstrating the fast resonance (from 1808 to 1340 ppm), attribut- were carried out on AlFFIVE-1-Ni for a H2O vapor
water vapor desorption rate (fig. S11). able to the creation of the fivefold coordinated (single component) and in the presence of N2,
A microscopic picture of the dehydration pro- Al3+ cation (Fig. 1J). This prediction strongly CH4, and CO2 adsorbates using a similar total
cess can be obtained from the analysis of the suggests that a structural relaxation of the host flow of 23 cm3/min and an RH of 75%. The H2O
DFT geometry-optimized structures of AlFFIVE- framework assists the removal of the last co- retention times in the column were relatively sim-
1-Ni obtained by sequentially removing 1 H2O ordinated H2O molecule. ilar (500 to 600 min/g within experimental error)
per formula unit. Figure 1E first shows that the Rehydration allows Al cations to recover their and unaltered by the presence of CH4 and/or CO2
simulated arrangement of water in the fully hy- sixfold coordination state, as shown by the 27Al in the hydrated mixed-gas CO2/CH4: 1/99 mix-
drated material agrees with the SCXRD data MAS NMR spectrum (Fig. 1J, §) in which the res- ture (Fig. 4A and fig. S16B). After multiple cycles of
(Fig. 1B), with one H2O coordinated to the Al onance shifts to higher chemical shift (1640 ppm). the adsorption column breakthrough test with a
site and interconnected by H-bonding to the Noticeably, in the rehydrated sample (fig. S2, §), hydrated CO2/CH4: 1/99 gas mixture, the per-
other molecules that strongly interact with the the 19F MAS NMR spectrum contains three res- formance of AlFFIVE-1-Ni was unaltered (fig. S17). A
F atoms of pillars. However, during the de- onances of similar intensity, in contrast to the reduction in dehydration performance (nearly 50%)
sorption process, the last H2O molecule desorbed sharp resonance of the dehydrated sample, which was observed for zeolite 4A when the desorption
(Fig. 1G) still interacted with F atoms of pillars, confirms that the F atoms in the rehydrated form temperature was reduced from 250° to 105°C
but the Al3+ environment relaxed from its initial have a different environment compared with in (fig. S18). Increasing the CO2 concentration in
square bipyramidal geometry toward a trigo- the dehydrated form. This difference is explained the CO2/N2 mixture to 1, 10, and 50% (Fig. 4, B
nal bipyramidal-like structure that tends to by the introduction of the H2O molecule within to D) caused no noticeable changes on the H2O
push away H2O from the Al3+ sites. This step the coordination sphere of Al3+ cations. A 19F-19F vapor retention time in the column (500 to
leads to a dehydrated structure with a perfect 2D NMR correlation spectrum (fig. S3) shows 600 min/g within experimental error). Thus, the
trigonal bipyramidal environment for Al3+ that that the rehydrated form contains three inequi- same H2O vapor adsorption behavior and uptake
is reminiscent of the geometry previously re- valent F atoms and one H2O molecule in the occurred independently of the CO2 concentra-
ported for the [AlF5]2– complex (29, 30). This equatorial plane of the Al3+ octahedron. We in- tion and the composition of the evaluated gas
dehydration process is also confirmed by the vestigated the temperature of reactivation of mixtures (e.g., CO2/CH4: 1/99, CO2/N2: 1/99, CO2/
19
F and 27Al MAS NMR spectra of AlFFIVE-1- AlFFIVE-1-Ni using various isothermal adsorption- N2: 10/90, and CO2/N2: 50/50). However, although
Ni, recorded on a sample previously dehydrated desorption cycles and found that mild desorption CH4 (humid condition; Fig. 4A) and N2 (both dry
for 8 hours at 105°C under vacuum, which conditions (25°C under vacuum) (fig. S12) fully and humid conditions; fig. S16, A and C, and Fig.
shows a single 19F resonance (fig. S2, ‡), much removed H2O from the cavities, including those 4, B to D) did not show any noticeable uptake, the
narrower than that in the as-synthesized sam- connected to Al3+ cations; similar reactivation at retention time in the column for CO2 during
ple (fig. S2, †), and a strong shift of the 27Al room temperature of zeolite 4A (fig. S13) showed moisture-containing tests revealed a nominal

Cadiau et al., Science 356, 731–735 (2017) 19 May 2017 3 of 5

R ES E A RC H | R E PO R T

Downloaded from https://www.science.org at King Abdullah University of Science and Technology on November 08, 2022
Fig. 3. Distinct adsorption sites for H2O and CO2 molecules within after activation (105°C under N2 flow for 8 hours) and exposure to a mixture of
AlFFIVE-1-Ni. DFT calculations performed on AlFFIVE-1-Ni, showing the CO2 and H2O. (F) Comparison of two independent TG-DSC measurements
different adsorption sites for H2O and CO2: (A) AlFFIVE-1-Ni containing on activated AlFFIVE-1-Ni with a (1/99) CO2/N2 mixture and with humid
one CO2 molecule, (B) AlFFIVE-1-Ni containing one H2O molecule, and N2 (75% RH) at 25°C. Temperature-programmed desorption revealing the CO2/
(C) AlFFIVE-1-Ni containing both H2O and CO2 molecules. (D and E) Pro- H2O composition of (G) 0.67/0.33 at CO2 breakthrough time, (H) 0.24/0.76
jection of single-crystal structure along [001] and [010] of AlFFIVE-1-Ni at half H2O breakthrough time, and (I) 0.11/0.89 at H2O breakthrough time.

difference (within experimental error) when com-

pared with the corresponding dry tests for all
of the evaluated gas mixtures. Although the H2O
adsorption energetics in AlFFIVE-1-Ni particularly
favor the H2O adsorption, concomitant adsorption
of the CO2 molecules occurred in its presence.
Subsequent post–in situ temperature-programed
desorption studies were performed by progres-
sively heating the column to 100°C, initially sat-
urated with a CO2/H2O/N2: 9.2/2/88.8 gas feed
mixture, at three adsorbed phase states: (i) just
after the breakthrough time of CO2 ~ 25 min/g;
(ii) at half breakthrough time of H2O ~ 300 min/g;
and, finally, (iii) after full breakthrough time of
H2O ~ 600 min/g (Fig. 3, G to I). The H2O and CO2
molecules simultaneously desorbed from the col-
umn with CO2/H2O adsorbed phase compositions
of 0.67/0.33 (Fig. 3G), 0.24/0.76 (Fig. 3H), and
0.11/0.89 (Fig. 3I) at adsorbed phase states i,
ii, and iii, respectively, indicative of coadsorbed
CO2 and H2O in the pore system of the AlFFIVE-1-
Ni. Both H2O and CO2 could be simultaneously
removed with a H2O/CO2 selectivity ranging from
2 to 39. Single-crystal structure refinement of
AlFFIVE-1-Ni simultaneously exposed to air moist-
ure and pure CO2 after activation showed oc-
cupancies of 26 and 74% for CO2 and H2O,
respectively (Fig. 3B and table S3). DFT calcu-
lations were performed for AlFFIVE-1-Ni contain-
ing diverse amounts of H2O and CO2 adsorbed in
Fig. 4. Column breakthrough experiments. (A) Breakthrough test for a humid (75% RH) CO2/CH4: mixture (see supplementary materials). As a model
1/99 mixture. (B to D) Effect of CO2 concentration on water breakthrough tests. (B) humid (75% RH) mixture case (1 H2O + 1 CO2 per formula unit), Fig.
CO2/N2: 1/99. (C) Humid (75% RH) CO2/N2: 10/90. (D) Humid (75% RH) CO2/N2: 50/50. CA, 3A confirms that both species occupy the same
adsorbate concentration; C0, saturation concentration. adsorption sites as single components. Calorimetric

Cadiau et al., Science 356, 731–735 (2017) 19 May 2017 4 of 5

R ES E A RC H | R E PO R T

measurements of a hydrated gas mixture (CO2/ 6. V. Guillerm et al., Chem. Soc. Rev. 43, 6141–6172 (2014). 30. U. Gross, D. Müller, E. Kemnitz, Angew. Chem. Int. Ed. 42,
N2:1/99) showed similar CO2 and H2O Qst values 7. Y. He, B. Li, M. O’Keeffe, B. Chen, Chem. Soc. Rev. 43, 2626–2629 (2003).
5618–5656 (2014).
as compared with those for the single-adsorbate AC KNOWLED GME NTS
8. S. Kitagawa, R. Kitaura, S. Noro, Angew. Chem. Int. Ed. 43,
cases (Fig. 3F and fig. S8). 2334–2375 (2004). Research reported in this publication was solely performed
To ascertain the prospective molecular exclu- 9. N. C. Burtch, H. Jasuja, K. S. Walton, Chem. Rev. 114, at KAUST and was supported by KAUST funds, KAUST
sion cutoff for MFFIVE-1-Ni imposed by the 10575–10612 (2014). funding grants (CCF/1/1972-02-01, CCF/1/1972-8-01, and
adsorbent pore-aperture size and shape and sub- 10. J. Canivet, A. Fateeva, Y. Guo, B. Coasne, D. Farrusseng, OSR-2017-CPF-3325), and Aramco. The data reported in the
Chem. Soc. Rev. 43, 5594–5617 (2014). paper are presented in the main text and the supplementary
sequently modulating the access to the confined materials. Crystal structures of the as-synthesized AlFFIVE-1-Ni,
11. P. Küsgens et al., Micropor. Mesopor. Mater. 120, 325–330
square-shaped channels, we explored the ad- (2009). the dehydrated AlFFIVE-1-Ni, AlFFIVE-1-Ni (rehydrated),
sorption of slightly larger and relatively bulkier 12. H. Reinsch et al., Chem. Commun. 48, 9486–9488 AlFFIVE-1-Ni·1.48H2O·0.26CO2, and FeFFIVE-1-Ni are available
probe molecules such as n-C4H10, iso-C4H10, (2012). free of charge from the Cambridge Crystallographic Data Centre
13. Y.-K. Seo et al., Adv. Mater. 24, 806–810 (2012). (CCDC) under reference nos. CCDC 1538217, 1538215,
1-propanol, and isopropanol (figs. S19 and S20). 1538216, 1538219, and 1538218. A.C., Y.B., K.A., P.M.B., M.E.,
14. A. Khutia, H. U. Rammelberg, T. Schmidt, S. Henninger,
Both AlFFIVE-1-Ni and FeFFIVE-1-Ni (KAUST-8 C. Janiak, Chem. Mater. 25, 790–798 (2013). and KAUST have filed provisional patents (WO2016/162834A1

Downloaded from https://www.science.org at King Abdullah University of Science and Technology on November 08, 2022
and KAUST-9) showed no noticeable adsorption 15. G. Akiyama et al., Micropor. Mesopor. Mater. 157, 89–93 and WO2016/162835A1) pertaining to the results presented
for isobutane and isopropanol, indicative of pro- (2012). herein. K.A., M.E., and A.C. conceptualized the design and the
construction of the reported MOF materials; A.C. carried out
spective efficient dehydration of gases and vapors 16. H. Furukawa et al., J. Am. Chem. Soc. 136, 4369–4381
the materials synthesis; K.A., A.C., and A.S. conducted and
with equal and larger sizes than isobutane and (2014).
interpreted the crystallographic experiments; P.M.B., Y.B.,
17. A. Cadiau et al., Adv. Mater. 27, 4775–4780 (2015).
isopropanol via full molecular sieving. Overall, 18. M. F. de Lange, K. J. F. M. Verouden, T. J. H. Vlugt, J. Gascon,
and A.C. conducted and interpreted low-pressure adsorption
these results highlight the potential of the KAUST-8 experiments; Y.B. and P.M.B. contributed to conceptualizing,
F. Kapteijn, Chem. Rev. 115, 12205–12250 (2015).
designing, conducting, and interpreting calorimetric and mixed
adsorbent-desiccant to congruently adsorb CO2 19. A. Cadiau, K. Adil, P. M. Bhatt, Y. Belmabkhout, M. Eddaoudi,
adsorption experiments; R.S.P. and G.M. performed and
and H2O and could be used for the on-demand Science 353, 137–140 (2016).
analyzed the DFT calculations; C.M.-C. conducted and analyzed
simultaneous removal of CO2 and H2O by varying 20. P. M. Bhatt et al., J. Am. Chem. Soc. 138, 9301–9307 the solid-state NMR experiments; and K.A., Y.B., A.C., and M.E.
(2016). wrote the manuscript.
the adsorption-desorption cycle times. 21. O. Shekhah et al., Nat. Commun. 5, 4228 (2014).
22. P. Nugent et al., Nature 495, 80–84 (2013).
RE FE RENCES AND N OT ES 23. O. Shekhah et al., Chem. Commun. 51, 13595–13598 (2015). SUPPLEMENTARY MATERIALS
1. A. Karimi, M. A. Abdi, Chem. Eng. Process. 48, 560–568 24. K. Adil, M. Leblanc, V. Maisonneuve, J. Fluor. Chem. 130,
www.sciencemag.org/content/356/6339/731/suppl/DC1
(2009). 1099–1105 (2009).
Materials and Methods
2. P. Gandhidasan, A. A. Al-Farayedhi, A. A. Al-Mubarak, Energy 25. A. Cadiau et al., Solid State Sci. 13, 151–157 (2011).
Supplementary Text
26, 855–868 (2001). 26. A. Cadiau, S. Auguste, F. Taulelle, C. Martineau, K. Adil,
Figs. S1 to S26
3. S. S. B. M. Rohani, Natural Gas Dehydration Using Silica Gel: CrystEngComm 15, 3430–3435 (2013).
27. A. Le Bail, H. Duroy, J. L. Fourquet, Mater. Res. Bull. 23, Tables S1 to S5
Fabrication of Dehydration Unit (University of Malaysia Pahang,
2009). 447–452 (1988). References (31–48)
4. T. Devic, C. Serre, Chem. Soc. Rev. 43, 6097–6115 28. L.-Z. Zhang, H.-X. Fu, Q.-R. Yang, J.-C. Xu, Energy 65, 430–440 Data S1 to S5
(2014). (2014).
5. H. Furukawa, K. E. Cordova, M. O’Keeffe, O. M. Yaghi, Science 29. F. Bouyer, G. Picard, J.-J. Legendre, Int. J. Quantum Chem. 23 January 2017; accepted 25 April 2017
341, 1230444 (2013). 52, 927–934 (1994). 10.1126/science.aam8310

Cadiau et al., Science 356, 731–735 (2017) 19 May 2017 5 of 5

 Hydrolytically stable fluorinated metal-organic frameworks for energy-efficient
dehydration
Amandine CadiauYoussef BelmabkhoutKarim AdilPrashant M. BhattRenjith S. PillaiAleksander ShkurenkoCharlotte
Martineau-CorcosGuillaume MaurinMohamed Eddaoudi

Downloaded from https://www.science.org at King Abdullah University of Science and Technology on November 08, 2022
Science, 356 (6339), • DOI: 10.1126/science.aam8310

Drying natural gas efficiently

Natural gas must be purified before it can be transported. The preparation process also includes a drying step to
remove water. Microporous adsorbents such as zeolites are used for this purpose, but they often need to be heated
to temperatures up to 250°C to remove the water so that they can be reused. Cadiau et al. describe a fluorinated
metal-organic framework containing nickel metal centers that can remove water from gas streams but that can be
regenerated by heating to only 105°C.
Science, this issue p. 731

View the article online

https://www.science.org/doi/10.1126/science.aam8310
Permissions
https://www.science.org/help/reprints-and-permissions

Use of this article is subject to the Terms of service

Science (ISSN 1095-9203) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW,
Washington, DC 20005. The title Science is a registered trademark of AAAS.
Copyright © 2017, American Association for the Advancement of Science