Computational design of metal-organic frameworks based on stable zirconium building units for storage and delivery of methane

Article abstract of DOI:10.1021/cm502304e

A metal-organic framework (MOF) with high volumetric deliverable capacity for methane was synthesized after being identified by computational screening of 204 hypothetical MOF structures featuring (Zr₆O₄)(OH)₄(CO₂)_n inorganic building blocks. The predicted MOF (NU-800) has an fcu topology in which zirconium nodes are connected via ditopic 1,4-benzenedipropynoic acid linkers. Based on our computer simulations, alkyne groups adjacent to the inorganic zirconium nodes provide more efficient methane packing around the nodes at high pressures. The high predicted gas uptake properties of this new MOF were confirmed by high-pressure isotherm measurements over a large temperature and pressure range. The measured methane deliverable capacity of NU-800 between 65 and 5.8 bar is 167 cc(STP)/cc (0.215 g/g), the highest among zirconium-based MOFs. High-pressure uptake values of H₂ and CO₂ are also among the highest reported. These high gas uptake characteristics, along with the expected highly stable structure of NU-800, make it a promising material for gas storage applications. (Figure Presented).

Full text of DOI:10.1021/cm502304e

Article
pubs.acs.org/cm
Computational Design of Metal−Organic Frameworks Based on
Stable Zirconium Building Units for Storage and Delivery of Methane
†
‡
§,∥
†
Diego A. Gomez-Gualdron, Oleksii V. Gutov, Vaiva Krungleviciute, Bhaskarjyoti Borah,
‡
,‡
,§,∥
,‡,⊥
Joseph E. Mondloch, Joseph T. Hupp,* Taner Yildirim,* Omar K. Farha,*
,
†
and Randall Q. Snurr*
†
‡
Department of Chemical and Biological Engineering and Department of Chemistry and International Institute for Nanotechnology,
Northwestern University, 2145 Sheridan Road, Evanston, Illinois 60208, United States
§
NIST Center for Neutron Research, National Institute of Standards and Technology, Gaithersburg, Maryland 20899, United States
∥Department of Materials Science and Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104, United States
⊥Department of Chemistry, Faculty of Science, King Abdulaziz University, Jeddah 22254, Saudi Arabia
*
S Supporting Information
ABSTRACT: A metal−organic framework (MOF) with high
volumetric deliverable capacity for methane was synthesized after
being identified by computational screening of 204 hypothetical
MOF structures featuring (Zr O )(OH) (CO ) inorganic
6
4
4
2 n
building blocks. The predicted MOF (NU-800) has an fcu
topology in which zirconium nodes are connected via ditopic 1,4-
benzenedipropynoic acid linkers. Based on our computer
simulations, alkyne groups adjacent to the inorganic zirconium
nodes provide more efficient methane packing around the nodes
at high pressures. The high predicted gas uptake properties of
this new MOF were confirmed by high-pressure isotherm
measurements over a large temperature and pressure range. The
measured methane deliverable capacity of NU-800 between 65 and 5.8 bar is 167 cc(STP)/cc (0.215 g/g), the highest among
zirconium-based MOFs. High-pressure uptake values of H and CO are also among the highest reported. These high gas uptake
2
2
characteristics, along with the expected highly stable structure of NU-800, make it a promising material for gas storage
applications.
1
. INTRODUCTION
nanoporous material should be able to store and deliver 263
cc(STP)/cc of methane. Storage should occur at a pressure in
the range of 35 to 65 bar and ambient temperature, and the
delivery pressure must be 5.8 bar as determined by the engine
inlet pressure. Additionally, the nanoporous material should
retain its performance after numerous adsorption/desorption
cycles, which implies material stability and resistance to
poisoning by higher hydrocarbons in NG such as ethane and
propane.
Natural gas (NG), which is mainly composed of methane, is an
attractive midterm solution to problems associated with energy
security and greenhouse gas emissions due to its abundant
19
1
reserves and low CO emissions. Nonetheless, for on-board
2
storage in vehicular applications, the low volumetric energy
density of methane at standard temperature and pressure
creates the need to implement densification strategies such as
compressed natural gas (CNG) (250 bar) or liquefied natural
gas (LNG) (111 K). These strategies present challenges
associated with cost and practicality that have prevented them
from being widely adopted. Therefore, alternative densification
strategies such as adsorbed natural gas (ANG), in which natural
gas is stored in nanoporous materials, are being actively
Metal−organic frameworks (MOFs) − which are crystalline
nanoporous materials assembled through the modular combi-
nation of organic and inorganic building blocks−are promising
materials for methane storage and delivery. One of their most
attractive features is that different combinations of building
1
1,20−23
2
−18
blocks can give rise to millions of different structures.
explored.
This fact opens the possibility of tuning the pore structure to
optimize material performance. A major concern for the
practical application of MOFs is the poor stability of some
The success of ANG technologies depends on the ability of
the nanoporous material to store and deliver methane in
quantities that allow a practical driving range in a cost-efficient
fashion. Recently, the Advanced Research Projects Agency −
Energy (ARPA-E) established a goal that ANG systems should
Received: June 26, 2014
Revised: September 14, 2014
19
match the volumetric energy density of CNG (9.2 MJ/L).
Assuming no packing losses, this target implies that the
©
XXXX American Chemical Society
A
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
MOFs. However, MOFs with exceptional thermal, chemical,
2
4−27
and hydrolytic stability,
6
such as zirconium-based UiO-
6, have now been synthesized. Furthermore, a variety of
MOFs having similar (Zr O )(OH) (CO ) inorganic building
2
8
6
4
4
2 n
blocks have been found to be highly stable, suggesting that their
stability is linked to the presence of this type of building
2
3,29−37
unit.
To date, lack of hydrolytic and chemical stability (against
impurities present in NG) has hindered the performance of
MOFs when subject to numerous adsorption/desorption
cycles. MOFs with the highest measured volumetric methane
deliverable capacity (storage at 65 bar and delivery at 5.8 bar)
3
8
39
to date are MOF-519 (203 cc(STP)/cc), UTSA-76 (187
Figure 1. Four net topologies compatible with zirconium-based
MOFs.
3
17
15
cm (STP/cc), MOF-5 (185 cc(STP)/cc), HKUST-1 (184
13
cc(STP)/cc), and NU-125 (174 cc(STP)/cc), which feature
less stable inorganic building blocks such as Zn O(CO ) or
4
2 6
Cu (CO ) . Improvement in the stability of, for instance,
2
2 4
copper-based MOFs has been explored through the use of
bulky hydrophobic groups, which have been added postsyn-
40
thesis via plasma treatment or prior to MOF synthesis as part
41
of the organic building blocks. However, occupation of pore
volume by bulky groups can be detrimental for storage
applications. Therefore, designing MOFs with inherent stability
and high deliverable capacities is a very attractive goal.
From an abstract perspective, creating new MOFs is as
simple as selecting appropriate organic and inorganic building
units and connecting them to form a three-dimensional
network. However, experimentally, creating new MOFs can
be a time-consuming process that includes optimizing the
solvothermal reaction conditions for obtaining a crystalline
product and optimizing the solvent evacuation procedure to
fully “activate” the pores of the MOF. Molecular simulation can
be a valuable tool to accelerate the discovery of materials for
applications such as gas storage because it can provide valuable
Figure 2. Basic building blocks used to construct zirconium-based
hypothetical MOFs. The color-coded translucent dots indicate
connection points. In the names of the ditopic and peripheral building
blocks, P indicates a phenyl ring and T indicates a triple bond.
1
1,42−45
insights to guide material design
and can be used to
47
the RSCR database based on having the same connectivity of
screen candidates and directly identify desirable candidates for
the zirconium SBUs as in the csq net. MOFs of csq and scu
topological nets have 8-coordinated inorganic nodes, whereas
the fcu and ftw topological nets have 12-coordinated ones.
Additionally, some ftw mixed-ligand and selected functionalized
MOFs were built (detailed description in the SI). Notice that
the criterion for selecting topological nets was based on
previously shown synthetic accessibility, for fcu, ftw, and csq
nets, or their compatibility with experimentally observed
zirconium SBU geometries (see Figure S3.1a), for the simple
edge-transitive scu net. This was done to increase the
probability for the hypothetical structures to be synthetically
accessible.
1
1
synthesis.
Here we describe our efforts in computationally designing
and screening 204 hypothetical MOFs featuring highly stable
(
Zr O )(OH) (CO ) inorganic building blocks, with the goal
6 4 4 2 n
of identifying and synthesizing a zirconium based MOF with
high volumetric deliverable capacity for methane. These efforts
provided valuable insights into structural features that improve
methane deliverable capacity and led to the synthesis,
activation, and testing of the identified top zirconium-based
MOF in our screening: NU-800.
2. COMPUTATIONAL METHODS
fcu MOFs feature octahedral cages and smaller tetrahedral
ones, and ftw MOFs feature cubic cages and smaller octahedral
ones. On the other hand, csq MOFs have one-dimensional
hexagonal channels and smaller triangular ones, and scu MOFs
Construction of Hypothetical MOFs. We used a reverse
46
topological approach, where the four topological nets shown
in Figure 1 were used as “templates” for the construction of
zirconium-based hypothetical MOFs using the building blocks
shown in Figure 2. Twelve ditopic organic building blocks were
combined with zirconium secondary building units (SBUs) to
construct MOFs with a fcu topological net, which is the
46
have diamond channels. As noted by Bureekaew and Schmid,
when the building blocks have lower symmetry than their
corresponding “vertex” in the network template, one can
observe isoreticular isomerism, which corresponds to structures
with the same topological net and same chemical formula, but
different arrangements of the building blocks. Here, one type of
isoreticular isomerism, which we refer to as type I, occurs for
csq and scu MOFs, because the somewhat rectangular
tetratopic building blocks (with the exception of the perfectly
square porphyrin-based building blocks) can be placed in two
different orientations in the network template, resulting in
isoreticular isomeric MOF pairs that differ in their channel sizes
28
topological net of UiO-66. The four central and nine
peripheral building blocks were combined into 36 tetratopic
building blocks following the formula Y-4X, where Y refers to
central building blocks and X to peripheral ones. These were in
turn combined with zirconium SBUs to construct MOFs with
ftw, csq, and scu topological nets. ftw is the topological net of
2
9
33
MOF-525 and NU-1100, while csq is the topological net of
2
9
32
MOF-545 and NU-1000. The scu net was selected from
B
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
see illustration in Figure S3.3). Note that the tetratopic
Article
50
51
51
(
methane, nitrogen, 5 and carbon dioxide, and the Levesque
2
building blocks were allowed to adopt their somewhat
rectangular shape in mixed-ligand ftw MOFs but were
constrained to a square shape in single-ligand ftw MOFs.
Another type of isoreticular isomerism, which we refer to as
type II, occurs for csq, scu, fcu, and ftw MOFs when the
organic building blocks are isomers among themselves. For
instance, the ditopic building blocks PTTP and TPPT are
isomers, as are the tetratopic building blocks Py-4PTT, Py-
model for hydrogen. Lorentz−Berthelot mixing rules were
used to obtain the interaction parameters between MOF atoms
and gas molecules. During the simulations, heats of adsorption
53
were calculated using the fluctuation method. Surface areas
were calculated geometrically as the area of the surface created
54
by rolling a nitrogen-size probe over the MOF atoms. Void
fractions were calculated via Widom insertions of a helium
probe, where the void fraction was equal to the average Widom
55
4TPT, and Py-4TTP (Figure 3).
factor. Pore size distributions were calculated geometrically
56
using the method of Gelb and Gubbins, where for a large
number of random points in the pore volume, the largest
sphere that can enclose each point without overlapping
framework atoms is determined.
3
. RESULTS
Property-Performance Relationships. We characterized
the structural properties of the ∼200 hypothetical MOFs and
simulated their volumetric deliverable capacities for methane
between 65 and 5.8 bar. The results revealed some relationships
which are important as guidelines for material design. In fact, as
these relationships were emerging from the screening (e.g., a
relationship between deliverable capacity and the location of
triple bonds in the structures), we used them to guide the
selection of new building blocks until the final set illustrated in
Figure 2 was obtained. Figure 4a shows that the best volumetric
deliverable capacities are obtained with hypothetical MOFs
containing void fractions in the range of 0.78−0.85. It is
apparent that MOFs with fcu and ftw topologies globally
exhibit better volumetric deliverable capacities than MOFs with
csq and scu topologies. Although the highest deliverable
capacity did not correspond to the MOF with the highest
volumetric surface area, in general these two variables are
linearly correlated as shown in Figure 4b. Moreover, the values
of the volumetric surface area explain much of the difference in
performance among the four topologies explored here, as the
similarity of Figures 4a and 4c suggests. These figures show
that, for the csq topology, upon variations in void fraction, the
Figure 3. Example of a triad of isomeric tetratopic building blocks Y-
4X, with Y = Py, and X = PTT, TPT, and TTP.
The type II isoreticular isomerism is interesting, because it
results in sets of MOFs with the same chemical formula that
feature virtually equal unit cell sizes and, thus, virtually equal
pore sizes and specific pore volumes. Remarkably, however, we
found that the adsorption properties can differ among these
types of isoreticular isomers, most notably for ftw and fcu
topological nets, as we discuss in the next section. It is worth
emphasizing that type II isoreticular isomerism can be readily
controlled during synthesis through ligand selection. However,
control over type I isoreticular isomerism (and sometimes
topology) can be difficult to achieve, possibly requiring a careful
tuning of synthesis conditions.
After preliminary structures of the hypothetical MOFs were
obtained, they were subjected to structural optimization using
48
the Universal force field to describe the energetics of the
structure (see details in the SI). We found that this procedure
predicts the experimental lattice constants of IRMOF-1,
HKUST-1, UiO-66, and MOF-525 within 2.1%, 0.3%, 2.1%,
and 0.62% accuracy, respectively. Furthermore, we have
previously used this procedure to create preliminary MOF
structures to assess structural, textural, and adsorption proper-
2
volumetric surface area remained within the 1000−1450 m /
3
cm range, and the deliverable capacity remained within the
106−142 cc(STP)/cc range.
As mentioned above, some insights into the effect of MOF
structural features (i.e., location of triple-bonds) also emerged
from our analysis. Notice that for a set of isomeric building
blocks, such as the ones illustrated in Figure 3, the building
block can be terminated in a triple bond or in a phenyl group.
3
3,49
ties of materials of interest prior to their synthesis.
Gas adsorption was calculated in these MOFs using grand
48
canonical Monte Carlo simulations using Universal force field
parameters for MOF atoms, TraPPE force field parameters for
Figure 4. Performance-property and property-property relations for 200 hypothetical zirconium-based MOFs color-coded by topology. a)
Deliverable capacity between 65 and 5.8 bar vs void fraction. b) Deliverable capacity between 65 and 5.8 bar vs volumetric surface area. c)
Volumetric surface area vs void fraction. Pink: fcu topology; green: ftw topology; blue: scu topology; red: csq topology.
C
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
This posed an interesting design question: Does it matter
which isomer is selected for synthesis. In previous work, some
based on linkers TPT, DPD, and PTP, which have similar
lengths. The structure with the terminal triple bonds, TPT,
outperforms the structure with the terminal phenyl rings, PTP.
As a matter of fact, we found that the fcu MOF based on the
TPT building block was predicted to have the highest
deliverable capacity between 65 and 5.8 bar among all of the
hypothetical zirconium-based MOFs explored here. The
predicted value is 197 cc(STP)/cc, which is among the best
values for predicted deliverable capacities at these operation
conditions among all types of MOFs. This material was, thus,
selected for synthesis, activation, and experimental testing.
Before proceeding to synthesis, however, we tried to
understand why the “terminal” triple-bond configuration was
beneficial for the cases illustrated in Figure 5. In Figure 6 we
5
7
of us determined that incorporating triple bonds into the
organic building blocks is more efficient than incorporating
phenyl rings to increase the surface area, which as shown in
Figure 4b correlates well with the performance of the MOF.
Our simulations here suggest that the position of the triple
bonds can also be important as apparent from comparing the
methane deliverable capacities among MOFs of an isoreticular
44
(
type II) isomeric set (Figure 5a,b). Figure 5a shows this
Figure 6. Simulation-averaged methane density maps at low (1 bar)
and high (65 bar) pressure. The horizontal color bar indicates the scale
from low density (black) to high density (orange). Regions
inaccessible to methane are not colored. a) Comparison between
ftw MOFs Py-4TP and Py-4PT. b) Comparison between fcu MOFs
PTP and TPT. For clarity, the core of the zirconium-based node is
delineated in blue.
Figure 5. Comparison of deliverable capacity between 65 and 5.8 bar
among: a) triads of isoreticular isomeric ftw MOFs formed from
tetratopic building blocks Y-4X (where Y = TPE, Por, Py, and P). b)
pairs of isoreticular isomeric MOFs formed from tetratopic building
blocks Y-4X (ftw MOFs with Y = TPE, Por, Py, and P) and ditopic
building blocks TPPT and PTTP (fcu MOFs). c) fcu MOFs built with
linkers of similar lengths, including the MOF with the highest
predicted deliverable capacity among all MOFs studied here (TPT).
show simulation-averaged methane density maps, in the low
pressure (1 bar) and high pressure (65 bar) regimes, comparing
the ftw MOFs Py-4TP and Py-4PT and the fcu MOFs TPT and
PTP. Notice in the density maps at low pressure in Figure 6
that the high-density red spots are located around the
zirconium-based nodes. This observation agrees with the
preferred adsorption site of methane in UiO-66 determined
comparison for triads of ftw isoreticular isomeric structures
such as the one illustrated in Figure 3. Notice that the bars
representing structures with triple-bond termination (both sets
of colored bars) are 10−20 cc(STP)/cc higher than the bars
representing the corresponding phenyl-terminated isomers
58
by neutron diffraction. In Figure 6a, at low pressure, the
difference between methane density around the nodes of ftw
MOFs Py-4PT and Py-4TP is not significant, but at 65 bar the
high density spots are larger and closer to the nodes of Py-4PT
than for Py-4TP. This indicates that at high pressure there is a
more efficient packing of methane around the zirconium-based
nodes in the triple-bond terminated Py-4PT. A similar
observation can be made in Figure 6b for the fcu MOFs PTP
and TPT, where the high density spots at 65 bar are larger and
extend closer to the nodes of TPT than for PTP. Thus, the
higher deliverable capacities of the structures with terminal
triple bonds are due to higher adsorption at the storage
pressure.
(
white bars).
An analogous trend can be seen in Figure 5b for ftw and fcu
isoreticular isomeric pairs, where the bars for the structure with
triple-bond termination are 10−38 cc(STP)/cc higher than
those for the corresponding isomer with phenyl ring
termination. For instance, the calculated deliverable capacity
of TPPT is 38 cc(STP)/cc higher than that of PTTP and that of
Py-4PT is 15 cc(STP)/cc higher than that of Py-4TP. We
should note that for the ftw topology, the position of the triple
bond also helps boost the geometric surface areas (see more
details in Figure S3.4). For instance, the hypothetical isomeric
ftw MOFs TPE-4TTP, TPE-4TPT, and TPE-4PTT show
Figure 7 shows the simulated isotherms for the MOFs
discussed in Figure 6. Notice how the isotherms of Py-4PT and
Py-4TP are similar at low pressure but start to diverge in favor
of Py-4PT when the pressure increases. A similar trend is
observed when comparing TPT and PTP.
MOF Synthesis. Based on the simulation results, we
proceeded to synthesize the zirconium-based fcu MOF based
on the TPT ligand, which features terminal triple bonds. We
2
gravimetric surface areas of 5950, 6300, and 6550 m /g,
respectively.
The positive effect of the terminal triple bond can also be
extended to other MOFs. Indeed, seeking to design MOFs that
may take advantage of this effect was a motivation to add the
ditopic TPT linker as an organic building block. Figure 5c
shows the deliverable capacity comparison between fcu MOFs
D
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
There was good agreement between the PXRD pattern of the
synthesized material and the simulated pattern of the NU-800
model (i.e., hypothetical fcu MOF TPT), confirming the
formation of the targeted structure (Figure 8). Nevertheless,
Figure 7. Simulated methane adsorption isotherms for the ftw
isoreticular isomeric MOFs Py-4TP (empty triangles) and Py-4PT
(filled triangles) and fcu MOFs PTP (empty circles) and TPT (filled
circles).
successfully synthesized the targeted MOF structure and refer
to it as NU-800. To the best of our knowledge, there are no
other MOFs to date based on the terminal triple-bond
5
9
configuration, with the exception of IRMOF-0, although
this MOF is based on a simple acetylene-dicarboxylic acid
ligand. We synthesized the organic ligand 3 (1,4-benzenedi-
propynoic acid) from the commercially available 1,4-benzene-
diacrylic acid (ligand 1) using a facile two-step procedure
illustrated in Scheme 1.
Scheme 1. Synthesis Procedure of Ligand 3 (TPT)
Figure 8. Top: Measured (purple) PXRD for NU-800 and simulated
PXRD for originally predicted (black) and reoptimized (blue) NU-800
model structures. Bottom: View of the structure of the fcu MOF NU-
8
00, with the purple spheres indicating the large octahedral cavities,
and the blue spheres indicating the small tetrahedral cavities.
the lattice constant of NU-800 was found to be 28.75 Å via
60
Pawley refinement, indicating a 3.5% underprediction of the
lattice constant by our structure generation procedure. With the
experimental value of the lattice constant, the NU-800 model
was reoptimized, and the resulting structure was used to
simulate complete adsorption isotherms for various gases.
Figure 8 also shows a view of the NU-800 structure,
illustrating the two cages characteristic of zirconium-based
MOFs with the fcu topological net. Accordingly, NU-800
features large octahedral cages (indicated by purple spheres)
formed by 12 linkers and six zirconium nodes, surrounded by
small tetrahedral cages (indicated by blue spheres) formed by
six linkers and four zirconium nodes. The largest spherical
probes that can fit in these pores are approximately 16.5 and 7.5
Å, respectively, as shown by the pore size distribution shown as
an inset in Figure 9.
We must note that application of synthesis protocols used for
other zirconium-based MOFs did not result in a crystalline
product when applied with ligand 3. Thus, screening of suitable
synthesis conditions was necessary. Based on the number of
conditions involving different solvents, modulators, reactant
concentrations, and reaction temperatures and times tested
until conditions leading to a pure crystalline material were
found (see details in the SI), it is worth noting again that, in the
search for the best materials for methane storage and delivery,
computational efforts are able to drastically reduce synthetic
efforts by suggesting the most promising structures.
We were able to activate around 90% of the porosity of NU-
800 as indicated by a comparison of the simulated and
measured nitrogen isotherms shown in Figure 9. Supercritical
61
CO activation was necessary to get as close as possible to the
2
ideal surface area and pore volume, which is critical to achieve
6
2
The best reaction conditions corresponded to a 120 h
solvothermal reaction of ligand 3 and ZrOCl ·H O in DMF
the best adsorption performance of a given MOF material.
2
The measured BET surface area was ∼3150 m /g, which
2
2
with formic acid as a modulator at 70 °C. The generality of the
procedure was suggested by the successful synthesis of UiO-67
corresponds to ∼89% of the simulated BET surface area of NU-
2
800 (∼3560 m /g), and similarly the activated pore volume was
(
1
based on 4,4′-dibenzoic acid) and NU-801 (based on ligand
1.34 cc/g, which corresponds to ∼92% of the simulated pore
) also using the reaction conditions above.
volume (1.44 cc/g).
E
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
volumetric hydrogen adsorption capacity is 44.4 g/L, which is
among the highest reported at these conditions (i.e., 35 g/L for
7
63
63
PCN-68, 36 g/L for MOF-200, 41 g/L for MOF-210, 43
33
49
g/L for NU-1100, 47 g/L for NU-100, 49 g/L for NU-
11 and MOF-177 ). We note that among these MOFs only
14
64
1
NU-1100 is based on highly stable zirconium building blocks.
The high-pressure adsorption of carbon dioxide in NU-800 is
also significant, corresponding to 26 mmol/g at 35 bar. These
results clearly show that structurally stable zirconium-based
MOFs can compete with the copper paddle-wheel MOFs that
have large pore volumes and surface areas but are structurally
27
weaker.
The room temperature methane uptake properties of NU-
00 are shown in Figure 10 (and Figure 12) in gravimetric (and
8
volumetric) units. The methane deliverable capacity between
5 and 5.8 bar was measured as 167 cc(STP)/cc, which is
10% lower than the simulated value of 187 cc(STP)/cc. We
6
∼
Figure 9. Simulated (blue) and measured (purple) nitrogen isotherms
at 77 K for NU-800. The inset shows the geometrically obtained pore
size distribution.
attribute this discrepancy to incomplete activation of our
sample. With a different activation procedure it is quite possible
that the methane uptake values of NU-800 could come closer
to the simulated ones. We note, however, that even this
nonoptimized deliverable capacity of the zirconium-based MOF
Gas Adsorption Measurements. In order to validate the
predicted adsorption characteristics of NU-800, we performed
isotherm measurements over a large pressure and temperature
range for methane. Additionally, we performed hydrogen
adsorption measurements, which are also relevant for vehicular
applications. Figure 10 shows the comparison between the
14
NU-800 is equal to or better than Ni-MOF-74 (121
1
4
14
cc(STP)/cc), PCN-14 (149 cc(STP)/cc), and UTSA-20
162 cc(STP)/cc), and it is only 15% lower than the current
(
38
best MOF-519 (203 cc(STP)/cc) and 10% lower than the
other top MOFs UTSA-76, HKUST-1, and MOF-5
∼185 cc(STP)/cc). In addition, the measured NU-800
39
15
17
(
gravimetric deliverable capacity of 0.216 g/g is higher than
that of the MOFs mentioned above: MOF-519 (0.149 g/g),
UTSA-76 (0.191 g/g), HKUST-1 (0.150 g/g), MOF-5 (0.213
g/g), UTSA-20 (0.128 g/g), PCN-14 (0.129 g/g), and Ni-
MOF-74 (0.075 g/g). It should be noted that MOF-519 has
been reported focusing on storage at higher pressures, with a
deliverable capacity between 80 and 5.8 bar of 220 cc(STP)/cc.
The simulated value for NU-800 at those operation conditions
is 203 cc(STP)/cc.
In order to obtain additional insight about the guest−host
interactions in NU-800, we determined the heats of adsorption
(
Q ) for methane and hydrogen from the temperature
st
dependent experimental isotherms in Figure 10 using the
Clausius−Clapeyron equation (see the SI for details). Figure 11
shows that the agreement between the simulated and measured
Qst values is quite good for hydrogen and methane. For
hydrogen, the Q initially remains roughly constant at ∼5 kJ/
Figure 10. Measured (blue lines with filled circles) and simulated
absolute adsorption isotherms (dashed red lines) for methane and
hydrogen at several temperatures.
st
mol and then slowly decreases down to 3 kJ/mol with
simulated and measured isotherms. Overall there is good
agreement between the simulated and experimental results,
with some discrepancies at high pressure partly due to
somewhat incomplete activation of the NU-800 sample. The
methane saturation loading at 125 K was measured as 34.3
mmol/g, which corresponds to a pore volume of 1.37 cc/g, in
good agreement with the pore volume determined by nitrogen
adsorption. A similar pore volume (1.30 cc/g) is also
determined by carbon dioxide adsorption at 220 K (Figure
S7.1).
The excess adsorption isotherm of hydrogen at 77 K (Figure
S7.3) shows a very shallow maximum at ∼35 bar, with a nearly
flat region between 20 and 65 bar. At 65 bar and 77 K, the
measured gravimetric hydrogen adsorption is 0.08 g/g. The
Figure 11. Experimental (black lines from a spline and red lines from a
virial fit) and simulated (scattered points) heats of adsorption.
F
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
increasing loading. For methane, the initial Q is relatively low
with values around 11 kJ/mol (simulated) and 12.5 kJ/mol
over a large number of gas adsorption cycles indicated no
sample degradation, consistent with the highly stable structures
of zirconium-based MOFs. Hydrogen adsorption in NU-800
was also high with uptake values of 0.08 g/g and 44.4 g/L at 65
st
(
measured). A material with high methane deliverable capacity
and a relatively low Q may facilitate thermal management in
st
ANG on-board systems and may also make the material less
prone to “poisoning” by hydrocarbon impurities in natural gas
during adsorption/desorption cycles.
Figure 12 illustrates the stability of NU-800 during
adsorption/desorption cycles. No evidence of sample degrada-
bar and 77 K. Similarly the CO uptake at 35 bar was quite high
2
(26 mmol/g). These promising gas uptake characteristics along
with its expected superior stability make NU-800 a promising
material for gas adsorption applications.
ASSOCIATED CONTENT
Supporting Information
■
*
S
General simulation, synthesis, and measurement methods.
Simulation models. Details of MOF construction. Snapshots
of top hypothetical MOFs based on various metrics. Details of
NU-800 synthesis and activation. Nitrogen sorption and PXRD
data for NU-800, NU-801, and UiO-67. Detailed simulated
and measured CH , H , and CO adsorption isotherms for NU-
4
2
2
8
00. Heats of adsorption. Excel file with tabulated screening
data. Compressed file with 204 cif files for hypothetical
zirconium MOFs (including NU-800 molecular model). This
material is available free of charge via the Internet at http://
pubs.acs.org.
AUTHOR INFORMATION
Corresponding Authors
E-mail: j-hupp@northwestern.edu.
E-mail: taner@seas.upenn.edu.
E-mail: o-farha@northwestern.edu.
E-mail: snurr@northwestern.edu.
■
Figure 12. Experimental absolute methane adsorption isotherms for
NU-800 at 298 K over a dozen adsorption/desorption cycles. Inset
shows adsorption at 65 and 35 bar. The volumetric deliverable capacity
of NU-800 between 65 and 5.8 bar is indicated by a vertical arrow.
*
*
*
*
Author Contributions
tion was found over a dozen cycles. The variation of the
adsorption values at 65 bar and 298 K is within the 2%
experimental error. Furthermore, we checked the pore volume
and surface area of the sample after a large number of isotherm
measurements were done with H , CO , and CH and did not
The manuscript was written through contributions of all
authors. All authors have given approval to the final version of
the manuscript. D.A.G.-G. and O.V.G. contributed equally.
Notes
2
2
4
The authors declare the following competing financial
interest(s): Omar Farha, Joseph Hupp, and Randall Snurr
have a financial interest in the start-up company NuMat
Technologies, which is seeking to commercialize metal-organic
frameworks.
see any decrease in pore volume or surface area (Figure S6.5).
CONCLUSIONS
■
We have computationally constructed and screened over 200
hypothetical MOFs based on highly stable zirconium building
blocks. Our simulations suggest that alkyne groups adjacent to
the zirconium corner are more beneficial for methane storage
and delivery than phenyl rings, because they allow an efficient
packing of methane around the zirconium corner at high
pressure. It follows that this beneficial effect might be applicable
to other MOFs where alkyne groups allow more efficient
packing around regions where methane/MOF interactions are
strong. Among the hypothetical zirconium-based materials
studied, the MOF with the fcu topological net based on the
ditopic ligand 3 1,4-benzenedipropynoic acid was predicted to
have the best methane deliverable capacity. The MOF NU-800
based on this ligand was successfully synthesized and
characterized. The experimental structure was found to be in
good agreement with the preliminary NU-800 computer
model, as indicated by PXRD measurements. Upon super-
ACKNOWLEDGMENTS
■
The Northwestern University team acknowledges that the work
presented herein was funded by the Advanced Research
Projects Agency-Energy (ARPA-E), U.S. Department of
Energy, under Award Number DE-AR0000248. T.Y. acknowl-
edges support by the U.S. Department of Energy through BES
Grant No. DE-FG02-08ER46522. The computations were
made possible by the high performance computing system,
QUEST, at Northwestern University, and the National Energy
Research Scientific Computing Center’s Carver Cluster.
REFERENCES
■
(1) EIA U.S Crude Oil and Natural Gas Proved Reserves, 2011. http://
www.eia.gov/naturalgas/crudeoilreserves/ (accessed Sept 13, 2014).
(2) Wegrzyn, J.; Gurevich, M. Appl. Energy 1996, 55, 71.
critical CO activation, gas adsorption measurements confirmed
2
(3) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; O’Keeffe,
the predicted high capacity methane adsorption properties of
NU-800. In particular, the volumetric and gravimetric methane
deliverable capacities (between 65 and 5.8 bar) were measured
as 167 cc(STP)/cc (simulation: 187 cc(STP)/cc) and 0.215 g/
g (simulation: 0.245 g/g), respectively. We determined that
NU-800 has a relatively low methane heat of adsorption (12.5
kJ/mol measured and 11.6 kJ/mol simulated). Stability tests
M.; Yaghi, O. M. Science 2002, 295, 469.
4) Ma, S.; Sun, D.; Simmons, J. M.; Collier, C. D.; Yuan, D.; Zhou,
H.-C. J. Am. Chem. Soc. 2007, 130, 1012.
(
(5) Wu, H.; Zhou, W.; Yildirim, T. J. Am. Chem. Soc. 2009, 131, 4995.
(6) Furukawa, H.; Yaghi, O. M. J. Am. Chem. Soc. 2009, 131, 8875.
(7) Yuan, D.; Zhao, D.; Sun, D.; Zhou, H.-C. Angew. Chem., Int. Ed.
2010, 49, 5357.
G
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX

Chemistry of Materials
Article
(
8) Klein, N.; Senkovska, I.; Baburin, I. A.; Gru
̈
nker, R.; Stoeck, U.;
(38) Gan
Soc. 2014, 136, 5271.
́
dara, F.; Furukawa, H.; Lee, S.; Yaghi, O. M. J. Am. Chem.
Schlichtenmayer, M.; Streppel, B.; Mueller, U.; Leoni, S.; Hirscher, M.;
Kaskel, S. Chem. − Eur. J. 2011, 17, 13007.
(39) Li, B.; Wen, H.-M.; Wang, H.; Wu, H.; Tyagi, M.; Yildirim, T.;
Zhou, W.; Chen, B. J. Am. Chem. Soc. 2014, 136, 6207.
(40) Decoste, J. B.; Peterson, G. W.; Smith, M. W.; Stone, C. A.;
Willis, C. R. J. Am. Chem. Soc. 2012, 134, 1486.
(41) Makal, T. A.; Wang, X.; Zhou, H.-C. Cryst. Growth Des. 2013,
13, 4760.
(42) Duren, T.; Sarkisov, L.; Yaghi, O. M.; Snurr, R. Q. Langmuir
̈
2004, 20, 2683.
(43) Simon, C. M.; Kim, J.; Lin, L.-C.; Martin, R. L.; Haranczyk, M.;
Smit, B. Phys. Chem. Chem. Phys. 2014, 16, 5499.
(
9) Mendoza-Cortes, J. L.; Pascal, T. A.; Goddard, W. A. J. Phys.
Chem. A 2011, 115, 13852.
10) Stoeck, U.; Krause, S.; Bon, V.; Senkovska, I.; Kaskel, S. Chem.
Commun. 2012, 48, 10841.
11) Wilmer, C. E.; Leaf, M.; Lee, C. Y.; Farha, O. K.; Hauser, B. G.;
Hupp, J. T.; Snurr, R. Q. Nat. Chem. 2012, 4, 83.
12) Duan, X.; Yu, J.; Cai, J.; He, Y.; Wu, C.; Zhou, W.; Yildirim, T.;
Zhang, Z.; Xiang, S.; O’Keeffe, M.; Chen, B.; Qian, G. Chem. Commun.
013, 49, 2043.
13) Wilmer, C. E.; Farha, O. K.; Yildirim, T.; Eryazici, I.;
(
(
(
2
(
́ ́
(44) Gomez-Gualdron, D. A.; Wilmer, C. E.; Farha, O. K.; Hupp, J.
Krungleviciute, V.; Sarjeant, A. A.; Snurr, R. Q.; Hupp, J. T. Energy
Environ. Sci. 2013, 6, 1158.
T.; Snurr, R. Q. J. Phys. Chem. C 2014, 118, 6941.
(45) Rana, M. K.; Koh, H. S.; Zuberi, H.; Siegel, D. J. J. Phys. Chem. C
2014, 118, 2929.
(
14) Peng, Y.; Srinivas, G.; Wilmer, C. E.; Eryazici, I.; Snurr, R. Q.;
Hupp, J. T.; Yildirim, T.; Farha, O. K. Chem. Commun. 2013, 49, 2992.
15) Peng, Y.; Krungleviciute, V.; Eryazici, I.; Hupp, J. T.; Farha, O.
K.; Yildirim, T. J. Am. Chem. Soc. 2013, 135, 11887.
16) He, Y.; Zhou, W.; Yildirim, T.; Chen, B. Energy Environ. Sci.
013, 6, 2735.
(
(
46) Bureekaew, S.; Schmid, R. CrystEngComm 2013, 15, 1551.
47) O’Keeffe, M.; Peskov, M. A.; Ramsden, S. J.; Yaghi, O. M. Acc.
(
Chem. Res. 2008, 41, 1782.
48) Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A.;
Skiff, W. M. J. Am. Chem. Soc. 1992, 114, 10024.
49) Farha, O. K.; Yazaydin, A. O.; Eryazici, I.; Malliakas, C. D.;
(
(
2
(
(
17) Mason, J. A.; Veenstra, M.; Long, J. R. Chem. Sci. 2014, 5, 32.
Hauser, B. G.; Kanatzidis, M. G.; Nguyen, S. T.; Snurr, R. Q.; Hupp, J.
T. Nat. Chem. 2010, 2, 944.
(
18) Barin, G.; Krungleviciute, V.; Gomez-Gualdron, D. A.; Sarjeant,
A. A.; Snurr, R. Q.; Hupp, J. T.; Yildirim, T.; Farha, O. K. Chem. Mater.
014, 26, 1912.
19) ARPA-E Methane Opportunities for Vehicular Energy(MOVE),
012. http://arpa-e-foa.energy.gov(DE-FOA-000672) (accessed Sept
3, 2014).
(
(
(
50) Martin, M. G.; Siepmann, J. I. J. Phys. Chem. B 1998, 102, 2569.
51) Potoff, J. J.; Siepmann, J. I. AIChE J. 2001, 47, 1676.
52) Levesque, D.; Gicquel, A.; Darkrim, F. L.; Kayiran, S. B. J. Phys.:
2
(
2
1
Condens. Matter 2002, 14, 9285.
53) Nicholson, D.; Parsonage, N. G. Computer simulation and the
statistical mechanics of adsorption; Academic Press: London, 1982.
(
(
20) O’Keeffe, M.; Yaghi, O. M. Chem. Rev. 2011, 112, 675.
(
21) Martin, R. L.; Lin, L.-C.; Jariwala, K.; Smit, B.; Haranczyk, M. J.
̈
(
54) Bae, Y.-S.; Yazaydın, A. O.; Snurr, R. Q. Langmuir 2010, 26,
Phys. Chem. C 2013, 117, 12159.
22) Bon, V.; Senkovskyy, V.; Senkovska, I.; Kaskel, S. Chem.
Commun. 2012, 48, 8407.
23) Feng, D.; Chung, W.-C.; Wei, Z.; Gu, Z.-Y.; Jiang, H.-L.; Chen,
Y.-P.; Darensbourg, D. J.; Zhou, H.-C. J. Am. Chem. Soc. 2013, 135,
7105.
24) Fer
J.; Surble,
5
475.
(
(55) Talu, O.; Myers, A. L. AIChE J. 2001, 47, 1160.
(56) Gelb, L. D.; Gubbins, K. E. Langmuir 1998, 15, 305.
(
(57) Farha, O. K.; Wilmer, C. E.; Eryazici, I.; Hauser, B. G.; Parilla, P.
A.; O’Neill, K.; Sarjeant, A. A.; Nguyen, S. T.; Snurr, R. Q.; Hupp, J. T.
J. Am. Chem. Soc. 2012, 134, 9860.
1
(
́
ey, G.; Mellot-Draznieks, C.; Serre, C.; Millange, F.; Dutour,
S.; Margiolaki, I. Science 2005, 309, 2040.
(
58) Wu, H.; Chua, Y. S.; Krungleviciute, V.; Tyagi, M.; Chen, P.;
Yildirim, T.; Zhou, W. J. Am. Chem. Soc. 2013, 135, 10525.
59) Tranchemontagne, D. J.; Hunt, J. R.; Yaghi, O. M. Tetrahedron
008, 64, 8553.
́
(
̂ ́
25) Park, K. S.; Ni, Z.; Cote, A. P.; Choi, J. Y.; Huang, R.; Uribe-
(
2
Romo, F. J.; Chae, H. K.; O’Keeffe, M.; Yaghi, O. M. Proc. Natl. Acad.
Sci. U. S. A. 2006, 103, 10186.
(60) Pawley, G. J. Appl. Crystallogr. 1981, 14, 357.
(
26) Colombo, V.; Galli, S.; Choi, H. J.; Han, G. D.; Maspero, A.;
(61) Nelson, A. P.; Farha, O. K.; Mulfort, K. L.; Hupp, J. T. J. Am.
Palmisano, G.; Masciocchi, N.; Long, J. R. Chem. Sci. 2011, 2, 1311.
Chem. Soc. 2008, 131, 458.
62) Chowdhury, P.; Bikkina, C.; Meister, D.; Dreisbach, F.; Gumma,
S. Microporous Mesoporous Mater. 2009, 117, 406.
63) Furukawa, H.; Ko, N.; Go, Y. B.; Aratani, N.; Choi, S. B.; Choi,
(
(
27) Wu, H.; Yildirim, T.; Zhou, W. J. Phys. Chem. Lett. 2013, 4, 925.
28) Cavka, J. H.; Jakobsen, S.; Olsbye, U.; Guillou, N.; Lamberti, C.;
(
Bordiga, S.; Lillerud, K. P. J. Am. Chem. Soc. 2008, 130, 13850.
29) Morris, W.; Volosskiy, B.; Demir, S.; Gandara, F.; McGrier, P.
L.; Furukawa, H.; Cascio, D.; Stoddart, J. F.; Yaghi, O. M. Inorg. Chem.
012, 51, 6443.
30) DeCoste, J. B.; Peterson, G. W.; Schindler, B. J.; Killops, K. L.;
Browe, M. A.; Mahle, J. J. J. Mater. Chem. A 2013, 1, 11922.
31) Zhang, W.; Huang, H.; Liu, D.; Yang, Q.; Xiao, Y.; Ma, Q.;
Zhong, C. Microporous Mesoporous Mater. 2013, 171, 118.
32) Mondloch, J. E.; Bury, W.; Fairen-Jimenez, D.; Kwon, S.;
(
(
́
̈
E.; Yazaydin, A. O.; Snurr, R. Q.; O’Keeffe, M.; Kim, J.; Yaghi, O. M.
Science 2010, 329, 424.
2
(
(64) Furukawa, H.; Miller, M. A.; Yaghi, O. M. J. Mater. Chem. 2007,
1
7, 3197.
(
(
DeMarco, E. J.; Weston, M. H.; Sarjeant, A. A.; Nguyen, S. T.; Stair, P.
C.; Snurr, R. Q.; Farha, O. K.; Hupp, J. T. J. Am. Chem. Soc. 2013, 135,
10294.
(
33) Gutov, O. V.; Bury, W.; Gomez-Gualdron, D. A.; Krungleviciute,
V.; Fairen-Jimenez, D.; Sarjeant, A. A.; Snurr, R. Q.; Hupp, J. T.;
Yildirim, T.; Farha, O. K. Chem. − Eur. J. 2014, DOI: 10.1002/
chem.201402895.
(
34) Feng, D.; Gu, Z.-Y.; Li, J.-R.; Jiang, H.-L.; Wei, Z.; Zhou, H.-C.
Angew. Chem., Int. Ed. 2012, 51, 10307.
35) Jiang, H.-L.; Feng, D.; Wang, K.; Gu, Z.-Y.; Wei, Z.; Chen, Y.-P.;
Zhou, H.-C. J. Am. Chem. Soc. 2013, 135, 13934.
36) Mondloch, J. E.; Katz, M. J.; Planas, N.; Semrouni, D.; Gagliardi,
L.; Hupp, J. T.; Farha, O. K. Chem. Commun. 2014, 50, 8944.
37) Nickerl, G.; Leistner, M.; Helten, S.; Bon, V.; Senkovska, I.;
Kaskel, S. Inorg. Chem. Front. 2014, 1, 325.
(
(
(
H
dx.doi.org/10.1021/cm502304e | Chem. Mater. XXXX, XXX, XXX−XXX