go framework experimental synthesis
TRANSCRIPT
Supporting Information
� Wiley-VCH 2010
69451 Weinheim, Germany
Graphene Oxide Framework Materials: Theoretical Predictions andExperimental Results**Jacob W. Burress, Srinivas Gadipelli, Jamie Ford, Jason M. Simmons, Wei Zhou, andTaner Yildirim*
anie_201003328_sm_miscellaneous_information.pdf
Disclaimer: Certain commercial suppliers are identified in this paper to foster understanding. Such identification does not
imply recommendation or endorsement by the National Institute of Standards and Technology, nor does it imply that the
materials or equipment identified are necessarily the best available for the purpose.
Materials:
All chemicals are obtained from commercial sources (Sigma-Aldrich, Alfa Aesar, etc.) and are used without further
purification.
Graphene Oxide
Synthesis:
Graphene oxide was prepared with a modified Hummer’s method. 10 g of graphite powder (synthetic, 1-2 micron) is stirred
with cold concentrated sulfuric acid (230 mL at 0 °C) in an ice bath. Then, potassium permanganate (30 g) is added slowly to
the suspension to prevent a rapid rise in temperature (less than 20 °C). The reaction mixture is then cooled to 2 °C. After
removal of the ice-bath, the mixture is stirred at room temperature for 15 min. Distilled water (230 mL) is slowly added to the
reaction vessel to keep the temperature under 98 °C. The diluted suspension is stirred for an additional 15 minutes and further
diluted with distilled water (1.4 L), before adding hydrogen peroxide (100 mL). The mixture is left overnight. Graphene oxide
particles are separated from the excess liquid by centrifugation followed by decantation. The graphene oxide is washed several
times with water (roughly 14 L) before it is then freeze-dried to remove water before being used in the GOF synthesis.
Graphene Oxide Framework Materials
COF-like synthesis:
In a nitrogen environment, we suspended 200 mg GO and 165.75 mg benzene-1,4-diboronic acid (B14DBA) in a 10 ml:10 ml
mixture of mesitylene and dioxane in a 25 ml teflon-lined autoclave. The mixture was sonicated for about 15 min, after which
the autoclave is closed and put in a furnace at 85 °C for four days. The product was filtered, washed with acetone and air dried
before outgassing and H2 storage measurements.
Figure S1. PXRD results for COF-like synthesis. The second peak in the “GOF” sample is an impurity phase likely due to
COF-1.
Methanol solvothermal synthesis and the effect of other solvents:
Approximately 100 mg of GO was mixed in 10 mL of methanol with 10 mg to 200 mg of B14DBA. Control samples using
only GO (GO-control) and only B14DBA (B14DBA-control) were also prepared. The solutions were placed on a roller mixer
until all the B14DBA dissolved. The solution was kept in an 85 °C to 90 °C oven for several days. The solution was shaken
every few hours as the GO settled. The samples were then allowed to cool and placed in a centrifuge to get all the GOF to
settle. The excess methanol was removed using a syringe. The sample was washed repeatedly with more methanol in order to
remove unreacted B14DBA. The resulting GOFs were tested by PXRD and outgassed at 125 °C for 24 hours before being
analyzed for gas storage. As discussed below there is an optimum ratio of linker to GO that results in the maximum hydrogen
uptake around the 100 mg B14DBA concentration, i.e. 1:1 initial mass of linker to GO. This concentration leads to significant
expansion of the interlayer spacing, enabling access to hydrogen, while maintaining sufficiently large spacing between ligands
so as not to fill much of the newly generated pore volume between the GO layers.
5 10 15 20 25 30 35
0
200
400
600
800
Inte
nsi
ty (
cou
nts
per
sec
on
d)
2 (°)
Freeze-dried GO
GOF13
Figure S2. PXRD plots of freeze-dried GO and GOF (1:1 mass ratio) prepared by methanol solvothermal synthesis. We note
that there are no COF-1 or graphite peaks in the pattern, indicating GOF materials are single phase. A Lorentzian was fit to
each peak and the FWHM was used with the Scherrer formula (see below for details) to determine the average number of
layers. We estimated the lower bound for the number of graphene layers in GO and GOF as 12 and 11 layers, respectively.
It is possible to estimate the c-axis correlation length (i.e. number of ordered graphene layers) from the x-ray spectra shown in
Fig. S2 using the Scherrer formula:[1,2]
where K is the shape factor, λ is the x-ray wavelength, β is the line broadening at half the maximum intensity (FWHM) in
radians, and θ is the Bragg angle; τ is the mean size of the ordered (crystalline) domains, which may be smaller or equal to the
grain size. The dimensionless shape factor has a typical value of about 0.9, but varies with the actual shape of the crystallite.
The Scherrer equation is limited to nano-scale particles. It is not applicable to grains larger than about 0.1 μm. It is important to
realize that the Scherrer formula provides a lower bound on the particle size. The reason for this is that a variety of factors can
contribute to the width of a diffraction peak; besides particle size, the most important of these are usually inhomogeneous
strain and instrumental effects. If all of these other contributions to the peak width were zero, then the peak width would be
determined solely by the particle size and the Scherrer formula would apply. If the other contributions to the width are non-
zero, then the particle size can be larger than that predicted by the Scherrer formula, with the "extra" peak width coming from
the other factors.
(Black color)
(Black color)
(Brown color)
Figure S3. (a) Low pressure hydrogen sorption as a function of temperature on a GOF sample synthesized with 1:1 linker to
GO ratio in methanol. (b) The best hydrogen adsorption on GOFs synthesized using different solvents (methanol/MeOH,
ethanol/EtOH and N,N-dimethylmethanamide/DMF), indicating that methanol is the best solvent. GO-control samples for each
solvent are also shown.
The effect of initial linker to GO ratio on the interlayer spacing, BET surface area and gas adsorption properties of GOFs:
We have studied interlayer spacing, BET surface area and hydrogen uptake capacity of a large number of GOF matarials using
different linker:GO concentration in the synthesis.
Figure S4. XRD patterns of outgassed (at 125 °C for 24 hour) GOFs with different levels of linker loading. Note the
successive increase in the interlayer distance of graphene oxide layers with increasing linker concentration up to a certain level
(i.e. 1:1), above which there is little change in interlayer distance. For GOFs with low-linker concentration, the outgassing
broadens the peaks (compare to Figure 3) and moves it to a value close to bare GO. This suggests the low-level linker GOFs
are less stable compared to high-level linker GOFs. The 125 °C for 24 hour heating of the samples may partially reduce/de-
oxygenate GO and affect the interlayer spacing.
Figure S5. (a) Nitrogen sorption isotherms of GO precursor with a BET specific surface area (SSA) of 16 m2/g and GOFs
which have BET SSA’s of 74 m2/g, 138 m
2/g, 280 m
2/g, 470 m
2/g and 275 m
2/g for the initial GO to linker weight ratios of
1:0.1, 1:0.25, 1:0.5, 1:1, and 1:2, respectively. (b) The corresponding hydrogen adsorption isotherms. Given that the GO alone
has almost no usable surface area (< 20 m2/g), there is essentially no H2 adsorption. However, it is clear that the addition of the
diboronic acid linker progressively increases the accessible surface area and thereby increases the H2 uptake. Eventually,
increased linker concentrations reduce the accessible surface area due to the filling up of the interlayer space, which also
reduces the H2 uptake. The BET area of nearly 500 m2/g obtained for the 1:1 initial mass loading is considerably higher from
the graphene oxide sheets.
Figure S6. Comparison of H2 uptake (wt.%) vs BET of GOFs with generalized H2 uptake in wt.% vs BET (red dotted line) at
1 atm and 77 K. Inset shows the maximum H2 uptake at 77 K against BET. Our GOFs are showing significantly more H2
uptake for a given BET due to large isosteric heat of adsorption. At 77 K and 1 atm, high surface area activated carbon
materials and all other porous materials typically achieve about 1 wt.% for every 1000 m2/g.
[3] GOFs exceed this trend, having
as much as twice the H2 uptake for a given SSA. At high pressure, the excess hydrogen uptake is generally linearly dependent
on SSA, with ≈1 wt.% for every 500 m2/g for carbons and MOFs that depend only on van der Waals binding.
[4] Again, the
maximum uptake of H2 in our GOFs exceeds this generalized BET vs H2max
trend.
Other Characterization Measurements of Graphene Oxide Framework Materials
Thermogravimetric Analysis (TGA):
TGA spectra were collected on a TA Instruments Q600 SDT using alumina pans at 2 °C/min under flowing N2. TGA plots of
air-dried GO precursor, cross-linker and one of the GOFs with saturated cross-linker are shown in Figure S8, which clearly
indicates the relative structural stability of the GOF compounds. The exfoliation temperature is roughly increased by 100 °C in
GOFs, implying stronger bonds between B14DBA and the GO sheets than between the GO sheets themselves. After
exfoliation, GOFs do not lose as much mass as the GO control during carbonization because the linker remains associated with
the sample. TGA curves of GO and GOF samples show significant weight loss (10-15 wt %) before 100 °C, likely due to
evaporation of trapped water in GO and trapped solvent (e.g. methanol) in GOFs, respectively. The second portion of large
weight loss (≈25 wt %) is attributed to the removal of labile oxygen functional groups.
0 100 200 300 40050
60
70
80
90
100 B14DBA
GO
GOF
Wei
gh
t p
erce
nt
(%)
Temperature (°C)
Figure S7. Thermogravimetric analsysis of B14DBA, GO, and GOF. The exfoliation temperature increased from GO to GOF.
Also, the dehydration drop for the B14DBA at ≈200 °C is not seen in the GOF. This implies that the B14DBA had already
reacted with the surface functionalties of the graphene oxide.
Fourier Transform Infrared Spectroscopy (FT-IR):
The FT-IR spectra of B14DBA, COF-1, GOF, and GO are shown in Figure S9. The characteristic phenyl C=C stretch at 1522
cm-1
and the B-O deformation at 675 cm-1
of B14DBA are observed in both COF-1 and GOF, indicating successful
incorporation into the networks. The COF-1 spectrum also contains a new peak at 690 cm-1
corresponding to the out-of-plane
deformation of a B3O3 ring system. This peak is very weak in the GOF spectrum, implying that the formation of a planar
boroxine ring network between the graphene oxide sheets is a minor impurity phase. In addition, we observe a strong
absorbance band between 1250 and 1400 cm-1
in GOF. We attribute this to the variety of B-O stretches formed between
B14DBA and a myriad of functional C-O groups on the graphene oxide surfaces.[5,6]
Figure S8. FTIR spectra of Graphene Oxide, GOF, COF-1, and B14DBA.
Neutron Prompt Gamma Activation Analysis (PGAA):
PGAA was used to determine the elemental concentrations in the GO and GOF samples. PGAA is a non-destructive, bulk
sampling technique in which incident neutrons are absorbed by the nucleus of an element, afterwhich the element decays by
emitting gamma rays with characteristic energies.[7]
When properly calibrated by running standard samples, the intensity of the
gamma ray emission provides a quantitative measure of the amount of a given element in the neutron beam. PGAA is
particularly sensitive to the amount of boron in the sample, but is insensitive to the presence of oxygen. For each sample, c.
100 mg is sealed in a teflon bag and suspended in a vacuum chamber in the neutron beam.
0 1000 2000 30001E-3
0.01
0.1
1
10
100
1000
GOF 1:4
GOF 1:2
GOF 1:1
Inte
nsi
ty (
a.u
.)
Energy (keV)
(a)
1255 1260 1265 1270
0.08
0.10
0.12
0.14 GOF 1:4
GOF 1:2
GOF 1:1
Inte
nsi
ty (
a.u
.)
Energy (keV)
(b)
(a) (b)
(c) (d)
0 1000 2000 30001E-3
0.01
0.1
1
10
100
1000
GOF 1:4
GOF 1:2
GOF 1:1
Inte
nsi
ty (
a.u
.)
Energy (keV)
(a)
1255 1260 1265 1270
0.08
0.10
0.12
0.14 GOF 1:4
GOF 1:2
GOF 1:1
Inte
nsi
ty (
a.u
.)
Energy (keV)
(b)
(a) (b)
(c) (d)
Figure S9. (a) Survey spectra showing the wealth of information provided by PGAA. Samples are labeled by the intial mass
ratios of linker (B14DBA) to GO. In addition to carbon hydrogen and boron, minor impurities (less than one impurity per
2000 C atoms) such as manganese, nitrogen, sodium, potassium and sulfur are also detected, each coming the initial GO
synthesis. (b) and (c) Spectra are normalized to the 1261 keV carbon emission, such that the relative amount of boron in each
sample is clearly seen in (c). (d) Quantitative determination of the final B14DBA loading in each GOF, expressed as the
number of graphene carbons per linker. This number is calculated using the relative ratios of the boron and carbon intensities,
after subtracting the carbon from the teflon (CF2) bag.
X-ray photoelectron spectroscopy (XPS):
X-ray photoelectron spectra (XPS) were collected on a Kratos AXIS Ultra DLD spectrometer equipped with a monochromatic
Al source. All samples were thoroughly degassed (75 oC for GO, 100
oC for the GOFs) and stored in an inert glove box prior
to measurement, but were briefly exposed to air for introduction into the XPS system. Spectra were obtained at three different
locations for each sample to test the uniformity of the samples. Atomic concentrations were calculated from survey spectra,
collected over a wide binding-energy range, with a 0.5 eV step size, a pass energy of 160 eV and a 100 ms dwell time. High
resolution oxygen and carbon core level spectra (O 1s and C 1s, respectively) were acquired with a 20 eV pass energy and
0.1 eV steps. All energies were normalized to the C 1s transition of the graphene host, present at 285.0 eV.[8]
Figure S10. XPS core level spectra of GO and GOFs with three different linker concentration. The intensities are normalized
to graphite peak at 285.0 eV.
Sample C (wt.%) O(wt.%) GO
stoichiometry
GO* 62.7 ± 1.1 35.6 ± 0.6 CO0.43
0.25DBA:GO 70.9 ± 0.5 28.6 ± 0.4 CO0.30
0.5DBA:GO 68.1 ± 0.4 30.4 ± 0.3 CO0.34
1.0DBA:GO 67.0 ± 0.2 31.2 ± 0.2 CO0.35
Table S1. The carbon/oxygen ratios for GO and GOFs as obtained from XPS shown in Fig. S9. * Notes: Error bars represent
standard deviation from three locations on the sample.
Detailed chemical information is present in the respective core levels for each sample, shown in Figure S10. XPS provides the
oxygen stoichiometry that is not available from the PGAA measurements, but is less sensitive to the boron concentration than
PGAA. Thus we have used XPS to determine the O/C ratios and PGAA for the B/C ratios. The carbon and oxygen
composition for each sample is presented in table S1. The C 1s core level spectra for the starting GO is dominated by roughly
equal amounts of graphene-like carbons, represented by the peak at 285.0 eV, and a feature at 287.0 eV which is consistent
with singly-bonded C-O chemistries, accompanied by a small feature near 289.0 eV corresponding to carboxylic groups on the
edges of the graphene sheet.[9]
The position of this C-O peak is weighted towards that seen for epoxides and -hydroxy
functionalities (that is hydroxyl species on adjacent carbon atoms), indicating relatively little isolated hydroxyl (C-O-H)
species. The conclusion that epoxides/ -hydroxyls are the predominant carbon-oxygen functionalities is supported by the
532.8 eV O 1s peak position that is correlated with and epoxide ring.[10]
Core level spectra for three different GOFs are also
shown in Figure S10. From this figure, a partial reduction of the GO sheet is apparent as a decrease in the overall oxygen
intensity and by a decrease in the epoxide-like C 1s core level at 287.0 eV, from 0.4 oxygens per carbon (O/C) in the GO to 0.3
O/C in the 0.25DBA:GO GOF. With increasing DBA addition, the level of GO reduction decreases (XPS intensity increases),
suggesting that the solvothermal synthesis initially reduces some of the GO but that the addition of DBA inhibits the further
reduction by binding the oxygens in more energetically favorable environments. For the 1:1 GOF, the stoichiometry of the
graphene oxide sheet recovers to 0.35 O/C. It is also interesting to note that GOF formation does not appreciably affect the
concentration of carboxylic species which are commonly assigned to the edges of the graphene planes, supporting our
conclusion that the dominant chemistry is being performed near the center of the graphene oxide sheets rather than at the edges
and generating the pillared graphene oxide framework.
Inelastic Neutron Spectroscopy (INS):
Inelastic neutron spectroscopy data on GO and the GOF-1:1 samples were collected at the NIST Center for Neutron Research
using the Filter-Analyzer Neutron Spectrometer (FANS) at the BT4 beamline.[11,12]
Each sample was about 1 gram. The
samples were loaded in special Al-cans in a He-glove box. The sample holder has a gas-line for hydrogen loading. Due to very
large neutron absorption of boron, the spectra from GOF was not easy to collect and each spectra took about 2 days to provide
reasonable statistics. Measurements were taken at 4.4 K. INS measurements of the GO and GOF host lattices were collected
using a Cu(220) monochromator with in-pile collimation of 40 minutes of arc and post-monochromator collimation of 40
minutes of arc. Data were collected over the energy range of 34-200 meV. The low energy hydrogen rotational modes were
collected using a PG(002) (PG=Pyrolitic Graphite) monochromator with in-pile collimation of 20 minutes of arc and post-
monochromator collimation of 20 minutes of arc. Data were collected over the energy range of 5-45 meV. The bare GOF
samples were first measured and this spectra was subtracted from the H2+GOF spectra to obtain the modes arising only from
the adsorbed hydrogen.
Figure S11. Inelastic neutorn spectra (INS) of GO and GOF 1:1. The top panel (a) shows the INS for GO and GOF 1:1 and the
calculated spectrum for ideal GOF-32 structure shown in Fig. 1. Considering the disorder present in GOF sample, the agrement
between data and calculation is not bad, further supporting that the GOF 1:1 sample is very close to ideal GOF-32 structure.
The bottom panel (b) shows the difference INS spectra of GOF 1:1 with three different hydrogen loadings. The bare GOF 1:1
spectra (i.e. no hydrogen) is substracted from these spectra to yield pure hydrogen modes trapped in the nanoporous GOF
structure. At the high loading of 2 wt% H2, we start to see a new peak near 14.7 meV appearing (indicated by *), corresponing
non-hindered hydrogen rotation. The low energy peak near 10 meV indicates strongly hindered rotational dynamics for the H2
molecule, suggesting strong gas-host interactions in GOF sample.
The neutron vibrational spectra were measured for GO and GOF 1:1 samples. INS can directly probe the vibrational density of
states and is particularly sensitive to the hydrogen vibrational modes (Figure S11). In the case of GO, there are two broad
peaks near 68 meV and 170 meV. The first of these peaks is typical for trapped water libration modes as well as OH functional
groups in GO. The second peak near 170 meV is associated with phonons arising mainly from carbon motion and is typical of
graphitic systems. A detailed analysis of the INS spectra of GO along with first-principles model calculations are in progress.
Here we note that the spectra of GO is totally different than GOF 1:1 which exhibits many sharp features indicating the
diboronic acid linkers are successfully incorporated into the GOF structure. The bottom, blue curve in the top panel (Fig. S11a)
shows the first-principles calculated neutron spectrum for the ideal GOF-32 structure. Considering the large amount of disorder
present in GOF 1:1 sample, the agreement is not bad. The calculated phonon modes (associated with linker C-H motions) are
all present in the measured spectra.
We also study the ortho-para transition (i.e. rotational dynamics of hydrogen quantum rotor) for three different hydrogen
loadings in the GOF 1:1 sample, which is shown in Fig. S11b. The spectra shown in Fig. S11b are difference spectra after
subtracting the bare GOF 1:1 spectra (i.e. without any hydrogen loading). The neutron ortho-para transition occurs near 14.7
meV if the hydrogen molecules are almost free 3D rotors (i.e. unhindered), and is an indication of weak guest-host interactions.
However, if the interaction of hydrogen with the adsorption sites is strong, then one sees large deviations from this 14.7 meV
peak position. For a 2D confinement, the peak position occurs at around B=7.35 meV. From Fig. S11b it is clear that the
hydrogen molecules in the GOF channels are rotationally hindered significantly. Due to disorder, the peaks in the spectrum are
very broad. However the overall shape and energy scales are very similar to the INS spectrum of the H2-HKUST system,
which suggests that the H2-GOF interaction is of about the same order as MOFs with open metal sites that also exhibit peaks
around 10 meV.[13]
This is totally consistent with the high heat of absorption that we obtained from the adsorption isotherms
(shown in Fig. 4a of the main text). Fig. S11b shows that at higher loadings (such as 2wt%), we start seeing free 3D hydrogen
rotation around 14.7 meV which is indicated by “*”.
GO and Other Linkers
In order to prove that the boronic-ester linkage is the key, we tested other similarly sized molecules that are commonly used in
MOF/COF chemistry. These included benzene-1,4-dicarboxylic acid, benzene-1,3,5-tricarboxylic acid, benzene-1,2,3,4,5,6-
hexacarboxylic acid, 4,5-dicyanoimidazole, 2-methylimidazole and imidazole. Samples were prepared the same way as GOF
samples using the methanol solvothermal synthesis. We found that none of these linkers significantly intercalated into the
graphene oxide, Figures S12-S13, and none show any hydrogen adsorption. The fact that benzene-1,4-dicarboxylic acid
(Figure S11-b) does not intercalate the GO layers is quite interesting as it is of similar size and shape as B14DBA but with
different chemistry. All these results strongly support our theory that the diboronic acid interacts with the OH-groups on
graphene oxide to form some kind of ester bonding, interlinking the planes and forming a 3D framework structure.
Figure S12. PXRD plots of a) GO, b) GO-
benzene-1,4-dicarboxylic acid, c) GO-
benzene-1,3,5-tricarboxylic acid and d) GO-
benzene-1,2,3,4,5,6-hexacarboxylic acid.
Samples were prepared the same way as the
GOF sample with MeOH solution. This shows
that these linkers do not intercalate into the
GO. The peaks higher than 13° are peaks from
the crystalline linker. Note that the benzene-
1,4-carboxylic acid (b) does not go into GO
layers despite to the fact that it has basically
the same size and shape as the B14DBA linker,
further supporting our theory of boronic-ester
formation in GOFs.
Figure S13. PXRD plots of a) GO, b) GO-4,5-dicyanoimidazole, c) GO-2-methylimidazole and d) GO-imidazole. Samples
were prepared same way as the GOF samples with MeOH solution. This shows that these linkers do not intercalate into the
GO.
GOF-8 GOF-16 GOF-32
GOF-64 GOF- GOF-128
Details of our theoretical modeling and GCMC simulations on GOFs
Model systems of GOFs were built based on graphene oxide layers and rigid diboronic acid linkers. In our ideal GOF
structures, we only consider oxygen atoms that are covalently shared by graphene planes and the diboroinic linkers. To obtain
the optimized hydrogen adsorption, we assume all other functional groups such as epoxy-oxygen, etc. are removed from the
system. We label these structure by GOF-n where n is the number of GO carbon per linker. Various n=carbon:linker ratios
(n= , 128, 64, 32, 16, and 8) were adopted and uniform linker distribution was assumed, leading to a series of structures with
different pore size, pore volume and surface areas as shown below.
Figure S14. The ideal GOF-n structures with various n=GO-carbon/linker ratios of , 128, 64, 32, 16, and 8.
These structures were then geometrically optimized by first-principles calculations based on density-functional theory
(DFT). From these structural optimizations, we obtained the optimum interlayer spacing of ≈11 Ǻ. We used the PWSCF
package,[14]
Vanderbilt-type ultrasoft pseudopotentials, and the local-density approximation (LDA) with the Perdew-Zunger
exchange correlation. A cutoff energy of 544 eV and a gamma-point k sampling were sufficient for the total energy to
converge within 0.5 meV/atom. Both lattice parameters and atomic positions of the GOF model structures were fully relaxed.
In the optimized structures, the graphene layers are buckled due to the covalent bonding with the linker molecules. This is as
expected and similar to the case of graphene oxide. To evaluate the stability of the model structures, we performed latticed
dynamics calculations using the supercell method with finite displacements. The model structures with relatively high linker
concentrations (GOF-32, GOF-16, and GOF-8) were found to be stable. Several imaginary phonon modes were found in
structures with too low concentrations of linkers, suggesting the structure would distorted into a denser one with reduced
graphene interlayer distance. The hypothetical graphite structure with 11 Ǻ interlayer separation is apparently the least stable
one. These non-stable structures are considered solely for theoretical comparison purpose.
GCMC simulations of H2 adsorption in these GOF model systems were performed using our own code based on
standard algorithms.[15]
In the simulation, the chemical potential (directly related to pressure), temperature, volume are
imposed at given values. Both the H2 molecules and the frameworks are treated as rigid bodies. The Metropolis Monte Carlo
method was used and four different move types were considered: insertions, deletions, translations, and rotations. At each step,
a move type is selected at random, using equal weight. The dimension of the simulation box is ≈ 20×20×20 Å3. Periodic
boundary conditions were applied. For each (P, T) state, 2×107 total steps were used for equilibration and additional 2×10
7
steps were used to sample the desired thermodynamic properties. We used the standard universal force field (UFF) to describe
the hydrogen-framework interaction and the H2-H2 interaction. Note that in general, GCMC method based on empirical force
fields is not highly accurate. To calibrate the method that we used, we compared our simulation results on graphite (with 11 Ǻ
interlayer separation) and MOF-5 with the literature data. Our simulated H2 uptake in graphite (with 11 Ǻ interlayer
separation) at 298K and 100 bar is 2.65 wt%, reasonably close to the reported ≈3 wt% uptake derived using a more
sophisticated simulation method.[16]
Our simulated H2 uptake in MOF-5 at 77K and 1 bar is 1.67 wt%, reasonably close to the
reported ≈1.3 wt% experimental value.[17]
These general agreements suggest that the results derived from our GCMC
simulations based on UFF would give us a reasonable, qualitative evaluation of the H2 storage potential of our GOF materials.
In addition to the H2 adsorption isotherm data (77 K, 0-1 bar) shown in the main text, here we include the data at 77 K
and 298 K for a larger pressure range (0-100 bar).
0 10 20 30 40 50 60 70 80 90 1000.0
0.5
1.0
1.5
2.0
2.5 298 K
Absolu
te a
mount of H
2 a
dsorb
ed (
wt%
)
Pressure (bar)
VI
V
IV
III
II
I
0 10 20 30 40 50 60 70 80 90 100
0
2
4
6
8
10
12
VI
V
IV
III
II
77 K
Absolu
te a
mount of H
2 a
dsorb
ed (
wt%
)
Pressure (bar)
I
Figure S15. H2 adsorption isotherm data (298 K and 77 K), obtained for the GOF model systems from GCMC simulations.
[1] P. Scherrer, Göttinger Nachrichten Gesell., 2, 98 (1918)
[2] A. Patterson, "The Scherrer Formula for X-Ray Particle Size Determination". Phys. Rev. 56, 978 (1939)
[3] K. M. Thomas, Catalysis Today 120 389–398 (2007)
[4] M. Hirscher and B. Panella, Scripta Materialia, 56 809-812 (2007)
[5] W. Gerrard, The Organic Chemistry of Boron, Academic Press, New York, 1961, p. 223
[6] G. Socrates, Infrared Characteristic Group Frequencies, John Wiley & Sons, Chichester, 1980, p. 130
[7] R. M. Lindstrom, J. Res. Natl. Inst. Stand. Technol. 98, 127 (1993)
[8] D. Briggs and G. Beamson, Anal. Chem. 64, 1729 (1992))
[9] Dreyer et al., Chem. Soc. Rev. 39, 228 (2010)
[10] D. Briggs and G. Beamson, Anal. Chem. 65, 1517 (1993)
[11] http://www.ncnr.nist.gov/instruments/fans
[12] J.R.D. Copley, D. A. Neumann and W. A. Kamitakahara, Can. J. Phys. 73, 763 (1995)
[13] C. M. Brown, Y. Liu, T. Yildirim, V. K. Peterson, and C. J. Kepert, Nanotechnology 20, 204025 (2009)
[14] P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococcioni, I.
Dabo, A. Corso Dal, S. Fabris, G. Fratesi, S. de Gironcoli, R. Gebauer, U. Gerstmann, C. Gougoussis, A. Kokalj, M.
Lazzeri, L. Martin-Samos, N. Marzari, F. Mauri, R. Mazzarello, S. Paolini, A. Pasquarello, L. Paulatto, C. Sbraccia, S.
Scandolo, G. Sclauzero, A. P. Seitsonen, A. Smogunov, P. Umari, R. M. Wentzcovitch, J. Phys.: Condens. Matter 21,
395502 (2009)
[15] D. Frenkel, B. Smit, Understanding Molecular Simulation: From Algorithms to Applications. San Diego: Academic Press,
2002
[16] S. Patchkovskii, J. Tse, S. Yurchenko, L. Zhechkov, T. Heine, G. Seifert, Proc. Nat. Acad. Sci. 102, 10439 (2005)
[17] W. Zhou, H. Wu, M. R. Hartman, T. Yildirim, J. Phys. Chem. C 111, 16131 (2007)