Published on Web 02/26/2009

Post-Synthesis Alkoxide Formation Within Metal-Organic Framework

Materials: A Strategy for Incorporating Highly Coordinatively Unsaturated
Metal Ions
Karen L. Mulfort,, Omar K. Farha, Charlotte L. Stern, Amy A. Sarjeant, and Joseph T. Hupp*,
Department of Chemistry, Northwestern UniVersity, 2145 Sheridan Road, EVanston, Illinois 60208, and DiVision of
Chemical Sciences and Engineering, Argonne National Laboratory, Argonne, Illinois 60439
Received December 21, 2008; E-mail: j-hupp@northwestern.edu

Metal-organic framework (MOF) materials have received

considerable attention as potential high-performance, multifunctional
molecular sorbents.1 The attention derives in part from their
typically very high internal surface areas, low densities, and
permanent microporosity. Also highly attractive is their (typically)
crystalline nature, a characteristic that ensures complete uniformity
of channel sizes (for a given MOF) and allows one to determine
the position of every atom composing the framework. In turn, the
detailed positional information allows for high-quality computational modeling of observed or anticipated sorption behavior.
Additionally, the hybrid nature of the materials facilitates immense
structural and chemical variety. Taken together, these features point
to the opportunity to initially design (and/or modify after synthesis2)
the steric and chemical properties of the pores in order to tune
host-guest interactions with sufficient precision to render the
sorbent materials highly functional for specialized applications such
as chemical separations, gas storage, and selective catalysis.
On the basis of both experimental and computational studies, it
has become increasingly clear that the presence of coordinatively
unsaturated metal centers can greatly enhance the performance of
MOFs in the above-mentioned applications.3 The approaches to
introducing accessible metal centers include (a) exploitation of
incidental structural defects that leave metal-containing nodes
incompletely coordinated,4 (b) use of metal complexes (porphyrins,
salens, etc.) as organic struts,5 (c) electrostatic encapsulation of
metal complexes or solvated (or unsolvated6) metal cations by
anionic frameworks,7 (d) incorporation of metal-containing nodes
featuring thermally removable solvent molecules as ligands,8 (e)
binding of metal salts to reactive sites (e.g., silver nitrate attachment
to strut alkyne functionalities),9 and (f) photochemical attachment
of organometallic complexes to aromatic components of struts.10
Here we discuss an attractive alternative approach based on
conversion of pendant alcohols to metal alkoxides (Scheme 1) and
investigate its application to reversible uptake of molecular hydrogen.11 In contrast to all of the above except (c), the pendant-alcohol
strategy readily allows for incorporation of alkali metal ions.
Lithium ions in particular have attracted considerable attention in
theoretical investigations of MOFs because of their potential for
engendering high heats of adsorption for H2.12-15
We turned to the alkoxide approach after previously exploring
chemical reduction of MOF struts as a means of incorporating alkali
metal cations.16 While initial findings were encouraging (e.g.,
enhancements of H2 uptake by up to 75%), limitations to the
reduction approach subsequently became evident, at least for the
MOFs examined. Briefly, it appears that the incorporated cations
(a) localize around carboxylates rather than the reduced portions

Northwestern University.

Argonne National Laboratory.

3866

J. AM. CHEM. SOC. 2009, 131, 38663868

Scheme 1. Metal Alkoxide Formation within a Porous Framework

of the struts13 and (b) are largely shielded from direct interaction
with H2, exerting their effects instead by facilitating favorable
displacement of catenated frameworks. With these problems in
mind, we sought an alternative approach that would avoid catenation
and anchor ions far from carboxylates or nodes.
Catenation was addressed by using the recently developed octaoxygen ligand, L1.17 In contrast to most other carboxylate ligands
used in pillared-paddlewheel structures, L1 often yields noncatenated structures. As Figure 1 shows, the combination of L1 with
a Zn(II) source and the diol-containing strut, L2, gives, after 2 days
of heating, colorless block crystals of 1, which we have termed
DO-MOF. Single-crystal X-ray measurements confirmed that 1
consists of only a single network containing large cavities with
readily accessible alcohol functionalities. Application of the
SQUEEZE routine in PLATON revealed a remarkable 76% solventaccessible void volume.18 Characterization by thermal gravimetric

Figure 1. (A) Chemical structures of L1 and L2 and the crystal structure

of 1 (DO-MOF). Gray, carbon; blue, nitrogen; red, oxygen; yellow

tetrahedra, zinc. Hydrogens and solvent molecules have been omitted for
clarity. (B) Packing diagram of 1 down the (left) a and (right) b axes.
10.1021/ja809954r CCC: $40.75 2009 American Chemical Society

COMMUNICATIONS
Table 1. Summary of Adsorption Properties of 1 and 1-M
material

M/Zn2a

BET surface
area (m2/g)

pore volume
(cm3/g)c

H2 uptake (wt %)
at 1 atm, 77 K

Qst (kJ/mol)
at 0-1 atm

solvent/Me

1 (DO-MOF)
1-Li0.20
1-Li2.62d
1-Mg0.86
1-Mg2.02

H+
Li+
Li+
Mg2+
Mg2+

2b
0.20 ( 0.01
2.62 ( 0.05
0.86 ( 0.02
2.02 ( 0.02

810
840
270
820
510

0.35
0.46
0.20
0.40
0.29

1.23
1.32
0.77
1.16
1.01

6.3-4.7
6.3-6.6
5.6-0.5
6.2-6.9
7.3-5.0

n/a
0.40f
0.13f
0.08g
1.10h

a
Determined via inductively coupled plasma analysis of dissolved samples. b From the crystal structure. c Measured at P/Po 0.95. d Sample showed
substantial loss of crystallinity, suggesting degradation. e Total number of residual solvent molecules per M atom, as determined by 1H NMR analysis of
the evacuated material (see the SI). f DMF + THF. g DMF + THF + (methanol or methoxide). h 0.19 DMF + 0.04 THF + 0.88 (methanol or
methoxide).

Figure 2. N2 adsorption isotherms of 1, 1-Li0.20, and 1-Mg0.86. Closed

symbols, adsorption; open symbols, desorption.

analysis [see the Supporting Information (SI)] gave a value of 55%.

Solvent loss ended at 150 C, with degradation occurring only above
300 C. N2 adsorption (77 K) for the solvent-evacuated version of
1 yielded a type-I isotherm (Figure 2), indicating microporosity,
and a BET surface area of 810 m2/g19 (Table 1).
Once we had established the structural stability of this largepore material, we set out to convert alcohol functionalities to lithium
alkoxides. After preliminary experiments indicated that harsh
reagents (e.g., methyllithium) degraded the MOF, we turned to a
much milder reagent, lithium t-butoxide.20 Exchange of hydroxyl
protons was achieved by replacing (via soaking) the initially present
guest solvent molecules (DMF) with more volatile THF molecules
and then stirring DO-MOF in an excess of Li+[O(CH3)3-] in
CH3CN/THF (see the SI). The extent of lithium loading was
controlled by adjusting the stirring rate and time.
Samples of 1-Li were activated by heating at 200 C under
reduced pressure for 24 h. 1H NMR (dissolved samples; Table 1)
established that activation removes nearly all of the solvent, while
N2 adsorption measurements (Figure 2) showed that 1 remains
microporous after lithiation. At low loading [0.20 Li/Zn2 (1-Li0.20)],
the MOF retains its sizable surface area (Table 1). However, in
the extreme of high loading (1-Li2.62), both the surface area and
micropore volume diminish. Additionally, crystallinity is lost. The
effects are tentatively ascribed to partial displacement of zinc by
lithium, as simple ROH to ROM conversion should limit the Li/
Zn2 ratio to 2. Notably, no unreacted Li+[O(CH3)3-] was detected.
As Figure 3 shows, low-pressure adsorption of H2 by 1 is
reversible at 77 K and reaches 1.23 wt % at 1 atm (Table 1). 1-Li0.20
exhibits only slightly greater uptake (1.32 wt % at 1 atm).
Nevertheless, the increase corresponds to two additional H2 per Li+.
This finding is broadly consistent with computational predictions
that an exposed lithium cation on carbon or MOF materials can
(depending on pressure) directly bind up to six H2 molecules.21
Unfortunately, extension of the measurements to the highly lithiated
sample (1-Li2.62) yielded inferior sorption behavior, provisionally
ascribed to partial framework degradation.

H2 uptake was also examined at 87 K. Fits of 77 and 87 K

isotherms to a virial equation (see the SI) enabled pressuredependent isosteric heats of adsorption, Qst, to be determined.22
As Figure 3B shows, unreacted 1 displays more-or-less typical
MOF behavior, i.e., a modest initial Qst value that decreases with
increasing H2 sorption; this is expected if the first molecules to
enter the material bind at the sites offering the highest interaction
energy. Though the initial value (6.3 kJ/mol) is well below that
necessary for practical H2 storage, on the basis of binding-relevant
factors such as pore size, 1 compares well to similarly structured
materials.1d 1-Li0.20 exhibits much more unusual behavior. While
the value of Qst at very low H2 loading is similar to that for 1, it
increases at higher H2 loading. While rare, behavior of this kind
has occasionally been described, most notably for a series of Ti(III)decorated porous silicas.23 There the observation was rationalized
in terms of changes in the mode of binding to Ti(III) with increasing
number of hydrogens, with the change facilitated by the ability of
Ti(III) to engage in d-orbital-based Kubas interactions with H2.24
In the case of 1-Li, the metals cannot deploy Kubas binding.
Nevertheless, they appear to engender specific interactions that are
absent in the parent MOF material.
Reasoning that replacement of Li+ by a more highly charged
cation might increase the heat of adsorption (for example, via greater
local field strength15 or enhancement of charge-quadrupole
interactions14,25), we also examined Mg2+-containing versions of
DO-MOF. These were prepared by reacting 1 with methanolic
solutions of Mg(OMe)2 and then activating as described above.
Materials containing either 1 or 2 magnesium ions per glycol
strut (1-Mg0.86 or 1-Mg2.02, respectively) were obtained, depending
on the reaction conditions (see the SI).
1-Mg0.86 is assumed to contain individual dications that that are
bound to pairs of L2 alkoxide oxygens but are otherwise free of
ligands, consistent with 1H NMR data for the dissolved material
(Table S2 in the SI). Figure 3A shows that Mg2+ incorporation has
surprisingly little effect on H2 uptake at 77 K but does alter the
binding, eliciting the same unusual increase in Qst with H2 loading
as found for 1-Li. That the absolute Qst values are so similar for
1-Mg0.86 and 1-Li0.20, however, suggests that the presence of an
additional L2 alkoxide oxygen for Mg2+ very effectively diminishes
the charge and/or field experienced by proximal H2 molecules.
For the more highly loaded material 1-Mg2.02, we assume that
every L2 alkoxide site binds a dication independently (necessitating
a second charge-balancing anion for each Mg2+, presumably a
methoxide anion). Repeated attempts to obtain single crystals after
magnesium functionalization (and thereby confirm the coordination)
were unsuccessful. Nevertheless, 1H NMR characterization of
dissolved samples (Table S2 in the SI) is consistent with retention
of 1 methoxide per Mg2+.
In contrast to the case of 1-Mg0.86, the N2-accessible surface area
of 1-Mg2.02 is somewhat diminished relative to the parent MOF.
The H2 uptake is also diminished (see Table 1 and the SI), and the
J. AM. CHEM. SOC.

VOL. 131, NO. 11, 2009

3867

COMMUNICATIONS

(2)

(3)

(4)
(5)

(6)
(7)

Figure 3. (A) Low-pressure H2 adsorption isotherms of 1, 1-Li0.20, and

1-Mg0.86. Closed symbols, adsorption; open symbols, desorption. (B) H2
isosteric heats of adsorption of 1, 1-Li0.20, and 1-Mg0.86.

unusual increase of Qst with H2 loading is absent in the more highly

metalated MOF. Evidently, the presence of ligands other than struts
is detrimental to the performance of added metal ions as sorption
sites. The appropriate use of diol-containing struts (as opposed to
monoalcohols) therefore appears to be an important design consideration when using MOF-based alkoxides to incorporate dications.
In summary, we have introduced a noncatenated hydroxylfunctionalized MOF and exchanged the hydroxyl protons for lithium
and magnesium cations via solution methods. At low to intermediate
levels of cation substitution, the activated metals appear to be naked,
apart from alkoxide (L2 strut) anchoring, resulting in unusual Qst
behavior and modest enhancement of H2 sorption (2 additional
H2 per added Li+ at 77 K and 1 atm). While the focus here has
been on metal ions that may improve hydrogen sorption, the strategy
may well prove to be a general one that is also suitable for metals
that facilitate chemical catalysis11 or separations. We are currently
investigating these possibilities as well as continuing investigations
of H2 sorption.
Acknowledgment. We thank Dr. Andy Ott for assistance with
crystallography and Patrick Ryan for the calculation of theoretical
surface area. We gratefully acknowledge the U.S. Department of
Energy (Grant DE-FG02-08-15967) and the Northwestern NSEC
for financial support. K.L.M. gratefully acknowledges a LaboratoryGrad Fellowship from Argonne National Laboratory.

(8)
(9)

(10)
(11)

(12)
(13)
(14)
(15)
(16)

(17)
(18)
(19)

(20)

Supporting Information Available: Full synthesis details for 1,

crystallographic data for 1 in CIF format, preparation details for all
1-M compounds, N2 adsorption isotherms of 1-Li2.62 and 1-Mg2.02, H2
adsorption isotherms, details of isosteric heat of adsorption calculations,
and details of 1H NMR analysis of evacuated 1 and 1-M. This material
is available free of charge via the Internet at http://pubs.acs.org.

(1) (a) Ferey, G. Chem. Soc. ReV. 2008, 37, 191214. (b) Collins, D. J.; Zhou,
H.-C. J. Mater. Chem. 2007, 17, 31543160. (c) Lin, X.; Jia, J.; Hubberstey,

J. AM. CHEM. SOC.

(22)
(23)

(24)
(25)

References

3868

(21)

VOL. 131, NO. 11, 2009

P.; Schroder, M.; Champness, N. R. CrystEngComm 2007, 9, 438448.

(d) Zhao, D.; Yuan, D.; Zhou, H.-C. Energy EnViron. Sci. 2008, 1, 222
235.
(a) Gadzikwa, T.; Lu, G.; Stern, C. L.; Wilson, S. R.; Hupp, J. T.; Nguyen,
S. T. Chem. Commun. 2008, 54935495. (b) Farha, O. K.; Mulfort, K. L.;
Hupp, J. T. Inorg. Chem. 2008, 47, 1022310225. (c) Wang, Z.; Cohen,
S. M. Angew. Chem., Int. Ed. 2008, 47, 46994702. (d) Tanabe, K. K.;
Wang, Z.; Cohen, S. M. J. Am. Chem. Soc. 2008, 130, 85088517. (e)
Morris, W.; Doonan, C. J.; Furukawa, H.; Banerjee, R.; Yaghi, O. M. J. Am.
Chem. Soc. 2008, 130, 1262612627. (f) Hwang, Y. K.; Hong, D.-Y.;
Chang, J.-S.; Jhung, H.; Seo, Y.-K.; Kim, J.; Vimont, A.; Daturi, M.; Serre,
C.; Ferey, G. Angew. Chem., Int. Ed. 2008, 47, 41444148.
(a) Bae, Y.-S.; Farha, O. K.; Spokoyny, A. M.; Mirkin, C. A.; Hupp, J. T.;
Snurr, R. Q. Chem. Commun. 2008, 41354137. (b) Dinca, M.; Dailly, A.;
Liu, Y.; Brown, C. M.; Neumann, D. A.; Long, J. R. J. Am. Chem. Soc.
2006, 128, 1687616883. (c) Chen, B.; Ockwig, N. W.; Millward, A. R.;
Contreras, D. S.; Yaghi, O. M. Angew. Chem., Int. Ed. 2005, 44, 47454749. (d) Dinca, M.; Long, J. R. Angew. Chem., Int. Ed. 2008, 47, 6766
6779.
Fujita, M.; Kwon, Y. J.; Washizu, S.; Ogura, K. J. Am. Chem. Soc. 1994,
116, 11511152.
(a) Cho, S.-H.; Ma, B.; Nguyen, S. T.; Hupp, J. T.; Albrecht-Schmitt, T. E.
Chem. Commun. 2006, 25632565. (b) Chen, B.; Zhao, X.; Putkham, A.;
Hong, K.; Lobkovsky, E. B.; Hurtado, E. J.; Fletcher, A. J.; Thomas, K. M.
J. Am. Chem. Soc. 2008, 130, 64116423. (c) Smithenry, D. W.; Wilson,
S. R.; Suslick, K. S. Inorg. Chem. 2003, 42, 77197721.
Yang, S.; Lin, X.; Blake, A. J.; Thomas, K. M.; Hubberstey, P.; Champness,
N. R.; Schroder, M. Chem. Commun. 2008, 61086110.
(a) Liu, Y.; Kravtsov, V. C.; Larsen, R.; Eddaoudi, M. Chem. Commun.
2006, 14881490. (b) Alkordi, M. H.; Liu, Y.; Larsen, R. W.; Eubank,
J. F.; Eddaoudi, M. J. Am. Chem. Soc. 2008, 130, 1263912641. (c) Liu,
Y.; Kravtsov, V. C.; Eddaoudi, M. Angew. Chem., Int. Ed. 2008, 47, 8446
8449.
In some cases, alteration of the chemical identity of node-localized metal
ions has proven feasible via cation exchange. For example, see: Dinca,
M.; Long, J. R. J. Am. Chem. Soc. 2007, 129, 1117211176.
Emberger, G. A.; Bae, Y.-S.; Nguyen, S. T.; Hupp, J. T.; Broadbelt, L. J.;
Snurr, R. Q. In Proceedings of the 8th International Conference on
Characterization of Porous Solids (COPS VIII), Edinburgh, U.K., June 1013, 2008.
Kaye, S. S.; Long, J. R. J. Am. Chem. Soc. 2008, 130, 806807.
A precedent of sorts should be noted: Wu, C.-D.; Hu, A.; Zhang, L.; Lin,
W. J. Am. Chem. Soc. 2005, 127, 89408941. These authors used
binaphthanol-containing structures to immobilize catalytic Ti(IV) centers
and accompanying charge-compensating ligands. For condensed-phase
catalysis, charge-compensating species present little difficulty since they
may dissociate from the active site. For gas sorption, on the other hand,
our experience has been that these extra species inhibit the interaction
of sorbates with metal centers.9 Additionally, while our work was in
progress, a computational study of H2 binding by a (hypothetical) lithium
alkoxide-containing MOF appeared15.
(a) Han, S. S.; Goddard, W. A., III. J. Am. Chem. Soc. 2007, 129, 8422
8423. (b) Blomqvist, A.; Araujo, C. M.; Srepusharawoot, P.; Ahuja, R.
Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2017320176.
Dalach, P.; Frost, H.; Snurr, R. Q.; Ellis, D. E. J. Phys. Chem. C 2008,
112, 92789284.
Belof, J. L.; Stern, A. C.; Eddaoudi, M.; Space, B. J. Am. Chem. Soc. 2007,
129, 1520215210.
Klontzas, E.; Mavrandonakis, A.; Tylianakis, E.; Froudakis, G. E. Nano
Lett. 2008, 8, 15721576.
(a) Mulfort, K. L.; Hupp, J. T. J. Am. Chem. Soc. 2007, 129, 96049605.
(b) Mulfort, K. L.; Hupp, J. T. Inorg. Chem. 2008, 47, 79367938. (c)
Mulfort, K. L.; Wilson, T. M.; Wasielewski, M. R.; Hupp, J. T. Langmuir
2009, 25, 503508.
Farha, O. K.; Mulfort, K. L.; Hupp, J. T. Inorg. Chem. 2008, 47, 10223
10225.
Spek, A. L. J. Appl. Crystallogr. 2003, 36, 713.
Notably, the calculated maximum N2-accessible surface area is 4000 m2/
g. The disparity between this and the experimental value of 800 m2/g
presumably is a reflection of either partial channel collapse or pore blockage
at particle-particle interfaces.
Hanna, T. A.; Liu, L.; Angeles-Boza, A. M.; Kou, X.; Gutsche, C. D.;
Ejsmont, K.; Watson, W. H.; Zakharov, L. N.; Incarvito, C. D.; Rheingold,
A. L. J. Am. Chem. Soc. 2003, 125, 62286238.
Barbatti, M.; Jalbert, G.; Nascimento, M. A. C. J. Chem. Phys. 2001, 114,
22132218.
Czepirski, L.; Jagiello, J. Chem. Eng. Sci. 1989, 44, 797801.
(a) Hu, X.; Skadtchenko, B. O.; Trudeau, M.; Antonelli, D. M. J. Am. Chem.
Soc. 2006, 128, 1174011741. (b) Hamaed, A.; Trudeau, M.; Antonelli,
D. M. J. Am. Chem. Soc. 2008, 130, 69926999. (c) Hu, X.; Trudeau, M.;
Antonelli, D. M. Inorg. Chem. 2008, 47, 24772484.
Kubas, G. J. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 69016907.
Lochan, R. C.; Head-Gordon, M. Phys. Chem. Chem. Phys. 2006, 8,
13571370.

JA809954R

Supporting Information for

Post-Synthesis Alkoxide Formation Within Metal-Organic

Framework Materials: A Strategy for Incorporating Highly
Coordinatively Unsaturated Metal Ions
Karen L. Mulfort,, Omar K. Farha, Charlotte L. Stern, and Joseph T. Hupp*

Department of Chemistry, Northwestern University, 2145 Sheridan Road, Evanston IL 60208

The Chemical Sciences and Engineering Division, Argonne National Laboratory, 9700 South
Cass Avenue, Argonne IL 60439

S1

General methods and materials.

All commercial reagents were of ACS grade and purchased from Sigma-Aldrich unless
otherwise noted. The synthesis of L1 has been previously reported.1 L2 was obtained from TCI
America and used without further purification. Tetrahydrofuran and acetonitrile were purified
using a two-column solid-state purification system (Glasscontour System, Jeorg Meyer, Irvine,
CA). 500 MHz 1H NMR was performed on a Varian INOVA model spectrometer and
referenced to the residual solvent peak. Powder X-ray diffraction (PXRD) patterns were
recorded with a Rigaku XDS 2000 diffractometer using nickel-filtered Cu K radiation ( =
1.5418 ) over a range of 5 < 2 < 40 in 0.1 steps with a 1-s counting time per step. Powder
samples were placed in the diffractometer mounted on a stainless steel holder with double-sided
tape. Thermogravimetric analyses (TGA) were performed on a Mettler-Toledo
TGA/SDTA851e. Samples (3-5 mg) in alumina pans were heated from 25oC to 700oC at
10oC/minute under N2. Elemental analysis was performed by Atlantic Microlabs, Inc. (Norcross,
GA). Inductively coupled plasma (ICP) spectroscopy was conducted on a Varian model ICP
spectrometer that is equipped to cover the spectral range from 175 to 785 nm. Samples (3-5 mg)
were digested in 1:1 H2SO4:H2O2 and heated at 120oC until the solution became clear and
colorless and no further vapor was produced. An aliquot of this concentrated acid solution was
diluted to 5% with DI H2O and analyzed for M (Li at 610.365 nm, Mg at 279.800 nm) and Zn
(206.200 nm) content as compared to standardized solutions.
Low-pressure nitrogen and hydrogen adsorption measurements were carried out on an
Autosorb 1-MP from Quantachrome Instruments. Ultra-high purity grade He, H2, and N2 were
used for all adsorption measurements. Prior to analysis, samples of 1 were soaked in THF for
approximately 24 hours, exchanging the supernatant solvent several times. Dry materials were
loaded into a sample tube of known weight and heated at 200oC under dynamic vacuum for ~24
hours to completely remove guest solvent. After evacuation, the sample and tube were reweighed to obtain the precise mass of the evacuated sample. Care was taken in handling
evacuated samples of 1 and 1-M to minimize exposure to air and water before adsorption
measurements. N2 adsorption isotherms were measured at 77K and H2 adsorption isotherms
were measured at both 77K and 87K in order to calculate the isosteric heat of adsorption.

S2

Synthesis of 1.
Single crystals of 1 were obtained upon heating Zn(NO3)26H2O (50 mg, 0.17 mmol), L1 (100
mg, 0.18 mmol), and L2 (25 mg, 0.12 mmol) in 25 ml DMF plus one drop of concentrated HCl.
This solution was divided equally between five 4 dram screw cap vials and heated to 80oC for 3
days at which time clear colorless block crystals had formed. Large scale materials were
obtained by simply scaling up the single crystal preparation. A sample preparation follows:
Zn2(NO3)6H2O (100 mg, 0.34 mmol), L1 (200 mg, 0.36 mmol), L2 (50 mg, 0.24 mmol), and 50
ml DMF were added to a 50 ml erlenmeyer flask and sonicated to disperse. After about 10
minutes, 2 drops concentrated HCl was added to the suspension and sonicated for an additional 5
minutes upon which time the suspension dissolved to give a clear pale yellow solution. The
solution was then divided equally among ten 4 dram screw cap vials and placed in an 80oC oven
for three days. The clear colorless crystals were removed from the flask, isolated via filtration,
washed with DMF, and allowed to dry in air. Zn2(L1)(L2) (1) recovered: ~30 mg, 5% yield
based on zinc. Anal. calcd. for 12H2O, C46H34N2O12Zn2: C, 58.93; H, 3.63; N, 2.99. Found: C,
58.78; H, 3.10; N, 1.93. Phase purity was verified by PXRD and TGA (Figure S1).

S3

Crystal structure of 1.
Single crystals of 1 were mounted on a BRUKER APEX2 V2.1-0 diffractometer equipped
with a graphite-monochromated MoKa ( = 0.71073 ) radiation source in a cold nitrogen
stream. All crystallographic data were corrected for Lorentz and polarization effects (SAINT).
Data diffracted out to 0.84 Angstroms but fit poorly at low resolution suggesting limitations in
the proposed model. The structures were solved by direct methods and refined by the full-matrix
least-squares method on F2 with appropriate software implemented in the SHELXTL program
package. Most of the guest DMF solvent molecules within the pores are severely disordered,
which hindered satisfactory development of the model; therefore, the SQUEEZE routine in
PLATON2 was applied to remove the contributions of electron density from disordered solvent
molecules. The outputs from the SQUEEZE calculations are attached to the CIF file. All of the
non-hydrogen atoms were refined anisotropically. Heavy restraints (esd 0.001) on the bond
distances and angles were used on L2 to idealize the pyridyl ring and to stabilize the solution of
1. Rigid-bond restraints (esd 0.001) were imposed on displacement parameters for the DPG
ligand similar amplitudes (esd 0.001) were imposed on displacement parameters for the sites on
the ligand separated by less than van der Waals radii. All efforts to reduce symmetry to avoid
twofold and mirror disorders failed.

S4

Table S1. Summary of crystallographic data of 1.

compound

128DMF

empirical formula a

C65H113N15O19Zn

formula weight

1474.07

crystal color, habit

colorless, tabular

crystal dimensions (mm3)

crystal system

orthorhombic

space group

Pmmm

a ()

11.3014 (11)

b ()

15.7553 (13)

c ()

15.7553(13)

a (deg)

90

b (deg)

90

g (deg)

90

V (3)

2805.3(4)

(calcd, g/cm3)

1.745

(mm-1)

0.541

goodness-of-fit on F2

0.793

Rb

0.1078

Rw
a

0.443 x 0.218 x 0.067

0.2575

The SQUEEZE routine in PLATON was employed to mask diffuse electron density in the cavities due to

disordered solvent (DMF) molecules. The outputs from the SQUEEZE routine gave 28 DMF solvent molecules per
unit cell; these have been added to the unit cell formula. bR(F) = (Fo Fc)/Fo. cRw(Fo2) = [[w(Fo2
Fc2)2]/wFo4]1/2.

S5

normalized intensity

Figure S1. PXRD (A) and TGA (B) of 1.

as synthesized

simulation

10

15

20

25

30

35

two theta

1A

100

80

% mass

54.5%
60

40

20

0
100

200

300

400

500
o

temperature ( C)

1B

S6

600

700

Alkoxide formation.
In general, to perform the proton exchange and create the alkoxide frameworks, crystalline
samples of 1 were submerged in THF for 1 day to replace the pore-filling DMF. The THFexchanged samples were then stirred vigorously in dry THF/CH3CN with a metal source
overnight. The solid was then isolated by filtration, washed with copius amounts of THF to
remove any weakly physisorbed ions, dried briefly in air, and quickly transferred to an
adsorption sample tube. This procedure was tested on the L2 ligand alone and the complete
removal of the hydroxyl proton verified by 1H NMR, Figure S2.
Stoichiometric loading is defined here as the amount of alkoxide formation per L2 strut. The
crystal structure of 1 gives 1 L2 strut per Zn2 cluster, stoichiometric loading here is 2 Li(I) per 2
Zn(II). Specific preparation procedures follow:
Synthesis of 1-Li0.20: 1 (~10 mg) was added to a small scintillation vial, covered with 3 ml
CH3CN, and stirred gently with 2 ml LiOtBu solution (1.0M THF solution) overnight. The solid
was washed with THF several times and isolated via filtration. ICP analysis for Li/Zn returned
0.20 0.01 Li / Zn2 unit.
Synthesis of 1-Li2.62: 1 (13 mg) was added to a 100 ml flask, covered with 20 ml THF, and
stirred vigorously. Lithium-tert-butoxide (4 mg, 0.05 mmol) was added to the flask and stirred at
room temperature overnight. ICP analysis for Li/Zn returned 2.62 0.05 Li / Zn2 unit.
Synthesis of 1-Mg0.86: 1 (~30 mg) was added to a small conical vial, covered with 3 ml THF
and stirred vigorously to create a strong dispersion. Mg(OMe)2 (60 l, 6-10% solution in
MeOH) was added to the vial, which was then capped and allowed to stir overnight. ICP
analysis for Mg/Zn returned 0.86 0.02 Mg / Zn2 unit.
Synthesis of 1-Mg2.02: 1 (~30 mg) was added to a small conical vial, covered with 5 ml THF
and stirred vigorously to create a strong dispersion. Mg(OMe)2 (200 l, 6-10% solution in
MeOH) was added to the vial, which was then capped and allowed to stir overnight. ICP
analysis for Mg/Zn returned 2.02 0.02 Mg / Zn2 unit.

S7

Figure S2. 500 MHz 1H NMR in DMSO-d6 of L2 (bottom, black) and L2-2Li+ (top, blue).
Hydroxyl protons at 5.7 ppm are not present in spectrum of alkoxide-functionalized ligand.

S8

Figure S3. N2 adsorption isotherms of 1 and 1-Li (A) and 1 and 1-Mg (B). Closed symbols,
adsorption; open symbols, desorption.

400

1
1-Li0.20

350

1-Li2.62

volume adsorbed (cm /g)

450

300
250
200
150
100
50
0
0.0

0.2

0.4

0.6

0.8

1.0

P / Po

3A

300

volume adsorbed (cm /g)

400

200

100

1
1-Mg0.86
1-Mg2.02

0
0.0

0.2

0.4

0.6

P / Po

3B

S9

0.8

1.0

Figure S4. 77K and 87K H2 adsorption isotherms of 1 (A), 1-Li0.20 (B), 1-Li2.62 (C), 1-Mg0.86
(D), 1-Mg2.02 (E). Closed symbols, adsorption; open symbols, desorption.

1.4
1.2
1.0

wt %

0.8
0.6
0.4

77K
87K

0.2
0.0
0.0

0.2

0.4

0.6

0.8

1.0

P (atm)

4A

1.4
1.2

wt %

1.0
0.8
0.6
0.4

77K
87K

0.2
0.0
0.0

0.2

0.4

0.6

P (atm)

4B

S10

0.8

1.0

0.8

wt %

0.6

0.4

0.2

77K
87K
0.0
0.0

0.2

0.4

0.6

0.8

1.0

P (atm)

4C

1.2
1.0

wt %

0.8
0.6
0.4
0.2

77K
87K

0.0
0.0

0.2

0.4

0.6

P (atm)

4D

S11

0.8

1.0

1.0

wt %

0.8

0.6

0.4

0.2

77K
87K
0.0
0.0

0.2

0.4

0.6

P (atm)

4E

S12

0.8

1.0

Figure S5. Details of fitting and calculation of isosteric heat of adsorption for 1(A), 1-Li0.20 (B),
1-Li2.62 (C), 1-Mg0.86 (D), and 1-Mg2.02 (E).
Both hydrogen isotherms (77K, 87K) for each material were fit to a virial equation of the form
given in Equation 1.3 The heat of adsorption is then calculated from the fitting parameters using
Equation 2. The isotherms and fitting parameters are shown in Figure S5. Red lines are virial
equation fit to data.
ln p = ln N +

1
T

a N + b N
i

(1)

i=0

i=0
m

qst ( N ) = R ai N i

(2)

i =0

77K
87K
6

ln P (torr)

Chi /DoF = 0.01979

2
R = 0.99736

a0 = -756.02427 31.15921
a1 = 67.82439 13.37311
a2 = -19.05574 7.38794
a3 = 3.13151 1.54678
a4 = -0.2381 0.13837
a5 = 0.00673 0.00444
b0 = 11.06465 0.37528

-2
0

N, mg/g

5A

S13

10

12

14

77K
87K

ln P (torr)

Chi /DoF = 0.02714

2
R = 0.99635

a0 = -780.62964 67.40931
a1 = 27.0409 13.07371
a2 = -7.94879 3.10377
a3 = 0.76295 0.35668
a4 = -0.02531 0.01318
b0 = 10.91304 0.8191
b1 = 0.2849 0.12696

-2
0

10

12

14

N, mg/g

5B

77K
87K

Chi /DoF = 0.04966

2
R = 0.99369

ln P (torr)

a0 = -669.12954 80.23388
a1 = 353.43436 56.88392
a2 = -247.5793 65.36447
a3 = 103.6587 32.66379
a4 = -21.82458 7.72871
a5 = 2.23554 0.86588
a6 = -0.08875 0.03696
b0 = 10.37112 0.9703
b1 = -0.74008 0.23481

-2
0

N, mg/g

5C

S14

8
77K
87K

ln P (torr)

Chi /DoF = 0.02504

2
R = 0.99672

a0 = -772.48033 64.86774
a1 = 52.73039 21.37485
a2 = -28.33613 9.49021
a3 = 5.21312 2.11209
a4 = -0.43229 0.20039
a5 = 0.01318 0.00682
b0 = 11.24135 0.78551
b1 = 0.39617 0.16078

-2
0

10

12

N, mg/g

5D

77K
87K

ln P (torr)

Chi /DoF = 0.00309

2
R = 0.99959

a0 = -874.85671 23.15925
a1 = 44.33304 5.80591
a2 = -4.21387 1.77951
a3 = 0.58996 0.26291
a4 = -0.02792 0.01262
b0 = 12.33106 0.28064
b1 = -0.09992 0.04975

-2
0

N, mg/g

5E

S15

10

Figure S6. Isosteric H2 heat of adsorption of 1 and 1-Li (A) and 1 and 1-Mg (B).

8
7

Qst, kJ/mol

6
5
4
3
2

1
1-Li0.20

1-Li2.62

0
0

10

N, mg/g

6A

8
7

Qst, kJ/mol

6
5
4
3
2

1
1-Mg0.86

1-Mg2.02

0
0

N, mg/g

6B

S16

10

Figure S7. 1H NMR analysis of evacuated samples to quantify residual solvent content of 1-Li
and 1-Mg.
Following complete evacuation and N2 and H2 adsorption measurements, samples of 1, 1-Li and
1-Mg were dissolved in D2SO4 (96-98 wt% in D2O, 99.5 atom% D) in order to detect residual
solvent in the pores or coordinated to Li+ and Mg2+. For consistency, the spectra were referenced
to the first solvent peak of DMF ( = 3.01 ppm in H2O). The proton peaks for DMF (two -CH3
groups, 6H), THF (only one of two peaks, 4H), t-butanol (1-Li samples, 9H) and methanol (1Mg samples, 3H) were integrated and referenced to proton peaks for the dissolved MOF ligands
(specifically those of the -protons to the hydroxyl group of L2). Figure S7 shows the NMR
spectra for each sample. Table S2 outlines the calculations. See also reference 16 in the text for
previous use of this method.

1-Li0.20

1-Li2.62

1-Mg0.86

1-Mg2.02
9.0

8.5

8.0

7.5

7.0

6.5

6.0

5.5

5.0

4.5

S17

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

Table S2. Quantification of solvent content in 1-Li and 1-Mg.

M/Zn2
1-Li0.20
1-Li2.62
1-Mg0.86
1-Mg2.02

0.20
2.62
0.86
2.02

L2 proton
integration
20
20
20
20

M/Zn2

# L2
ligands
10
10
10
10

# DMF / M

DMF
integration
2.6
17.8
2.2
22.6

# THF / M

#
DMF
0.43
2.97
0.37
3.77

THF
integration
1.2
1.3
0.7
2.9

# R-OH / M

0.20
0.22
0.15
1-Li0.20
2.62
0.11
0.01
1-Li2.62
0.86
0.04
0.02
1-Mg0.86
2.02
0.19
0.04
1-Mg2.02
[a] R-OH (1-Li) = t-butanol, R-OH (1-Mg) = methanol.

0
0.00
0.02
0.88

S18

#
THF
0.30
0.33
0.18
0.73

total # of solvent
molecules / M
0.37
0.12
0.08
1.10

R-OH
integration[a]
0
0.6
0.4
53.1

#
R-OH
0
0.67
0.13
17.7

total residual
mass loss (%)
0.28
0.10
0.05
0.48

References
1) Farha, O. K.; Mulfort, K. L.; Hupp, J. T., Inorg. Chem. 2008, 47, 10223-10225.
2) Spek, A. L., J. Appl. Crystallogr. 2003, 36, 7-13.
3) Czepirski, L.; Jagiello, J., Chemical Engineering Science 1989, 44, 797-801.

S19