Angewandte
Chemie
DOI: 10.1002/anie.201108462
Covalent Organic Frameworks
Internal Functionalization of Three-Dimensional Covalent Organic
Frameworks**
David N. Bunck and William R. Dichtel*
Covalent organic frameworks (COFs) represent an emerging
class of porous crystalline materials composed of light
elements,[1] typically C, N, O, and/or B, that crystallize into
two-dimensional (2D) layered structures or three-dimen-
sional (3D) networks.[2,3] The nearly eclipsed structures[4,5] of
most 2D COFs give rise to high intrinsic charge mobilities[6]
and their recent synthesis as oriented thin films[7] portends
their use in optoelectronic and energy-storage devices. In
contrast, few 3D COFs have been crystallized, and despite
exhibiting exceptionally high surface areas (> 4000 m2 gÀ1)
and record low densities (0.17 gcmÀ3), these networks have no
well-developed applications.[8] Functionalizing the interior of
3D COFs might harness these desirable properties to provide
structurally precise platforms for catalysis,[9] separations,[10]
and the storage and release of molecular payloads.[11] How-
ever, no functionalized 3D COFs have been reported, while
the functionalization of 2D COFs has been limited to alkyl
chains.[12] Postsynthetic functionalization of related metal–
organic frameworks (MOFs) relies on incorporating reactive
groups,[13] such as alkynes[14] or amines,[15] on the organic
linkers, but these moieties are not readily incorporated onto
symmetric, polyvalent 3D COF building blocks.
Herein we report a general approach to functionalize 3D
COFs using a new monomer-truncation strategy. A tetrahe-
dral building block, which self-condenses to form the 3D
network known as COF-102[2] (Scheme 1a), was modified
such that one of its four arylboronic acid moieties is replaced
with an arbitrary functional group. The resulting trigonal
tris(boronic acid) is co-condensed with the parent tetrahedral
monomer to provide functionalized COF-102 (Scheme 1b).
The degree of functionalization is determined by the feed
ratio of the two monomers and tolerates relatively high
loadings of the truncated monomer (> 30%), while the
crystallinity, permanent porosity, and high surface area of
the unfunctionalized material are maintained. This method
also requires no modification of the solvothermal growth
conditions used to crystallize COF-102. The truncated
monomer is incorporated throughout the lattice, rather than
on the crystallite surface, which might be unexpected given
the reversible bond-forming conditions employed in COF
synthesis.[16] However, growth conditions that produce crys-
talline materials are optimized empirically, and COF nucle-
ation and growth processes are poorly understood. Our
results indicate that boroxine hydrolysis is too slow to liberate
truncated monomers from the COF-102 interior, thus their
pendant functionality is distributed throughout the mate-
rial.[17]
Dodecyl-functionalized COF-102 (COF-102-C12) was
obtained by condensing mixtures of 1 and 2 under solvother-
mal conditions (mesitylene/1,4-dioxane 1:1 v/v, 908C, 24 h).
Samples of COF-102-C12 were isolated as microcrystalline
powders by filtration and were activated under vacuum at
908C for 13 h. Fourier transform infrared (FTIR) spectra of
activated COF-102 and COF-102-C12 were indicative of
À
boroxine-linked materials, as judged by the intense B O
stretch at 1343 cmÀ1 and attenuated O H stretch relative to
À
the analogous bands of their boronic acid precursors (Fig-
ure 1a). These spectra are consistent with those previously
reported. The spectrum of a COF-102-C12 sample containing a
27% loading of the dodecyl-functionalized monomer 2
showed stretches for C(sp3)–H bonds at 2900 cmÀ1 not
observed in the unfunctionalized COF-102 samples. COF-
102-C12 samples also typically showed more intense residual
À
O H stretches, which we attribute to dangling boronic acid
moieties opposite the dodecyl chains in the lattice. Overall,
these spectra indicate that the co-condensation strategy
produces a dodecyl-functionalized, boroxine-linked material
similar to COF-102.
The percent incorporation of 2 into the COF-102-C12
1
lattice (TC12) was evaluated by H NMR spectroscopy after
the material was digested in CD3CN/D2O (3:1 v/v). The ratio
of 1:2 was determined by comparing the integration of the
resonance at 0.84 ppm, corresponding to the -CH3 protons of
the dodecyl chain, to that at 7.25 ppm, corresponding to an
aromatic proton found in both monomers (Figure 1b). The
TC12 values calculated using this approach were very close to
the feed ratios of the monomers used in the COF-102-C12
synthesis, up to feed ratios of 33%. Higher feed ratios (50%)
still produced crystalline COF-102-C12, albeit with TC12 levels
that did not exceed 37%. This loading level might represent
an upper limit for incorporating truncated monomers into the
COF-102 lattice, in which the truncated building block
comprises more than one-third of the network. In preliminary
experiments, we isolated 2 as an amorphous boroxine-linked
network instead of as a monomeric tris(boronic acid). When
co-condensed with 1, the boroxine form of 2 provided low TC12
values, further suggesting that boroxine hydrolysis is slow
[*] D. N. Bunck, Prof. Dr. W. R. Dichtel
Department of Chemistry and Chemical Biology, Cornell University
Baker Laboratory, Ithaca, NY 14853 (USA)
E-mail: wdichtel@cornell.edu
[**] This research was supported by startup funds provided by Cornell
University, the NSF CAREER award (CHE-1056657), and a 3M
Nontenured Faculty Award. We also made use of the Cornell Center
for Materials Research (CCMR) facilities which are supported by the
NSF Materials Research Science and Engineering Centers (MRSEC)
program (DMR-1120296). D.N.B. acknowledges the award of a
Graduate Research Fellowship from the NSF.
Supporting information for this article is available on the WWW
Angew. Chem. Int. Ed. 2012, 51, 1885 –1889
ꢀ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1885