Water chains in hydrophobic crystal channels: Nanoporous materials as supramolecular analogues of carbon nanotubes

Article abstract of DOI:10.1002/anie.201002418

Pore design: Nanoporous steroidal crystals can be engineered for a range of channel properties. For example, the walls may be coated with aromatic groups to mimic the interior of carbon nanotubes (CNTs). The pores in these structures are occupied by water wires , as proposed for water in CNTs themselves. (Figure Presented).

Full text of DOI:10.1002/anie.201002418

Angewandte
Chemie
DOI: 10.1002/anie.201002418
Water Wires
Water Chains in Hydrophobic Crystal Channels: Nanoporous Materials
as Supramolecular Analogues of Carbon Nanotubes**
Ramalingam Natarajan, Jonathan P. H. Charmant, A. Guy Orpen, and Anthony P. Davis*
The behavior of water in narrow apolar pores has
attracted much recent interest. Although it might
seem that such channels should repel water,[1] it
transpires that they can be hydrated and moreover
that their “hydrophobic” nature promotes rapid water
flow. The phenomenon is observed in biology, where
the pores of aquaporins (water-transporting proteins)
are composed largely of hydrophobic amino acids.[2] It
is also seen for carbon nanotubes (CNTs) which,
according to both theoretical[3] and experimental[4]
studies, allow rapid passage of water molecules.
Water flow through CNTs can be selective[5] and
controllable,[6] suggesting applications in water-purifi-
cation[7] and nanofluidic devices.
To understand these systems, it is important to
gather structural information on water in hydrophobic
environments. Especially relevant is the “water wire”,
a one-dimensional hydrogen-bonded chain of water
molecules. The water in aquaporin channels adopts
this motif,[2] as does the water in the narrowest and
best-understood CNTs.[3a,b,e] However, while there are
many crystal structures which show water molecules in
single file,[8] there is limited information on water
wires in purely hydrophobic environments.[9] In par-
ticular there seem to be no structures in which water
wires are surrounded exclusively by p systems. We
now report the crystal engineering of channels
bounded by nonpolar aromatic units, and the struc-
tural characterization of linear chains of water mole-
Figure 1. a) Previously described steroidal bis-phenylureas forming nanoporous crystals.
b) The structure of 1b in the crystal, showing the bound water molecule. c) Packing
diagram for 1b viewed along the axis of the channels. A single steroid is highlighted in
the centre of the diagram, showing how the ester CH3 groups form part of the surface of
one channel while the C3-terminal NHPh groups protrude into another. The ester CH3 and
cules within these pores.
We have previously shown that steroidal bis-
phenylureas 1 (Figure 1a) crystallize as monohydrates
to form structures with hexagonal P61 symmetry,
containing one-dimensional channels of unusually C3 NHPh groups are shown in space-filling mode. Also highlighted are the water
large diameter (Figure 1b,c).[10] Groups R1 and R2
molecules bound to each unit of 1b.
are directed into the channels, so that in principle
they can be varied substantially without disrupting the
crystal packing. Potentially, this should allow a broad scope
for tuning of channel size and properties. Indeed, our original
report[10] described three structures 1a–c, in which group R1
was altered to give pore diameters of 14.3, 12.3, and 11.6 ꢀ,
respectively.
[*] Dr. R. Natarajan, Dr. J. P. H. Charmant, Prof. A. G. Orpen,
Prof. A. P. Davis
Having shown that R1 could be varied, we sought to
confirm that ester group CO2R2 could also be used to tune the
properties of the crystals. Included among our targets were
School of Chemistry, University of Bristol
Cantock’s Close, Bristol, BS81TS (UK)
Fax: (+44)117-929-8611
À
steroids 2–4, featuring large aromatic groups linked to C24 O
E-mail: anthony.davis@bristol.ac.uk
via two-carbon tethers. Modeling suggested that this arrange-
ment would allow the aromatic groups to lie roughly parallel
to the channel axis, lining the walls and reducing the
diameters to about 6 to 7 ꢀ. We could thus create an
environment not dissimilar to that inside a CNT.[11] Esters 2–4
[**] This work was supported by the EPSRC (EP/E021581/1). We thank
Drs. G. Whittell and M. F. Haddow for assistance with TGA and X-
ray crystallography, respectively.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201002418.
Angew. Chem. Int. Ed. 2010, 49, 5125 –5129
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5125

Communications
aromatic groups overlapped in a helical sequence, like scales
applied to a tightly curved surface. Despite the hydrophobic-
ity of the channels, no organic solvent was included. Instead,
the crystallography detected electron density in the channels
consistent with one water molecule per steroid. The presence
of water in the channels was confirmed by thermogravimetric
analysis (TGA) of the freshly grown air-dried crystals. In each
case the TGA trace revealed weight loss equivalent to two
molecules of water per steroid, occurring mainly between 140
and 1708C with stabilization at about 1808C.[12] Loss of two
water molecules per steroid is expected, one from the pore
and the other from the individual steroidal binding sites (see
Figure 1).
Considered in detail, the crystal structure of 2[13] features
naphthyl units making edge-to-face contact with an inter-
plane angle of 598 (Figure 2a). The structure appears to gain
were synthesized from 1b by methyl ester hydrolysis followed
by O alkylation of the resulting carboxylate.[12] In each case
needlelike crystals were obtained through slow evaporation
from acetone–water or methyl acetate–water mixtures, and
were subjected to X-ray crystallography at 100 K.
À
stabilization from C H···p interactions (dCÀH···p(centroid) = 2.8
and 2.9 ꢀ and qCÀH···p(centroid) = 133 and 1438). The pore
diameter ranges from 5.5 to 6.6 ꢀ. The naphthyl groups (the
smallest studied in this work) are not quite large enough to
provide a continuous coating for the pore, so that gaps of
about 2.4 ꢀ appear between turns of the helices (Figure 2d).
However these gaps do not reveal any polar groups, so that
The crystal structures obtained for 2–4 were isostructural
with those of 1. As expected, the ester aromatic units came
together in the channel region of the structures, essentially
defining the walls of the pores (see Figure 2). Adjacent
Figure 2. a–c) Crystal structures of 2, 3, and 4, respectively, viewed along the c axis. Thermal ellipsoids (50% probability level) are shown for one
asymmetric unit in each case, at the top left of each diagram. The aryl groups which line the channels, and the enclosed water molecules, are
shown in space-filling mode. Water-wire hydrogen atoms in 2 and 3 were inserted and positioned by modeling, as discussed in the text. In the
case of 4, all four water positions are shown as occupied. d,f,h) Crystal structures of 2, 3, and 4, respectively, viewed perpendicular to the c axis.
Three unit cells are represented in each case. Only channel-enclosed water and surrounding aryl groups are shown. e,g,i) Channel-enclosed water
positions for 2, 3, and 4, respectively, viewed perpendicular to the c axis. Two unit cells are represented in each case.
5126
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 5125 –5129

Angewandte
Chemie
the channel is entirely hydrophobic. The naphthyl units show
modest disorder or thermal motion (mean square atomic
displacement Ueq = 0.09–0.15 ꢀ2; see Figure S1 in the Sup-
porting Information), which is not unexpected considering
their position at the channel surface. The water molecules
within the pore form helical chains which follow the lines of
the interstices between the naphthyl helices. Displacement
parameters are large (Ueq of water oxygen atoms = 0.79 ꢀ2),
so the water molecules cannot be located precisely. However,
water wire in an aliphatic crystal environment.[9a] Modeling of
the water hydrogen atom positions, as for 2, yielded a
À
structure with O H···O angles of 1568 and hydrogen bonding
H···O distances of 1.64 ꢀ. As shown in Figure 3b, the water
À
wire takes part in O H···p interactions with the pyrene
p surfaces.[17] However this model is tentative, given the
uncertainties surrounding both water and pyrene positions.
Structures in which the water wire and pyrene units lack 6-
fold screw symmetry are also possible, as highlighted by the
case of 4 discussed below.
the distance between the centroids (3.15 ꢀ) is well within the
[9a]
normal range for O H···O hydrogen bonding. The water
In the crystal structure of 4,[18] the trans-diazobenzene
fragments overlap in a continuous fashion, defining a channel
of diameter 5.7 ꢀ (minimum) to 8.7 ꢀ (maximum) (Fig-
ure 2c,h). The channel-enclosed water molecules were found
in four positions in the asymmetric unit, each with 25%
occupancy (Figure 2i, Figure 4a). As shown in Figure 4a,
three of the positions form a roughly equilateral triangle in a
plane nearly perpendicular to the channel axis, while the
fourth lies at the approximate centre of this triangle. The
water positions were quite precisely defined, with an average
Ueq of 0.094 ꢀ2. Random occupation of these sites is
unrealistic, as many of the O···O distances between successive
asymmetric units would be too small (Table S1 in the
Supporting Information). Certain sequences must therefore
be preferred and, given the equal occupancy of all four
positions, a regular structure with an (H2O)4 repeat unit seems
probable. Of the six possible sequences, five require at least
one O···O distance of 2.2 ꢀ or less.[12] The remaining option
(CBDA, Figure 4b) was therefore considered the most likely.
To model the water wire, twelve oxygen atoms were
positioned according to this sequence, encompassing three
cycles of the water chain and two unit cells of the crystal.
Hydrogen atoms were added and their positions were energy-
minimized using periodic boundary conditions.[15] The result-
ing structure is shown in Figure 4c,d. The extra-chain hydro-
gen atoms are placed in a variety of situations, being directed
À
chain (Figure 2e) was modeled with the assistance of
Materials Studio software.[14] Water hydrogen atoms were
added to the oxygen atoms and then allowed to move[15] in an
energy minimization with periodic boundary conditions
employing the COMPASS 27 force field with Ewald option.
À
The minimization yielded a water wire with O H···O angles
of 1608 and hydrogen bonding distances (H···O) of 2.22 ꢀ.
The extra-chain hydrogen atoms were positioned in the
groove separating the turns of the naphthyl helix, roughly
equidistant (ca. 3 ꢀ) from three naphthyl hydrogen atoms
(Figure 3a).
Figure 3. Environments of water-wire molecules in 2 (a) and 3 (b)
from crystallography and modeling. In (a), dotted lines represent close
contacts between OH and naphthyl hydrogen atoms (2.9–3.1 ꢀ). In
=
towards the benzenoid aromatic surfaces, the N N p surface,
À
(b), the proposed O H···p contact between pyrene and water wire is
or the CH groups. All the environments are essentially
apolar; although there is potential for H-bonding to azo
nitrogen lone pairs, none of the OH groups is properly
positioned for such an interaction.
visualized in space-filling mode.
The crystal structure of pyrenylethyl ester 3[16] (Figure 2b)
reveals a similar assembly of aromatic units, with an angle of
58.548 between the pyrene planes. Again, the structure
In conclusion, we have shown that crystal engineering
with nanoporous steroids 1 can be extended to create pores
with aromatic internal surfaces. To an approximation, these
materials can be seen as supramolecular analogues of carbon
nanotubes, employing overlapping planar aromatic surfaces
to mimic the continuous curved p system inside a CNT.
Although the mimicry is by no means perfect, crystallinity
confers a key advantage in that detailed structures can be
obtained. The observation of water wires within the pores
illustrates the value of this approach. The properties of water
in CNTs have attracted much attention, driven by theoretical
interest and potential applications, but relevant structural
data have been scarce. This work has provided the first
characterization of water wires in environments similar to
CNT interiors (that is, surrounded by apolar aromatic units).
More generally it contributes to the understanding of water in
nonpolar environments, including the water chains found in
aquaporins. In future work we hope to show that nanoporous
À
appears to be stabilized by C H···p interactions between
neighboring aryl units. Unlike the naphthyl groups in 2 the
larger pyrene unit is capable of lining the pore surface without
leaving substantial gaps (Figure 2 f). The pore diameter
ranges from 6.4 to 7.6 ꢀ. Despite the apparently favorable
packing, the pyrene carbon atoms show large displacement
parameters (Ueq = 0.43–0.67 ꢀ2), indicating significant disor-
der or thermal motion (Figure S2 in the Supporting Informa-
tion). The water molecules occupying the pores form helical
chains (Figure 2g) and also show large displacement param-
eters (Ueq of water oxygen atoms = 0.66 ꢀ2). In contrast to 2,
the water wire passes close to the centroids of the aromatic
surfaces. The distance between the water oxygen atoms is also
shorter (2.6 ꢀ). Although this is low for hydrogen-bonded
water molecules, the same spacing has been proposed for a
Angew. Chem. Int. Ed. 2010, 49, 5125 –5129
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5127

Communications
[1] M. S. P. Sansom, P. C. Biggin, Nature 2001, 414, 156; S. Granick,
S. C. Bae, Science 2008, 322, 1477.
[2] P. Agre, Angew. Chem. 2004, 116, 4377; Angew. Chem. Int. Ed.
2004, 43, 4278; D. D. Kozono, M. Yasui, L. S. King, P. Agre, J.
Clin. Invest. 2002, 109, 1395; for a functional aquaporin model
see: M. S. Kaucher, M. Peterca, A. E. Dulcey, A. J. Kim, S. A.
Vinogradov, D. A. Hammer, P. A. Heiney, V. Percec, J. Am.
Chem. Soc. 2007, 129, 11698.
[3] a) G. Hummer, J. C. Rasaiah, J. P. Noworyta, Nature 2001, 414,
188; b) A. Kalra, S. Garde, G. Hummer, Proc. Natl. Acad. Sci.
USA 2003, 100, 10175; c) K. Koga, G. T. Gao, H. Tanaka, X. C.
Zeng, Nature 2001, 412, 802; d) D. Takaiwa, I. Hatano, K. Koga,
H. Tanaka, Proc. Natl. Acad. Sci. USA 2008, 105, 39; e) L. Wang,
J. J. Zhao, F. Y. Li, H. P. Fang, J. P. Lu, J. Phys. Chem. C 2009, 113,
5368.
[4] a) J. K. Holt, H. G. Park, Y. M. Wang, M. Stadermann, A. B.
Artyukhin, C. P. Grigoropoulos, A. Noy, O. Bakajin, Science
2006, 312, 1034; b) H. J. Wang, X. K. Xi, A. Kleinhammes, Y.
Wu, Science 2008, 322, 80; c) J. K. Holt, Adv. Mater. 2009, 21,
3542.
[5] F. Fornasiero, H. G. Park, J. K. Holt, M. Stadermann, C. P.
Grigoropoulos, A. Noy, O. Bakajin, Proc. Natl. Acad. Sci. USA
2008, 105, 17250.
[6] Z. K. Wang, L. J. Ci, L. Chen, S. Nayak, P. M. Ajayan, N.
Koratkar, Nano Lett. 2007, 7, 697; Y. Maniwa, K. Matsuda, H.
Kyakuno, S. Ogasawara, T. Hibi, H. Kadowaki, S. Suzuki, Y.
Achiba, H. Kataura, Nat. Mater. 2007, 6, 135; X. S. Li, G. Y. Zhu,
J. S. Dordick, P. M. Ajayan, Small 2007, 3, 595; X. J. Gong, J. Y.
Li, H. J. Lu, R. Z. Wan, J. C. Li, J. Hu, H. P. Fang, Nat.
Nanotechnol. 2007, 2, 709.
[7] M. A. Shannon, P. W. Bohn, M. Elimelech, J. G. Georgiadis, B. J.
Marinas, A. M. Mayes, Nature 2008, 452, 301.
[8] M. Mascal, L. Infantes, J. Chisholm, Angew. Chem. 2006, 118, 36;
Angew. Chem. Int. Ed. 2006, 45, 32.
[9] a) U. S. Raghavender, S. Aravinda, N. Shamala, Kantharaju, R.
Rai, P. Balaram, J. Am. Chem. Soc. 2009, 131, 15130; b) U. S.
Raghavender, Kantharaju, S. Aravinda, N. Shamala, P. Balaram,
J. Am. Chem. Soc. 2010, 132, 1075.
[10] A. L. Sisson, V. del Amo Sanchez, G. Magro, A. M. E. Griffin, S.
Shah, J. P. H. Charmant, A. P. Davis, Angew. Chem. 2005, 117,
7038; Angew. Chem. Int. Ed. 2005, 44, 6878.
Figure 4. Modeling of the water wire in 4. a) The asymmetric unit from
crystallography, viewed down the c axis, showing the four environ-
ments for water molecules within the pore (labelled A–D, each with
25% occupation). Non-hydrogen atoms are shown as thermal ellip-
soids. b) The sequence of water environments proposed for the water
wire, viewed perpendicular to the c axis. O···O distances are shown in
ꢀ. Water oxygen atoms are shown as thermal ellipsoids. c) Modeled
water wire, based on the CBDA sequence. The twelve water molecules
shown inhabit two unit cells of the crystal. d) One repeat unit of the
water wire with neighboring azobenzene groups, viewed along a line
close to the c axis, with the sequence CBDA heading out of the page.
Hydrogen bonds are shown as dotted lines.
[11] The internal diameter of a (6,6) single-walled carbon nanotube,
as featured in many simulations,[3] is ca. 5 ꢀ.
[12] For details see the Supporting Information.
[13] Crystal data for 2: C57H68N6O5 2H2O, M = 953.21; hexagonal,
P61, a = b = 28.9480(6), c = 11.4319(2) ꢀ, V= 8296.3(3) ꢀ3, Z =
6, qmax = 27.488, MoKa radiation, l = 0.71073 ꢀ, T= 100 K, m =
0.076 mmÀ1, 1calc = 1.142 MgmÀ3. 47396 reflections measured,
12306 unique (Rint = 0.0204), 658 parameters, and 9 restraints;
R1 = 0.0485
[I > 2s(I)],
wR2 = 0.1478,
GooF = 1.086.
CCDC 774502 contains the supplementary crystallographic
data for this compound. These data can be obtained free of
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data_request/cif.
steroidal crystals can be engineered for a range of other
characteristics, including corrugated surfaces and polar/
charged interiors.
[14] Materials Studio 4.4, Accelrys Software Inc.
[15] All atoms apart from the water-chain hydrogen atoms were kept
stationary during minimization.
[16] Crystal data for 3: C63H70N6O5·2H2O, M = 1027.29; hexagonal,
P61, a = b = 29.315(3), c = 11.5430(11) ꢀ, V= 8590.5(14) ꢀ3, Z =
6, qmax = 65.148, CuKa radiation, l = 1.54178 ꢀ, T= 100 K, m =
0.620 mmÀ1, 1calc = 1.189 MgmÀ3. 65842 reflections measured,
8353 unique (Rint = 0.0427), 713 parameters, and 260 restraints;
Received: April 23, 2010
Published online: June 22, 2010
R1 = 0.0639
[I > 2s(I)],
wR2 = 0.1801,
GooF = 1.127.
Keywords: crystal engineering · hydrogen bonding ·
hydrophobic effect · supramolecular chemistry · water chains
CCDC 774503 contains the supplementary crystallographic
data for this compound. These data can be obtained free of
.
5128
www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 5125 –5129

Angewandte
Chemie
charge from The Cambridge Crystallographic Data Centre via
www.ccdc.cam.ac.uk/data request/cif.
8401.5(6) ꢀ3, Z = 6, qmax = 27.488, MoKa radiation, l =
0.71073 ꢀ, T= 100 K, m = 0.079 mmÀ1 1calc = 1.192 MgmÀ3
,
.
[17] S. Suzuki, P. G. Green, R. E. Bumgarner, S. Dasgupta, W. A.
Goddard, G. A. Blake, Science 1992, 257, 942; S. Scheiner, T.
Kar, J. Pattanayak, J. Am. Chem. Soc. 2002, 124, 13257; S. Li,
V. R. Cooper, T. Thonhauser, A. Puzder, D. C. Langreth, J. Phys.
Chem. A 2008, 112, 9031.
47722 reflections measured, 12208 unique (Rint = 0.0369), 791
parameters, and 9 restraints; R1 = 0.0397 [I > 2s(I)], wR2 =
0.1115, GooF = 1.102. CCDC 774504 contains the supplemen-
tary crystallographic data for this compound. These data can be
obtained free of charge from The Cambridge Crystallographic
Data Centre via www.ccdc.cam.ac.uk/data_request/cif.
[18] Crystal data for 4: M: C59H70N8O5·2H2O, M = 1007.26; hexag-
onal,
P61,
a = b = 29.2924(12),
c = 11.3062(5) ꢀ,
V=
Angew. Chem. Int. Ed. 2010, 49, 5125 –5129
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
5129