Anomalous Water Molecules and Mechanistic ... - ACS Publications

Jan 26, 2010 - Trantham , E. C.; Rorschach , H. E.; Clegg , J. S.; Hazlewood , C. F.; Nicklow , R. M.; Wakabayashi , N. Biophys. J. 1984, 45, 927– 9...
0 downloads 0 Views 4MB Size
Download PDF
J. Phys. Chem. B 2010, 114, 2091–2099

2091

Anomalous Water Molecules and Mechanistic Effects of Water Nanotube Clusters Confined to Molecular Porous Crystals Makoto Tadokoro,*,† Takashi Ohhara,‡ Yuhki Ohhata,† Takaaki Suda,† Yuji Miyasato,† Takeshi Yamada,§ Tatsuya Kikuchi,§ Ichiro Tanaka,| Kazuo Kurihara,‡ Masaharu Oguni,⊥ Kazuhiro Nakasuji,# Osamu Yamamuro,§ and Kuroki Ryota‡ Department of Chemistry, Faculty of Science, Tokyo UniVersity of Science, Kagurazaka 1-3, Shinjuku-ku, Tokyo 162-8601, Japan; Neutron Science Research Center, Japan Atomic Energy Research Institute, 2-4 Shirane Shirakata, Tokai, Naka-gun, Ibaraki 319-1195, Japan; Neutron Science Laboratory, Institute for Solid State Physics, UniVersity of Tokyo, 106-1 Shirakata, Tokai, Naka-gun, Ibaraki 319-1106, Japan; Department of Biomolecular Functional Engineering, College of Engineering, Ibaraki UniVersity, 4-12-1 Nakanarisawa-cho, Hitachi, Ibaraki 316-8511, Japan; Department of Chemistry, Graduate School of Science and Engineering, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan; and Department of Subject of General Education, Fukui UniVersity of Technology, 3-6-1 Gakuen, Fukui City, Fukui 910-8505 ReceiVed: July 22, 2009; ReVised Manuscript ReceiVed: NoVember 27, 2009

The movement of water molecules in the limited space present within nanoscale regions, which is different from the molecular motion of bulk water, is significantly affected by strong interfacial interactions with the surrounding outer walls. Hence, most of the water molecules that are confined to nanochannel spaces having widths less than ca. 2 nm can generally be classified together as “structural water”. Since the motions of such water molecules are limited by interfacial interactions with the outer wall, the nature of structural water, which is strongly influenced by the interactions, will have different characteristics from normal water. For our investigations on the characteristics of structural water, we have developed a nanoporous crystal with a diameter of ca. 1.6 nm; it was constructed from 1-D hydrophilic channels by self-organization of the designed molecules. A tubelike three-layered water cluster, called a water nanotube (WNT), is formed in each internal channel space and is regulated by H-bonds with the outer wall. The WNT undergoes a glass transition (Tg ) 107 K) and behaves as a liquid; it freezes at 234 K and changes into an icelike nanotube cluster. In this study, the structure of the WNT is investigated through neutron structure analysis, and it is observed to stabilize by a mechanistic anchor effect of structural water. Furthermore, from neutron-scattering experiments, it is seen that a few water molecules around the center of the WNT move approximately with the same diffusion constant as those in bulk water; however, the residence time and average jump length are longer, despite the restrictions imposed by the H-bonding with structural water. The behavior of mobile water within a WNT is investigated; this can be used to elucidate the mechanism for the effect of structural water on vital functions on the cell surface. 1. Introduction The movement of water molecules in the limited space present in nanometer-scale regions is significantly affected by strong interfacial interactions with the surrounding outer walls.1 For example, it has been theoretically predicted that, when water molecules come in contact with a hydrophobic interface inside carbon nanotubes, osmotic water transport through the carbon nanotube membranes immediately occurs; further, such water molecules will have a one-dimensional (1-D) chain hydrate in their internal space.2 Moreover, it has been previously reported that the critical water-ice phase transition point of the molecules depends on the tube width,3 and the molecules freeze into a novel ice polymorph with a multilayered helix structure.4 According to theoretical calculations, the melting point of each confined tubelike water cluster constructed using five- to eightmembered rings does not decrease with the pore sizes;5 however, †

Tokyo University of Science. Japan Atomic Energy Research Institute. University of Tokyo. | Ibaraki University. ⊥ Tokyo Institute of Technology. # Fukui University of Technology. ‡ §

it has been experimentally demonstrated that the melting point increases with the width of single-walled carbon nanotubes.6 The structure of a confined water nanotube (WNT), which is a tubelike three-layered water cluster, can be appropriately designed by using interfacial interactions such as those occurring in hydrophobic and hydrophilic areas,7 the dispositions of hydrogen bonding (H-bonding) groups on porous channel surfaces, and the widths, shapes, and dimensionality of such channels.8 In this manner, such design strategies have led to the development of water clusters such as WNTs9,10 with a unique phase and glass transitions,11 which are affected by interfacial interactions on the channel surface. We have previously developed a nanoporous crystal with a diameter of ca. 1.6 nm from 1-D hydrophilic channels by self-organizing the designed molecules.12 A WNT is thus formed in each internal channel space and is regulated by H-bonding with the outer wall. In supramolecular chemistry, crystal engineering has attracted considerable research interest because of the possibilities it has offered for controlling the crystal structure by self-organizing artificial molecular building blocks.13 In order to assemble thermodynamically stable molecular arrays, it is important to

10.1021/jp9069465  2010 American Chemical Society Published on Web 01/26/2010

2092

J. Phys. Chem. B, Vol. 114, No. 6, 2010

control the crystal arrangements of supramolecular crystals by using mutual intermolecular interactions such as complementary H-bonds and directional coordination bonds; this requires thermal equilibrium in the solution phase under moderate conditions. Some water clusters occupy the interstitial spaces of such supramolecular crystals. The study of this phenomenon can be developed into a new branch of chemistry, which will involve the determination of water mobility with respect to the spectrochemical characteristics14 and structural dynamics of the water clusters.15 In this study, we have introduced a tris-2,2′-biimidazole cobalt(III) complex ([CoIII(H2bim)3]3+) and a trimesate (TMA3-) as the building blocks for constructing a nanoporous molecular

crystal. The biimidazolate metal complex can not only form intermolecular H-bonds of the dual N-H · · · N type by partial deprotonation16 but also form complementary H-bonds with a carboxyl group. In addition, the difference between the acidities of the NH group and carboxylate group results in strong ionic H-bonds based on the two Nδ+sH · · · Oδ- type H-bonds.17 On the basis of this, we can now prepare a molecular crystal {[CoIII(H2bim)3](TMA) · 20H2O}n (1) with 1-D nanoporous channels by alternating stacking and adjusting each pore between the ∆- and Λ-hexagonal sheets with a (6,3)-net.18 The two chiral sheets of the molecular crystal are constructed by alternating H-bonding between the ∆- or Λ-[CoIII(H2bim)3]3+ and TMA3-. A large tubelike cluster of the WNT is formed from ca. 20 water molecules in the periodic unit and is stabilized in each nanochannel pore of 1. In previous works, we studied the structure of a frozen WNT in 1, or an ice nanotube (INT). We conducted X-ray crystal analysis at 198 K, which is below the phase transition temperature of 234 K, along the frozen-in direction. We obtained the following results. First, we found that the periodic unit of the INT was constructed from 60 water molecules. The length of the c axis of 1 with INT is approximately 3 times that with WNT because the periodicity of INT formation increases, while the same space group is retained. Only a limited number of water molecules in the secondary and tertiary regions of the WNT probably participate in the water-ice transition. Second, all the water molecules in the INT form H-bonds with each other, and the INT is separated into three-layered structures with primary and secondary layers and tertiary domains. The INT is also attached to most of the carboxylic O atoms of TMA3- in the channel framework. The H-bonded networks of water molecules in the primary layer are formed from multicyclic water polygons such as pentagons, hexagons, and octagons. From X-ray analysis, it is observed that TMA3- exists as 1.5 molecules in an asymmetric unit; therefore, the carboxylic O atoms exist as a total of nine atoms, individually. A carboxylic O(8) atom does not form an H-bond with water oxygen in the primary layer. Hence, no water oxygen exists within less than 4 Å around O(8). In the secondary layer and tertiary regions, a 1-D spirocyclic water molecular cluster is constructed from a

Tadokoro et al. periodic structure with alternating linkages between a dynamic disordered water molecule (O(40) or O(40)*) and a spirobihexagon (a water dodecamer) and stabilized. The dodecamer is recognized as one of the smallest models of a crystal embryo in the cubic ice-phase Ic during heterogeneous nucleation. A water molecule in the link between the dodecamers is disordered at two sites of O(40) and O(40)* in the tertiary domain, even at temperatures less than the freezing point. The disordered water molecule has an unusual two-handed H-bond and is also located at the center of the cyclic belt-like water hexagons belonging to the primary layer. On the other hand, from the X-ray crystal analysis conducted at 296 K, the electron densities of water oxygens are found to be located around the outer walls of the nanochannel; however, the electron densities cannot be accurately ascertained at this temperature due to the electron thermal instability of each water oxygen. (Figure 1) In this study, we focus on the structures and properties of a liquid-like WNT by conducting neutron structure analysis and neutron scattering. In a certain single-walled carbon nanotube with a diameter of ∼1.6 nm, which is similar to the porous size of the framework of 1, the WNT confined to the nanotube has been identified as a single-layered structure of INT. It also has a high melting point near room temperature.19 However, the WNT of 1 has the three-layered structure of INT due to H-bonds, despite the fact that its porous size is similar to that of a carbon nanotube. This is because the interfacial surface of the channel wall has different compositions: the interfacial surface in the carbon nanotube is formed from hydrophobic π-arene, whereas that of the channel wall of 1 is from hydrophilic H-bonding carboxyl groups. It is also well-known that in the case of hexagonal ice Ih, which is the most typical structure found in the atmosphere, the hydrogen atoms (H atoms) in the network are in a disordered state. Moreover, the freezing-in, or the glass transition, of the rearrangement of water molecules occurs at Tg ≈ 136 K or even at higher temperatures.20 No phase transition due to the ordering of the H atom positions is observed below Tg because of the immobility of the molecules. The phase transition is realized at 72 K only after hydroxide ions are introduced in place of water molecules in the network, or, in other words,21 phase transition is realized only after ice is doped with some L-type Bjerrum defects. The phase transition and glass transition behaviors of WNT in 1 have been studied using adiabatic calorimetry.22 The heat capacities of 1 are measured by repeating thermometry from 4 to 300 K along the melt-in direction; the energy supply yields a temperature increase from 1.0 to 2.5 K under adiabatic conditions. The glass transition of WNT in 1 is observed at Tg ) 107 K; this is discussed as a liquid phenomenon involving a small number of water molecules around the center of the WNT. 2. Experimental Section 2.1. Preparation of Crystals 1 and 2. An H2O solution (10 cm3) containing [Co(H2bim)3](NO3)3 (0.16 g, 0.24 mmol) was added to an H2O (5 cm3) solution containing dissolved H3TMA (0.026 g, 0.12 mmol) and LiOMe (0.012 g, 0.32 mmol) to yield orange powder precipitates. The suspension was then heated to 120 °C and dissolved completely. It was placed overnight at 45 °C to obtain orange hexagonal crystals 1. Orange crystal 2 was prepared by using the same method as that for 1 except that D2O was used instead of H2O. Elemental analyses of crystal 1 after vacuum drying at 100 °C; found C, 45.62%; H, 3.64%; N, 22.91%; calculated values for [CoIII(H2bim)3](TMA) · 2.5H2O: 45.45%, H, 3.67%, N, 23.56%. TG (thermogravimetric analysis): 35.03%, calcd at 35.02% for 20 H2O.

Mechanistic Effects of Water Nanotube Clusters

J. Phys. Chem. B, Vol. 114, No. 6, 2010 2093

Figure 1. Electron density map of the porous channel unit of 1 along the c axis. The electron densities of structural water in the primary layer are localized near the channel wall, but not observed around the central axis of WNT. Most electron densities of the water molecules belonging to the secondary layer and all those belonging to the tertiary domain in WNT could not be observed by X-ray structure analysis. Nonobservable water molecules around the central axis of WNT show significant movement into the channel maintaining free rotation and causing a liquid state unlike bulk water.

2.2. Single-Crystal Structure Analysis by Neutron Diffraction. The crystal 1 used for the neutron structure analysis was grown up to a size of 3.90 × 1.05 × 0.44 mm3 for several weeks, and data on Bragg’s reflections were gathered by a 2-D neutron data-collection apparatus, the BIX-III diffractometer,23 at the Japan Atomic Research Institute. Crystal 1 was placed in a closed glass capillary (φ 1.6 mm); this was then fixed on an aluminum pin and mounted on the BIX-III diffractometer, which was equipped with a 2-D neutron data-collection imaging plate24 and set up at the JRR-3 M reactor of the Japan Atomic Energy Research Institute (JAERI). The results of the neutron crystal analysis of 1 at 293 K are as follows: C27H61N12O26Co; FW ) 542.00; monoclinic structure of space group C2/c (#15); cell dimensions a ) 16.338(2) Å, b ) 29.335(2) Å, c ) 10.909(2) Å, β ) 90.336(10)°, V ) 5228.3(12) Å3, and Z ) 8; Dcalcd ) 1.377 gcm-3; neutron (λ ) 1.510 0 Å); R1 ) 12.33%, and wR2 ) 23.20%. The maximum and minimum highest peaks in the final differential map are 0.093 and -0.107 e-/Å3, respectively, and the goodness of fit for 397 parameters was equal to ) 1.238 (all, 2θ < 114°) for 3565 reflections. The neutron diffraction data were collected by using the ω scan method (oscillation range ∆ω ) 1.0°) at 293 K with perfect-silicon-crystalmonochromated neutron radiation (λ ) 1.510 0 Å). Since BIXIII is a single-axis cylindrical diffractometer, there is a large blind region around the rotation axis. To reduce the area of this blind region, the data (ω-scan (∆ω ) 0.5°)) were collected by

changing the angle of the aluminum pin (to approximately 180, 135, and 90°) instead of changing the c circle position of the ordinary X-ray diffractometer. The data on reflections were integrated in the Denzo program, and the data collections (without absorption collection) were performed using the Scalepack program.25 The positional parameters of the nonhydrogen atoms were constrained using the coordinates obtained by X-ray analysis of [Co(Hbim)3](TMA) · 20H2O as an initial model.12 The refinement was on F2 (a square of the structural factor) against all reflections by full-matrix least-squares using SHELXL-97. Hydrogen atoms were observed in difference Fourier maps (neutron scattering length density maps) as negative and positive peaks. All the water molecules in WNT had one constant O-H distance, 0.96(2) Å, while the distance between two H atoms was 1.51(3) Å. The parameters for O(8), O(14), H(4A), H(5A), H(5B), H(8A), H(8B), H(9B), H(14A), and H(14B) atom were calculated for isotropic refinement; in neutron analysis, the parameters for cobalt atom were also derived since it was observed to be present over a small cross section on neutron atomic scattering. Numerical absorption correction was performed using the face indices determined by the SMART CCD diffractometer with the ABSG program.26 2.3. Neutron-Scattering Analysis. The quasi-elastic neutron scattering (QENS) data of 1 were measured with a time-offlight (TOF) spectrometer AGNES at the Institute for Solid State Physics, The University of Tokyo. This instrument is installed

2094

J. Phys. Chem. B, Vol. 114, No. 6, 2010

at the cold neutron guide (C3-1) of JRR-3, Japan Atomic Energy Agency (Tokai, Ibaraki, Japan).27,28 Neutrons with a wavelength of 4.22 Å (for the standard mode) or 5.50 Å (for the high-resolution mode) were extracted with an array of five PG(002) monochromators and pulsed with a double Fermi chopper. The pulsed neutrons were scattered by a sample and detected with 328 3He tube detectors arranged in a wide detector bank with scattering angles of 10-130°. The energy resolution, energy range, and Q range at the standard mode, which was used in this experiment, are 120 µeV, -4 < ∆E < 20 meV, and 0.2 < Q < 2.7 Å-1, respectively. Small pieces of the crystals 1 were wrapped with aluminum foil and loaded in a concentric double-cylinder aluminum can (height ) 40 mm, inner diameter of the outer cylinder ) 14.0 mm, outer diameter of the inner cylinder ) 12.0 mm). The sample was then sealed by using an indium gasket. The amount of the sample was ca. 1.8 cm3, and the thickness of the sample was approximately 0.5 mm corresponding to the transmission of neutron beams of approximately 85%. The sample was set in a top-loading type cryostat and immediately cooled to 276 K. The elastic (fixedwindow) scan was first performed from 276 to 100 K in the cooling direction, and then from 100 to 300 K in the heating direction to investigate the effects of hysteresis of the transition. The temperature step was 5-10 K, and the duration of each measurement was 1 h. Full data sets of quasielastic and inelastic scattering were recorded at 150, 266, 256, and 246 K taking approximately 15 h for each point. Bulk water was also measured at 266, 276, 286, and 296 K for comparison. 3. Results and Discussion 3.1. Measurement of WNT Using Differential Scanning Calorimetry. Differential scanning calorimetry (DSC) was performed on the WNT of 1 between 190 and 290 K along both the melt-in and freeze-in directions. The reversible phase transitions of a freeze-in exothermic peak for the transition from WNT to INT at 234 K within a supercooled region as well those for a melt-in endothermic peak at 245 K were observed by the DSC. The temperature difference between the phase transitions with WNT substituted for heavy water molecules is almost comparable to the isotope effect on their freezing and melting processes. This similarity suggests that the DSC peaks of the phase transitions are related to the condensation and fusion of water molecules within the WNT, which indicates that only a limited number of water molecules in the channels participate in the water-ice transition (the heat of fusion ∆H ) 4.2 J · g-1). Therefore, the strongly H-bonded water molecules in the primary layer probably play a less important role in this transition; this is because, in the primary region, these molecules would be strongly H-bonded with the carboxylic oxygen of TMA3- on the surface of the channels. From the DSC measurements between 210 and 240 K, a weak broad endothermic peak of the premelting range is also observed before the phase transition. (Figure 2) The peak corresponds to the “programmed” movement of water molecules in the INT in such a manner that they reach the phase transition. Such a phenomenon is also observed at the phase transition to the melt-in of the water molecules confined to mesoporous silica.29 Such premelting phenomena have not been observed in bulk ice in which each freezing water molecule moves into the dispositions that can cause phase transitions during the melting of INT. Hence, H-bonding networks between each water molecule in INT, which is revealed at temperatures below 198 K by X-ray crystal analysis, are not suitable for the direct formation of the liquid state of WNT by the phase transition. By using neutron structure

Tadokoro et al.

Figure 2. DSC spectrum of crystal 1 confined to the WNT through heat-up: the INT melts from 245 K onward. From 210 to 240 K, a broad peak of the premelting phase is also observed. The peak from 273 K is due to melting and consequent transition to bulk water, which is added to stabilize crystal 1 in the aluminum pan. (velocity 5 K/min).

analysis, we would like to confirm that the structure of INT changes in the premelting region. 3.2. Crystal Structure of Structural Water in WNT by Neutron Diffraction Analysis. From the neutron crystal structure analysis at 298 K, the dispositions of static structural water in the WNT were determined. Considering the previous results of the X-ray structure analysis mediated by electron densities, which were conducted at 296 K, it is difficult to determine the precise positions of water oxygens in WNT; this is because all the electrons of the water molecules undergo thermally heavy fluctuations by sustaining the disorders and because of large temperatures. Detailed dispositions and Hbonding structures of structural water cannot be precisely determined despite the higher electron densities in structural water. Furthermore, the electron densities of water molecules limited to the secondary layer are hardly observed, while those belonging to the tertiary region around the center of the WNT could not be localized because of the diffusing of molecules particularly during the heavy movement of water molecules. Consequently, it is not possible to draw the entire WNT structure by using X-ray structure analysis based on the diffused electron densities. Hence, we performed the experiment on the basis of structure analysis by using neutron irradiation, which can be used to observe the diffraction intensity for an atomic nucleus as the electron has thermally more delocalized nature than the nucleus. Figure 3 shows a perspective view along the c axis of the WNT confined to a channel unit, and the dispositions of structural water are determined by neutron structure analysis at 293 K. All the water molecules near the channel wall are localized as structural water by H-bonds with the carboxyl O atom of TMA3- forming the channel frameworks. The nuclear densities of the water molecules are localized near the channel walls; however, the densities are not observed around the central axis of WNT (Figure 4). A water molecule comprises covalent bonds between two H atoms and an O atom. The former has negative scattering lengths for a nuclear density when a neutron is irradiated, whereas the latter has positive scattering lengths. Therefore, the heavy mobile water molecule cannot be discriminated by irradiating the neutron because the opposite atomic scattering lengths between H and O atoms nullify each other. We have not observed most of the nuclear densities of water molecules belonging to the secondary layer and all those belonging to the tertiary region in WNT. The nonobservable

Mechanistic Effects of Water Nanotube Clusters

J. Phys. Chem. B, Vol. 114, No. 6, 2010 2095

Figure 3. Perspective view of the structural water of WNT in a channel unit along the c axis: they are fixed by an anchor effect of hydrogen bonds with carboxylate oxygen atoms, thereby forming the channel framework. The blue closed lines and violet lines show [Co(Hbim)3]3+ and TMA3-, respectively, thereby forming the outer frameworks of the channel. The red spheres and white spheres show the oxygen atoms and hydrogen atoms of the structural water molecules, respectively. The dashed blue lines show the intermolecular hydrogen bonds forming the frameworks. The results of neutron crystal analysis of 1 at 293 K are C27H61N12O26Co, FW ) 542.00, monoclinic of space group C2/c (#15), with cell dimensions as follows: a ) 16.338(2) Å, b ) 29.335(2) Å, c ) 10.909(2) Å, β ) 90.336(10)°, V ) 52 228.3(12) Å3, and Z ) 8; Dcalcd ) 1.377 g · cm-3, neutron (λ ) 1.510 00 Å), R1 ) 12.33%, wR2 ) 23.20%, and GOF ) 1.238.

Figure 4. Atomic nucleus density map of a channel unit containing WNT along the c axis: the H and O atoms are shown by the white and violet shaded parts, respectively. Molecules are not observed around the center of the channel because of significant free rotation and translational motion of water molecules. The yellow trace lines show the shapes of each molecule and intermolecular hydrogen bonds based on the location of the atomic nucleus density map.

2096

J. Phys. Chem. B, Vol. 114, No. 6, 2010

Tadokoro et al.

Figure 5. Hydrogen-bonding networks with each atom (numbered) of structural water molecules in WNT and some hydrogen-bonded atoms observed by neutron structure analysis: the magenta and deep blue lines show [TMA]3- and [Co(Hbim)3]3+ frameworks, respectively. The H and O atoms of structural water are shown by the white and red circles, respectively. The yellow dotted lines show the hydrogen bonds. O(5) and O(8), which belong to the secondary hydrate layer, connect with two O(4) and O(14) through H(5A) and H(8B). H(4A), H(4B), and H(4C) are attached to the O(4) atom, since H(4B) and H(4C) are treated as disorders contributing to each occupancy factors of 0.327 46 and 0.673 54. O(4) is connected to another O(9) through H(4C) to form a sequence of a cyclic hydrogen-bonding network with O(1) and O(3).

nuclear densities of water molecules around the center of WNT imply that the water molecules are disordered as those in bulk water. The porous frameworks are confirmed in the structures obtained by X-ray crystal analysis. It is demonstrated that two H atoms participated in the formation of complementary dual H-bonds of Nδ+sH · · · Oδ- types between the carboxylic O atoms of TMA3- and the N atoms of Hbim- (N(4)sH(N4) · · · O(1)/ N(2)sH(N2) · · · O(2),N(6)sH(N6) · · · O(3)/ N(6)*sH(N6)* · · · O(3)*) [* ) -x + 1, y, -z + 3/2]. Two H atoms are localized to one side of the N atoms because of the difference in their acidities, and they form strong ionic H-bonds with the -1 charge of two carboxylic O atoms. Three water molecules with O(4), O(9), and O(14), which belong to the primary layer of WNT, form direct intermolecular H-bonds of O-H · · · O (O(4)sH(4A) · · · O(3), O(9)sH(9A) · · · O(1), and O(14)sH(14A) · · · O(2)) with three carboxylic O atoms of TMA3- of O(4), O(9), and O(14), respectively. (Figure 5) Their H-bonds belong to a medium-strength type with a double-well energy potential, which are different from a strong type with a single-well energy potential because each H atom localizes to one side of two water oxygen atoms.30 Two water molecules of O(5) and O(8), which belong to the secondary layer, bond with two water molecules of O(4) and O(14) through H(5A) and H(8B) to form two H-bonds of O(4)-H(5A) · · · O(5) and

O(14)-H(8B) · · · O(8), respectively. H(4A), H(4B) and H(4C) are attached to the O(4) atom, since H(4B) and H(4C) covalentbonded with O(4) are treated as disorders that contribute toward occupancy factors of 0.33 and 0.67. O(4) is H-bonded with another O(9) through H(4C) to form a sequence of a cyclic H-bonding network of O(3) · · · H(4A)sO(4)sH(4C) · · · O(9)s H(9A) · · · O(1), which inserts two water molecules to either side of O(3) and O(1). A comparison between the temperature factors of H and O atoms comprising structural water in WNT has indicated that each atom near the channel wall is successively smaller than that near the center of WNT. Thus, the structural water directly connected to the O atoms of TMA3- works as anchors to stabilize the WNT cluster and to restrain the fluctuation of water molecules present in it. For example, in two sequences A and B of H-bonded networks of O(2) · · · H(14A)sO(14) · · · H(8B)s O(8)sH(8A) and O(3) · · · H(4A)sO(4) · · · H(5A)sO(5)sH(5B), respectively, the temperature factors of the former has the following order of isotropic displacement parameters: Ueq of 102(3) < 230(14) < 304(15) < 450(40) < 620(50) < 110(200) [Å2 × 103]. These are defined as one-third of the trace of the orthogonalized Uij tensor. The latter also has the following order: 54(2) < 140(7) < 224(12) < 231(15) < 335(15) < 550(60) [Å2 × 103], both of which show a gradual increase in their values with increasing distance from the channel wall. In the same

Mechanistic Effects of Water Nanotube Clusters

Figure 6. (a) Schematic representations with each numbered atom, and A, B, and C sequences of hydrogen bonding networks are shown. (b) Each line graph of A, B, and C sequences describes the relationship between the number of atomic spheres from the carboxylic O atoms in the channels and the values of isotropic displacement parameters Ueq, which are defined as one-third of the trace of the orthogonalized Uij tensor. Ueq indicates the fluctuations for each atom.

manner, the temperature factors of the sequence C of H-bonding atoms of O(1) · · · H(9A)sO(9)sH(9B) also increase in the order of 77(2) < 213(11) < 347(14) < 430(40) [Å2 × 103], respectively (Figure 6). Interestingly, each TMA3- unit making up the framework simultaneously can stabilize the structural water of the three WNTs confined to the adjacent three different channels by using H-bonds with six carboxylic O atoms. Therefore, all the WNTs are chemically related to each other and must maintain the same periodic structure throughout the entire crystal. As a result, the varying structures of all the WNTs become standardized as a result of mechanistic anchor effect, and all WNTs simultaneously exhibit macroscopic motions and phase transitions. 3.3. Detection of Anomalous Water by Neutron-Scattering Analysis. The neutron scattering of crystal 1 is measured with an angle focusing cold neutron spectrometer (AGNES). Figure 7a shows that the elastic scattering intensity of WNT deviated from the nearly straight data for bulk ice above 210 K; this corresponds to the order-disorder transition of water molecules in WNT, which was observed at ca. 210 K in the calorimetric experiments. This temperature dependence of the elastic intensity is similar to that of a biocorrelated material such as protein.31 The gap in the elastic intensity of WNT at 273 K is due to the

J. Phys. Chem. B, Vol. 114, No. 6, 2010 2097

Figure 7. (a) Temperature dependence of the elastic intensity normalized by the intensity at 100 K. In bulk water (blue circles), the logarithmic intensity of ice is proportional to temperature as is expected from the harmonic oscillation approximation I/I0 ) exp[-1/3〈u2〉Q2], where Q [)4π(sin θ)/λ] is the momentum transfer (length of scattering vector) and 〈u2〉 is the mean-square displacement, which is a measure of the average amplitude of motions of water molecules. The elastic intensity changed abruptly at Tm ()273 K), which is the melting temperature of ice; molecules have considerably higher mobility in a liquid phase than in a solid phase. In WNTs (red circles for heating and yellow ones for cooling runs), there are also abrupt jumps around Tm because a small amount of bulk water was added to the sample to stabilize crystal 1. The gradual decrease in the intensity beginning from 210 K in the heating run is due to the order-disorder transition of water molecules in WNT. (b) Neutron quasielastic spectrum of crystal 1 measured at T ) 268 K and Q ) 1.57 Å-1. The observed data (circles) were fitted to the model function (red line), which is composed of elastic (green line) and quasielastic (blue line) components. The quasielastic component was further divided into translational (magenta) and rotational (light blue) parts. See text for details of the fitting.

melting of a small amount of bulk water that was added to the sample to stabilize WNT crystals. Therefore, the quasielastic scattering data were measured at several temperatures below 273 K to account for and avoid the effect of bulk water. The quasielastic spectra were analyzed using the method established by Teixeira et al.,32 which considers both translational jump diffusion and rotational motions of water molecules. At the beginning of the analysis, nine dynamic structure factors S(Q,E) were straightforwardly obtained from the inelastic neutron-scattering spectra with different momentum transfers (Q ) 0.31, 0.42, 0.57, 0.70, 0.88, 1.12, 1.35, 1.57, and 1.80 Å-1) and having the same temperature. The nine S(Q,E)’s were fitted simultaneously (i.e. global fitting) after taking into consideration the integrated intensity of an elastic peak for each Q value, integrated intensity of quasielastic peak at Q ) 0, mean square displacement, diffusion coefficient, residence time, and rotational relaxation time as flexible parameters. Figure 7b shows

2098

J. Phys. Chem. B, Vol. 114, No. 6, 2010

Tadokoro et al.

Figure 8. Half-width at half-maximum (hwhm) of the Lorentz function for the translational diffusion plotted as a function of squared momentum transfer. The red and blue curves represent the calculated values for bulk water and WNT, respectively. The calculation was performed based on the jump-diffusion model with the observed temperature of 268 K, and the parameters were determined by the fitting described above. Γ(Q) is represented as a function of the width of the QENS signal, which is an expression for a random-jump-diffusion model, where τ0 represents the residence time. The translational diffusion constant D is expressed in terms of the mean-square jump length [〈l2〉av]1/2 by D ) 〈l2〉av/6τ.

TABLE 1: Results of Neutron-Scattering Parameters at 268 K WNT in 1 bulk water (this work) bulk water (ref 33) WNT in CNT (ref 34)a a

D/Å2 s-1

τ0/s

[〈l2〉av]1/2/Å

9.44 × 1010 1.01 × 1011 0.85 × 1011 5.40 × 1010

4.70 × 10-11 4.82 × 10-12 4.66 × 10-12 12.5 × 10-11

5.15 1.73 1.54 6.40

Measured at 260 K.

an example of the fitting for the data at 268 K and 1.57 Å-1. As shown in this figure, the quasielastic component is fitted well by the two Lorentz functions for the zeroth and first terms of a spherical Bessel function,33 each corresponding to the translational diffusion and rotational motions of water molecules, respectively. The second and higher terms were negligible in the fitting. It is noteworthy that the quasielastic component was smaller than the elastic one due to the bulk ice coexisting with WNT and hydrogen atoms in the framework of 1. Because of this undesirable situation, the fitting was successful only for the data at 268 K, which has the largest quasielastic component. Figure 8 shows the half-width at half-maximum (hwhm) of the Lorentz function for the translational diffusion plotted as a function of squared momentum transfer. The hwhm values were calculated from the equation hwhm ) DQ2/(1 + DQ2τ0) based on the jump-diffusion model, and the parameters D and τ0 were determined by global fitting. The curve for the supercooled water, which was measured and analyzed using the same methods as that for WNT, is also shown for comparison. The diffusion coefficient of the translational mode D corresponding to the slope of the curve at Q ) 0, the residence time τ0 corresponding to the inverse hwhm at the high-Q limit, and the average jump length 〈l〉 calculated from the relation D ) 〈l2〉/ 6τ0 are given in Table 1. The quantities of bulk water obtained in this work are essentially the same as those in the previous work,32 demonstrating the reliability of the data and analysis of this work. The characteristic values for τ0 and 〈l〉 are also larger than those of bulk water, as shown in Table 1, and the value of D is almost identical to that of the bulk one. This result agrees with a

possible picture of fused WNT in the channelsthat is, the water molecules of WNT are hydrogen-bonded to the neighboring molecules for a longer time than those of bulk water; however, once the hydrogen bonds are broken, the water molecules of WNT jump a longer distance than those of bulk water. This picture may be a result of the situation that the density of the water molecules in WNT (0.81 g/cm3 at 296 K) is less than that of bulk water (0.97 g/cm3 at 298 K). The determined parameter 〈l〉 seems to be shorter than the actual jump distance, when the distance between the water molecules is taken into consideration on a 3-D motion as bulk water. This is probably because the actual jump of water molecules in WNT is anisotropic due to the surface effect and geometry of the channel. The relation D ) 〈l2〉/2τ0 for 1-D diffusion gives higher values than the relation D ) 〈l2〉/6τ0 for 3-D diffusion. From quasielastic neutron scattering for water in a single-walled CNT, the diameter is found to be approximately 1.6 Å, which is similar to the WNT of 1.34,35 4. Conclusions In conclusion, the anomalous water molecules at the center of a WNT are a novel example of the occurrence of a phase transition due to a small number of water molecules; further, a glass transition with a macroscopic behavior similar to that of a liquid phase is also indicated. On the other hand, a characteristic feature is that a freezing INT changes into a melting WNT at 245 K, which is lower than the phase change of bulk ice at 273 K, thereby influencing structural water H-bonding with the channel wall. Furthermore, before the melting point of the INT, a weak endothermic broad peak from 210-240 K is observed; this is probably related to the movement of the water molecules in INT. Such premelting reactions appearing prior to the phase transition is a unique characteristic of INTs but not of bulk ice. To our knowledge, employing neutron structure analysis, this study has for the first time confirmed the mechanistic effects of structural water in WNT confined to hydrophilic nanoscale channels with diameters of ca. 1.6 nm. The liquidlike domain in WNT with fast water molecules and the structure of static structural water have been identified simultaneously. The dispositions of structural water in WNT, roughly localized by X-ray analysis, are precisely determined by neutron structure analysis. We have performed neutron scattering on a small amount of anomalous water molecules in WNT. Water molecules restricted by structural water typically move slower than those restricted by bulk water36–38 because the H-bonds are highly oriented with the static structural water. However, it is clear that the anomalous water molecules in WNT have at least the same velocity as that of bulk water and move at a faster velocity because they are constructed from incomplete H-bonds and small densities in WNT. For example, in actin filaments interacting with an activated myosin motor domain, it is confirmed that the motion of water molecules in a muscular fiber is impacted by structural water, which moves approximately three times faster than bulk water.39 Furthermore, for each atom in the WNT, the thermal motion increases with increasing distance from the primary layer near the outer wall toward the secondary and tertiary domains in that order. Hence, this would result in higher flexibility in moving water molecules due to the minimization of the forming of structural water in pseudo-1-D channels. The WNT detected in the experiments may have one of the smallest structural units confined to hydrophilic nanoporous channels that have a liquid behavior and probably reveals a macroscopic phase transition with some water molecules. Further, the 1-D movement of water

Mechanistic Effects of Water Nanotube Clusters molecules at the center of the WNT is useful for understanding the model for elucidating the cohesion mechanism involved in transferring water to heights such as 100 m; for example, water transferred through a xylem conduit of a certain tree growing up to a height of approximately 100 m.40,41 Moreover, this movement is also useful as a biopore model for the transfer of water molecules through the inner and outer of cells such as aquaporin-142 of water transporting proteins. Lastly, we hope to find an unprecedented water phase in which water molecules infinitely bond pseudo-one-dimensionally with H-bonds in a future work. Acknowledgment. This work was supported by a Grant-inAid for Scientific Research (Nos. 18033049 and 20045018) on Priority Areas from the Ministry of Education, Science and Culture, Japan, and for REIMEI Research Project for JAERI. The authors thank the Analytical Center in Tokyo University of Science for the use of a CCD single-crystal X-ray diffractometer and 2H NMR spectrometer. Supporting Information Available: Detailed preparation of crystal 1, neutron crystal structure analysis (with CIF file), neutron-scattering analysis and supporting figures. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Ohba, T.; Kanoh, K.; Kaneko, K. J. Am. Chem. Soc. 2004, 126, 1560–1562. (2) Kalra, A.; Garde, S.; Hummer, G. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 10175–10180. (3) Tanak, H.; Koga, K. Bull. Chem. Soc. Jpn. 2006, 79, 1621–1644. (4) Bai, J.; Wang, J.; Zeng, X. C. Proc. Natl. Acad. Soc. U.S.A. 2006, 103, 19664–19667. (5) Akporiaye, D.; Hansen, E. W.; Schmidt, R.; Sto¨cker, M. J. Phys. Chem. 1994, 98, 1926–1928. (6) Maniwa, Y.; Kataura, H.; Abe, M.; Udaka, A.; Suzuki, S.; Achiba, Y.; Kira, H.; Matsuda, K.; Kadowaki, H.; Okabe, Y. Chem. Phys. Lett. 2005, 401, 534–538. (7) Schellman, J. A. Biophys. J. 1997, 73, 2960–2964. (8) Berg, O. G.; von Hippel, P. H. Annu. ReV. Biophys. Biophys. Chem. 1985, 14, 131–160. (9) Keutsch, F. N.; Saykally, R. J. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 10533–10540. (10) Ludwig, R. Angew. Chem., Int. Ed. 2001, 40, 1808–1827. (11) Ludwig, R. Angew. Chem., Int. Ed. 2006, 45, 3402–3405. (12) Tadokoro, M.; Fukui, S.; Kitajima, T.; Nagao, Y.; Ishimaru, S.; Kitagawa, H.; Isobe, K.; Nakasuji, K. Chem. Commun. 2006, 1274–1276. (13) Desiraju, G. R. Angew. Chem., Int. Ed. Engl. 1995, 34, 2311–2327. (14) Henry, M. Chem. Phys. Chem. 2002, 3, 607–616.

J. Phys. Chem. B, Vol. 114, No. 6, 2010 2099 (15) Janiak, C.; Scharmann, T. G. J. Am. Chem. Soc. 2002, 124, 14010– 14011. (16) Tadokoro, M.; Kanno, H.; Kitajima, T.; Shimada-Umemoto, H.; Nakanishi, N.; Isobe, K.; Nakasuji, K. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 4950–4955. (17) Atencio, R.; Chaco´n, M.; Gonza´lez, T.; Bricen˜o, A.; Agrifoglio, G.; Sierraalta, A. Dalton, Trans. 2004, 505–513. ¨ hrstro¨m, L.; Larsson, K. Dalton Trans. 2004, 347–353. (18) O (19) Maniwa, Y.; Matsuda, K.; Kyakuno, H.; Ogasawara, S.; Hibi, T.; Kadowaki, H.; Suzuki, S.; Achiba, Y.; Kataura, H. Nat. Mater. 2007, 6, 135–141. (20) Giovambattista, N.; Angell, C. A.; Sciortino, F.; Stanley, H. E. Phys. ReV. Lett. 2004, 93, 047801-1–4. (21) Tajima, Y.; Matsuo, T.; Suga, H. Nature 1982, 299, 810–812. (22) Watanabe, K.; Oguni, M.; Tadokoro, M.; Ohata, Y.; Nakamura, R. J. Phys.: Condens. Matter 2006, 18, 8427–8436. (23) Tanaka, I.; Kurihara, K.; Chatake, T.; Niimura, N. A. J. Appl. Crystallogr. 2002, 35, 34–40. (24) Haga, Y.; Kumazawa, S.; Niimura, N. J. Appl. Crystallogr. 1999, 32, 878–882. (25) Otwinnowski, Z.; Minor, W. Processing of x-ray diffraction data collected in oscillation mode, methods in enzymology; Carter, C. W., Jr., Sweet, R. M., Eds.; Academic Press: New York, 1997; Part A, Vol. 276, pp 307-326. (26) Spek, A. L. J. Appl. Crystallogr. 2003, 36, 7–13. (27) Kajitani, T.; Shibata, K.; Ikeda, S.; Kohgi, M.; Yoshizawa, H.; Nemoto, K.; Suzuki, K. Physica B 1995, 213/214, 872–874. (28) Yamamuro, O.; Inamura, Y.; Kawamura, Y.; Watanabe, S.; Asami, T.; Yoshizawa, H. Inst. Solid State Phys., UniV. Tokyo 2005, 12, 12–15. (29) Schreiber, A.; Ketelsen, I.; Findenegg, G. H. Phys. Chem. Chem. Phys. 2001, 3, 1185–1195. (30) Benoit, M.; Marx, D.; Parrinello, M. Nature 1998, 382, 258–261. (31) Ferrand, M.; Dianoux, A. J.; Petry, W.; Zaccai, G. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 9668–9672. (32) Teixeira, J.; Bellissent-Funel, M.-C.; Chen, S.-H.; Dianoux, A. J. Phys. ReV. A 1985, 31, 1913–1917. (33) Bee, M. Quasielastic neutron scattering; Adam Hilger: Bristol, UK, 1988. (34) Monontov, E.; Bumham, C. J. S.; Chen, H.; Moravsky, A. P.; Loong, C.-K.; deSouza, N. R.; Kolesnikov, A. I. J. Chem. Phys. 2006, 124, 194703-1–6. (35) Kolesnikov, A. I.; Zanotti, J. -M.; Loong, C.-K.; Thiyagarajan, P. Phys. ReV. Lett. 2004, 93, 0355031–0355034. (36) Takahara, S.; Sumiyama, N.; Kittaka, S.; Yamaguchi, T.; BellissentFunel, M.-C. J. Phys. Chem. B 2005, 109, 11231–11239. (37) Trantham, E. C.; Rorschach, H. E.; Clegg, J. S.; Hazlewood, C. F.; Nicklow, R. M.; Wakabayashi, N. Biophys. J. 1984, 45, 927–938. (38) Koenig, S. H.; Schillinger, W. E. J. Biol. Chem. 1969, 244, 3283– 3289. (39) Kabir, S. R.; Yokoyama, K.; Mihashi, K.; Kodama, T.; Suzuki, M. Biophys. J. 2003, 85, 3154–3161. (40) Tyree, M. T. Nature 2003, 423, 923. (41) Pockman, W. T.; Sperry, J. S.; O’Leary, J. W. Nature 1995, 378, 715–716. (42) Agre, P. Angew. Chem., Int. Ed. 2004, 43, 4278–4290.

JP9069465