NMR Studies of Structure and Dynamics of Liquid Molecules Confined

Jul 15, 2009 - We fabricated an NMR cell equipped with 10−100 nm scale spaces on a glass substrate (called extended nanospaces), and investigated mo...
1 downloads 11 Views 2MB Size
Download PDF
10808

J. Phys. Chem. B 2009, 113, 10808–10816

NMR Studies of Structure and Dynamics of Liquid Molecules Confined in Extended Nanospaces Takehiko Tsukahara,† Wataru Mizutani,‡ Kazuma Mawatari,‡ and Takehiko Kitamori*,‡ Research Laboratory for Nuclear Reactors, Tokyo Institute of Technology, 2-12-1-N1-32, O-Okayama, Meguro, Tokyo 152-8550 Japan, and Department of Applied Chemistry, School of Engineering, The UniVersity of Tokyo, 7-3-1, Hongo, Bunkyo, Tokyo 113-8656, Japan ReceiVed: April 9, 2009; ReVised Manuscript ReceiVed: June 11, 2009

We fabricated an NMR cell equipped with 10-100 nm scale spaces on a glass substrate (called extended nanospaces), and investigated molecular structure and dynamics of water confined in the extended nanospaces by 1H NMR chemical shift (δH) and 1H and 2H NMR spin-lattice relaxation rate (1H- and 2H-1/T1), 1H NMR spin-spin relaxation rate (1H-1/T2), and 1H NMR rotating-frame spin-lattice relaxation rate (1H-1/T1F) measurements of H2O and 2H2O. The δH and 1H- and 2H-1/T1 results showed that size-confinement produces slower translational motions and higher proton mobility of water, but does not affect the hydrogen-bonding structure and rotational motions. Such unique phenomena appeared in the space size of 40 to 800 nm. However, the 1H-1/T1 value at 40 nm was still different from that in 4 nm porous nanomaterial, because translational and rotational motions were inhibited for H2O molecules in the nanomaterial. By examining temperatureand deuterium-dependence of the 1H-1/T1 values, the molecular translational motions of the confined water were found to be controlled by protonic diffusion invoking a proton hopping pathway between adjacent water rather than hydrodynamic translational diffusion. Furthermore, we clarified that proton exchange between adjacent water molecules in extended nanospaces could be enhanced by the chemical exchange of protons between water and SiOH groups on glass surfaces, (tSiO- · · · H+ · · · H2O) + H2O f tSiO- + (H3O+ + H2O) f tSiO- + (H2O + H3O+), based on 1H-1/T2 measurements. An enhancement of proton exchange rate of water due to the reduction of space sizes was verified from the results of 1H-1/T1F values, and the rate of water in the 100 nm sized spaces is larger by a factor of more than ten from that of bulk water. Such size-confinement effects were distinctly observed for hydrogen-bond solvents with strong proton-donating ability, while they did not appear for aprotic and nonpolar solvent cases. Based on these NMR results, we suggested that an intermediate phase, in which protons migrate through a hydrogen-bonding network and the water molecules are loosely coupled within 50 nm from the surface, exists mainly in extended nanospaces. This model could be supported by a three-phase theory based on the weight average of three phases invoking the bulk, adsorbed, and intermediate phases. 1. Introduction Micro chemical systems on a chip have been receiving much attention for a variety of applications in chemical and biochemical analyses.1-4 Since microspaces have short molecular diffusion distances, large surface-to-volume ratio, small heat capacity, and so on, molecular and energy transport in microspaces can occur rapidly and highly efficiently compared to those in conventional bulk space. In order to utilize these characteristics skillfully, continuous flow chemical processing, consisting of micro unit operations such as solvent mixing, reaction, extraction, and separation and gas/liquid and liquid/liquid multiphase laminar flow networks, has been realized. Several chemical systems such as analysis, diagnosis, synthesis, immunoassay, and cell-handling have been integrated on a microchip, and their superior performances have been proven. However, the main advantage of such micro chemical systems is simply large surface-to-volume ratio of microspaces, and the physicochemical * Corresponding author. Tel: +81-3-5841-7231. Fax: +81-3-5841-6039. E-mail: [email protected]. † Tokyo Institute of Technology. ‡ The University of Tokyo.

properties of liquids in microspaces are not different from those in bulk; the operating principle is dominated by classical mechanics. On the other hand, as the space size shrinks to 1 nanometer scale, size-confinement characteristics such as quantum size and near-field effects appear.5-7 By utilizing 1 nm scale nanomaterials such as carbon nanotubes, porous silica, and polymers, evolutional molecular nanoelectronic and -photonic devices, machines, and sensors have been developed. Since the 1 nm scale space corresponds to a single individual molecule, the behaviors of liquids confined in the space also show specific features that cannot be observed in bulk.7-13 Examples of unique properties of water confined in 1 nm scale nanomaterials include the formation of icelike structures on the surfaces, the slowing down of molecular motions, and depression of the freezing point. Several studies have indicated that such confined water properties play an important role for the appearance of chemical and biological functions including chromatography separation of molecules, cell signaling mechanisms, and stability of protein hydration. However, 1 nm scale nanomaterial is too small to deal with liquid phase molecular clusters associated with intermolecular interactions, or to be used for analyzing molecular behavior in the liquid phase.

10.1021/jp903275t CCC: $40.75  2009 American Chemical Society Published on Web 07/15/2009

Liquid Molecules Confined in Extended Nanospaces When these micro- and nanotechnologies are classified according to space sizes, a 10-100 nm scale space can be expected to be an attractive space size for not only implementing novel micro- and nanofluidics devices but also gaining important insights into molecular clusters with a size range from 10 to 100 nm in a liquid phase. In particular, since a 10-100 nm scale space corresponds to a transition area of molecular behavior from a single individual molecule to a bulk condensed phase, this nanospace should make it possible to characterize the complicated collective behavior of liquid-phase molecular clusters. However, to our knowledge, there have been no experimental tools for studying the spaces with both width and depth dimensions on the 10-100 nm scale. Thus, we have christened the well-defined two-dimensional 10-100 nm scale space as an extended nanospace, and investigated the fabrication, fluidics control, and detection methods for extended nanospaces as well as physicochemical properties of liquids in the spaces. So far, we have developed extended nanospaces by means of top-down nanofabrication techniques, and we realized liquid mixing and chemical reactions in the extended nanospaces using pressure-driven flow.14,15 Furthermore, in order to clarify the macroscopic properties of water in extended nanospaces, we carried out capillarity and time-resolved fluorescent measurements. The results showed that the water confined in a 330 nm space had a higher viscosity and a lower dielectric constant compared with bulk water.16 We also measured NMR chemical shifts and relaxation times of water in 300-5000 nm spaces, and found that size-confinement produced slower translational motions and localized proton charge distributions of water with keeping water structures.17 Such phenomena appeared below around 800 nm, and the values shifted with surface modification. On the basis on these results, we have hypothesized that an intermediate phase differing from bulk phase and adsorbed phase, in which water molecules are migrating within about 50 nm from the glass surface, exists in extended nanospaces. Similar size-confinement phenomena have been recently established from experimental results of electrophoretic transport of ionic species, DNA separation due to electroosmotic flow, the filling kinetics, ion conductivities of fluidics, and diffusion coefficients of water in one-dimensional 10-100 nm scale spaces.18-25 However, it is still not clear about the mechanisms how the properties of water molecules change by size-confinement from bulk to 1 nm sized nanomaterials. Moreover, the existence of intermediate phase in extended nanospaces is no better than a working hypothesis. This is because no one has much knowledge on dynamical properties of liquids in the extended nanospaces. Accordingly, it is essential to elucidate the translational, rotational, and proton transfer dynamics of water and nonaqueous solvents in the size range of 10 to 100 nm scale extended nanospaces in detail, and to provide a careful discussion at the molecular level for the size-confinement mechanisms of liquids. An NMR study, which is sensitive to the variation of intermolecular interactions including liquid-liquid and liquidsurface interactions, should be helpful in determining the dynamical properties of water confined in a nanospace. Actually, NMR spin-echo techniques have been applied to evaluate the dynamics of water molecules inside 1 nm sized nanomaterials, and clarified that the proton relaxation rate can be enhanced by fast spin exchange between water adsorbed on surface and the surface.26-32 In the present study, we fabricate NMR sample cell with 40-5000 nm extended nanospaces on a glass substrate by using top-down nanofabrication techniques, and perform the

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10809 following five NMR measurements: (1) size-dependence of H NMR chemical shifts of water (H2O), (2) size-dependence of 1H NMR spin-lattice relaxation rate (1H-1/T1) of H2O, (3) 2H NMR spin-lattice relaxation rate (2H-1/T1) of heavy water (2H2O), (4) temperature and deuterium substitution effects for 1H-1/T1 of H2O, and (5) spin-spin relaxation rate (1H-1/T2) and rotating-frame spin-lattice relaxation rate (1H1/T1F) of H2O. We employ these NMR spectroscopy results to clarify the novel confinement-induced nanospatial effects for (1) hydrogen-bonding structure, (2) molecular motion, (3) rotation and translation, (4) proton mobility, and (5) proton exchange rate of H2O molecules. Furthermore, in order to evaluate the relationship between dynamical properties and hydrogen-bond donor-acceptor abilities in the extended nanospaces, we examine size-dependence of 1H-1/T1 of nonaqueous solvents. Based on these results, we develop a model for explaining the size-confinement mechanism of water molecules in combination with a theoretical framework. 1

2. Experimental Section 2.1. Fabrication of Extended Nanospaces for NMR Measurements. Our concept for fabrication of an NMR cell with extended nanospaces was centered on making an NMR cell that can be applied with commercial NMR probes. The fabrication procedures are diagrammed in Figure 1. A 0.7 mm thick synthetic quartz glass substrate (VIOSIL-SX, Shin-Etsu Quartz Co., Ltd.) with 30 × 70 mm sides was chosen as the NMR material, because of its nonmagnetic property, high stability, and high purities. It was confirmed that dissolution of silica into water from the synthetic quartz glass surface is negligible under our experimental conditions, because no silica species in sample water were observed for the less than detection limit (less than 1 ppb) due to the inductively coupled plasma atomic emission spectroscopy (ICP-AES). The positive-type electron beam resist (ZEP-520A, Zeon corp.) and conductive polymer (Espacer300, Showa Denko, Co., Ltd.) were spin coated onto a glass substrate by a spin coater, and prebaked at 180 °C. An electron beam lithography apparatus (ELS-7500, Elionix Co., Ltd.) was used to draw nanopatterns onto the substrate (Figure 1(a)). The drawn nanopatterns were developed with o-xylene for either 30 s or 5 min and rinsed with 2-propanol. The drawn nanopatterns were fabricated with an ICP-type etching apparatus (NE-550, ULVAC Co., Ltd.) with a mixture of CF6 and CHF3 gases (Figure 1(b)). The fabricated nanospaces were connected with a microspace that can be used to introduce samples. The 50 nm thick Cr layer was sputtered on the substrate using a sputtering apparatus (L-332S-FH, Canon Anelva Corp.), and a photoresist layer (OFPR-800, Tokyo Ohka Kogyo Co. Ltd.) was coated on the substrate covering the Cr layer. After exposure to UV light through a photo mask, the photoresist and Cr layers were developed. The micropatterns were etched by plasma etching, and microspaces, 200 µm wide and 8 µm deep, were obtained (Figure 1(c)). Then, the inlet holes were pierced through the fabricated substrate using a diamond coated drill. After that, the substrate was rubbed with melamine resin including surface-active agent (Cleanthrough-7008B, Kao Corp.), and was washed repeatedly in xylene, DMSO, pure water, and mixed solution of sulfuric acid and hydrogen peroxide (3:1). The fabricated substrate was thermally laminated with the cover substrate in a vacuum furnace at 1080 °C, and was cut with a diamond cutter for utilizing as a NMR sample cell (Figure 1(d-f)). The effects of impurities could be excluded by performing all operations in class-100 and -1000 cleanrooms. A typical scanning electron microscope (SEM) image of the drawn nanopatterns with 540 nm width, 440 nm depth and a

10810

J. Phys. Chem. B, Vol. 113, No. 31, 2009

Tsukahara et al.

Figure 1. Schematic diagram of micro- and nanofabrication procedures for the NMR spectral measurement cell of liquids confined in 40-5000 nm scale extended nanospaces. (a) Drawing nanospaces by electron beam lithography. (b) Plasma etching. (c) Using photolithography for fabrication of microspaces. (d) Thermal bonding. (e) Cutting with a diamond cutter. (f) Introducing liquids into nanospaces and making NMR measurements.

pitch of 2 µm and the SEM image of fabricated nanospaces in a section are shown in Figure S1 in the Supporting Information. The fabricated nanospace width was measured from SEM images, and the depth was determined by atomic force microscope (AFM) images. It was confirmed that the substrates could be bonded without deforming the shape of the fabricated extended nanospaces. Here, since the fabricated nanospaces that had W width (40-5000 nm), D depth (40-1500 nm), and 42 mm length were rectangular in shape, the rectangle-shaped nanospaces were converted as the cylindrical-shaped nanospaces. The cylindrical spaces with diameter R are replaced by 4/R using the surface-to-volume ratio, 2(D + W)/DW, and all results could be optimized as equivalent diameter R (40-5000 nm). 2.2. Sample Preparation. All ultrapure H2O samples were treated by a water purification system composed of a reverse osmosis membrane, ion exchange cylinder, and UV sterilizer (MINIPURE TW-300RU, Nomura Micro Science Co., Ltd.), yielding specific resistivity greater than 18.0 MΩ · cm. Heavy water (2H2O 99.8+%) was purchased from ISOTEC Inc. and used without further purification. A sample was degassed by means of a number of freeze-pump-thaw cycles, and then capillary force was used to fill the extended nanospaces on a NMR sample cell with the sample under an argon atmosphere. 2.3. NMR Measurement Conditions. All NMR spectra were measured with a JEOL ECA-500 spectrometer at 500 MHz without spinning and locking. The probe temperature was measured with a thermocouple placed just below the sample tube and controlled (accuracy: (0.1 °C) using a JEOL NMLVT temperature controller unit with a dry nitrogen stream from a liquid nitrogen supply. When a simple air flow was used, the signal of vapor containing in the air appeared at around 3.5 ppm (Figure S2 in the Supporting Information). Since the vapor adsorbed on the outside glass surface of NMR cell, the vapor signal was quite broadened. The baseline of H2O signal in the extended nanospaces was often distorted due to the appearance of the broadening vapor signal. Therefore, utilizing the dry nitrogen stream, only water signal in extended nanospaces, which had enough sensitivity to evaluate size-confinement phenomena, could be detected without a vapor signal appearing.

A 1H NMR spectrum of chloroform confined in a 1500 nm sized space at 22 °C was given to demonstrate the magnetic field homogeneity within the NMR cell with extended nanospaces (Figure S3 in the Supporting Information). The full line width at half-height (PW) and signal-to-noise (SN) ratio in the 1H NMR spectrum were 0.5 Hz and 100, respectively. The pulse sequence was applied using a 90° pulse width of 13.0 µs. The data were acquired for 32 scans (4048 data points) with a spectral width of 15 kHz and an acquisition time of 400 ms. The 1H-1/T1 and 2H-1/T1 values of H2O were measured with the inversion recovery pulse sequence, (π)x-τ-(π/2)x, while varying τ values from 1 ms to 20 s, and then expressed by a single exponential function. The 1H-1/T2 and -T1F measurements were performed by using the Carr-Purcell Meiboom-Gill pulse sequence, (π/2)x-τ-((π)x-2τ-(π)x)n, and spin-locking pulse sequence, (π/2)x-B(τ)y, respectively.33 3. Results and Discussion 3.1. Hydrogen-Bonding Structure. In order to examine the size-confinement effects of structures of H2O molecules, we carried out 1H NMR chemical shift (δH) measurements in the nanospace size of 40 to 5000 nm at 22.0 °C. If the molecular structures of the confined H2O differ from ordinary H2O, the δH values of H2O move to a higher or lower field than that of ordinary H2O (4.8 ppm). In fact, the δH values for H2O in 1 nm sized nanomaterials and ice have been determined as ∼5.8 ppm and ∼8 ppm, respectively.34-37 Figure 2 shows typical 1H NMR spectra of H2O confined in the space of 40 to 5000 nm. The δH values of confined H2O are almost constant at around 4.8 ppm regardless of space sizes. This result indicates that the H2O molecules confined in extended nanospaces retain the fourcoordinated hydrogen-bond structures without changing the O-O distance between H2O molecules as seen for ordinary liquid H2O. 3.2. Molecular Motions. Since the full line width at halfheight (PW) in the 1H NMR spectrum broadened with decreasing space sizes, the molecular motions of H2O would be hindered by size-confinement. Thus, we examined the size dependence of the 1H-1/T1 values for H2O in the same

Liquid Molecules Confined in Extended Nanospaces

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10811

Figure 2. 1H NMR spectra of water confined in various extended nanospaces at 500 MHz and 22.0 °C. The δH value of bulk water was adopted as the external reference.

temperature and space size as δH measurements. As shown in Figure 3 (a), although the 1H-1/T1 values are almost constant for space sizes of 800-5000 nm, the values change drastically below about R ) 800 nm and increase continuously with decreasing space sizes up to around 200 nm. The 1H-1/T1 value for a 212 nm space is 0.88 s-1, which is larger than bulk H2O (0.32 s-1) by a factor of about 3. Below about 200 nm, size-dependence of the 1H-1/T1 values disappeared. The 1H-1/T1 values are found to approach a constant value in the size of 200 to 40 nm. When the size is further reduced to 1 nm scale space, in which we determine the 1 H-1/T1 values as 8.0 s-1 based on the measurement of H2O in 4 nm controlled porous glass (#7930, Corning Incorporated), an increase in 1H-1/T1 values is observed again. The 1H-1/T1 value at 4 nm was quite consistent with previous reports.28-31 Three inflection points of 1H-1/T1 values at around 800, 200, and 40 nm suggest that the H2O behaviors in confined geometries would depend upon the weighted average of phases. Therefore, we compared our experimental 1H-1/T1 values with theoretical ones based on a three-phase theory stating that the H2O in confined geometries is composed of three phases: a bulk phase (SB), an intermediate phase (SP), and an adsorbed phase (Sa). The H2O molecules in the 800-5000 nm space sizes are dominated by the SB phase, and the 1H-1/T1 values do not depend on the size. Since the SP phase appears with decreasing sizes at around 800 nm, the 1H-1/T1 values begin to change according to the interfacial area ratio of SP phase to SB phase. The SP phase dominates H2O motions at less than about 200 nm space size, while the Sa phase with a thickness of 0.3 nm begins to affect H2O molecules with decreasing in size. Thus, the overall relaxation rate (1/T1EXP) is expressed as follows as the weighted average of these phases:

1 T1exp

)

λA1 1 εA2 1 1 + + T1B V1 T1P V2 T1a

(1 ) SP, 2 ) Sa)

(1)

Figure 3. (a) Log-Log plot of the 1H-1/T1 values for water confined in R ) 4-5000 nm spaces at 500 MHz and 22.0 °C. (b) Sizedependence of 1H-1/T1 values of water in the range of 200 to 5000 nm. (c) Size-dependence of 1H-1/T1 values of water for spaces below 200 nm. The solid lines in (b) and (c) represent fitting of the 1H-1/T1 values according to eq 1. In these cases, the thickness (ε) of the SP phase and the thickness (λ) of the Sa one were adopted as 50 and 0.3 nm, respectively.

where 1/T1B, 1/T1P, 1/T1a, λ, ε, and A/V are the 1/T1 of SB, 1/T1 of SP, 1/T1 of Sa, thickness of Sa, thickness of SP, and interfacial area ratio, respectively. When the thickness (ε) of SP was presumed to be 50 nm, the 1H-1/T1 values given by eq 1 are quite consistent with the measured 1H-1/T1 data (Figure 3 (b)). On the other hand, the experimentally measured 1H-1/T1 values in the space size of 4 to 200 nm can be expressed by a twophase model, which takes into account only the exchange of H2O between the SP and the Sa phases as shown in Figure 3(c). Thus, the marked changes in observed 1H-1/T1 values here indicate strongly the existence of an intermediate phase, consisting of H2O molecules within about 50 nm from the glass surfaces, in extended nanospaces differing from bulk and adsorbed phases.

10812

J. Phys. Chem. B, Vol. 113, No. 31, 2009

Tsukahara et al.

3.3. Determination of Rotation and Translation. The 1H1/T1 relaxation measurement is difficult to evaluate the detailed features of molecular interactions of H2O in SP phase, because the measured 1H-1/T1 values (1/T1meas) of H2O are related to both an intramolecular rotational component (1/T1intra) associated with molecular reorientational correlation time (τR) and an intermolecular translational one (1/T1inter) associated with proton’s self-diffusion coefficient (D). Namely, the 1/T1meas of H2O can be explained by eq 2 under the extreme narrowing limit as follows:33

1 T1meas

)

( ) ( ) 1

T1intra

+

1

T1inter

)

3 γ4p2 π Nγ4p2 τR + 6 2 r 5 aD

(2)

where p is the Planck constant, r is the distance between the hydrogen atoms in a H2O, γ is the proton gyromagnetic ratio, a is the hydrodynamic radius, N is the number of spins per unit volume, kB is the Boltzmann constant, and T is the temperature. The τR is related to the viscosity (η) through the StokesEinstein-Debye hydrodynamic equation as shown in eq 3:

τR )

4πa3η 1 ) 3kBT 2DR

(3)

where DR is the rotational diffusion coefficient. On the other hand, the 2H-1/T1 measurement of heavy water (2H2O) with a quadrupole moment enables the extraction of the 1/T1intra, because the obtained 2H-1/T1 values are simply expressed by the molecular reorientational correlation time (τRD) for the motion of deuterium electric field gradient tensor through eq 4:33

( )

1 3π2 e2Qq 2 D ) τR T1 2 h

(4)

where e2Qq/h is the quadrupole coupling constant (QCC; 256 kHz for the ambient water) which is connected with the quadrupole moment of the nucleus (eQ) and the electric field gradient at the nucleus (eq). Therefore, we examined sizedependence for 2H-1/T1 values of 2H2O in the space size of 40 to 5000 nm, and found that the 2H-1/T1 values did not depend on the sizes. The τRD values could be estimated by using eq 4. Since an isotope effect on viscosity is predicted by the following simple relationship, η(2H2O)/η(H2O) ) 1.4, the τR values of H2O are determined 1.4 times as large as τRD ones of 2H2O.38,39 As a result, the τR values of H2O were obtained in the extended nanospaces, and the size-dependence of 1/T1inter and 1/T1intra for H2O were divided according to eq 2 by substituting the obtained τR values. As shown in Figure 3(a), we found that 1/T1intra was almost constant regardless of the sizes, and the main sizeconfinement effect appeared in 1/T1inter. It was suggested that only molecular translational motions of H2O were influenced by size-confinement without changing structures and rotations. It follows from this result that the dynamical properties of H2O molecules confined in extended nanospaces are quite different from those in 1 nm sized nanomaterials, because not only translation but also rotation is inhibited for H2O molecules in 1 nm sized nanomaterials. 3.4. Evaluation of Proton Mobility. As seen from eq 2, the translational motions of H2O were strongly influenced by

Figure 4. The size-dependence of Ea values of water.

changes of proton’s diffusion coefficient D, which can be generally contributed from hydrodynamic translational diffusion (DS). Based on hydrodynamic Stokes-Einstein law, translational motion associated with DS should be strongly coupled with rotational diffusion DR, because both motions depend on changes of viscosity. However, as shown in the section 3.3, translational motion alone changed in extended nanospaces. This fact indicated that the size-dependence of 1/T1inter values of H2O was dominated by processes other than DS. Protonic diffusion (DH+), of which excess proton transfers between adjacent H2O molecules via Grotthuss proton transfer mechanism H3O+ + H2O f H2O + H3O+, is most likely to control the translational motions in extended nanospaces.38-40 The relation of DS and DH+ is simply expressed by

D ) χSDS + χH+DH+ ) χS

kBT kBTλH+ + χH+ 2 2 6πaη zF

(5)

where z is the protic charge, F is Faraday’s constant, χS is molar fraction of H2O, χH+ is molar fraction of proton, and λH+ is the proton mobility. Since in the extended nanospaces the specific interface area of the glass surfaces becomes very high, the charged surface invoking ionizable silanol (SiOH) groups makes it possible to induce proton hopping along a linear O · · · H-O hydrogen-bonding chain between adjacent H2O molecules, i.e., (tSiO- · · · H+ · · · H2O) + H2O f tSiO- + (H3O+ + H2O) f tSiO- + (H2O + H3O+). The excess proton given in H2O molecules can produce an effect on the χH+DH+ term in eq 5 rather than the χSDS one. Similarly, when stable protonated H2O molecules such as H5O2+ are formed, e.g., in ice and in acid solution at the limit of infinite dilution, the DH+ of H2O molecules has been posited to become dominant.40-42 In order to support the validity of the interpretation about the DH+, we measured the temperature-dependence of the 1H1/T1 values in the range of 4 to 50 °C. We examined the sizedependence of the apparent activation energy (Ea) values taken from the Arrhenius plots for 1H-1/T1. If DS is dominantly affected by size-confinement, the Ea values should increase because of a raise in the potential energy barrier due to the inhibition of translational motion. However, as shown in Figure 4, the Ea value of bulk H2O is 18 kJ/mol and the value decreases to 10 kJ/mol with decrease in sizes at R ) 320 nm. This loss of 8 kJ/mol was quite consistent with the difference in Ea values between dielectric relaxation based on hydrodynamic diffusion and proton hopping times calculated from proton conductivity data for bulk H2O.38,39 The results support that the potential energy barrier to the DH+ of excess proton for H2O was reduced

Liquid Molecules Confined in Extended Nanospaces

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10813

Figure 5. Deuterium concentration dependence (1H/(1H + 2H)) of 1H1/T1 values of water confined in the extended nanospaces (9, 5000 nm; b, 1500 nm; 2, 418 nm; and 1, 212 nm).

by size-confinement, and that the proton mobility of H2O could be quite high in extended nanospaces compared with bulk H2O. When protonic diffusion is the dominant mechanism, the proton mobility between H2O molecules should be restricted by deuterium substitution. Therefore, we examined the deuteriumconcentration dependence of 1H-1/T1 values for H2O + 2H2O mixtures in the extended nanospaces. As seen from plotting 1H1/T1 vs deuterium concentration (Figure 5), the 1H-1/T1 values decreased with increasing deuterium concentration in extended nanospaces, but did not depend on any deuterations in 1500-5000 nm spaces. When the molar fraction of 2H2O approached 0.4, the size-confinement effects for 1H-1/T1 values disappeared. These results are evidence that proton mobility of water molecules in extended nanospaces was reduced by the addition of deuterium, because the exchange rate of 1H-2H nuclei becomes slow by a factor of more than ten from that of 1H-1H nuclei. 3.5. Proton Exchange Rate. In order to clarify sizeconfinement processes for the proton mobility of the confined H2O in detail, we focused on the measurements of spin-spin relaxation rate (1H-1/T2) and rotating-frame spin-lattice relaxation rate (1H-1/T1F). In non viscous liquids, the relationship of 1 H-1/T1 ) 1H-1/T2 usually holds. However, the 1H-1/T1 values provide insight into the faster motions, i.e., higher frequency component, whereas the 1H-1/T2 values are sensitive to the slower motions related to the lower frequency component.43 Namely, since the contribution from slow motions of H2O molecules adsorbed near and on glass surfaces can be strongly reflected in the 1H-1/T2 values, the experimental 1H-1/T2 values of adsorbed H2O tend to be quite long compared with the 1H1/T1 ones.28,32 When the experimental 1H-1/T2 is mentioned as the effective transverse relaxation rate 1H-1/T2*, the differences between the 1H-1/T2* and 1H-1/T1 values are related with proton lifetime that a proton exchanges on a H2O molecule according to the following simple expression:44

1 1 ≈ Aτe T2* T1

(6)

where τe is the proton exchange correlation time and A corresponds to a measure of the strength of the intermolecular interactions that depends on 1H-17O spin-spin scalar coupling constant (pure H2O containing 17O comprise only 0.037% of the total). Therefore, in order to estimate correlation between faster and slower motions for H2O confined in extended nanospaces, we measured the 1H-1/T2 values of the confined

Figure 6. Plot of 1/ T2* - 1/T1 values vs space sizes from 4 to 5000 nm at 22.0 °C.

Figure 7. 1H-1/T1p values as functions of B1 for water confined in various extended nanospaces. The solid lines are the curves fitted to eq 7.

TABLE 1: Values of A and τe Obtained by the Analysis of 1/T1G Data Using Eq 7 R/nm

A/104 s-2

τe/10-5 s

1/T1′/s-1

5000 1500 694 418 256 120

0.1 ( 0.1 0.7 ( 0.2 4.3 ( 1.6 8.7 ( 1.5 21.8 ( 4.2 75.7 ( 16.7

21.9 ( 7.2 11.4 ( 4.4 2.8 ( 0.1 2.5 ( 0.3 1.7 ( 0.4 1.4 ( 0.3

0.42 ( 0.0 0.70 ( 0.0 2.23 ( 0.1 2.77 ( 0.1 5.23 ( 0.2 5.06 ( 0.7

H2O and compared them with the 1H-1/T1 data. Figure 6 plots the differences between 1H-1/T1 and 1H-1/T2* values of H2O against R. Size-confinement increases the 1H-1/T2* values just as it does for 1H-1/T1, while the 1/T2* - 1/T1 values become greater with reductions in the space sizes. As expected, the 1/T2* - 1/T1 values increase more than 2 orders of magnitude by changes from 5000 to 40 nm, and the differences approach the value of adsorbed H2O. Accordingly, the increase in the 1/T2* - 1/T1 values can be regarded as the proton exchange rates being enhanced by the chemical exchange of protons between H2O and SiOH group on charged glass surface (tSiO- · · · H+ · · · H2O) + H2O f tSiO- + (H3O+ + H2O) f tSiO- + (H2O + H3O+). To characterize the A and τe based on proton exchange processes, we measured the rf field amplitude (B1) dependence of 1H-1/T1F values based on changes of NMR signal intensity against spin-locking times, because the 1H-1/T1F values depend on frequency in the spin rotating frame (ω1 ) γB1).45-47 The results are shown in Figure 7. We found that the 1H-1/T1F values increase with decreasing B1, and that the B1 dependence was enhanced by size-confinement. The τe values were obtained according to eq 7:

10814

J. Phys. Chem. B, Vol. 113, No. 31, 2009

Aτe 1 1 ) ′ + T1F T1 1 + 4ω12τe2

Tsukahara et al.

(7)

where ω1 is the Larmor frequency in the proton spin rotating frame and γ is the proton gyromagnetic ratio. The term 1/T1′ in eq 7 includes the relaxation due to mechanisms other than the exchange, i.e., fast rotational and translational motions in the confined water. In the limit ω1 ) 0, the 1H-1/T1F value will approach the 1H-T2* one. As listed in Table 1, the τe values are reduced by size-confinement. The τe value of 1.4 × 10-5 s for a 120 nm space is reduced by about a factor of 1/20 from a space size of 5000 nm. The obtained τe value was close to 0.8 × 10-5 s for pore surface H2O. Moreover, the coefficient A values were found to increase considerably with decreasing space sizes. These results mean that the intermolecular spin-spin interactions of H2O in confined geometries are getting more superior rather than those of bulk H2O, and proton hopping pathway along a hydrogen-bonded chain becomes ratedetermining step in extended nanospaces. Such interpretation agrees with earlier theoretical observation, in which the behavior

Figure 8. Schematic pictures illustrating the three-phase model. In SB, the water molecules have an ordinary liquid structure and free translation and rotation. In SP, the water molecules have the following properties: (1) keeping four-coordinated H2O structure, (2) slower molecular motion compared with bulk H2O, (3) size-confinement effect of only intermolecular translation without changing rotation, (4) higher proton mobility due to proton hopping along a linear O · · · H-O hydrogen-bonding chain, and (5) fast proton transfer rate induced by chemical exchange between H2O and SiOH groups on surfaces. In Sa, the water is similar to icelike bilayer structure and both translation and rotation are inhibited.

of H2O molecules with well-ordered hydrogen-bonding networks such as ice are controlled by monopole-dipole type interactions between H3O+ and H2O.42 In other word, the proton transfer mechanism in extended nanospaces would be analogous to that in ice unlike liquid H2O associated with large rearrangement of the hydrogen-bond network. The important findings in these NMR results of the confined H2O are summarized as follows: (1) keeping four-coordinated ordinary H2O structure, (2) slower molecular motion compared with bulk H2O, (3) size-confinement effect for only intermolecular translation, (4) higher proton mobility due to proton hopping along a linear O · · · H-O hydrogen-bonding chain between adjacent H2O molecules, and (5) fast proton transfer rate induced by chemical exchange of protons between H2O and SiOH groups on surfaces according to (tSiO- · · · H+ · · · H2O) + H2O f tSiO- + (H3O+ + H2O) f tSiO- + (H2O + H3O+). In this case, the H2O molecules are arranged in a linear O · · · H-O hydrogen-bonding line, and loosely coupled within about 50 nm via hydrogen bonds in a direction perpendicular to the glass surfaces as illustrated in Figure 8. The existence of such long-range ordering of hydrogen-bonded solvents on a surface has been recently suggested from the results of molecular dynamics (MD) simulation.48 Then, we determine the H2O molecules in extended nanospaces as a proton transfer phase (SP). The SP is quite different from both a bulk phase (SB) having ordinary water structure and free translational and rotational motions and an adsorbed phase (Sa) having icelike bilayer structure and slower translational and rotational motions. Specifically, we can conclude that the H2O behavior in nano confining geometries depends upon the weighted average of the three phases: the SB, Sa, and SP phases. 3.6. Hydrogen-Bond Donor-Acceptor Ability. The δH and 1 H-1/T1 values were examined for various solvents in extended nanospaces. Since the δH values for nonaqueous solvents such as methanol, ethanol, dimethyl sulfoxide (DMSO), acetonitrile, benzene, and hexane were almost constant in the whole sizes just as for H2O, the molecular structures of these solvents in extended nanospaces were found to be quite consistent with those in bulk (Figure S4 in the Supporting Information). The 1 H-1/T1 values at bulk for every solvent were normalized to unity, and the relative 1H-1/T1 values in extended nanospaces (1/T1nano/1/T1bulk) are plotted against the space sizes as shown in Figure 9(a). We found that the 1/T1nano/1/T1bulk values in hydrogen-bonded solvents such as H2O, methanol, and ethanol increased drastically with decreasing sizes at around 800 nm, while the size-confinement did not affect the 1/T1nano/1/T1bulk

Figure 9. (a) Size-dependence of the relative 1H-1/T1 values (1/T1nano/1/T1bulk) of water (9), methanol (b), ethanol (2), dimethyl sulfoxide (DMSO) (1), acetonitrile (right pointing 2), benzene (left pointing 2), and hexane ([). (b) The DN (9) and AN (0) dependence of the 1/T1nano/1/T1bulk values at 254 nm. The dashed and solid lines correspond to the least-squares fitting for DN and AN, respectively.

Liquid Molecules Confined in Extended Nanospaces values in aprotic solvents such as DMSO and acetonitrile and nonpolar solvents such as hexane and benzene. Therefore, we tried to evaluate the hydrogen-bond donor-acceptor ability by plotting the obtained 1/T1nano/1/T1bulk values at 254 nm against electron donor number (DN) and electron acceptor one (AN), because DN and AN can be applied to a measure of protonaccepting ability and proton-donating one, respectively.49 The plot of 1/T1nano/1/T1bulk values vs AN was found to give a linear relationship with the correlation coefficient of 0.93 as shown in Figure 9(b), but the 1/T1nano/1/T1bulk values did not depend on the DN. This means that the variation of 1H-1/T1 values in extended nanospaces was induced by change of proton-donating ability. It is supported our explanations that an excess proton sharing between adjacent hydrogen-bonded solvents plays an important role in controlling dynamical properties of liquids confined in extended nansospaces. 4. Conclusions In the present study, the molecular structural and dynamical properties of liquids confined in 40-5000 nm extended nanospaces were investigated by the measurements of (1) sizedependence of δH of water, (2) size-dependence of 1H-1/T1 of water, (3) comparison between size-dependence of 1H-1/T1 and 2 H-1/T1 of water, (4) temperature and deuterium substitution effects for 1H-1/T1 of water, (5) size-dependence of 1H-1/T2 and 1 H-1/T1F of water, and (6) size-dependence of 1H-1/T1 for various nonaqueous solvents. From the NMR results, the following important findings were obtained. 1. The hydrogen-bonding structure of the confined water was almost consistent with that of ordinary water. 2. The size-dependence of the 1H-1/T1 values showed three inflection points at around 800, 200, and 40 nm. Concretely, the 1H-1/T1 values increased with decreasing space sizes from 800 to 200 nm, while they did not change in the size of 800 to 5000 nm. Below about 200 nm, the 1H-1/T1 values settled to a constant value. When the space sizes arrived at around 40 nm, the 1H-1/T1 values began to increase again and approached that in 4 nm porous nanomaterial. These phenomena indicated that molecular motions of water were inhibited by size-confinement, and that the motions were controlled by the weight average of bulk, adsorbed, and intermediate phases. 3. Size-confinement effects on the 2H-1/T1 values of 2H2O were quite small differing from 1H-1/T1 results. Namely, only translational motions of water could be modulated with decreasing sizes without changing rotational motions of water. 4. Changes of translational motion of the confined water resulted from protonic diffusion invoking a proton hopping pathway between adjacent water molecules rather than hydrodynamic translational diffusion. Such results were attributable to an enhancement of proton mobility of water by the reduction of space sizes. 5. The differences between 1H-1/T2* and 1H-1/T1 values of water increased continuously by a decrease in sizes of the confinement area. An increase in the 1/T2* - 1/T1 values meant that fast proton exchange between adjacent water molecules in extended nanospaces could be induced by the chemical exchange of protons between water and SiOH groups on the glass surfaces according to (tSiO- · · · H+ · · · H2O) + H2O f tSiO- + (H3O+ + H2O) f tSiO- + (H2O + H3O+) mechanism. Based on the results of 1H-1/T1F values, we confirmed that the proton exchange rate of water in the 100 nm sized extended nanospaces could be larger by a factor of more than ten from that of bulk water.

J. Phys. Chem. B, Vol. 113, No. 31, 2009 10815 6. Such size-confinement phenomena were characteristics common to hydrogen-bond solvents, while they did not appear in the case of aprotic and nonpolar solvents. Specifically, we presented that a proton transfer phase (SP), consisting of loosely coupled hydrogen-bonded molecules such as water within about 50 nm from the glass surfaces, exists mainly in extended nanospaces. The validity of this model was qualitatively supported by a three-phase theory where the water motions in extended nanospaces were controlled by the weighted average of three phases such as the SB having ordinary water structure and free translational and rotational motions, the Sa having icelike bilayer structure and slower translational and rotational motions, and the SP. Finally, we concluded that proton mobility plays an important role in controlling dynamical properties of hydrogen-bonded molecules confined in extended nansospaces. We believe that these findings will have important implications for the understanding of basic (bio)chemistry of confined liquids, and for the evolution in the fields of micro- and nanofluidic devices and fuel-cell devices. Acknowledgment. The present work was partially supported by the Grant of Core Research for Evolutional Science and Technology (CREST) from Japan Science and Technology Agency (JST). Supporting Information Available: Figures depicting SEM images and 1H NMR spectra. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Dittrich, P. S.; Tachikawa, K.; Manz, A. Anal. Chem. 2006, 78, 3887–3907. (2) Atencia, J.; Beebe, D. J. Nature 2005, 437, 648–655. (3) Gunther, A.; Jensen, K. F. Lab. Chip 2006, 6, 1487–1503. (4) Kitamori, T.; Tokeshi, M.; Hibara, A.; Sato, K. Anal. Chem. 2004, 76, 52A–60A, and references therein. (5) Avouris, P.; Chen, Z.; Perebeinos, V. Nat. Nanotechnol. 2007, 2, 605–615. (6) Ohtsu, M., Ed. Progress in Nano-Electro-Optics; Springer-Verlag: Berlin, 2006. (7) Goldberger, J.; Fan, R.; Yang, P. Acc. Chem. Res. 2006, 39, 239– 248. (8) Buch, V., Devlin, J. P., Eds. Water in Confining Geometries; Springer-Verlag: Berlin, 2003. (9) Israelachvili, J. N. Intermolecular and Surface Forces, 2nd ed.; Academic: London, 1992. (10) (a) Bruni, F.; Ricci, M. A.; Soper, A. K. J. Chem. Phys. 1998, 109, 1478–1485. (b) Soper, A. K.; Bruni, F.; Ricci, M. A. J. Chem. Phys. 1998, 109, 1486–1494. (11) Farrer, R. A.; Fourkas, J. T. Acc. Chem. Res. 2003, 36, 605–612. (12) Yang, J.; Wang, E. G. Phys. ReV. B 2006, 73, 035406-1–035406-7. (13) Ball, P. Chem. ReV. 2008, 108, 74–108. (14) Tamaki, E.; Hibara, A.; Kim, H.-B.; Tokeshi, M.; Kitamori, T. J. Chromatogr. A 2006, 1137, 256–262. (15) Tsukahara, T.; Mawatari, K.; Hibara, A.; Kitamori, T. Anal. Bioanal. Chem. 2008, 391, 2745–2752. (16) Hibara, A.; Saito, T.; Kim, H.-B.; Tokeshi, M.; Ooi, T.; Nakao, M.; Kitamori, T. Anal. Chem. 2002, 74, 6170–6176. (17) Tsukahara, T.; Hibara, A.; Ikeda, Y.; Kitamori, T. Angew. Chem., Int. Ed. 2007, 46, 1180–1183. (18) Eijkel, J. C. T.; van den Berg, A. Microfluid. Nanofluid. 2005, 1, 249–267. (19) Kaji, N.; Tezuka, Y.; Takamura, Y.; Ueda, M.; Nishimoto, T.; Nakanishi, H.; Horiike, Y.; Baba, Y. Anal. Chem. 2004, 76, 15–22. (20) Daiguji, H.; Yang, P.; Majumdar, A. Nano Lett. 2004, 4, 137–142. (21) Pu, Q.; Yun, J.; Temkin, H.; Liu, S. Nano Lett. 2004, 4, 1099– 1103. (22) Stein, D.; Kruithof, M.; Dekker, C. Phys. ReV. Lett. 2004, 93, 035901–035901-4. (23) Plecis, A.; Schoch, R. B.; Renaud, P. Nano Lett. 2005, 5, 1147– 1155. (24) Kaji, N.; Oguma, R.; Oki, A.; Horiike, Y.; Tokeshi, M.; Baba, Y. Anal. Bioanal. Chem. 2006, 386, 759–764.

10816

J. Phys. Chem. B, Vol. 113, No. 31, 2009

(25) Xu, X.-H.; Yeung, E. S. Science 1998, 281, 1650–1653. (26) Gun’ko, V. M.; Turov, V. V.; Bogatyrev, V. M.; Zarko, V. I.; Leboda, R.; Goncharuk, E. V.; Novza, A. A.; Turov, A. V.; Chuiko, A. A. AdV. Colloid Interface Sci. 2005, 118, 125–172. (27) Mattea, C.; Kimmich, R.; Ardelean, A.; Wonorahardjo, S.; Farrher, G. J. Chem. Phys. 2004, 121, 10648–10656. (28) Overloop, K.; Van Gerven, L. J. Magn. Reson. A 1993, 101, 147– 156. (29) D’Orazio, F.; Tarczon, J. C.; Halperin, W. P.; Eguchi, K.; Mizusaki, T. J. Appl. Phys. 1989, 65, 742–751. (30) Hirama, Y.; Takahashi, T.; Hino, M.; Sato, T. J. Colloid Interface Sci. 1996, 184, 349–359. (31) Korb, J.-P.; Whaley; Hodges, M.; Gobron, Th.; Bryant, R. G. Phys. ReV. E 1999, 60, 3097–3106. (32) Allen, S. G.; Stephenson, P. C. L.; Strange, J. H. J. Chem. Phys. 1997, 106, 7802–7809. (33) Farrar, T. C.; Becker, E. D. Pulse and Fourier Transform NMR; Academic Press: New York, 1971. (34) Kinney, D. R.; Chuang, I.-S.; Maciel, G. E. J. Am. Chem. Soc. 1993, 115, 6786–6794. (35) Pfrommer, B. G.; Mauri, F.; Louie, S. G. J. Am. Chem. Soc. 2000, 122, 123–129. (36) Ghosh, S.; Ramanathan, K. V.; Sood, A. K. Europhys. Lett. 2004, 65, 678–684.

Tsukahara et al. (37) We observed that the 1H NMR peak of water inside molecular sieves of 4 Å was 5.77 ppm, when the bulk water was adopted as the external reference. (38) Agmon, N. J. Phys. Chem. 1996, 100, 1072–1080. Erratum: Agmon, N. J. Phys. Chem. 1997, 101, 4352. (39) Cohen, B.; Huppert, D. J. Phys. Chem. A 2003, 107, 3598–3605. (40) Kornyshev, A. A.; Kuznetsov, A. M.; Spohr, E.; Ulstrup, J. J. Phys. Chem. B 2003, 107, 3351–3366. (41) Roberts, N. K.; Zundel, G. J. Phys. Chem. 1980, 84, 3655–3660. (42) Kobayashi, C.; Saito, S.; Ohmine, I. J. Chem. Phys. 2000, 113, 9090–9100. (43) Meiboom, S. J. Chem. Phys. 1961, 34, 375–388. (44) Knispel, R. R.; Pintar, M. M. Chem. Phys. Lett. 1975, 35, 238–240. (45) Lamb, W. J.; Brown, D. R.; Jonas, J. J. Phys. Chem. 1981, 85, 3883–3887. (46) Graf, V.; Noack, F.; Bene, G. J. J. Chem. Phys. 1980, 72, 861–863. (47) Holly, R.; Peemoeller, H.; Choi, C.; Pintar, M. M. J. Chem. Phys. 1998, 108, 4183–4188. (48) Andoh, Y.; Kurahashi, K.; Sakuma, H.; Yasuoka, K.; Kurihara, K. Chem. Phys. Lett. 2007, 448, 253–257. (49) Linert, W.; Fukuda, Y.; Camard, A. Coord. Chem. ReV. 2001, 218, 113–152.

JP903275T