Structure at the air/water interface in the presence of phenol: a study using heterodyne-detected vibrational sum frequency generation and molecular dynamics simulation

Ryoji Kusaka§ a, Tatsuya Ishiyama b, Satoshi Nihonyanagi ac, Akihiro Morita de and Tahei Tahara *ac
aMolecular Spectroscopy Laboratory, RIKEN, 2-1 Hirosawa, Wako 351-0198, Japan. E-mail:
bDepartment of Applied Chemistry, Graduate School of Science and Engineering, University of Toyama, Toyama 930-8555, Japan
cUltrafast Spectroscopy Research Team, RIKEN Center for Advanced Photonics (RAP), 2-1 Hirosawa, Wako 351-0198, Japan
dDepartment of Chemistry, Graduate School of Science, Tohoku University, Sendai 980-8578, Japan
eElements Strategy Initiative for Catalysts and Batteries (ESICB), Kyoto University, Kyoto 615-8520, Japan

Received 30th July 2017 , Accepted 10th October 2017

First published on 27th October 2017

Many kinds of organic compounds pollute the aquatic environment, and they change the properties of the water surface due to their high surface affinity. Chemical reactions at the water surface are key in environmental chemistry because, for instance, reactions occurring at the surface of aqueous aerosols play essential roles in the atmosphere. Therefore, it is very important to elucidate how organic compounds affect the properties of water surfaces. Here, we choose phenol as an organic pollutant prototype and report how phenol affects the molecular-level structure of the air/water interface. Interface-selective vibrational spectra, i.e., the imaginary part of second-order nonlinear susceptibility (Im[thin space (1/6-em)]χ(2)), of the air/water–phenol mixture interface in the OH stretch region were collected using heterodyne-detected vibrational sum frequency generation (HD-VSFG) spectroscopy, and the observed Im[thin space (1/6-em)]χ(2) spectra were interpreted with the aid of molecular dynamics (MD) simulation. The Im[thin space (1/6-em)]χ(2) spectra observed via HD-VSFG drastically change as a function of phenol concentration in water, and exhibit two isosbestic points. In the spectra, a positive OH band appears at 3620 cm−1, which is assigned to an OH group of water that forms an OH⋯π hydrogen-bond (H-bond) with the aromatic ring of phenol, and a strong negative OH band appears around 3200 cm−1, which is attributed to a water that accepts a H-bond from the phenol OH, while pointing its OH groups toward the bulk water side. It was concluded that two types of unique water molecules hydrate a phenol molecule: (1) water that forms an OH⋯π H-bond; and (2) water that accepts a H-bond from a phenol OH group. Each phenol molecule adsorbed at the air/water forms a specific hydration structure, which causes a large change in the interfacial water structure. The present study provides a clear example demonstrating that even such a simple organic pollutant as phenol can drastically alter the interfacial water structure.

1. Introduction

Air/water interfaces are ubiquitous in nature, and are found at the surface of oceans, lakes, atmospheric aqueous aerosols, and fog/cloud droplets. These air/water interfaces provide reaction fields for the various heterogeneous chemical reactions that take place on Earth. When organic compounds are present in the aquatic environment as pollutants, they usually come to the surface and change surface properties, such as polarity and hydrogen-bonding (H-bonding) structures, which is expected to affect the chemical reactivity of adsorbates at the surface.1 Therefore, it is very important to elucidate how the presence of organic molecules changes the molecular-level structure of the air/water interface for a better understanding of atmospheric and environmental chemistry.

Many studies on the air/water interface in the presence of organic compounds, as well as inorganic salts, have been conducted using vibrational sum frequency generation (VSFG) spectroscopy.2–8 Although VSFG spectroscopy is a powerful technique for obtaining interface-selective vibrational spectra, conventional VSFG can only provide information relating to the absolute square of the second-order nonlinear susceptibility (|χ(2)|2). On the other hand, heterodyne-detected VSFG (HD-VSFG), which determines the phase and amplitude of the VSFG signal, enables us to obtain complex vibrational χ(2) spectra.9,10 In particular, the imaginary part of the χ(2)(Im[thin space (1/6-em)]χ(2)) spectrum provides direct information about the vibrational resonance of interfacial molecules. Furthermore, vibrational Im[thin space (1/6-em)]χ(2) spectra give decisive information about the up/down orientation of interfacial molecules.11–17 Although HD-VSFG is very advantageous compared to conventional VSFG, it has not been intensively applied yet to the study of air/water interfaces in the presence of neutral organic compounds.18,19

In this study, we chose phenol as a prototype of neutral amphiphilic organic compounds and studied the structure of the air/water–phenol interface with HD-VSFG, as well as with complementary MD simulation. In our world, the production and use of phenols are huge, and phenols are also generated through the pyrolysis of lignin, which is a major component of wood tissue. In fact, phenol derivatives are one of the most abundant organic pollutants, and they are widely observed in aquatic systems.20 It has been reported that cloud and fog processing of phenolic compounds might be an important pathway for the formation of the low-volatility and highly oxygenated organic species that make up secondary organic aerosols (SOAs);21 SOAs are particulate matter composed of organic compounds formed from the atmospheric transformation of organic species.22 It is also known that the existence of phenol at the sea surface influences chemical reactions taking place at the sea interface.23 Actually, due to its amphiphilic nature, phenol tends to come to the air/water interface,24 and hence the elucidation of its effects on the interfacial structure of water is essential. Here, we report that the addition of phenol to water drastically changes the interface-specific vibrational Im[thin space (1/6-em)]χ(2) spectra in the OH stretch region. On the basis of detailed spectral analyses with MD simulation, it is revealed that the interfacial water H-bonding structure is greatly altered by the presence of phenol, even though phenol is a simple and neutral molecule.

2. Experimental & computational methods

2.1. Samples

High-purity H2O (18.2 MΩ cm resistivity, Millipore, Milli-Q Advantage A10), D2O (NMR grade, 99.9%, Wako), phenol (purity > 99.0%, Wako), and phenol-d5 (deuterated phenol, C6D5OH, purity > 98%, Cambridge Isotope Laboratories) were used to prepare aqueous solutions of phenol. The phenol concentrations in the solutions were 25, 57, and 117 mM, and the surface excesses were reported to be 0.7, 1.6, and 2.7 molecules nm−2, respectively.24 These surface excesses correspond to concentrations of 1.2, 2.7, and 4.5 M at the interface, which are calculated by assuming a 1 nm thickness for the interface. The pH values of the sample solutions were measured to be in the range between 5.1 and 5.4. No bands assignable to a water molecule that hydrates a phenolate anion were observed in this study. This indicates that neutral phenol is predominant at the surface, and phenolate anion concentration is negligible. This is consistent with a previous study that carried out surface tension measurements and VSFG studies in the C–O stretch region of phenol and phenolate.25 The temperature was kept at 25 °C during the measurements.

2.2. HD-VSFG spectroscopy

The HD-VSFG setup used in the present study has been described previously.26–28 Briefly, a Ti:sapphire regenerative amplifier (Spectra Physics, Spitfire Ace) was used as the light source (800 nm, 5 mJ, 80 fs, 1 kHz). A part of the output (1 mJ) was attenuated using a variable neutral density (VND) filter and spectrally narrowed (ω1) using a bandpass filter (center wavelength: 795 nm; bandwidth: 20 cm−1). Another part of the output (2 mJ) was converted to a tunable and broadband infrared beam (ω2, bandwidth: ca. 300–400 cm−1) using an optical parametric amplifier with a difference frequency generation unit (Light Conversion, TOPAS-C).

For the present HD-VSFG measurements, the ω1 and ω2 beams were first focused into a thin y-cut quartz crystal (thickness: 10 μm), where the local oscillator (ωLO) was generated via the sum frequency generation of ω1 and ω2. A plano-convex lens (f = 500 mm) and an off-axis parabolic mirror (f = 100 mm) were used for focusing ω1 and ω2, respectively. The ωLO beam and transmitted ω1 and ω2 beams were then refocused using a spherical concave mirror onto a sample surface to generate the sum frequency light (ω3) of ω1 and ω2 from the sample surface. Before reaching the sample, the ωLO beam was transmitted through a silica plate (3 mm thickness) to be delayed by ∼5 ps with respect to the ω1 and ω2 pulses. The incident angle and the polarization direction were ∼40° and s for ω1 and ∼50° and p for ω2, whereas the s-polarized components of ω3 and ωLO were isolated using a Glan–Taylor prism for detection (i.e., ssp polarization combination). For heterodyne detection, the ω3 and ωLO beams were collinearly introduced into a polychromator, frequency-dispersed, and finally detected using a liquid nitrogen cooled charge coupled device (CCD; Princeton Instruments, Spec-10). The interference fringe recorded with the CCD was analysed in the way that we previously reported to obtain the χ(2) spectrum.11

All experimental Im[thin space (1/6-em)]χ(2) spectra presented in this paper were normalized using z-cut quartz as a reference (i.e., χ(2)sample/χ(2)quartz), and were carefully obtained by checking image file: c7cp05150f-t1.tif/χ(2)quartz and image file: c7cp05150f-t2.tif/χ(2)quartz spectra to ensure that their imaginary parts were nearly flat and zero in the 2900–3900 cm−1 region (for image file: c7cp05150f-t3.tif/χ(2)quartz) and 2550–2900 cm−1 region (for image file: c7cp05150f-t4.tif/χ(2)quartz). This ensures that phase error in χ(2)sample/χ(2)quartz spectra is negligible in the present study. This experimental condition was achieved through cleaning the surface of the quartz, setting the surface of the z-cut quartz as flat as possible with the aid of a displacement sensor (Keyence, SI-10), and avoiding bad spots (e.g., scratches) on the quartz surface during measurement. Three spectra were measured with ω2 tuned at around 2800, 3200, and 3600 cm−1, and they were combined to obtain an Im[thin space (1/6-em)]χ(2) spectrum over the 2550–3900 cm−1 region.

2.3. Computational procedures

2.3.1 Molecular model. The charge response kernel (CRK) model29 was employed to calculate the structure and VSFG spectrum at the aqueous phenol solution surface in MD simulation. Here we briefly describe an outline of the present model, which incorporates fluctuating partial charges in response to solvation and vibration. The CRK model is a polarizable model, in which the partial charge at site a of the ith molecule, Qai, is determined through the electrostatic potential exerted on the site b, Vbi, as follows:29
image file: c7cp05150f-t5.tif(1)
image file: c7cp05150f-t6.tif(2)
where Q0ai is the site charge for an isolated molecule, Kabi is the CRK of the ith molecule, rai,bj is the distance between ai and bj sites, and fai,bj is the damping function, moderating the Coulomb interaction at a short distance.30Q0ai and Kabi are a function of the molecular conformation (internal coordinate S) as follows:
image file: c7cp05150f-t7.tif(3)
image file: c7cp05150f-t8.tif(4)
where Qeq and Keq are the values of equilibrium molecular conformation in the gas phase.

The total potential energy of the CRK model system Utotal consists of the intramolecular part Uintra, the van der Waals part UvdW, and the Coulomb part UC, as follows:

Utotal = Uintra + UvdW + UC(5)
In the CRK model, UC is formulated as follows:
image file: c7cp05150f-t9.tif(6)

We employed the CRK water model developed in our previous study,30 while the CRK model for phenol molecules is newly developed, and thus is briefly described in the following text. The potential parameters for intra- and intermolecular van der Waals interactions, Uintra + UvdW for phenol molecules were taken from the AMBER ff99 model,31 whereas the intermolecular Coulombic interactions were modeled with the CRK method. All CH bond lengths of phenol were fixed in the present MD simulation. The static charge Q and the CRK K for phenol atomic sites were determined using the modified GAMESS-UK program package32 at the B3LYP level of theory with the aug-cc-pVDZ basis set. Since we focus on the OH stretching vibration in this study, the geometrical dependence of charge and CRK, Q(S) and K(S), was accounted for using the function S = ΔrOH, where ΔrOH is the deviation of the OH length from the equilibrium value. The intramolecular potential for OH stretching, uintrarOH), takes account of anharmonicity using the same functional form as for water (see eqn (2) of ref. 30):

image file: c7cp05150f-t10.tif(7)
The potential parameters k2k6 and ∂Q/∂S for OH in phenol were empirically determined to reproduce the experimentally observed difference IR spectrum of the OH stretching region between phenol solution and water (not shown). The final parameter sets for k2k6, and Qeq, Keq, ∂Q/∂S, and ∂K/∂S for the phenol model are summarized in Tables SI–SV in the ESI.

The damping parameter ξai (the width of the Gaussian charge distribution)30 to regulate f in eqn (2) is set to 0.593 Å for water and phenol.

2.3.2 MD simulation. MD simulations of aqueous phenol solution were performed in a rectangular simulation box with the dimensions Lx × Ly × Lz = 25 Å × 25 Å × 150 Å. A liquid slab consisting of 600 H2O and 25 phenol molecules (2 molecules per nm2 phenol in water) was formed in the center of the cell, where two vapor/solution interfaces were produced perpendicular to the z axis (see Fig. S1 in the ESI).

The MD simulations were carried out under 3D periodic boundary conditions. The initial configuration of each constituent molecule was prepared with random position and orientation within an Lx × Ly × Lz = 25 Å × 25 Å × 40 Å box, using the PACKMOL package.33 In the equilibration process, an NVT ensemble simulation with the Nosè–Hoover thermostat34,35 was carried out for 4 ns (time step Δt = 2.0 fs) with the GROMACS package (version 4.6.1)36 to speed up equilibration, where the TIP3P model37 for water and AMBER ff99 model31 for phenol were employed. The temperature was kept at T = 298.15 K. After the equilibration run, a short NVT ensemble simulation over 1 ps (time step Δt = 0.61 fs) using the CRK model was carried out followed by an NVE ensemble simulation for production. An ensemble average of 120 ps per one MD trajectory was taken in 32 parallel computations with different initial configurations, generating about a 4 ns average. Long-range electrostatic interactions were corrected using the Ewald method and the cutoff length of Lennard-Jones, and the real space part of the Ewald sum was taken to be 10 Å.

2.3.3 Calculation of the SFG spectrum. The resonant second order nonlinear susceptibility χR is calculated using the time correlation function formalism38
image file: c7cp05150f-t11.tif(8)
where ωIR is the IR frequency, kB is the Boltzmann constant, T is the temperature, t is the time, and A and M are the system polarizability and dipole moment, defined as sums of the individual ith molecular “effective” polarizability αi,xx and molecular dipole moment μi,z:
image file: c7cp05150f-t12.tif(9)
In eqn (8), Tc was set to 1.2 ps. “Effective” means that α incorporates the local field correction by definition.39 The molecular polarizability and dipole moment of water in an instantaneous configuration were calculated using the CRK model.30 The effective polarizability and dipole moment were calculated with the molecular values by taking account of the self-consistent polarization interaction.30 The calculation of χR in eqn (8) combined with eqn (9) consists of the self correlation terms image file: c7cp05150f-t13.tif and the cross correlation terms image file: c7cp05150f-t14.tif, generating a large amount of noisy signal. We confirmed that the latter cross correlation terms are insignificant in the present case, and hence we neglected the cross terms in the present χR calculation. This enables us to employ decomposition analysis for χR, in which the contributions of water and phenol are classified at t = 0 for the time correlation function.40 Multiplying χRxxz in eqn (8) by appropriate Fresnel factors gives χ(2) in ssp polarization. For the air/water interface, band analysis of Im[thin space (1/6-em)]χ(2) in ssp can be performed with the calculated Im[thin space (1/6-em)]χRxxz because of the small frequency dependence of the Fresnel factors.

3. Results & discussion

3.1. Experimental vibrational Im[thin space (1/6-em)]χ(2) spectra

The black solid line in Fig. 1 shows the vibrational Im[thin space (1/6-em)]χ(2) spectrum of the air/H2O interface in the 2550–3900 cm−1 region. This spectrum is in good agreement with those previously reported with D2O correction.41,42 The positive sharp band at 3700 cm−1 is due to the OH stretch vibration of free OH pointing upward (toward air) in the topmost layer, and the negative broad band in the 3100–3600 cm−1 region is assigned to hydrogen-bonded (H-bonded) OH groups pointing downward (toward the bulk water phase).41,43,44 The intensity in the 2550–3000 cm−1 region is zero, within the signal-to-noise ratio. Note that it was recently clarified that the positive band below 3200 cm−1, which was previously reported,11,44 actually arises from inaccurate phase calibration.41
image file: c7cp05150f-f1.tif
Fig. 1 Im[thin space (1/6-em)]χ(2) spectra of the air/H2O–phenol interface obtained with HD-VSFG. The concentrations of phenol in the bulk aqueous solution are given in the figure.

With an increase in phenol concentration, the Im[thin space (1/6-em)]χ(2) spectrum drastically changes and four spectral features appear, as shown in Fig. 1: (1) a positive band at 3620 cm−1; (2) a negative broad band in the 3100–3500 cm−1 region peaking at around 3200 cm−1; (3) a negative sharp band at 3090 cm−1; and (4) a negative tail below 3000 cm−1, which extends to 2550 cm−1. Simultaneously, the spectral features of the pure water component, peaking at 3500 and 3700 cm−1, become weak, and they almost disappear in the spectrum obtained with 117 mM phenol. The disappearance of the 3700 cm−1 band indicates that water does not retain free OH in the topmost layer in the presence of phenol. The spectral features of the water–phenol surface shown in Fig. 1 are significantly different from those reported so far for charged monolayer/water,11 air/electrolyte solution,4,8 silica/water,45,46 and air/water–DMF (N,N-dimethylformamide) mixture interfaces.18 This implies that the water–phenol surface has a specific interfacial structure. We note that, to the best of our knowledge, the OH stretch band extending as low as 2550 cm−1 has not been reported so far, even in bulk phenol aqueous solutions.

A key observation from the Im[thin space (1/6-em)]χ(2) spectra is the presence of two isosbestic points observed at around 3370 and 3660 cm−1. The appearance of the isosbestic points indicates that the Im[thin space (1/6-em)]χ(2) spectra in Fig. 1 consist solely of two spectral components. This demonstrates an advantage of HD-SFG, because isosbestic points can be observed in Im[thin space (1/6-em)]χ(2) (not |χ(2)|2) spectra measured with a high signal-to-noise ratio.47–49 Singular value decomposition (SVD) analysis yields only two dominant singular values (as shown in Fig. S2 in the ESI), which further confirms that all vibrational Im[thin space (1/6-em)]χ(2) spectra in Fig. 1 can be expressed through the linear combination of only two spectral components. The fact that the Im[thin space (1/6-em)]χ(2) spectra in Fig. 1 consist of two spectral components leads to two important insights into this interface: (1) each phenol molecule is solvated by water without clustering with other phenol molecules; and (2) the solvation structure of the phenol molecules is the same at any of the phenol concentrations examined. These points are concluded because the formation of different sizes of phenol clusters (dimer, trimer, etc.) and/or gradual changes in the solvation structure depending on phenol concentration is expected to cause a shift in vibrational bands, which never gives rise to isosbestic points. These molecular pictures deduced from the observation of isosbestic points are consistent with a previous study into the polarization dependence of second harmonic generation (SHG),24 which reported that the orientation angle of phenol at the air/water interface does not change at any concentration. Therefore, the spectral features that appear with an increase in phenol concentration are attributable to phenol molecules solvated by water with a specific solvation structure.

The negative intense sharp band at 3090 cm−1 is assigned to the CH stretch of phenol, because this band is also observed in the Im[thin space (1/6-em)]χ(2) spectrum of the air/D2O–phenol interface (Fig. S3 in ESI). This CH stretch is assignable to the ν20a mode, because this mode is both IR and Raman active and is observed at 3087 cm−1 in the gas phase.50 The negative sign of the CH stretch band indicates that the phenol OH points toward the water phase (see Fig. S4 in the ESI for details). An additional weak positive band observed at 3030 cm−1 in the air/D2O–phenol spectrum (Fig. S3 in the ESI) is assignable to the ν13 mode of the phenol CH stretch.51

The positive band at 3620 cm−1, the negative intense band in the 3100–3500 cm−1 region with a peak at around 3200 cm−1, and the extremely broad negative feature extending to 2550 cm−1 are due to OH stretches, as they are absent in the air/D2O–phenol spectrum (Fig. S3 in the ESI). The 3620 cm−1 band is assignable to an OH stretch in water molecules that weakly interact with the aromatic ring of phenol with a H-up orientation. This type of interaction between water and an aromatic ring is called an OH⋯π H-bond, and has been observed in Raman spectra of aqueous solutions of benzene and phenol (∼3610 cm−1)52 as well as in IR spectra of benzene–water clusters in the gas phase (3636–3657 cm−1).53,54 The relatively narrow feature of the 3620 cm−1 band suggests that the OH group forming the OH⋯π H-bond only interacts with the aromatic ring and not with other water molecules.

As for the negative OH band in the 2550–3500 cm−1 region, it includes not only a water OH but also a phenol OH, because phenol has an OH group with a H-down orientation. In order to identify the OH stretches of phenol and water in the air/H2O–phenol spectrum in Fig. 1, Im[thin space (1/6-em)]χ(2) spectra of HOD solutions were measured. Here, HOD represents isotopically diluted water with H/(H + D) = 25% (H2O[thin space (1/6-em)]:[thin space (1/6-em)]HOD[thin space (1/6-em)]:[thin space (1/6-em)]D2O = 1[thin space (1/6-em)]:[thin space (1/6-em)]6[thin space (1/6-em)]:[thin space (1/6-em)]9), where HOD molecules predominantly contribute to the OH stretch region. Fig. 2(a) shows Im[thin space (1/6-em)]χ(2) spectra of air/HOD–phenol (red) and air/HOD–phenol-d5 (blue) interfaces measured at a bulk phenol concentration of 117 mM. The air/D2O spectrum (black) represents the noise level in the 2800–3800 cm−1 region (note that the positive band at 2730 cm−1 is due to the free OD stretch). The spectrum of the air/HOD–phenol-d5 interface (blue) is also shown in order to clarify the spectral features of the negative OH band without overlapping from the CH bands of phenol. Fig. 2(b) shows the same spectra in an expanded view, as well as the corresponding H2O spectra (broken lines) which are scaled by 0.25 for comparison. The negative OH stretch band in the 3200–3500 cm−1 region of the H2O spectra is significantly red-sifted compared with the band observed in the HOD spectra. This frequency shift indicates that the Im[thin space (1/6-em)]χ(2) signal in this region contains a large amount of water OH, because it is known that the OH stretch frequency of H2O is significantly influenced by intra/intermolecular vibrational couplings of the OH stretch to other vibrational modes.13 In the region below 3000 cm−1, on the other hand, the H2O and HOD spectra overlap quite well with each other. This suggests that the frequency region below 3000 cm−1 is likely attributed to the phenol OH stretch ([phenol OH]/[phenol OD] ≈ 0.25 in HOD, because of efficient H/D exchange), although the HOD spectrum below 2800 cm−1 is governed by the OD stretch of water forming an OD⋯π H-bond. In the above discussion, we assume that the frequency of the phenol OH that is H-bonded to the oxygen of water is not sensitive to the H/D substitution of water molecules. This point is verified by MD simulations (vide infra).

image file: c7cp05150f-f2.tif
Fig. 2 (a) Im[thin space (1/6-em)]χ(2) spectra at the air/HOD–phenol (red) and air/HOD–phenol-d5 (blue) interfaces, obtained with HD-VSFG. The Im[thin space (1/6-em)]χ(2) spectrum of the air/D2O interface (black) is also shown for comparison. HOD represents isotopically diluted water, where H/(H + D) = 25% (H2O[thin space (1/6-em)]:[thin space (1/6-em)]HOD[thin space (1/6-em)]:[thin space (1/6-em)]D2O = 1[thin space (1/6-em)]:[thin space (1/6-em)]6[thin space (1/6-em)]:[thin space (1/6-em)]9). Phenol-d5 stands for deuterated phenol, C6D5OH. (b) The three spectra from (a) are shown in an expanded view. Im[thin space (1/6-em)]χ(2) spectra at the air/H2O–phenol (green broken line) and air/H2O–phenol-d5 (orange broken line) interfaces are also shown for comparison, scaled by a factor of 0.25. All phenol concentrations are 117 mM in the bulk.

In the gas phase, phenol–(H2O)1–4 clusters have been studied using IR spectroscopy, and each OH stretch band observed in experiment has been rigorously assigned to a phenol OH or water OH on the basis of quantum chemical calculations.50,55 In these assignments, phenol OH is always H-bonded to the oxygen atom of water and the frequency of the phenol OH is always lower than that of water OH. For example, the phenol–(H2O)4 cluster shows a phenol OH band at 3167 cm−1, and water H-bonded OH bands appear in the 3220–3430 cm−1 region. This tendency is consistent with the present assignment for the Im[thin space (1/6-em)]χ(2) spectra of the air/water–phenol interface. However, we note that the frequencies of the phenol OH stretch of a phenol–(H2O)4 cluster and even larger phenol–(H2O)19–49 clusters do not reach values below 3000 cm−1.56 This suggests that, for observing phenol OH stretches extending well below 3000 cm−1, the number of interacting water molecules is not crucial, and rather inhomogeneous broadening caused by thermal energy at room temperature may be important, because phenol–water clusters in the gas phase are frozen at an extremely low temperature.

3.2. MD simulation

In order to elucidate the structure of the water–phenol surface based on the obtained Im[thin space (1/6-em)]χ(2) spectra, we carried out MD simulations and theoretically simulated the Im[thin space (1/6-em)]χ(2) spectra for comparison. The calculated Im[thin space (1/6-em)]χ(2) spectrum of the air/H2O interface (gray solid line in Fig. 3(a)) nicely reproduces the experimental Im[thin space (1/6-em)]χ(2) spectrum from Fig. 1, which suggests that water molecules and water–water interactions are well-modelled in the present MD simulation.30
image file: c7cp05150f-f3.tif
Fig. 3 Computed Im[thin space (1/6-em)]χ(2) spectra of the air/H2O (gray solid) and air/H2O–phenol (green solid) interfaces in (a). In (a), the spectrum of the air/water–phenol interface (green solid line) is decomposed into phenol OH (red broken line) and H2O (blue broken line) components. In (b), the H2O component (blue broken line) in (a) is further decomposed into red and green dotted spectra. Each label, A, B, C and D, represents interfacial water species sketched in Fig. 5. The CH stretch vibrations of phenol are not taken into account in this simulation.

Then, the MD simulation was carried out in the presence of phenol. The calculated density profiles of water and phenol showed that phenol is adsorbed at the interface with its OH groups pointing at the bulk water side in the simulation (Fig. S1 and S5 in the ESI). The simulated Im[thin space (1/6-em)]χ(2) spectrum of the air/H2O–phenol interface is shown in Fig. 3(a) with a green solid line (note that the CH stretch vibrations of phenol are eliminated in this simulation because the CH bond lengths are fixed). In this spectrum, the negative band due to H-bonded OH shows a peak at around 3350 cm−1, which is lower in frequency than for the air/H2O interface (around 3450 cm−1). The amplitude of this negative band is substantially enhanced in the presence of phenol. These spectral features of the computed negative H-bonded OH band are a qualitative reproduction of those of the experimentally observed band at 3200 cm−1 (Fig. 1). A further quantitative comparison of the peak position is beyond the scope of the present MD simulation, which does not account for the quantum effects of vibrational couplings.57 We note that the computed peak position (∼3350 cm−1) is closer to that of the experimental HOD spectrum, in which the vibrational couplings are suppressed (Fig. 2). The computed Im[thin space (1/6-em)]χ(2) spectrum also elucidates other important features from the experimental Im[thin space (1/6-em)]χ(2) spectrum: (i) the almost complete disappearance of the positive band at 3700 cm−1 due to free OH; (ii) the appearance of the positive band at around 3600 cm−1 due to OH forming an OH⋯π H-bond; and (iii) the appearance of a long negative tail extending to 2600 cm−1. These agreements between the theoretical and experimental spectra suggest the high reliability of the structure of the air/water–phenol interface obtained using the present MD simulation.

The simulated air/H2O–phenol spectrum (green solid line) in Fig. 3(a) is decomposed into phenol (red broken line) and water (blue broken line) components. This spectral decomposition reveals that phenol OH is responsible for the extremely broad feature in the 2600–3500 cm−1 region, and that the negative band at 3350 cm−1 and the positive band at 3600 cm−1 are mainly due to water OH. Moreover, because the phenol OH band in the computed air/D2O–phenol (C6H5OH) spectrum is as broad as that in the H2O spectrum (Fig. S7 in the ESI), the extremely broad feature of the phenol OH is attributable simply to inhomogeneous broadening (the broad distribution of H-bonds). All these features are in excellent agreement with the assignment based on the isotope dilution experiment. The signal from water OH (blue broken line) is analysed in Fig. 3(b) to clarify the influence of phenol, where the water component (blue broken line) is further decomposed into water species accepting a H-bond from a phenol OH (red dotted line) and the remaining water (green dotted line). This decomposition indicates that water species accepting a H-bond from the phenol OH (water type (B) in Fig. 5) result in a negative band in the 3200–3600 cm−1 region, and that this water species is responsible for the enhanced amplitude of the negative band at around 3350 cm−1 in the air/water–phenol spectrum compared to the air/H2O interface.

In order to characterize the orientation of water molecules that interact with phenol, we plotted the orientational probability, P′(cos θdip,φ), in Fig. 4(a) and (b).30,58 As defined in Fig. 4(c), θdip is the angle between the water permanent dipole vector and the surface normal. φ is the dihedral angle between the molecular plane and the plane containing the surface normal and the permanent dipole. P′ = P − 1/4π is defined as the deviation of the probability distribution P from the isotropic average 1/4π. Fig. 4(a) shows a plot of P′ for water molecules located at a depth 1.5–3.0 Å lower than the Gibbs dividing surface. The water molecules in this depth range prefer to point an OH group upward because of the formation of an OH⋯π H-bond with phenol. It is known that water molecules having free OH at the air/water interface exist in the region that is 0.0–3.0 Å higher than the Gibbs dividing surface. Compared to this, the water molecules forming an OH⋯π H-bond with phenol are present in a rather deep region in the water phase. This is because the phenol ring is tilted toward the interface plane and an OH⋯π H-bond is formed below. Fig. 4(b) shows a P′ plot for water species that accept a H-bond from the phenol OH. This water clearly shows a preference for cos[thin space (1/6-em)]θdip ≈ −1, corresponding to an orientation with two OH groups vertically downward. This vertically-downward orientation is the reason why this water species gives rise to a large negative signal in the Im[thin space (1/6-em)]χ(2) spectra, because χ(2)ssp for the symmetric OH groups of water reaches a maximum when cos[thin space (1/6-em)]θdip = −1.11

image file: c7cp05150f-f4.tif
Fig. 4 Orientational probability distributions P′(cos[thin space (1/6-em)]θdip, φ) of water molecules that (a) are located in a region 1.5–3.0 Å lower than the Gibbs dividing surface and that (b) accept a H-bond from the phenol OH. A definition of the θdip and φ angles is shown in (c), where z is the normal coordinate to the surface. The blue regions refer to higher probabilities than the isotropic average (P′ > 0), while the red regions refer to lower probabilities (P′ < 0).

The interfacial structures concluded in the present study are sketched in Fig. 5. At the air/water–phenol interface, four types of water molecules exist at the interface: (A) water that forms an OH⋯π H-bond with the aromatic ring of phenol and orients the other OH toward the bulk; (B) water that accepts a H-bond from phenol OH and points its OH groups toward the bulk; (C) water that has free OH and orients the other OH toward the bulk; and (D) water that only has H-bonded OH groups pointing toward the bulk and does not interact with phenol. Water species (A) and (B) form a specific solvation structure around a phenol molecule. The population of water type (C) drastically decreases as the concentration of phenol increases. Water type (D) is present no matter whether phenol is present at the interface. In the Im[thin space (1/6-em)]χ(2) spectra of the air/water–phenol interface, the free OH of type (C) or the OH⋯π H-bonded OH of type (A) give rise to the positive OH bands, whereas the down-oriented H-bonded OH groups of type (A), (B), (C) and (D), as well as phenol OH, contribute to the negative bands. It is noted that water species (A) also contributes to the negative signal through the H-down OH that forms a H-bond with water lower down in the bulk.

image file: c7cp05150f-f5.tif
Fig. 5 Schematic diagram of the structure at the air/water–phenol interface, as elucidated in the present study. There are four types of water molecules at the interface: (A) water that forms an OH⋯π H-bond with the aromatic ring of phenol and orients the other OH toward the bulk; (B) water that accepts a H-bond from phenol OH and points its OH groups toward the bulk; (C) water that has free OH in the topmost layer and orients the other OH toward the bulk; and (D) water that has H-bonded OH groups pointing toward the bulk and does not interact with phenol. The population of water type (C) drastically decreases as the concentration of phenol increases.

4. Conclusions

We studied molecular-level structure at the air/water interface in the presence of phenol through vibrational Im[thin space (1/6-em)]χ(2) spectra in the OH stretch region. As a result of a rigorous comparison of experimental spectra obtained via HD-VSFG with spectra computed using MD simulation, it was clarified that phenol molecules are adsorbed at the air/water interface with the phenolic OH groups pointing toward the bulk water, and drastically change the interfacial water structure through forming a specific hydration structure. The present study demonstrated that even a simple neutral organic pollutant such as phenol significantly affects the interfacial water structure.

Conflicts of interest

There are no conflicts of interest to declare.


This work was supported by JSPS KAKENHI Grant numbers JP25104005, JP25104003, JP26288003, and 16H04095. The MD calculations were performed using the supercomputers at the Research Center for Computational Science, Okazaki, Japan. R. K. thanks the Special Postdoctoral Researchers (SPDR) program of RIKEN.


  1. D. J. Donaldson and V. Vaida, Chem. Rev., 2006, 106, 1445–1461 CrossRef CAS PubMed.
  2. D. Simonelli, S. Baldelli and M. J. Shultz, Chem. Phys. Lett., 1998, 298, 400–404 CrossRef CAS.
  3. G. Ma and H. C. Allen, J. Phys. Chem. B, 2003, 107, 6343–6349 CrossRef CAS.
  4. C. S. Tian, S. J. Byrnes, H. L. Han and Y. R. Shen, J. Phys. Chem. Lett., 2011, 2, 1946–1949 CrossRef CAS.
  5. S. T. Ota and G. L. Richmond, J. Am. Chem. Soc., 2012, 134, 9967–9977 CrossRef CAS PubMed.
  6. C. M. Johnson and S. Baldelli, Chem. Rev., 2014, 114, 8416–8446 CrossRef CAS PubMed.
  7. L. Piatkowski, Z. Zhang, E. H. G. Backus, H. J. Bakker and M. Bonn, Nat. Commun., 2014, 5, 4083 CAS.
  8. W. Hua, D. Verreault and H. C. Allen, J. Am. Chem. Soc., 2015, 137, 13920–13926 CrossRef CAS PubMed.
  9. S. Nihonyanagi, J. A. Mondal, S. Yamaguchi and T. Tahara, Annu. Rev. Phys. Chem., 2013, 64, 579–603 CrossRef CAS PubMed.
  10. Y. R. Shen, Annu. Rev. Phys. Chem., 2013, 64, 129–150 CrossRef CAS PubMed.
  11. S. Nihonyanagi, S. Yamaguchi and T. Tahara, J. Chem. Phys., 2009, 130, 204704 CrossRef PubMed.
  12. J. A. Mondal, S. Nihonyanagi, S. Yamaguchi and T. Tahara, J. Am. Chem. Soc., 2010, 132, 10656–10657 CrossRef CAS PubMed.
  13. S. Nihonyanagi, S. Yamaguchi and T. Tahara, J. Am. Chem. Soc., 2010, 132, 6867–6869 CrossRef CAS PubMed.
  14. J. A. Mondal, S. Nihonyanagi, S. Yamaguchi and T. Tahara, J. Am. Chem. Soc., 2012, 134, 7842–7850 CrossRef CAS PubMed.
  15. M. Okuno and T. A. Ishibashi, J. Phys. Chem. C, 2015, 119, 9947–9954 CAS.
  16. P. Balzerowski, K. Meister, J. Versluis and H. J. Bakker, Phys. Chem. Chem. Phys., 2016, 18, 2481–2487 RSC.
  17. S. Devineau, K. Inoue, R. Kusaka, S. Urashima, S. Nihonyanagi, D. Baigl, A. Tsuneshige and T. Tahara, Phys. Chem. Chem. Phys., 2017, 19, 10292–10300 RSC.
  18. S. K. Mondal, K. Inoue, S. Yamaguchi and T. Tahara, Phys. Chem. Chem. Phys., 2015, 17, 23720–23723 RSC.
  19. J. A. Mondal, V. Namboodiri, P. Mathi and A. K. Singh, J. Phys. Chem. Lett., 2017, 8, 1637–1644 CrossRef CAS PubMed.
  20. S. Rayne, K. Forest and K. J. Friesen, Environ. Int., 2009, 35, 425–437 CrossRef CAS PubMed.
  21. Y. L. Sun, Q. Zhang, C. Anastasio and J. Sun, Atmos. Chem. Phys., 2010, 10, 4809–4822 CAS.
  22. J. H. Kroll and J. H. Seinfeld, Atmos. Environ., 2008, 42, 3593–3624 CrossRef CAS.
  23. O. S. Ryder, N. R. Campbell, M. Shaloski, H. Al-Mashat, G. M. Nathanson and T. H. Bertram, J. Phys. Chem. A, 2015, 119, 8519–8526 CrossRef CAS PubMed.
  24. J. M. Hicks, K. Kemnitz and K. B. Eisenthal, J. Phys. Chem., 1986, 90, 560–562 CrossRef CAS.
  25. Y. Rao, M. Subir, E. A. McArthur, N. J. Turro and K. B. Eisenthal, Chem. Phys. Lett., 2009, 477, 241–244 CrossRef CAS.
  26. K. Matsuzaki, S. Nihonyanagi, S. Yamaguchi, T. Nagata and T. Tahara, J. Phys. Chem. Lett., 2013, 4, 1654–1658 CrossRef CAS PubMed.
  27. K. Inoue, S. Nihonyanagi, P. C. Singh, S. Yamaguchi and T. Tahara, J. Chem. Phys., 2015, 142, 212431 CrossRef PubMed.
  28. K. Matsuzaki, R. Kusaka, S. Nihonyanagi, S. Yamaguchi, T. Nagata and T. Tahara, J. Am. Chem. Soc., 2016, 138, 7551–7557 CrossRef CAS PubMed.
  29. A. Morita and S. Kato, J. Am. Chem. Soc., 1997, 119, 4021–4032 CrossRef CAS.
  30. T. Ishiyama and A. Morita, J. Chem. Phys., 2009, 131, 244714 CrossRef PubMed.
  31. J. Wang, P. Cieplak and P. A. Kollman, J. Comput. Chem., 2000, 21, 1049–1074 CrossRef CAS.
  32. T. Ishida and A. Morita, J. Chem. Phys., 2006, 125, 074112 CrossRef PubMed.
  33. L. Martinez, R. Andrade, E. G. Birgin and J. M. Martinez, J. Comput. Chem., 2009, 30, 2157–2164 CrossRef CAS PubMed.
  34. S. Nosé, J. Chem. Phys., 1984, 81, 511–519 CrossRef.
  35. W. Hoover, Phys. Rev. A, 1985, 31, 1695–1697 CrossRef.
  36. B. Hess, C. Kutzner, D. V. D. Spool and E. Lindahl, J. Chem. Theory Comput., 2008, 4, 435–447 CrossRef CAS PubMed.
  37. W. L. Jorgensen, J. Chandrasekhar, J. D. Madura, R. W. Impey and M. L. Klein, J. Chem. Phys., 1983, 79, 926–935 CrossRef CAS.
  38. A. Morita and T. Ishiyama, Phys. Chem. Chem. Phys., 2008, 10, 5801–5816 RSC.
  39. A. Morita and J. T. Hynes, J. Phys. Chem. B, 2002, 106, 673–685 CrossRef CAS.
  40. T. Ishiyama, D. Terada and A. Morita, J. Phys. Chem. Lett., 2016, 7, 216–220 CrossRef CAS PubMed.
  41. S. Nihonyanagi, R. Kusaka, K. Inoue, A. Adhikari, S. Yamaguchi and T. Tahara, J. Chem. Phys., 2015, 143, 124707 CrossRef PubMed.
  42. S. Yamaguchi, J. Chem. Phys., 2015, 143, 034202 CrossRef PubMed.
  43. Q. Du, R. Superfine, E. Freysz and Y. R. Shen, Phys. Rev. Lett., 1993, 70, 2313–2316 CrossRef CAS PubMed.
  44. N. Ji, V. Ostroverkhov, C. S. Tian and Y. R. Shen, Phys. Rev. Lett., 2008, 100, 096102 CrossRef CAS PubMed.
  45. V. Ostroverkhov, G. A. Waychunas and Y. R. Shen, Phys. Rev. Lett., 2005, 94, 046102 CrossRef PubMed.
  46. A. Myalitsin, S. Urashima, S. Nihonyanagi, S. Yamaguchi and T. Tahara, J. Phys. Chem. C, 2016, 120, 9357–9363 CAS.
  47. S. Yamaguchi, K. Bhattacharyya and T. Tahara, J. Phys. Chem. C, 2011, 115, 4168–4173 CAS.
  48. S. Yamaguchi, A. Kundu, P. Sen and T. Tahara, J. Chem. Phys., 2012, 137, 151101 CrossRef PubMed.
  49. A. Kundu, S. Yarnaguchi and T. Tahara, J. Phys. Chem. Lett., 2014, 5, 762–766 CrossRef CAS PubMed.
  50. T. Watanabe, T. Ebata, S. Tanabe and N. Mikami, J. Chem. Phys., 1996, 105, 408–419 CrossRef.
  51. Y. Yamada, T. Ebata, M. Kayano and N. Mikami, J. Chem. Phys., 2004, 120, 7400–7409 CrossRef CAS PubMed.
  52. K. P. Gierszal, J. G. Davis, M. D. Hands, D. S. Wilcox, L. V. Slipchenko and D. Ben-Amotz, J. Phys. Chem. Lett., 2011, 2, 2930–2933 CrossRef CAS.
  53. R. N. Pribble and T. S. Zwier, Science, 1994, 265, 75–79 CAS.
  54. C. J. Gruenloh, J. R. Carney, C. A. Arrington, T. S. Zwier, S. Y. Fredericks and K. D. Jordan, Science, 1997, 276, 1678–1681 CrossRef CAS.
  55. H. Watanabe and S. Iwata, J. Chem. Phys., 1996, 105, 420–431 CrossRef CAS.
  56. T. Hamashima, K. Mizuse and A. Fujii, J. Phys. Chem. A, 2011, 115, 620–625 CrossRef CAS PubMed.
  57. B. M. Auer and J. L. Skinner, J. Chem. Phys., 2008, 128, 224511 CrossRef CAS PubMed.
  58. T. Ishiyama and A. Morita, J. Phys. Chem. C, 2007, 111, 721–737 CAS.


Electronic supplementary information (ESI) available: A snapshot of MD simulations, SVD analysis, a vibrational Im χ(2) spectrum of air/D2O–phenol, the relation between χ(2) of the CH stretch and phenol orientation, a density profile of phenol and water at the air/water–phenol interface obtained via MD simulation, and a computed Im χ(2) spectrum of the air/D2O–phenol (C6H5OH) interface. See DOI: 10.1039/c7cp05150f
R. Kusaka and T. Ishiyama contributed equally to this work.
§ Present address: Nuclear Science and Engineering Center, Japan Atomic Energy Agency (JAEA), 2-4 Shirakata, Tokai, Ibaraki 319-1195, Japan.

This journal is © the Owner Societies 2018