Development of single-channel heterodyne-detected sum frequency generation spectroscopy and its application to the water/vapor interface Shoichi Yamaguchi
Citation: J. Chem. Phys. 143, 034202 (2015); doi: 10.1063/1.4927067 View online: http://dx.doi.org/10.1063/1.4927067 View Table of Contents: http://aip.scitation.org/toc/jcp/143/3 Published by the American Institute of Physics
THE JOURNAL OF CHEMICAL PHYSICS 143, 034202 (2015)
Development of single-channel heterodyne-detected sum frequency generation spectroscopy and its application to the water/vapor interface Shoichi Yamaguchia) Department of Applied Chemistry, Graduate School of Science and Engineering, Saitama University, 255 Shimo-Okubo, Sakura, Saitama 338-8570, Japan
(Received 22 May 2015; accepted 8 July 2015; published online 20 July 2015) Single-channel heterodyne-detected sum frequency generation (HD-SFG) spectroscopy for selectively measuring vibrational spectra of liquid interfaces is presented. This new methodology is based on optical interference between sum frequency signal light from a sample interface and phase-controlled local oscillator light. In single-channel HD-SFG, interferometric and spectrometric measurements are simultaneously carried out with an input IR laser scanned in a certain wavenumber range, which results in a less task than existing phase-sensitive sum frequency spectroscopy. The real and imaginary parts of second-order nonlinear optical susceptibility ( χ(2)) of interfaces are separately obtained with spectral resolution as high as 4 cm−1 that is approximately six times better than existing multiplex HD-SFG. In this paper, the experimental procedure and theoretical background of single-channel HD-SFG are explicated, and its application to the water/vapor interface is demonstrated, putting emphasis on the importance of a standard for the complex phase of χ(2). C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4927067]
INTRODUCTION
Structure and dynamics of molecules at liquid interfaces are of fundamental importance for many fields of chemistry.1 Microscopic view of interfacial molecules is crucial to understand the chemical function of liquid interfaces,2–9 but it cannot be readily gained because conventional spectroscopic techniques such as NMR, UV-visible-IR absorption, Raman scattering, and fluorescence lack intrinsic interface selectivity. Sum frequency generation (SFG) spectroscopy enables the microscopic investigation of liquid interfaces as thin as a few molecular diameter thickness.6,10–15 The interface selectivity of SFG is based on the tensorial property of the secondorder nonlinear optical susceptibility ( χ(2)) under the electric dipole approximation: χ(2) is zero in the centrosymmetric bulk, but it can be nonzero at an anisotropic interface. Secondorder nonlinear polarization given by the product of χ(2) and two input electric fields is localized at the interface, and this polarization is the source of sum frequency (SF) signal light. One can examine the structure and dynamics of interfacial molecules using the information of χ(2) conveyed by this SF signal light. Conventional SFG just measures the intensity of the signal light, which results in the absolute square of χ(2) (| χ(2)|2).6,12,14 There are three serious problems in a | χ(2)|2 spectrum: First, interference between multiple resonant peaks and/or between a resonant peak and nonresonant background inevitably deforms band shape in a | χ(2)|2 spectrum, which disables one to read the spectrum so straightforwardly as bulk spectra such as UVvisible, IR, and Raman. Second, | χ(2)|2 lacks the information of the sign of χ(2) that can inherently indicate “up” vs “down” alignment of interfacial molecules or functional groups. Third, a)[email protected]
0021-9606/2015/143(3)/034202/7/$30.00
because | χ(2)|2 is proportional to the square of the density of interfacial molecules, conventional SFG is not very suitable for quantitative spectrometric experiments where an isosbestic point can be a good marker for measurement reliability.16–18 Phase-sensitive SFG19–21 and multiplex heterodynedetected SFG (HD-SFG)22–27 have solved these problems of conventional SFG. In these methods, optical interference between SF signal light and local oscillator (LO) light permits the separate acquisition of the real and imaginary parts of χ(2) (Re χ(2) and Im χ(2)). An Im χ(2) spectrum is free from the nonresonant background,28 and it can be read as straightforwardly as bulk spectra also corresponding to the imaginary part of linear or nonlinear optical susceptibility.29 The sign of Im χ(2) directly indicates “up” vs “down” alignment of interfacial molecules.30–32 χ(2) is proportional to the density of interfacial molecules, which facilitates quantitative spectrometric experiments.16–18 Multiplex HD-SFG developed by the Tahara group takes advantage of multichannel detection using a femtosecond broad-band IR pulse, a polychromator, and a CCD.22,31,33–42 Interference fringes between SF signal and LO lights are recorded in the frequency domain. In this sense, interferometric and spectrometric measurements are both benefited by the multichannel detection in multiplex HD-SFG, which effectively reduces measurement time. Phase-sensitive SFG developed by the Shen group utilizes a picosecond narrow-band IR pulse, a monochromator, and a photomultiplier tube (PMT).19,43–48 Interferometric measurements are performed not in the frequency domain but in the time (or space) domain by tilting a phase-modulator plate step by step. Spectrometric measurements require the narrow-band IR pulse to be scanned in a certain wavenumber range. Interferometric and spectrometric measurements are both carried out with single-channel detection separately in a step-by-step
143, 034202-1
© 2015 AIP Publishing LLC
034202-2
Shoichi Yamaguchi
manner, inevitably resulting in much longer measurement time than multiplex HD-SFG. This article demonstrates a new method named singlechannel HD-SFG in which a picosecond narrow-band IR pulse, a monochromator, and a PMT are used. Interferometric measurements are performed not in the time domain but in the frequency domain at the same time as spectrometric measurements, which brings about a less task than phase-sensitive SFG. The spectral resolution of single-channel HD-SFG in the present study is as high as 4 cm−1 that is approximately six times better than multiplex HD-SFG. (In conventional multiplex SFG, it is not difficult to have spectral resolution better than 4 cm−1, but in multiplex HD-SFG, it is still very difficult because the interval of the interference fringes for heterodyne detection has to be much smaller than 4 cm−1.) This article is constructed as follows: In the section titled Experimental, the schematic of a single-channel HD-SFG setup is presented. The principle of single-channel HD-SFG is described in the section titled Theoretical. Data at each measurement stage and χ(2) spectra of the water/vapor interfaces are shown in the section titled Results. Complex-phase correction for the χ(2) spectrum of the H2O/vapor interface is carried out, and thus, finalized data are discussed in the section titled Discussion.
EXPERIMENTAL
Figure 1 shows the schematic of the optical setup of singlechannel HD-SFG. The light source is a Nd:YAG laser (EKSPLA, PL2231-50). A part of its fundamental output (wavelength: 1064 nm, pulse width: 28 ps, repetition rate: 50 Hz) is frequency-doubled to 532 nm by a second harmonic generation unit (EKSPLA, SFGH500-2H). An optical parametric amplifier (EKSPLA, PG501-DFG1) is pumped by a part of the 532nm output, and a tunable IR pulse (wavelength: 2.3–10 µm) is provided by difference frequency generation between the
J. Chem. Phys. 143, 034202 (2015)
1064-nm pulse and the idler output of the optical parametric amplifier. The other part of the 532-nm output and the IR pulse is used as ω1 and ω2 input pulses for SFG, respectively. The ω1 and ω2 pulses propagate nearly collinearly with a small vertical shift after a mirror (M1 in Figure 1). Time delay between the ω1 and ω2 pulses is controlled by an automatic translation stage (not shown here). They are focused into a y-cut quartz crystal (thickness: 10 µm) mounted on an automatic rotation stage to generate a SF (ω1 + ω2) pulse used as LO. The y-cut quartz has to be so thin because the LO intensity should be comparable with the intensity of SF signal light from a sample. A 5-µm glass plate (G1 in Figure 1) is inserted into (or removed from) the optical paths of the ω1, ω2, and LO pulses after the y-cut quartz crystal by an automatic translation stage. The ω1, ω2, and LO pulses are focused onto a sample interface or a reference left-handed z-cut quartz surface (Furuuchi Chemical). The positive x-axis direction of the z-cut quartz, which is already known from a piezoelectric measurement, is set parallel with the positive direction of the laboratory X axis defined in Figure 1. This results in positive effective χ(2) for the z-cut quartz in the present experiments.28 The incident angles of the ω1 and ω2 pulses are 57.5◦ and 56.6◦, respectively. The sample interface is set at the same vertical height as the reference z-cut quartz surface within the accuracy of 1 µm using a displacement sensor (Keyence, SI-F10) and an automatic vertical translation stage. A SF (ω1 + ω2) signal pulse generated from the sample interface (or the reference z-cut quartz surface) and the LO pulse propagates collinearly and enters a monochromator (Princeton Instruments, SP-2155). Monochromatically filtered light is detected by a PMT (Hamamatsu, R11568) followed by a current pre-amplifier (NF Corporation, LI-76) and a 16bit analog-to-digital (AD) converter (National Instruments, NI USB-6341). A commercial software IGOR Pro (WaveMetrics) is employed to control the optical parametric amplifier, the monochromator, the displacement sensor, the AD converter, and all the automatic stages.
FIG. 1. Schematic of the optical setup of single-channel HD-SFG. The ω 1 and ω 2 lights propagate nearly collinearly with a small vertical shift after the mirror M1. A light-blue thin line drawn between the glass plate G1 and the sample interface represents the LO light. The sample can be replaced with the z-cut quartz by the xyz automatic translation stages. F stands for optical filters. The positive direction of the P polarization of the ω 2 pulse at the incidence of the sample interface is defined to be along the positive directions of the laboratory X and Z axes. The inset shows relation between the polarizations of the ω 1, ω 2, and LO pulses and the crystal axes of the y-cut quartz.
034202-3
Shoichi Yamaguchi
Wavenumber calibration of the ω2 pulse was carried out by comparing IR transmittance spectra of a polystyrene film and water vapor measured with the single-channel HD-SFG setup and those with a FTIR spectrometer (PerkinElmer, System 2000). With the single-channel HD-SFG setup, the IR spectrum of the polystyrene film was obtained by normalizing a SFG intensity spectrum of the reference z-cut quartz surface with the film on the optical path of the ω2 pulse by that without the film. The IR spectrum of water vapor was obtained by normalizing the SFG intensity spectrum without N2 purge by that with it. The wavenumber of the ω2 pulse was calibrated with the accuracy better than 1 cm−1. Wavenumber resolution of the present setup was estimated by comparing the IR transmittance spectra of water vapor shown in Figure 2. The spectrum by the single-channel HDSFG setup (Figure 2(c), after the wavenumber calibration) is nearly identical with that by the FTIR spectrometer with the resolution set at 4 cm−1 (Figure 2(b)) and exhibits poorer resolution than that by FTIR with 2-cm−1 resolution (Figure 2(a)). The spectra in Figure 2 suggest that the wavenumber resolution of the single-channel HD-SFG setup is approximately 4 cm−1. This means that the bandwidth of the ω2 pulse is approximately 4 cm−1. Specific experimental conditions in the present study are as follows: Ultrapure water (resistivity: 18.2 MΩ cm, TOC (total organic carbon):
Citation: J. Chem. Phys. 143, 034202 (2015); doi: 10.1063/1.4927067 View online: http://dx.doi.org/10.1063/1.4927067 View Table of Contents: http://aip.scitation.org/toc/jcp/143/3 Published by the American Institute of Physics
THE JOURNAL OF CHEMICAL PHYSICS 143, 034202 (2015)
Development of single-channel heterodyne-detected sum frequency generation spectroscopy and its application to the water/vapor interface Shoichi Yamaguchia) Department of Applied Chemistry, Graduate School of Science and Engineering, Saitama University, 255 Shimo-Okubo, Sakura, Saitama 338-8570, Japan
(Received 22 May 2015; accepted 8 July 2015; published online 20 July 2015) Single-channel heterodyne-detected sum frequency generation (HD-SFG) spectroscopy for selectively measuring vibrational spectra of liquid interfaces is presented. This new methodology is based on optical interference between sum frequency signal light from a sample interface and phase-controlled local oscillator light. In single-channel HD-SFG, interferometric and spectrometric measurements are simultaneously carried out with an input IR laser scanned in a certain wavenumber range, which results in a less task than existing phase-sensitive sum frequency spectroscopy. The real and imaginary parts of second-order nonlinear optical susceptibility ( χ(2)) of interfaces are separately obtained with spectral resolution as high as 4 cm−1 that is approximately six times better than existing multiplex HD-SFG. In this paper, the experimental procedure and theoretical background of single-channel HD-SFG are explicated, and its application to the water/vapor interface is demonstrated, putting emphasis on the importance of a standard for the complex phase of χ(2). C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4927067]
INTRODUCTION
Structure and dynamics of molecules at liquid interfaces are of fundamental importance for many fields of chemistry.1 Microscopic view of interfacial molecules is crucial to understand the chemical function of liquid interfaces,2–9 but it cannot be readily gained because conventional spectroscopic techniques such as NMR, UV-visible-IR absorption, Raman scattering, and fluorescence lack intrinsic interface selectivity. Sum frequency generation (SFG) spectroscopy enables the microscopic investigation of liquid interfaces as thin as a few molecular diameter thickness.6,10–15 The interface selectivity of SFG is based on the tensorial property of the secondorder nonlinear optical susceptibility ( χ(2)) under the electric dipole approximation: χ(2) is zero in the centrosymmetric bulk, but it can be nonzero at an anisotropic interface. Secondorder nonlinear polarization given by the product of χ(2) and two input electric fields is localized at the interface, and this polarization is the source of sum frequency (SF) signal light. One can examine the structure and dynamics of interfacial molecules using the information of χ(2) conveyed by this SF signal light. Conventional SFG just measures the intensity of the signal light, which results in the absolute square of χ(2) (| χ(2)|2).6,12,14 There are three serious problems in a | χ(2)|2 spectrum: First, interference between multiple resonant peaks and/or between a resonant peak and nonresonant background inevitably deforms band shape in a | χ(2)|2 spectrum, which disables one to read the spectrum so straightforwardly as bulk spectra such as UVvisible, IR, and Raman. Second, | χ(2)|2 lacks the information of the sign of χ(2) that can inherently indicate “up” vs “down” alignment of interfacial molecules or functional groups. Third, a)[email protected]
0021-9606/2015/143(3)/034202/7/$30.00
because | χ(2)|2 is proportional to the square of the density of interfacial molecules, conventional SFG is not very suitable for quantitative spectrometric experiments where an isosbestic point can be a good marker for measurement reliability.16–18 Phase-sensitive SFG19–21 and multiplex heterodynedetected SFG (HD-SFG)22–27 have solved these problems of conventional SFG. In these methods, optical interference between SF signal light and local oscillator (LO) light permits the separate acquisition of the real and imaginary parts of χ(2) (Re χ(2) and Im χ(2)). An Im χ(2) spectrum is free from the nonresonant background,28 and it can be read as straightforwardly as bulk spectra also corresponding to the imaginary part of linear or nonlinear optical susceptibility.29 The sign of Im χ(2) directly indicates “up” vs “down” alignment of interfacial molecules.30–32 χ(2) is proportional to the density of interfacial molecules, which facilitates quantitative spectrometric experiments.16–18 Multiplex HD-SFG developed by the Tahara group takes advantage of multichannel detection using a femtosecond broad-band IR pulse, a polychromator, and a CCD.22,31,33–42 Interference fringes between SF signal and LO lights are recorded in the frequency domain. In this sense, interferometric and spectrometric measurements are both benefited by the multichannel detection in multiplex HD-SFG, which effectively reduces measurement time. Phase-sensitive SFG developed by the Shen group utilizes a picosecond narrow-band IR pulse, a monochromator, and a photomultiplier tube (PMT).19,43–48 Interferometric measurements are performed not in the frequency domain but in the time (or space) domain by tilting a phase-modulator plate step by step. Spectrometric measurements require the narrow-band IR pulse to be scanned in a certain wavenumber range. Interferometric and spectrometric measurements are both carried out with single-channel detection separately in a step-by-step
143, 034202-1
© 2015 AIP Publishing LLC
034202-2
Shoichi Yamaguchi
manner, inevitably resulting in much longer measurement time than multiplex HD-SFG. This article demonstrates a new method named singlechannel HD-SFG in which a picosecond narrow-band IR pulse, a monochromator, and a PMT are used. Interferometric measurements are performed not in the time domain but in the frequency domain at the same time as spectrometric measurements, which brings about a less task than phase-sensitive SFG. The spectral resolution of single-channel HD-SFG in the present study is as high as 4 cm−1 that is approximately six times better than multiplex HD-SFG. (In conventional multiplex SFG, it is not difficult to have spectral resolution better than 4 cm−1, but in multiplex HD-SFG, it is still very difficult because the interval of the interference fringes for heterodyne detection has to be much smaller than 4 cm−1.) This article is constructed as follows: In the section titled Experimental, the schematic of a single-channel HD-SFG setup is presented. The principle of single-channel HD-SFG is described in the section titled Theoretical. Data at each measurement stage and χ(2) spectra of the water/vapor interfaces are shown in the section titled Results. Complex-phase correction for the χ(2) spectrum of the H2O/vapor interface is carried out, and thus, finalized data are discussed in the section titled Discussion.
EXPERIMENTAL
Figure 1 shows the schematic of the optical setup of singlechannel HD-SFG. The light source is a Nd:YAG laser (EKSPLA, PL2231-50). A part of its fundamental output (wavelength: 1064 nm, pulse width: 28 ps, repetition rate: 50 Hz) is frequency-doubled to 532 nm by a second harmonic generation unit (EKSPLA, SFGH500-2H). An optical parametric amplifier (EKSPLA, PG501-DFG1) is pumped by a part of the 532nm output, and a tunable IR pulse (wavelength: 2.3–10 µm) is provided by difference frequency generation between the
J. Chem. Phys. 143, 034202 (2015)
1064-nm pulse and the idler output of the optical parametric amplifier. The other part of the 532-nm output and the IR pulse is used as ω1 and ω2 input pulses for SFG, respectively. The ω1 and ω2 pulses propagate nearly collinearly with a small vertical shift after a mirror (M1 in Figure 1). Time delay between the ω1 and ω2 pulses is controlled by an automatic translation stage (not shown here). They are focused into a y-cut quartz crystal (thickness: 10 µm) mounted on an automatic rotation stage to generate a SF (ω1 + ω2) pulse used as LO. The y-cut quartz has to be so thin because the LO intensity should be comparable with the intensity of SF signal light from a sample. A 5-µm glass plate (G1 in Figure 1) is inserted into (or removed from) the optical paths of the ω1, ω2, and LO pulses after the y-cut quartz crystal by an automatic translation stage. The ω1, ω2, and LO pulses are focused onto a sample interface or a reference left-handed z-cut quartz surface (Furuuchi Chemical). The positive x-axis direction of the z-cut quartz, which is already known from a piezoelectric measurement, is set parallel with the positive direction of the laboratory X axis defined in Figure 1. This results in positive effective χ(2) for the z-cut quartz in the present experiments.28 The incident angles of the ω1 and ω2 pulses are 57.5◦ and 56.6◦, respectively. The sample interface is set at the same vertical height as the reference z-cut quartz surface within the accuracy of 1 µm using a displacement sensor (Keyence, SI-F10) and an automatic vertical translation stage. A SF (ω1 + ω2) signal pulse generated from the sample interface (or the reference z-cut quartz surface) and the LO pulse propagates collinearly and enters a monochromator (Princeton Instruments, SP-2155). Monochromatically filtered light is detected by a PMT (Hamamatsu, R11568) followed by a current pre-amplifier (NF Corporation, LI-76) and a 16bit analog-to-digital (AD) converter (National Instruments, NI USB-6341). A commercial software IGOR Pro (WaveMetrics) is employed to control the optical parametric amplifier, the monochromator, the displacement sensor, the AD converter, and all the automatic stages.
FIG. 1. Schematic of the optical setup of single-channel HD-SFG. The ω 1 and ω 2 lights propagate nearly collinearly with a small vertical shift after the mirror M1. A light-blue thin line drawn between the glass plate G1 and the sample interface represents the LO light. The sample can be replaced with the z-cut quartz by the xyz automatic translation stages. F stands for optical filters. The positive direction of the P polarization of the ω 2 pulse at the incidence of the sample interface is defined to be along the positive directions of the laboratory X and Z axes. The inset shows relation between the polarizations of the ω 1, ω 2, and LO pulses and the crystal axes of the y-cut quartz.
034202-3
Shoichi Yamaguchi
Wavenumber calibration of the ω2 pulse was carried out by comparing IR transmittance spectra of a polystyrene film and water vapor measured with the single-channel HD-SFG setup and those with a FTIR spectrometer (PerkinElmer, System 2000). With the single-channel HD-SFG setup, the IR spectrum of the polystyrene film was obtained by normalizing a SFG intensity spectrum of the reference z-cut quartz surface with the film on the optical path of the ω2 pulse by that without the film. The IR spectrum of water vapor was obtained by normalizing the SFG intensity spectrum without N2 purge by that with it. The wavenumber of the ω2 pulse was calibrated with the accuracy better than 1 cm−1. Wavenumber resolution of the present setup was estimated by comparing the IR transmittance spectra of water vapor shown in Figure 2. The spectrum by the single-channel HDSFG setup (Figure 2(c), after the wavenumber calibration) is nearly identical with that by the FTIR spectrometer with the resolution set at 4 cm−1 (Figure 2(b)) and exhibits poorer resolution than that by FTIR with 2-cm−1 resolution (Figure 2(a)). The spectra in Figure 2 suggest that the wavenumber resolution of the single-channel HD-SFG setup is approximately 4 cm−1. This means that the bandwidth of the ω2 pulse is approximately 4 cm−1. Specific experimental conditions in the present study are as follows: Ultrapure water (resistivity: 18.2 MΩ cm, TOC (total organic carbon):
