Surface Selection Rule of Infrared Diffuse Reflection Spectrometry for

Mar 24, 2014 - Here, let us remember that the DR component is dominant at a small angle on a rough surface (Figure 5). On DR, the incident light is li...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Surface Selection Rule of Infrared Diffuse Reflection Spectrometry for Analysis of Molecular Adsorbates on a Rough Surface of a Nonabsorbing Medium Seiya Morimine, Shingo Norimoto, Takafumi Shimoaka, and Takeshi Hasegawa* Laboratory of Solution and Interface Chemistry, Division of Environmental Chemistry, Institute for Chemical Research, Kyoto University, Gokasho, Uji, Kyoto 611-0011, Japan S Supporting Information *

ABSTRACT: The surface selection rule (SSR) for discussing the molecular orientation in a thin film adsorbed on a rough surface is determined by analyzing a surface monolayer by defining the angle of incidence and polarizations. As the standard sample, a highly organized self-assembled monolayer (SAM) on a rough alumina surface is employed. By introducing crossed-Nicol polarizers in the incident and detection paths, the specular reflection and diffuse reflection components are readily separated. To fully understand the spectra of the SAM, a new idea is proposed that the incidental light can be excluded from the discussion when the angle of incidence is small, which is named the pseudotransmission (pd-Tr) model. Another important idea is that a part of a spectrum is degraded in the signal-to-noise ratio by the suppression of incidental light on the rough surface via a deconstructive interference, which can experimentally be revealed by the crossed-Nicol measurements of single-beam spectra depending on the angle of incidence. Through the experiments of all the combinations of polarizations and angles of incidence, the pd-Tr model and the light suppression are found to be an important base to fully understand the SSR of molecular adsorbates on a rough surface of a nonabsorbing medium.

T

molecular orientation is still difficult to discuss. To get over the limitation of the conventional spectroscopy, a new SSR for the rough surface is necessary. To measure absorption spectra of the molecular adsorbates on a rough surface, diffuse reflection spectrometry19 has conveniently been used thus far. Although this technique is powerful to measure the spectra with a high sensitivity, the spectra are difficult to be used for “characterization” of the adsorption structure. The conventional equipment of the diffuse reflection spectrometry involves a large concave mirror to cover a wide solid angle, so that a high-throughput spectroscopy is realized.20 This idea, in fact, works powerfully to measure the minute chemical species adsorbed on a powder bed. Nevertheless, at the same time, this idea spoils the important parameter of “angle of incidence”. An interesting spectrum is presented in Figure 1. This is a diffuse reflection spectrum of a self-assembled monolayer (SAM) of octadecyl silane prepared on a rough alumina surface at an angle of incidence of 20°, which has positive and negative peaks, which cannot be explained by the conventional SSRs. In the present study, a technique involving a well-defined angle of incidence and polarization is introduced to decompose the complicated diffusely reflected light into the specular reflection, diffuse reflection, and deconstructively interfering light components. In addition, a new concept of pseudotransmission has been added, which enables us to fully

he molecular adsorption structure is mostly discussed via the orientation and conformation on a “flat surface” when infrared (IR) absorption spectroscopy is employed. The surface can be metallic1,2 and dielectric matters,3−6 which involve even a liquid surface.7−9 On a flat surface, a surface selection rule (SSR) is available for every optical configuration such as transmission,2 external reflection,2−6 and internal reflection.2,10 For example, the transmission measurements are performed by using the normal-incident light; only the surface-parallel component of a transition moment appears as a band in the spectrum, 2 which is called the “SSR of transmission spectrometry”. The SSR is in fact quite convenient to discuss the molecular orientation of a molecule adsorbed on a flat surface. When we have both transmission and reflection− absorption (RA)2,11 spectra, in particular, we have information of both surface-parallel and -perpendicular transition moments, which enables us to reveal the orientation angle of each chemical group. Multiple-angle incidence resolution spectrometry (MAIRS) is another option to have both transmission and RA spectra at a time.12−15 When molecules are adsorbed on a “rough surface”, however, the situation becomes highly complicated. Here, the rough surface means that the roughness is comparable to the wavelength of the probe light. When the roughness is more than ca. 1 μm, the incident IR light is diffused by the rough surface, which brings us far away from the conventional SSRs. In practice, however, molecular adsorbates on a rough surface or on powder often play an important role. For example, dye molecules of a ruthenium complex adsorbed on titania particles play a key role in a dye-sensitized solar cell,16−18 but the © 2014 American Chemical Society

Received: November 12, 2013 Accepted: March 24, 2014 Published: March 24, 2014 4202

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208

Analytical Chemistry

Article

An alumina plate was oxidized under an ozone atmosphere to prepare a high-density SAM on the surface. The oxidization was performed by using a BioForce (Ames, IA, U.S.A.) PC440 UV/ ozone ProCleaner Plus. The total duration time of UVirradiation was 10 min. The oxidized alumina plate was dipped in the SAMLAY-A solution for 10 min, followed by rinsing with organic solvents: ethanol, acetone, and dichloroethane. To check the molecular density of the SAM, another SAM was prepared on a flat alumina surface. The flat surface was obtained by polishing the rough surface on a Maruto (Tokyo, Japan) Dia-Lap ML-150P polisher. The flatness was confirmed by AFM, and the surface ratio was 1.001. The flat surface exhibited the reflectance of nil for p-polarization at the angle of incidence of 60° (Brewster’s angle), which means that the refractive index of the surface in the IR region is 1.7. The SAM was characterized as follows. We have already obtained an ideally prepared SAM21 having nearly perpendicular orientation on “silicon”, and the absorbance of the CH2 antisymmetric stretching vibration band was 0.00214. With the refractive indices of silicon (n = 3.4)21 and alumina surface (n = 1.7), the absorbance of the monolayer on alumina can theoretically be predicted. Absorbance, A, at a wavelength of λ is correlated to the electric permittivity of the thin film, ε2, when the angle of incidence is zero (normal incidence):2

Figure 1. An unpolarized infrared reflection spectrum of a SAM of ODS deposited on a rough alumina surface at an angle of incidence of 20°, which is the same as the top spectrum in Figure 6.

understand the reflection spectra of a monolayer coating on a rough surface for all the angle of incidence and polarizations.

■

EXPERIMENTAL SECTION Chemicals. An alumina plate (30 × 30 mm2) with a thickness of 0.47 ± 0.05 mm with a surface roughness was purchased from Nippon Carbide Industries Co., Inc. (Toyama, Japan). For modifying the surface of the alumina plate with a SAM, octadecyl trimethoxy silane (ODS) was purchased from Nippon Soda Co., Ltd. (Tokyo, Japan) as a solution (SAMLAY-A) of an organic solvent involving methylethylbenzene. For other details of SAM preparation, the reader is referred to the literature.21 The water was obtained by a Millipore (Molsheim, France) Elix UV-3 pure-water generator and a Yamato (Tokyo, Japan) Autopure WT100U water purifier, which is a compatible model with Milli-Q. The water exhibited an electric resistivity higher than 18.2 MΩ cm, and the surface tension was 72.8 mN m−1 at room temperature (25 °C), which guaranteed that the water was free from contaminants. IR ER Measurements. IR diffuse reflection measurements were performed on a Thermo Fischer Scientific (Madison, WI, USA) Magna550 FT-IR spectrometer equipped with a Harrick (Pleasantville, NY, USA) The Seagull. The polarization measurements were performed by mounting Harrick PWGU1R wire-grid polarizers. The modulation frequency of FT-IR was 60 kHz, and the IR ray was detected by a liquid-N2 cooled MCT detector. The wavenumber resolution was 4 cm−1. The number of accumulation of the interferogram was 2000. AFM Measurements. The AFM images of the SAMs were measured on a Seiko Instruments Inc. (Chiba, Japan) NanoNavi IIs Probe Station equipped with a probing microscope unit, Nanocute, and the probe unit was put on an antivibration stage. The dynamic (tapping) mode was used for the AFM scanning. The cantilever was made of crystalline silicon tip, and its force constant was 40 N m−1. Sample Preparation. In the present study, two types of measurements were performed: single-beam and absorbance measurements. The single-beam (light intensity) measurements were carried out by using a rough (surface ratio: 1.200) and bare alumina surface to check the reflectivity depending on the surface roughness, whereas the absorbance measurements were done by using a rough alumina surface covered with a SAM to investigate the SSR. Here, the surface ratio is a ratio between the areas of the rough surface and ideally the flat surface.

A=

1 8πd 2m1m3 Im(ε2) ln 10·λ m1 + m3

(1)

Here, mj ≡ 1/cnj where j is a layer index for air (1), thin film (2), and substrate (3) and c is the light velocity. With this equation, two absorbance values can easily be compared with each other by eq 2, if the monolayers on the two substrates have a common structure: A alumina nSi + 1 = ASi nalumina + 1

(2)

Therefore, if the SAM on alumina is also ideally prepared as that on silicon, the calculated absorbance should agree with the experimental value. The absorbance is calculated to be: A alumina = A Si

nSi + 1 3.4 + 1 = 0.00214 × = 0.0035 nalumina + 1 1.7 + 1 (2′)

The IR transmission spectrum of the SAM on the flat alumina surface exhibited the band with absorbance of 0.0035 (Figure S-1, Supporting Information). This perfect agreement strongly suggests that the SAM on alumina has almost the same molecular density and orientation found in the ideally prepared SAM on silicon. This guarantees that the present SAM on the alumina surface has surface-perpendicular orientation, which is a good standard to discuss the spectra.

■

RESULTS AND DISCUSSION Discrimination of Specular Reflection and Diffuse Reflection. Figure 2 presents unpolarized IR “single-beam” spectra of a bare alumina plate with no SAM measured at the angles of incidence from 20° to 80°. Below ca. 50°, the intensity decreases with an increase of the angle. When the angle increases over ca. 50°, the intensity turns into an increasing trend and goes up rapidly showing a different spectral shape. In these measurements, both the specular reflection (SR) and diffuse reflection (DR) components coexist, which makes it 4203

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208

Analytical Chemistry

Article

Figure 4 presents the DR-filtered IR “single-beam” spectra of the bare alumina surface using the s-polarization as the incident

Figure 2. Infrared unpolarized (u:u) reflection “single-beam” spectra of a bare rough alumina surface as a function of the angle of incidence.

difficult to understand the spectral variation. At the moment, however, the spectrum at a grazing angle of incidence (∼80°) can be approximated as the SR component, since the light does not go deep inside the micro valley of the surface, which makes the DR component very weak. Therefore, the spectrum at 80° in Figure 2 can be approximated as the SR component. Here, SR has important characteristics: (1) the reflection angle is equal to the angle of incidence and (2) the plane of polarization of the incident path is preserved in the reflection path.22,23 Therefore, SR occurs limitedly on the frontier surface of the rough topography, which allows only a limited portion of the IR light reach the detector. On the other hand, DR has totally different characteristics: the light reaches the inside of each microvalley due to the multiple reflections in the valley especially for a small angle of incidence (∼20° or less), during which the light is depolarized. Another important characteristic is that the emitted light by DR is scrambled, and no specific reflection angle is available. With these differences, SR and DR can be separated experimentally. A schematic of the separation experiments is presented in Figure 3. In this example, a linear polarizer generating s-

Figure 4. Infrared crossed-Nicol (s:p) reflection “single-beam” spectra of a bare rough alumina surface as a function of the angle of incidence.

light (denoted as (s:p)) as a function of the angle of incidence. The SR component is largely suppressed for all the angles of incidence by introducing the DR filter, and the spectral intensity is thus entirely decreased as shown by the change of the full scales between Figures 2 and 4. When the angle of incidence is larger than ca. 60° in Figure 4, the spectral shape becomes similar to that of the SR component in Figure 2. This indicates that the DR component is largely suppressed at a large angle and the “leaked” SR component through the second polarizer reaches the detector. Therefore, the spectrum at 20° in Figure 4 can temporarily be assigned to the DR component. These approximations are confirmed as follows. The experimentally obtained two spectra (SR ad DR) are row-wisely stored in the matrix, K, which is used to reveal the quantity variation, C, extracted from the unpolarized thirteen spectra in Figure 2, A, by employing the CLS regression calculation:24 ⎛ SR spectrum ⎞ ⎟⎟ + R ≡ CK + R A = C⎜⎜ ⎝ DR spectrum ⎠

(3)

Here, the analytical residuals are discarded in R. The quantity variation is calculated by eq 4 as the least-squares solution,24 and the normalized results are shown in Figure 5. C = AKT(KKT)−1 Figure 3. A schematic of the DR filter consisted of two mutually orthogonal (crossed-Nicol) polarizers. The incidental and detection angles are set to be equal to each other. The solid circle (●) and the two-direction arrow (↔) indicate the s- and p-polarizations, respectively. Only the DR component passes through the second polarizer.

(4)

As expected, DR is dominant at 20°, whereas SR is dominant at 80°, and each variation is monotonous with the angle. The temporal assignments have thus proved to be reasonable. These two spectra are quite useful to understand the absorbance spectra of the SAM on a rough alumina surface, as follows. “Pseudo-Transmission (pd-Tr) Model” for a Small Angle of Incidence. Figure 6 presents unpolarized (denoted as (u:u)) IR “absorbance” spectra of the SAM on the rough alumina surface as a function of the angle of incidence (20− 80°). Each background measurement was performed by using the spectra in Figure 2. At a small angle of incidence below ca. 40°, the CH2 antisymmetric and symmetric stretching vibration (νa(CH2) and νs(CH2)) bands at 2916 and 2850 cm−1, respectively, appear as “positive” peaks with a nearly constant intensity. As mentioned in the Experimental Section, the SAM

polarization is set in the incident light path at the angle of incidence of θi, and the detection angle, θd, is set to satisfy θi = θd. Another linear polarizer is set in the detection path with the polarization orthogonal (p-polarization) to the incident polarization. With this optical configuration, the SR component should effectively be rejected, and only the depolarized light by the DR path would be passed through the second polarizer. This crossed-Nicol experiment thus works as a “DR filter”. 4204

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208

Analytical Chemistry

Article

and instead, the rough surface can be regarded as an illumination source. For example at 20°, the incident light virtually comes from the alumina surface, and the “transmitted” light through the SAM goes to the detector. The positive absorbance bands are readily understood in this model. In this manner, for a “small angle of incidence”, the SSR on a rough surface resembles that of transmission spectrometry. Exactly speaking, this is of course a rough approximation, since the incident light once passes through the SAM before the rough surface works as the illumination source. Therefore, this model is named “pseudo transmission (pd-Tr)” model. Fortunately, however, this approximation works reasonably as discussed below. The absorbance of the νa(CH2) band found in Figure 1 is 0.0097. As stated in the Experimental Section, the absorbance of the genuine transmission measurement of the SAM on the flat alumina surface is 0.0035, which means that pd-Tr enables us to measure the minute adsorbates on a rough surface with a high sensitivity. As a result of the enhancement, the initial incidence before the virtual illuminator becomes minor and the shape of a pd-Tr spectrum become close to that of the genuine transmission spectrum. The enhancement factor depends on the surface topography, which is highly difficult to be modeled by physical calculation. In this manner, the square (A) in Figure 6 has been found to be useful for discussing the molecular orientation in the adsorbates on a rough surface. When the angle becomes larger than 60°, the DR component becomes smaller as presented in Figure 5, and SR becomes dominant, in which the pd-Tr model alone does not work. Of note here is that the reflectance of spolarization is apparently larger than the p-polarization, especially near Brewster’s angle (Figure S-3, Supporting Information). As a result, this region (enclosed by a square (B) in Figure 6) is almost governed by the SSR of s-polarized ER spectrometry. On the other hand, the fingerprint region (below 2000 cm−1; enclosed by a square (C)) of the spectra in Figure 6 is characterized by a very poor signal-to-noise (SN) ratio especially for the small angles of incidence, in which the DR component is dominant. When referring to the DR intensity spectra in Figure 4, the light intensity in the range of 1600− 1000 cm−1 is particularly suppressed down to nearly zero. This significant suppression can be understood by taking the surface topography into account. Figure 7 presents a representative AFM image of the alumina surface. The rough surface has a highly complicated topography, but the size of an average particle-like protrusion is about 5−10 μm, which agrees with the wavelength of the suppressed region. In other words, the incident light should interfere with the rough surface in a deconstructive manner as a result, and the surface does not work as the virtual illuminator in the wavenumber region. Therefore, the very dark “leaked” SR component through the second polarizer can be a major component. In this case, the SN ratio becomes very poor and the SSR obeys the ER spectrometry. In this manner, the two crossed-Nicol (s:p) “single-beam” spectra at 20° and 80° that correspond to the DR and SR components (Figure 4) are useful to separate the wavenumber region to choose an appropriate SSR. As found in Figure 4, the DR spectrum (20°) and the SR one (80°) intersect in the range of 2000−1600 cm−1. Therefore, the region above ca. 2000 cm−1 (enclosed by a square (A) in Figure 6) is governed by the SSR of pd-Tr spectrometry, whereas the region below ca. 2000 cm−1

Figure 5. Normalized quantity variations of the SR (open circle) and DR (solid circle) components calculated via the CLS regression.

Figure 6. Infrared unpolarized (u:u) reflection “absorbance” spectra of a SAM of ODS on the rough alumina surface as a function of the angle of incidence.

molecules are revealed to stand nearly perpendicularly to the surface, which results in nearly parallel orientations for the two modes. Here, the “parallel orientation” can be considered because the size of the mode (dipole) is adequately smaller than curvature of the roughness in a μm scale. In this situation, the positive absorbance bands cannot be understood in the conventional SSR of ER spectrometry.2−6 Magnified spectra of Figure 6 are available in Figure S-2, Supporting Information. In the range of angles below Brewster’s angle (ca. 60°), the p-polarization component should give “negative” absorbance for a transition moment nearly parallel to the surface. In addition, the s-polarization component always gives a negative absorbance irrespective of the angle of incidence for a surface parallel transition moment.2−6 The unpolarized (s + p) measurements should thus definitely yield negative bands. Therefore, another mechanism is necessary to elucidate the observed positive peaks. Here, let us remember that the DR component is dominant at a small angle on a rough surface (Figure 5). On DR, the incident light is literally diffused by the surface roughness, and as a result, a depolarized emission is generated from the surface, which is “impervious to the incident angle” as found in Figure 6 (20−40°). This “angle-independent character” leads us to an idea that the incident light can be excluded from the discussion, 4205

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208

Analytical Chemistry

Article

Figure 8. Infrared crossed-Nicol (s:p) reflection “absorbance” spectra of a SAM of ODS on the rough alumina surface as a function of the angle of incidence.

Figure 7. An AFM image of a rough alumina surface.

In the fingerprint region, the baseline caves in largely. When referring to the polarized “single-beam” spectra in Figure 4, the reason is apparent: the intensity in the region is close to zero, which makes the analytical accuracy largely degraded. In other words, the large negative band should be attributed to an artifact, which should be removed from the discussion. In this manner, the newly proposed SSR for diffuse reflection spectrometry is a combination of the pd-Tr model, ER spectrometry, and the surface roughness effect, which has proved to be powerful to understand the structure of molecular adsorbates on a rough surface. The rest of the cases using different combination of polarizations is discussed in the Appendix.

is governed by the SSR of ER spectrometry. In this framework, the understanding of the weak negative band (the CH2 scissoring vibration mode; δ(CH2)) at 20° in Figure 1 is easy, as described below. Since the spectrum in Figure 1 is a result of unpolarized measurement at the angle of 20°, the DR component is diminished by the surface roughness in the low-wavenumber region, and the leaked unpolarized SR components reach the detector (Figure 4). By referring to the SSR of ER spectrometry,2−6 both s- and p-polarizations would yield negative absorbance for a surface-parallel transition moment (δ(CH2)). Therefore, in the unpolarized (s + p) spectrum, the band appears as a negative peak. When the angle increases (Figure S-3), on the other hand, p-polarization is quickly suppressed, since the angle is close to the Brewster angle, which makes the SN ratio largely poorer. In addition, the s-polarized “absorbance” monotonously decreases due to SSR.2−6 As a result, the δ(CH2) band appears only when the angle of incidence is small, which perfectly agrees with the results in Figure 6. In this manner, the unpolarized spectra of the SAM on a rough alumina surface (Figure 6) have readily been understood by the pd-Tr model and the SSR of ER spectrometry after taking the surface roughness into account. Understanding of Polarized Spectra. Polarized IR spectra of SAM on a rough alumina are analyzed in a similar manner. First, the crossed-Nicol (s:p) spectra of the same SAM in Figure 8 are analyzed. In the same manner as the discussion of unpolarized spectra, the pd-Tr model is employed for the region above 2000 cm−1. As expected, both νa(CH2) and νs(CH2) modes appear as positive peaks until the angle is increased up to 60°. After increasing the angle beyond 60°, the SN ratio becomes too poor to find a peak. This is understandable by referring to Figure 5. When the angle is beyond 60°, DR becomes minor, and only the leaked spolarized SR component with very weak single-beam intensity (Figure 4) becomes a major signal, which makes the SN ratio very poor. According to the SSR of s-polarized ER spectrometry, in addition, the absolute intensity becomes smaller with the angle of incidence.2−6 Therefore, in the region of 60°−80°, both peak intensity and SN ratio are becoming poorer simultaneously, which explains the invisible peaks at a large angle.

■

CONCLUSION The SSR of unpolarized diffuse reflection spectrometry for surface adsorbates on a rough surface is summarized as follows. When the angle of incidence is small and the DR component is not suppressed by the roughness (the square (A) in Figure 6), the pd-Tr model should be employed. This region is useful thanks to a good SN ratio. When the angle is larger than Brewster’s angle (the square (B) in Figure 6), the SSR of spolarized ER spectrometry works, since the p-polarization is diminished near Brewster’s angle. However, the band intensity is much weaker than that in the region (A). In the wavenumber region corresponding to the surface roughness (the square (C) in Figure 6), the SSR of ER spectrometry is employed. Unfortunately, however, this region is very noisy due to the extremely low throughput, which is useless. As a result, the region (A) is found most useful for characterizing the surface adsorbates. The pd-Tr model is quite useful to understand the diffuse reflection spectra after choosing an appropriate region by using the DR filter. Since the pd-Tr spectrum has almost the same shape as the genuine transmission spectrum, the SSR of pd-Tr spectra can be considered to be the same as the SSR of the transmission one. With this technique, the molecular orientation of the surface adsorbates can be discussed, as if the conventional transmission spectrometry was employed. As found in the Appendix, all the combinations of polarizations and angles can fully be explained, but the best SN ratio is obtained by the (u:u) measurements. Therefore, 4206

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208

Analytical Chemistry

Article

(p:u) Measurements

unpolarized spectra are most useful for discussing the adsorption structure of molecular adsorbates on a rough surface using an appropriate SSR after selection by using the DR filter.

■

Figure S-8, Supporting Information, presents the results for the (p:u) measurements. A similar thing between (s:s) and (s:u) happens. In addition, however, the reflectivity of the ppolarization is smaller than the s-polarization especially near Brewster’ angle. Therefore, the SN ratio of (p:u) is still poor near the Brewster angle.

APPENDIX

■

(p:s) Measurements

Crossed-Nicol (p:s) experiments were performed as presented in Figure S-4, Supporting Information. When the angle of incidence is small, the pd-Tr model should be employed. In fact, positive peaks are found in the region. Although the νa(CH2) and νs(CH2) bands may look weaker than those in Figure 8, the intensities are almost equal to each other when the scale bar is taken into account. In the present case, however, the SN ratio is very poor because the incident light is ppolarized. Since the reflectance of p-polarization on a dielectric matter is apparently less than the s-polarization especially about the Brewster angle, the SN ratio is rapidly degraded with an angle of incidence. Although the (p:s) measurements are not useful in this manner, the understanding of the spectra is within the pd-Tr model.

ASSOCIATED CONTENT

S Supporting Information *

Eight figures as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +81 774 38 3074. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The polisher to prepare a flat alumina surface was kindly provided by Dr. Hiroyuki Yoshida, Institute for Chemical Research, Kyoto University, for which the authors’ thanks are due. This work was financially supported by Grant-in-Aid for Scientific Research (B) (No. 23350031) from Japan Society for the Promotion of Science and Priority Areas (23106710) from the Ministry of Education, Science, Sports, Culture, and Technology, Japan. We also thank Sumitomo Foundation for the financial support for our study (#120384).

(s:s) Measurements

Figure S-5, Supporting Information, presents the results of the parallel Nicol (s:s) experiments. With the optical configuration, the SR component is allowed to pass the second polarizer. Therefore, when the angle of incidence is large, the DR is suppressed (Figure 5) and the SR component becomes dominant. In fact, the νa(CH2) and νs(CH2) bands appear as negative bands due to the SSR of the ER spectrometry. On the other hand, when the angle is small (20°−35°), the DR component becomes larger (Figure 5), which makes the bands positive because the pd-Tr bands win the negative SR bands.

■

REFERENCES

(1) Umemura, J. Reflection−Absorption Spectroscopy of Thin Films on Metallic Substrates. In Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R., Eds.; John Wiley: Chichester, 2002; Vol. 2, pp 982−998. (2) Tolstoy, V. P.; Chernyshova, I. V.; Skryshevsky, V. A. Handbook of Infrared Spectroscopy of Ultrathin Films; John Wiley: Chichester, 2003. (3) Hasegawa, T.; Umemura, J.; Takenaka, T. J. Phys. Chem. 1993, 97, 9009−9012. (4) Hasegawa, T.; Takeda, S.; Kawaguchi, A.; Umemura, J. Langmuir 1995, 11, 1236−1243. (5) Hoffmann, H.; Mayer, U.; Krischanitz, A. Langmuir 1995, 11, 1304−1312. (6) Kattner, J.; Hoffmann, H. External Reflection Spectroscopy of Thin Films on Dielectric Substrates. In Handbook of Vibrational Spectroscopy; Chalmers, J. M., Griffiths, P. R., Eds.; John Wiley: Chichester, 2002; Vol. 2, pp 1009−1027. (7) Wang, Y.; Du, X.; Guo, L.; Liu, H. J. Chem. Phys. 2006, 124, 134706. (8) Muro, M.; Harada, M.; Okada, T.; Hasegawa, T. J. Phys. Chem. B 2012, 116, 3148−3154. (9) Itoh, Y.; Muro, M.; Hasegawa, T. Appl. Spectrosc. 2010, 64, 1374−1378. (10) Harrick, N. J. Internal Reflection Spectroscopy; Harrick Scientific: New York, 1987. (11) Umemura, J.; Kamata, T.; Kawai, T.; Takenaka, T. J. Phys. Chem. 1990, 94, 62−67. (12) Hasegawa, T. J. Phys. Chem. B 2002, 106, 4112−4115. (13) Hasegawa, T. Anal. Chem. 2007, 79, 4385−4389. (14) Hasegawa, T. Anal. Bioanal. Chem. 2007, 388, 7−15. (15) Hasegawa, T. Appl. Spectrosc. Rev. 2008, 43, 181−201. (16) Grätzel, M. J. Photochem. Photobiol. C: Photochem. Rev. 2003, 4, 145−153.

(p:p) Measurements

Similar (p:p) experiments were performed, and largely different results are obtained as presented in Figure S-6, Supporting Information: the two bands always appear as positive peaks. When the angle is small, the two bands obey the pd-Tr rule. When the angle is near the Brewster angle, for example at 65°, the SR component is largely reduced because of the very low reflectivity of the p-polarization (Figure S-3, Supporting Information). At the same time, the DR intensity also becomes weak (Figure 5), which results in the poor SN ratio. At a large angle of incidence, the surface-parallel transition moment should appear as a positive band due to the SSR of ER spectrometry, which explains the positive peaks all through the angles. (s:u) Measurements

These measurements are basically similar to the (s:s) measurements. Only the difference is that the DR component goes to the detector fully for (s:u) because the second polarizer is removed. The (s:u) spectra are presented in Figure S-7, Supporting Information. As expected, the DR component appears dominantly for a small angle of incidence, which results in the positive bands in the C−H stretching vibration region because of the pd-Tr model. This is impressive particularly at θi = 50° that the bands appear as positive, whereas they are negative bands at θi = 50° in the (s:s) spectra in Figure S-4, Supporting Information. The SN ratio is largely improved in this region because of the high throughput after the removal of the second polarizer. For a large angle, the bands turn negative, since the DR component is largely lost. 4207

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208

Analytical Chemistry

Article

(17) Hagfeldt, A.; Boschloo, G.; Sun, L.; Kloo, L.; Pettersson, H. Chem. Rev. 2010, 110, 6595−6663. (18) Wei, D. Int. J. Mol. Sci. 2010, 11, 1103−1113. (19) Dahm, D. J.; Dahm, K. D. Interpreting Diffuse Reflectance and Transmittance; IM Publications: West Sussex, U.K., 2007. (20) Yang, P. W.; Mantsch, H. H. Appl. Opt. 1987, 26, 326−330. (21) Norimoto, S.; Morimine, S.; Shimoaka, T.; Hasegawa, T. Anal. Sci. 2013, 10, 979−984. (22) Landau, L. D.; Lifshitz, E. M. Electrodynamics of Continuous Media; Elsevier: Amsterdam, 2008. (23) Jackson, J. D. Classical Electrodynamics, 3rd ed.; Wiley: Hoboken, 1999. (24) Hasegawa, T. Principal-Component Regression and Partial Least Squares Regression. In Handbook of vibrational spectroscopy; Chalmers, J. M.; Griffiths, P. R., Eds.; Wiley Interscience: New York, 2002; Vol. 3, pp 2293−2312.

4208

dx.doi.org/10.1021/ac5005025 | Anal. Chem. 2014, 86, 4202−4208