1. Introduction

Multilayer coatings exploiting the effects of interference are widely applied in the production of filters, mirrors, and antireflection coatings. For such applications, layer thicknesses, refractive indices, and transmission range are the crucial material properties determining the quality and optical properties of the final device [1]. The resulting devices can be applied anywhere the need exists for light to be guided and manipulated.

Here, a very special application of optical interference coatings is shown: the use in analytical spectroscopy in order to increase transmission of light through thin metal films. The underlying effect used is induced transmission through a metal film, which in a symmetric geometry is exploited in induced transmission filters and induced absorption filters [2, 3, 4]. The application described here aims to assist in the understanding of molecular structures and the paths of chemical reactions at metal/liquid interfaces. Some of the great challenges of modern society, e.g., in modern energy conversion, rely on solving chemical problems of catalysis or electrocatalysis on solid/liquid interfaces. The investigation of such interfaces suffers from the so-called “pressure gap,” the lack of applicability of many analytical techniques using electrons, as those electrons would need to originate from, or reach to, buried interfaces.

Optical spectroscopies do not suffer from similar drawbacks. They can be used in different wavelength ranges, from the deep ultraviolet to the far-IR, to study condensed phases. Reflection techniques are especially suitable for the study of interfaces. Here, at least one of the examined media must be transparent at the wavelengths employed to deliver the light to the interface under study. A further issue is the separation of signals originating from the bulk over signals originating from the interface [5].

Here, the focus will be on the mid-IR wavelength range, where many molecular vibrations are available to study details of chemical transformations near interfaces. When a metal/liquid interface should be investigated, the light can be delivered to the surface under study through the molecular liquid. This method has several disadvantages: in the IR, many liquids are strong absorbers. Furthermore, in most applications, the liquid contains dissolved species that show a certain absorption spectrum, superimposing with the absorptions from species on the surface. In delivery of the light through the metal, the latter disadvantage disappears, while the former becomes worse: metals are much stronger IR absorbers than molecular liquids. However, they are usually broadband absorbers of IR light, so that the overall irradiance at the interface under study is reduced, as opposed to molecular liquids, where only parts—but usually, important parts—of the spectrum are absorbed [5].

Attenuated total internal reflection (ATR) spectroscopy is a well-established nondestructive, interface-sensitive method frequently used for studying organic molecules on solid surfaces [6, 7]. In ATR spectroscopy, a transparent, high-refractive index medium of incidence is used, with an incident angle above the critical angle of total reflection. Because metal films mostly reflect but hardly transmit light, working with thin metal layers is necessary, because otherwise no light will reach the sample surface [8, 9]. It should be noted that one way around the problems with metals is the use of surface enhanced infrared absorption spectroscopy (SEIRAS) using metal island films or small particles [10, 11]. In this work, the focus is on planar surfaces, which offer several advantages. The biggest advantage for mechanistic studies is the possibility to obtain information about the orientation of the molecular species under study by using the linear dichroism [7]. When thin metal films are used, there is no limitation to a total reflection geometry: the high reflectivity of metals ensures a sufficient intensity of light even below the critical angle of total reflection.

Recently, the concept of dielectric interlayers introduced by Hooper et al. [4] has been extended into the IR in an ATR geometry [12]. The interlayers induce transmission of light through the metal film [2, 4]. Therefore, the interlayers facilitate the delivery of light to the metal/liquid interface under investigation. As in working with standard metals in the mid-IR, the effect exploited here is not related to plasmon-induced transparency [4, 13].

In absorption spectroscopy, usually a background measurement with the optical system without a sample is performed before the sample is put into the beam path. The introduction of the dielectric interlayers between incident medium and metal films leads not only to increased absorption at certain wavelengths in the final absorbance spectrum, but also to the presence of interference fringes, if there is even a slight mismatch in the refractive index between sample and reference [12]. Here, a systematic study of these interference fringes by both calculations and by experiments is presented. Such results are needed for an application of these fringes in analytical spectroscopy.

The main purpose of the use of the transmission-inducing interlayers is to increase the absorbance of molecules adsorbed to or generated near the metal films [12]. Increase of absorbance is achieved by an increase in electric field strength at the metal/ sample interface. While the field strength is decaying into the sample medium, in reflection spectroscopy, mainly molecules in a region with a strong refractive index change contribute to the absorbance spectrum, as such regions have a strong contribution to the overall reflected fields. Naturally, strong changes in refractive index occur across interfaces. The interference fringes that are additionally observed in the spectra when using an interlayer offer additional opportunities to detect molecules near the metal at wavelengths away from their resonant absorption. Here, however, only the bulk solvent acetonitrile is used as model sample to investigate the effects on the absorbance spectrum induced by the presence of the interlayer.

The paper is organized as follows. After introducing the materials and methods used, a short discussion of previous results is presented. In the main part, the effects of thickness of the interference coating, refractive index, and angle of incidence on reflectivities and resulting absorbance spectra are discussed, comparing calculated and experimental spectra.

2. Materials and Methods

2A. Simulations

To compute reflectivities, a self-developed computer program was used [14], which is based on optical multilayer theory [15, 16]. In all cases, literature values of dependence of the optical constants on the wavenumber $\tilde{\nu }$ have been used as input [17, 18]. IR reflectivity spectra $R_{\text{ref}}^{(q)}(\tilde{\nu })$, where q indicates the linear polarization (s, perpendicular; p, parallel) were calculated for the stratified systems, as defined below, with air as an exit medium. In a second computation, the spectra $R_{\text{smp}}^{(q)}(\tilde{\nu })$ in contact with the sample acetonitrile were calculated. As in the measurement process of absorption spectroscopy, these two quantities have been used to calculate the reflectance absorbance spectrum $A(\tilde{\nu })$ as [5]

2B. Experiments

In a high-vacuum physical vapor deposition chamber (Leybold Univex 450) working at $\sim 5\times {10}^{-6}\mathrm{mbars}$, a layer of Ge (Umicore, Balzers, Liechtenstein) was evaporated onto a ${\mathrm{CaF}}_2$ hemisphere using electron-beam evaporation with an evaporation rate of $\approx 10\AA /\mathrm{s}$. On top of the Ge layer, a closed layer of Au (Wieland Dental, Pforzheim, Germany) was evaporated by thermal evaporation with an evaporation rate of $\approx 2\AA /\mathrm{s}$. All evaporation processes were monitored by a quartz crystal microbalance. No additional adhesive or nucleating layer was used, as it has been found that Ge itself acts as a nucleating layer for Au. Closed, continuous films of Au are formed above $d_{\text{Au}}=20\mathrm{nm}$, as has been verified by scanning electron microscopy [12].

A commercial Fourier transform infrared spectrometer Biorad FTS3000 (Varian, Palo Alto, California) equipped with a midband liquid-nitrogen-cooled mercury cadmium telluride detector was used for the measurements, fitted with a Harrick Seagull Unit (Harrick Scientific Products, Pleasantville, New York) that allows adjustment of the angle of incidence. The hemispheric ${\mathrm{CaF}}_2$ crystals with the evaporated layers were placed inside the Seagull unit. A schematic view of the setup is shown in Fig. 1.

As a model sample, the standard laboratory solvent acetonitrile (pro analysi; VWR International) was used and placed in direct contact with the Au layer underneath the ${\mathrm{CaF}}_2$ crystal. The entire setup was purged with nitrogen for 30 min before measurement.

3. Design and Rationale of the Optical Setup

Inspired from [4], a stratified system for use in IR reflection spectroscopy with the structure “incidence medium–high-index transparent interlayer–metal layer sample” (see also Fig. 1) was developed. A Ge interlayer in the system “ZnSe–Ge–Au–acetonitrile” has been shown to increase the transparency of the metal, yielding stronger absorbance of samples compared to the absorbance in the absence of such an interlayer [12]. The increase in absorbance is observed at certain wavenumbers with reduced reflectivity. These wavenumbers are determined by the interference in the interlayer.

Previously, reflectivity spectra were analyzed depending on the medium of incidence [12]. At an angle of incidence of $45°$, for incidence media with refractive indices n ranging from 1 to 3.4, minima in the reflectivity curves are observed. The reflectivity values at those minima become lower with smaller n.

The lower the reflectivity minima, the higher the probability of light tunneling through the metal layer and subsequently reaching the metal/sample interface under study. Therefore, the strongest absorption enhancement is expected when the refractive index of the incidence medium is lowest. Thus, in this work, the focus is on ${\mathrm{CaF}}_2$, a water and solvent-resistant material with low refractive index ($n\sim 1.4$), as incidence medium.

The effect of the thickness of the metal layer on the reflectivity is nontrivial, however, its effect on the final absorbance is trivial: a thicker layer will result in a weaker absorbance, but the shape of the reflectivity curve is not altered. The layer thickness that is needed is usually given by constraints in the manufacturing process, or by the application of the layer system. Therefore, no detailed discussion on the effect of the metal layer thickness will be attempted, and a fixed thickness of the closed, continuous Au layer at $d_{\text{Au}}=20\mathrm{nm}$ is chosen throughout this work.

In the measured as well as the computed absorbance spectra presented previously [12], the presence of interference fringes was noted. Here, these interference fringes are studied systematically with the aim of elucidating their value for analytical spectroscopy.

4. Influence of Interlayer Thickness and Incidence Angle on Reflectivity

This section is dedicated to the study of the effect of thickness of the interlayer and chosen incidence angle on the reflectivities. For this purpose, an exit medium with refractive index of 1.34 was chosen, which is typical for a number of solvents, including acetonitrile, which was used for this work.

Figure 2 depicts the reflectivity spectra $R^{(s)}$ [s polarization, Figs. 2a, 2b] and $R^{(p)}$ [p polarization, Figs. 2c, 2d] of the system “${\mathrm{CaF}}_2$–Ge–Au sample” with Ge interlayer thicknesses $d_{\text{Ge}}=1000\mathrm{nm}$ [Figs. 2a, 2c] and $d_{\text{Ge}}=2500\mathrm{nm}$ [Figs. 2b, 2d] over the complete range of the incident angles from $0°-90°$.

Regions with low reflectivities can be recognized in the graphs as dark areas; regions with high reflectivities are bright areas. First, the graphs for s polarization [Figs. 2a, 2b] shall be discussed. Regions with low reflectivities are shown as stripes at certain wavenumbers in the respective diagrams. These stripes are separated by $\approx 1100{\mathrm{cm}}^{-1}$ at $d_{\text{Ge}}=1000\mathrm{nm}$, drawing closer to each other at a separation of $\approx 500{\mathrm{cm}}^{-1}$ at a layer thickness of Ge $d_{\text{Ge}}=2500\mathrm{nm}$.

The width of the stripes decreases with increasing interlayer thickness. The wavenumber of the minima of the curves depends only weakly on the angle of incidence, while the exact value of the reflectivity as well as the width of the minima change drastically. For p polarization [Figs. 2c, 2d] at angles of incidence below $70°$, the curves are very similar to those with s polarization. Width and wavenumber of the minima in the curves agree for s and p polarization.

The region around $70°$ is special in p polarization, but not in s polarization. This region is the location of the principal angle (defined as the Brewster angle of a system with high-index incidence medium and low-index exit medium in [6], Chap. 2.B.1) of the system ${\mathrm{CaF}}_2/\text{sample}$. Near the principal angle, the reflectivity changes from almost 0 to almost 1, as also encountered in systems without metal and interlayer. At regions above $80°$, the “stripe pattern” is again seen, with the same spacing as before, but shifted by half of the spacing with respect to the regions below the critical angle. That shift implies that, above the principal angle, the minima in the reflectivity curves for s and p polarization occur at different wavenumbers. Nevertheless, the reflectivity values at the minima are below 0.1 for both polarizations. It must be noted that not every region with low reflectivity is a region where enhanced absorbance from molecules near the metal/sample interface are observed. In particular, the angular regions of low reflectivity around the principal angle in p polarization do not contribute to an enhancement of absorbance.

5. Comparison of Calculated and Measured Absorbance Spectra

The exact value of the minimum reflectivity values also crucially depends on the medium in contact with Au, i.e., the sample. In the procedure for absorbance measurements using Eq. (1), differences in the reflectivity at the minima will consequently lead to peaks in the spectra not directly associated with the absorption of light in the sample. For the system “${\mathrm{CaF}}_2$–Ge–Au–acetonitrile” with $d_{\text{Ge}}=910\mathrm{nm}$ and $d_{\text{Au}}=20\mathrm{nm}$, the dependence on the angle of incidence of these “interference fringe” features in the spectra has been studied. In this work, results presented will be limited to the region between 2000 and $6000{\mathrm{cm}}^{-1}$, where, with few characteristic exceptions, only relatively weak absorptions from vibrational modes can be observed.

Figure 3 shows calculated [Figs. 3a, 3c] and measured [Figs. 3b, 3d] spectra with both s [Figs. 3a, 3b] and p polarization [Figs. 3c, 3d]. The measured spectra show only weak absorption peaks from atmospheric contribu tions, mainly from uncompensated water vapor ($\approx 3500{\mathrm{cm}}^{-1}$, carbon dioxide ($\approx 2300{\mathrm{cm}}^{-1}$), and acetonitrile vapor ($\approx 3050{\mathrm{cm}}^{-1}$). Here, the main focus is on the overall shape of the curves, caused by the interference in the interlayers.

Because of interference in the interlayer, the spectra show maxima in s polarization [Figs. 3a, 3b] and maxima as well as minima in p polarization [Figs. 3c, 3d]. In all cases, the wavenumbers of these extrema agree quite well between prediction and measurement. At wavenumbers above $5500{\mathrm{cm}}^{-1}$, there is considerable disagreement in the shape of the curves, because this region is close to the bandgap of Ge, and the bulk optical constants used for the computations do not capture all features encountered in the electronic structure of a thin Ge film near the bandgap. Nevertheless, for p polarization, even the height of the interference fringe peaks matches nicely with the experiments below $5500{\mathrm{cm}}^{-1}$. For s polarization, the peaks in the computation are about a factor of 2 higher than in the experiment. This behavior could be due to inhomogeneities in the Ge layer. Also, the increase in peak height with increasing angle of incidence seen in the calculations is not clearly reproduced in the experiments.

A further interesting feature is the reversal in the direction of the peak encountered for p polarization from a positive peak at lower incidence angles to a negative peak at higher angles. The angle of this reversal at $\simeq 34°$ is seen both in computations and in experiments.

At higher angles of incidence, the situation becomes more involved. Figure 4 shows examples for the behavior at angles of incidence above $40°$. For s polarization [Figs. 4a, 4b], the peaks are always positive. However, the calculated peaks are always higher than the experimentally observed peaks. Especially between $40°$ and $50°$, the differences are large. At higher angles of incidence, the difference between calculated values and observed values is not as pronounced. Again, in all cases, the wavenumber positions of the extrema agree well with the calculations.

In p polarization [Figs. 4c, 4d], negative peaks of comparable magnitude at $44°$are found in both simulation as well as some experiments. At higher angles of incidence, these peaks become positive. While in experiments, the transition happens at $52°$, the simulations predict it to be at $46°$. Higher angles of incidence show positive peaks in both simulation and experiments. As in the case of s polarization, higher values of absorbance are predicted than are found in experiments, but again, the wavenumbers where the extrema occur are correctly identified from the experiments.

Overall, it can be noted that at low angles and high angles the computations for the system agree with the measurements. Figure 5 gives an overview of the absorbance maximum (for positive peaks) or minimum (for negative peaks) of the feature around $4500{\mathrm{cm}}^{-1}$ as a function of angle of incidence for both s and p polarization. Experimental data sets for two different realizations with the same sample preparation parameters are given to show the limiting cases achieved in the preparation so far. For s polarization [Fig. 5a], the theoretical and experimental curves show differences. In s polarization, all peaks are always positive both in experiment and in theory. A special angle in both theory and experiment is around $46°$. However, in the simulations, the absorbance values reach a maximum there, while they reach a minimum in experiments. At angles of incidence above $55°$, in both theory and experiment, the absorbance decreases with angle of incidence, but the absolute values are still quite different. For p polarization [Fig. 5b] at angles of incidence up to $45°$, the agreement is quite good in one sample, while differences start to appear at lower angles in another sample. Around $50°$, not even the sign of the peaks is correctly predicted, while at angles above $55°$, there is again a better agreement between experiment and theory.

The angle of $70°$ is another special angle in the experiment. In p polarization, an increase of the calculated absorbance of the interference modes is observed. The experimental point of one data set is not visible in Fig. 5b, as it has a value of $\sim 0.7$ and does not fit into the figure at its displayed scale. Computations also suggest a very strong perceptibility of the molecular absorption modes in the spectra, which is similar to experimental observation, but overall, experimental and simulated spectra show disagreement in certain features.

6. Discussion

By changing the thickness of the interlayer, the spectral regions with decreased reflectivity and hence increased sample absorbance can be tuned over the complete mid-IR range. For observation of a certain vibrational mode, an interlayer thickness needs to be chosen that yields enhanced absorbance at the respective wavenumber. Frequently, several vibrational modes are of interest in the study of complex molecules or molecular assemblies. The results here show that, for many possible combinations of modes, interlayer thicknesses exist that allow a concurrent enhancement of several vibrational modes.

Reflection properties of stratified media as used here are to a large extent governed by the incidence and exit medium and modified by the thin films. For the absorbance spectra measured here, it is important to note that one deals with a system with two critical angles of total internal reflection. The critical angle of the ${\mathrm{CaF}}_2/\text{air}$ system, which is relevant for the background measurements, is $\approx 46°$. The critical angle of ${\mathrm{CaF}}_2/\text{acetonitrile}$, relevant in the sample measurement, is $\approx 70°$, depending on the wavenumber.

The reflectivities of the ${\mathrm{CaF}}_2$–Ge–Au–acetonitrile system also show the critical angle as a special angle. In the curves for s polarization at high angles of incidence, between the reflectivity minima that are introduced through the presence of Ge and Au, regions with a reflectivity of 1, i.e., regions of total reflection, can be identified. In p polarization, such large values are reached only near $90°$. At the critical angle, very strong absorption modes of the sample can be observed in both experiment and computations, due to the presence of an evanescent wave with diverging penetration depth at the critical angle. Slightly below the critical angle is the principal angle, which determines the behavior of the reflectivities in p polarization. The location of the critical angles shows that there are three regimes in the spectroscopic experiments performed here. The first is the region of total internal reflection in the absence of the layer structure above $70°$. This region is the domain of attenuated total internal reflection spectroscopy and is discussed for a system based on ZnSe elsewhere [12]. Good semiquantitative agreement between computations and experiments is found for that system. For ${\mathrm{CaF}}_2$, angles above $70°$ are experimentally more difficult to reach, as the total incident intensity in the setup used here decreases at high angles of incidence.

The second regime is the regime where there is no total reflection in either background or reference, which happens at angles below $46°$. Here, for p polarization, quite good agreement can be found, which is almost quantitative for one sample (Fig. 3). For s polarization there are differences (Figs. 5, 3), though not in the overall behavior. Generally speaking, this regime is suitable for analytical applications.

The third regime is the region between the two critical angles, i.e., between $46°$ and $70°$. In this region, there is a considerable difference between sample and reference, leading to the observed disagreement between calculation and experiment. This dis agreement is particularly obvious around $46°$ (Figs. 4, 5). Above $60°$, the experimental trends are correctly reproduced in the computation. Close to the critical angles, deviation between computation and experiment is most likely to be present. Here, small deviations in the angle of incidence, the beam profile, and the degree of polarization are expected to have a huge impact on the results. The angles are correctly identified as special points in the angular dependence curves, both in experiments and theory; see Fig. 5.

Even though in this third regime the agreement between theory and experiments is less good, the sensitivity to the sample, i.e., the increase in “absorbance” in the interference mode with the presence of the sample, is higher. That means that empirical calibrations of the signals, as commonly applied in bioanalytics, can be used to achieve high sensitivity despite the lack of quantitative agreement with expected values.

The differences between different preparations, with two extreme cases shown in Fig. 5, are still problematic. However, the angles at which the sign of the bands change and where minima are observed are still the same, which means that the overall picture is valid, despite the differences in details. The differences may be due to inhomogeneities in the layers, which also play a role in the observed differences between experiments and computations.

7. Conclusions

The introduction of the high-index interlayer leads to the presence of interference bands in the absorbance spectra, if there is a change in refractive index in the exit medium in contact with the metal layer. The wavenumbers of these bands can be predicted, while the exact height shows some disagreement between calculations and experiment. In p polarization (Fig. 3), a peak inversion is noted, which can be predicted. This inversion is a crucial property, as the angle where this inversion occurs can be measured with high accuracy. From the angle, information about the refractive index about the sample in direct contact with the surface can be obtained. For a precise measurement, a cylindrical incidence medium is, however, better suited than the hemispheric one used here [6].

Thus, due to introduction of the interlayer, infrared absorption spectroscopy is enriched by information about the presence of nonabsorbing samples near the surface. Such information typically is obtained by surface plasmon resonance or electro chemical impedance spectroscopy studies. But, in addition to these interference modes, the absorption modes that yield information about the kind of molecules near the surface is present in infrared absorption spectroscopy.

It is worth noting that this additional information from the interference modes will also be there if a nonmetallic coating is used. Here, exclusively the use with metals is discussed, because there is a particular need for the technique for metals.

In this paper, the influences of various parameters were discussed, mainly incident angle and dielectric interlayer thickness on the wavenumber regions with reduced reflectivity that give rise to the interference modes in the absorbance spectra (Fig. 5). The incidence material used here was ${\mathrm{CaF}}_2$, a rather unusual material as medium of incidence in IR reflection spectroscopy, though a very common window material in IR transmission spectroscopy. The rationale for using ${\mathrm{CaF}}_2$ is that it does not show total internal reflection at accessible angles of incidence for common solvents such as water. The presence of the highly reflecting metal layer eliminates the need for total reflection.

The wavenumbers of the minima in the reflectivity curve and, therefore, the positions of the interference modes can be tuned, as is intuitively clear, by the thickness of the high-index dielectric interlayer [15, 16]. As in these interlayer modes, enhanced absorption is also observed [12], and the region for enhanced absorbance can be tailored to the analytical need by the right thickness of the interference coating.

Overall, the system reported here is promising for wider application in interface-sensitive analytical spectroscopy.

This work was funded by the German Science Foundation (Deutsche Forschungsgemeinschaft, DFG) under grant ER-601/1. M. Stratmann is acknowledged for his support.

 

Fig. 1 Schematic view of the setup used in the experiment, including the optical layer system.

Download Full Size | PDF

 

Fig. 2 Computed reflectivity spectra $R^{(s)}$ with s polarization [(a) and (b)] and $R^{(p)}$ with p polarization [(c) and (d)] for different angles of incidence for ${\mathrm{CaF}}_2$–Ge [$d_{\text{Ge}}=1000\mathrm{nm}$, (a) and (c), and $d_{\text{Ge}}=2500\mathrm{nm}$, (b) and (d)]–Au ($d_{\text{Au}}=20\mathrm{nm}$) sample. The reflectivities are shown on a gray scale as indicated on the right-hand side of the plots.

Download Full Size | PDF

 

Fig. 3 Absorbance spectra [s polarization, (a) and (b), and p polarization, (c) and (d)] for ${\mathrm{CaF}}_2$–Ge–Au–acetonitrile (reference: ${\mathrm{CaF}}_2$–Ge–Au–air), $d_{\text{Ge}}\simeq 900\mathrm{nm}$, $d_{\text{Au}}=20\mathrm{nm}$. Computed values [(a) and (c)] are compared with measured values [(b) and (d)] for angles of incidence between $30°$ and $40°$.

Download Full Size | PDF

 

Fig. 4 Absorbance spectra [s polarization, (a) and (b), and p polarization, (c) and (d)] for ${\mathrm{CaF}}_2$–Ge–Au–acetonitrile (reference: ${\mathrm{CaF}}_2$–Ge–Au–air), $d_{\text{Ge}}\simeq 900\mathrm{nm}$, $d_{\text{Au}}=20\mathrm{nm}$. Computed values [(a) and (c)] are compared with measured values [(b) and (d)] for angles of incidence from $44°$ to $64°$ as indicated in the graph.

Download Full Size | PDF

 

Fig. 5 Comparison of the respective maximum or minimum absorbance value of the peak around $4500{\mathrm{cm}}^{-1}$ between theoretical and experimental values from two different layer systems for (a) s polarization and (b) p polarization. Error bars indicate the errors for several measurements of the same layer preparation.

Download Full Size | PDF

1. P. Yeh, Optical Waves in Layered Media, 2nd ed. (Wiley, 2005).

2. P. H. Berning and A. F. Turner, “Induced transmission in absorbing films applied to band pass filter design,” J. Opt. Soc. Am. 47, 230–239 (1957). [CrossRef]  

3. B. V. Landau and P. H. Lissberger, “Theory of induced-transmission filters in terms of the concept of equivalent layers,” J. Opt. Soc. Am. 62, 1258–1264 (1972). [CrossRef]  

4. I. R. Hooper, T. W. Preist, and J. R. Sambles, “Making tunnel barriers (including metals) transparent,” Phys. Rev. Lett. 97, 053902 (2006). [CrossRef]   [PubMed]  

5. See, e.g., V. P. Tolstoy, I. V. Chernyshova, and V. A. Skryshevsky, Handbook of Infrared Spectroscopy of Ultrathin Films (Wiley, 2003). [CrossRef]  

6. N. J. Harrick, Internal Reflection Spectroscopy (Wiley, 1967).

7. E. Goormaghtigh, V. Raussens, and J.-M. Ruysschaert, “Attenuated total reflection infrared spectroscopy of proteins and lipids in biological membranes,” Biochim. Biophys. Acta 1422, 105–185 (1999). [PubMed]  

8. M. Boncheva and H. Vogel, “Formation of stable polypeptide monolayers at interfaces: controlling molecular conformation and orientation,” Biophys. J. 73, 1056–1072 (1997). [CrossRef]   [PubMed]  

9. A. Erbe, R. J. Bushby, S. D. Evans, and L. J. C. Jeuken, “Tethered bilayer lipid membranes studied by simultaneous attenuated total reflectance infrared spectroscopy and electrochemical impedance spectroscopy,” J. Phys. Chem. B 111, 3515–3524 (2007). [CrossRef]   [PubMed]  

10. R. F. Aroca, D. J. Ross, and D. Concepcion, “Surface-enhanced infrared spectroscopy,” Appl. Spectrosc. 58, 324A–338A (2004). [CrossRef]   [PubMed]  

11. K. Ataka and J. Heberle, “Biochemical applications of surface-enhanced infrared absorption spectroscopy,” Anal. Bioanal. Chem. 388, 47–54 (2007). [CrossRef]   [PubMed]  

12. M. Reithmeier and A. Erbe, “Dielectric interlayers increasing the transparency of metal films for mid-infrared attenuated total reflection spectroscopy,” Phys. Chem. Chem. Phys. 12, 14798–14803 (2010). [CrossRef]   [PubMed]  

13. S. A. Ramakrishna, “Physics of negative refractive index materials,” Rep. Prog. Phys. 68, 449–521 (2005). [CrossRef]  

14. A. Erbe, “Reflcalc,” http://home.arcor.de/aerbe/en/prog/a/reflcalc.html.

15. J. Lekner, Theory of Reflection, 1st ed. (Kluwer, 1987).

16. M. Schubert, “Polarization-dependent optical parameters of arbitrarily anisotropic homogeneous layered systems,” Phys. Rev. B 53, 4265–4274 (1996). [CrossRef]  

17. E. D. Palik, ed., Handbook of Optical Constants of Solids (Academic, 1998), Vols. I–III.

18. J. E. Bertie and Z. Lan, “Liquid water-acetonitrile mixtures at $25°\mathrm{C}$: the hydrogen-bonded structure studied through infrared absolute integrated absorption intensities,” J. Phys. Chem. B 101, 4111–4119 (1997). [CrossRef]