Infrared Reflection Absorption Spectroscopy (IRAS) on dielectric single crystals is challenging because the optimal incidence angles for light–adsorbate interaction coincide with regions of low IR reflectivity. Here, we introduce an optimized IRAS setup that maximizes the signal-to-noise ratio for non-metals. This is achieved by maximizing light throughput and by selecting optimal incidence angles that directly impact the peak heights in the spectra. The setup uses a commercial Fourier transform infrared spectrometer and is usable in ultra-high vacuum (UHV). Specifically, the optical design features sample illumination and collection mirrors with a high numerical aperture inside the UHV system and adjustable apertures to select the incidence angle range on the sample. This is important for p-polarized measurements on dielectrics because the peaks in the spectra reverse the direction at the Brewster angle (band inversion). The system components are connected precisely via a single flange, ensuring long-term stability. We studied the signal-to-noise ratio (SNR) variation in p-polarized IRAS spectra for one monolayer of CO on TiO2(110) as a function of incidence angle range, where a maximum SNR of 70 was achieved at 4 cm−1 resolution in a measurement time of 5 min. The capabilities for s polarization are demonstrated by measuring one monolayer D2O adsorbed on a TiO2(110) surface, where a SNR of 65 was achieved at a peak height ΔR/R0 of 1.4 × 10−4 in 20 min.

Infrared vibrational spectroscopy is a versatile and widely used technique to identify molecular species via their characteristic vibrational frequencies. This technique is particularly useful in catalysis research, as it can identify surface-bound reaction intermediates under operando conditions.1,2 While much work is performed on powder-based catalysts, their structural complexity makes it difficult to link the adsorbed species to a particular active site. For this reason, many groups utilize the so-called surface-science approach, where experiments are performed on well-defined single crystal samples in a tightly controlled ultra-high vacuum (UHV) environment.3 In such experiments, IR light is reflected once off the sample surface, and the spectrum is obtained with a Fourier transform infrared (FTIR) spectrometer.4,5 The method is called Infrared Reflection Absorption Spectroscopy (IRAS or IRRAS), Reflection Absorption IR Spectroscopy (RAIS or RAIRS), or External Reflection Spectroscopy (ERS). Initially, IRAS was developed for metal surfaces,6–8 where surface selection rules dictate that only vibrations with dipole changes perpendicular to the metal surface are infrared active.9–11 On metallic substrates, the formation of an image dipole enhances the interaction of the dipole moment with the electric field. This has a maximum at grazing incidence angles of the light12 and allows low coverages of adsorbates to be routinely detected. On dielectrics, however, IRAS experiments are less sensitive,13–15 partly because a weaker surface electric field interacts with the dipole and partly because the maximum of the electric field at the surface occurs at incidence angles where the infrared reflectivity is comparatively poor.9,12 An elegant way to circumvent this issue is to grow thin films of the dielectric in question on a metal support, and significant progress has been made using this approach.16–21 Often, this is not feasible, or the thin films are affected by the proximity of the metallic layer, both structurally and electronically, necessitating the study of bulk single crystals. Experiments performed on such samples have the advantage that dipoles parallel to the surface can be studied using s-polarized light.22 However, low signals (typically one to two orders of magnitude lower than on metals4,23) necessitate long measurement times,24 particularly for sub-monolayer adsorbate coverages.13–15 

In this paper, we describe an IRAS setup uniquely designed for studying low coverages of adsorbates on dielectric single crystals. While most current IRAS measurement systems employ fixed grazing incidence angles of 80°–85°, several studies25–28 demonstrated that altering the incidence angle of the IR light can enhance the IRAS peak heights and the signal-to-noise ratio (SNR). Our setup provides an adjustable incidence angle range, allowing us to maximize the peak heights obtained for each material we study. We also optimized the optical throughput, which effectively minimizes noise. Hence, our system achieves an SNR of ∼70 within minutes of measurement time for monolayer coverage.

This paper is organized as follows: The first section describes the underlying theory of IR reflectivity and the optical design. Next, we describe the mechanical design, focusing on its integration with an existing surface chemistry UHV chamber29 that utilizes a custom-built molecular beam30 for temperature-programmed desorption (TPD) experiments. This molecular beam, particularly its precise beam spot, imposes design constraints for the controlled illumination of the adsorbate-covered area on the sample. Finally, we evaluate our IRAS setup by measuring CO and D2O molecules adsorbed on a TiO2(110) surface, thus demonstrating the system capabilities.

IRAS measurements on metal single crystals are sensitive to adsorbate vibrations with the dipole moment perpendicular to the surface due to the dipole enhancement effect, a high vertical surface electric field at grazing incidence angles, and the near-unity reflectivity of both polarizations in the mid-IR regime. The adsorbate–light interaction of dipole moments oriented parallel to the surface is suppressed due to the absence of a horizontal surface electric field; this is also known as the surface selection rule.9–11 

For metal oxides and other dielectrics, the dipole enhancement effect is small or negligible. On these materials, the surface electric field is small compared to metals. The adsorbate–light interaction optimum is at lower incidence angles than on metals,9,12 where the reflectivity of the substrate is low (Fig. 1). By measuring s and p polarization, adsorbate vibrations with all orientations of the dipole moment can be detected.

FIG. 1.

Calculated angle-dependent Fresnel reflection on TiO2. S-polarized (blue) and p-polarized reflection (red) at ν̃= 2178 cm−1 for a TiO2 substrate with a complex refractive index31,32 n̂=2.27+0.002i. The dashed line indicates the Brewster angle θB.

FIG. 1.

Calculated angle-dependent Fresnel reflection on TiO2. S-polarized (blue) and p-polarized reflection (red) at ν̃= 2178 cm−1 for a TiO2 substrate with a complex refractive index31,32 n̂=2.27+0.002i. The dashed line indicates the Brewster angle θB.

Close modal

The incidence angle-dependent reflectivity R0(θ) of s- and p-polarized light on a flat surface can be calculated using the Fresnel equations and the complex refractive index n̂=n+ik. It is a material-specific quantity, where n is the real refractive index and k is the extinction coefficient. In Fig. 1, we show the results of the Fresnel equations for rutile TiO2 at a wavenumber ν̃=2178 cm−1 (this is a typical value for the CO stretch; see chapter V.B.1). It should be noted that TiO2 is birefringent,33 but in the following, this aspect is neglected in the calculations. For p-polarized light, the reflectivity is zero at the Brewster angle θB.

IRAS measures differences in the reflectivity of the adsorbate-covered surface (R) with respect to the pristine surface (R0). In Secs. III–V, we will refer to the normalized reflectivity difference defined as
ΔRR0=RR0R0,
(1)
as a function of the wavenumber ν̃ and the incidence angle θ. Langreth34 presented a way to calculate the incidence angle-dependent normalized reflectivity difference ΔR(θ,ν̃)/R0(θ,ν̃) for an adsorbate based on the surface polarizability. The adaptation of these formulas and expansion by the polarizability modeled with a Lorentzian oscillator according to Eq. (8) in Tobin’s work35 forms the basis for our calculation of the normalized reflectivity difference defined in Eq. (1) [see Eqs. (1)–(6) in the supplementary material].

For the calculations in this work, the static electronic polarizability of the adsorbate was taken as αe = 0 and the vibrational polarizability as αv = 3 × 10−26 cm3. The density of adsorbates was assumed to be N = 5.2 × 1014 cm−2, with a linewidth γ corresponding to 5 cm−1. We consider the normalized reflectivity difference at the resonance frequency (peak height), ν̃0=ν̃= 2178 cm−1. For the substrate, we assume the complex refractive index of TiO231,32 with with n̂=2.27+0.002i.

Figure 2 shows the calculated normalized reflectivity difference ΔR/R0 caused by the adsorbate as a function of the incidence angle θ under these assumptions. A dipole moment parallel to the surface but perpendicular to the incidence plane can be detected only with s polarization. For this case (green, dotted line), ΔR/R0 remains positive at all incidence angles; in other words, the adsorbate causes an increase in the reflected intensity at the resonance. With increasing incidence angle θ, ΔR/R0 decreases. In contrast, ΔR/R0 for p-polarized light shows a 1/x-like singularity at the Brewster angle θB for an ideal dielectric (extinction coefficient k → 0). This is valid for both horizontally and vertically oriented dipoles in the plane of incidence. The singularity arises because ΔR has a zero of order one and R0 has a zero of order two at the Brewster angle. The change of sign at the Brewster angle θB is known as band inversion.27,36–38

FIG. 2.

Calculated normalized reflectivity difference (peak height) ΔR/R0 of an adsorbate on a TiO2 surface. Three configurations of the surface adsorbate and the incoming light polarization are shown: horizontal dipole moment perpendicular to the plane of incidence and s-polarized light (green, dotted), horizontal dipole in the plane of incidence (blue, dashed), and vertical dipole moment (red); the latter two are detectable with p polarization. Due to the low signal, the s-polarized data were multiplied by 20. θB denotes the Brewster angle.

FIG. 2.

Calculated normalized reflectivity difference (peak height) ΔR/R0 of an adsorbate on a TiO2 surface. Three configurations of the surface adsorbate and the incoming light polarization are shown: horizontal dipole moment perpendicular to the plane of incidence and s-polarized light (green, dotted), horizontal dipole in the plane of incidence (blue, dashed), and vertical dipole moment (red); the latter two are detectable with p polarization. Due to the low signal, the s-polarized data were multiplied by 20. θB denotes the Brewster angle.

Close modal

In the case of materials with a non-negligible imaginary part k of the refractive index, the normalized reflectivity difference at the resonance frequency ν̃0 is zero close to the Brewster angle, and a Fano-like line shape is observed in the IR spectra instead of a Lorentzian peak.39 Then, ΔR/R0 at the resonance frequency remains finite for all angles but has maxima and minima with a large magnitude near the Brewster angle (not shown in Fig. 2). An asymmetric line shape occurs, and the absorption peak does not coincide with the resonance frequency. For p polarization, this is most obvious for incidence angles around the Brewster angle. Far away from the Brewster angle, the peak becomes more symmetric, and the peak position is close to the resonance frequency (see Fig. S1 of the supplementary material).

For p polarization and a perpendicular dipole moment, positive absorbance (ΔR/R0 < 0, as for adsorbates on metals) occurs only at θ < θB, not at grazing reflection, and ΔR/R0 approaches zero at both ends of the 0°–90° range. On dielectrics, p polarization can also detect a dipole moment oriented parallel to the surface and in the incidence plane. In this case, positive absorbance (ΔR/R0 < 0) occurs at grazing angles; ΔR/R0 again approaches zero at θ → 90°. ΔR/R0 is positive below the Brewster angle. At perpendicular incidence, the difference between s- and p-polarization becomes meaningless; thus, the reflectivity for a parallel dipole moment oriented along the direction of the electric field is the same for s and p polarization at θ = 0. Note that the signal of s-polarized light has been multiplied by 20 in Fig. 2. A comparison of horizontal and vertical dipole orientations for p polarization shows almost an order of magnitude stronger signals for the vertical orientation, except for small incidence angles.

Designing a highly sensitive IRAS setup relies on two fundamental objectives: minimizing noise levels and maximizing the peak heights in the spectrum. Improving both parameters results in a high SNR necessary for measuring low coverages of adsorbates. Sections III A and III B discuss our approaches to fulfill both requirements: low noise levels are primarily gained by maximizing the optical throughput, whereas increased peak heights are obtained by tailoring the light incidence angle on the sample.

Our IRAS setup is designed specifically to detect adsorbates with low coverages on dielectrics. To improve the SNR, one could conduct measurements with more scans, which extends the measurement time. However, further adsorption/desorption on the surfaces can happen during the measurement, or the background of the spectra can change, e.g., due to thermal drift. We have implemented several strategies to reduce noise while keeping the measurement time at a minimum. These strategies include optimizing the beam path and carefully selecting optical components leading to high intensities on the detector to improve the SNR. Furthermore, ensuring the mechanical stability of the system reduces drift caused by the movement of the focal spot on the detector or the sample and reduces noise due to environmental vibrations.

The optical throughput of the system defines the amount of light that reaches the detector after being emitted from the source and reflected off the sample surface. The implemented optimization approach can be divided into minimizing losses on the optical components and conservation of étendue.40,41 Étendue is a geometric quantity defined as the product of illuminated surface area A and the solid angle Ω spanned by the rays in the light beam at a focus on a surface perpendicular to the principal ray. It can be interpreted as the volume in phase space of the light beam. With a given diameter of an initially almost parallel beam, Ω is inversely proportional to the focal length. Thus, a small illuminated area A (restricted by the size of the crystal or, in our case, by the adsorbate-covered area) must be accompanied by a large solid angle, which requires a short focal length for illumination. For maximum throughput, the beam path should be designed without any obstructions. Ideally, it should capture the full area and as large as possible a solid angle of the light emitted from the IR source. All the collected light should be focused on the sample and subsequently onto the detector. In our case, the commercial spectrometer sets the boundary conditions for the source (area and light collection), while the existing UHV setup determines the desired size of the illuminated surface area, which should be the adsorbate-covered sample area (3.5 mm molecular beam spot diameter29). The layout of the UHV system also imposes constraints on the mirror arrangement. Étendue was then used to estimate the position and shape of optical components to optimize the illumination on the sample and find a good detector position.

The focal length of the illumination and collection mirror must be short enough to achieve the large solid angle required for a small illuminated spot. These considerations provide the basis for the final optimization performed by standard numerical ray tracing, which resulted in a short focal length of the illumination mirror of only a few centimeters, leading to a mirror arrangement with two mirrors in the UHV chamber. Placing mirrors outside of the UHV chamber would require a much larger focal length for the illumination mirror, accompanied by either unpractically large diameters of the mirrors and windows for the IR light or a loss of throughput by limiting the solid angle. The short focal length of the sample-facing mirrors is one of the key features characterizing our design; other systems usually use longer focal lengths.22,42–46 In addition, low-loss light transfer from the spectrometer is achieved using elliptical mirrors to overcome the distance of almost 1 m between the spectrometer and the sample. Furthermore, the ray tracing optimization took manufacturing tolerances into account, to avoid a design highly sensitive to small deviations from the nominal geometry. These optimization elements are essential in guaranteeing a high-performance IRAS system.

A schematic diagram of the IRAS system is shown in Fig. 3. The illumination path features three mirrors directing the IR light from the spectrometer exit to the sample focus. After the almost parallel beam exits the spectrometer (Bruker VERTEX 80v) through a wedged BaF2 window (3.5 mm thick), the light is focused into the polarizer focus by an off-axis parabolic mirror (“spectrometer-link mirror” in Fig. 3). A rotatable holographic wire-grid polarizer is placed after the focus point. The input mirror, an off-axis elliptical mirror, transfers the light through the shaping focus, a BaF2 window (3 mm thick), and two angle-selection plates to the illumination mirror. This off-axis elliptical mirror features a high numerical aperture, illuminating the sample with a cone angle of 39° due to its relatively short focal length. As a result, the illumination mirror creates a small, nearly circular illumination spot when the IR beam is directed at the sample at normal incidence.

FIG. 3.

Schematic top view of the IRAS system optics. The main optical components and focus points along the illumination (full red) and collection beam paths (red outline) are shown. Elliptical mirrors are filled in gray, and parabolic mirrors are black. BaF2 windows separate chambers with different pressure levels. The components are not shown to scale and are simplified to ensure clarity.

FIG. 3.

Schematic top view of the IRAS system optics. The main optical components and focus points along the illumination (full red) and collection beam paths (red outline) are shown. Elliptical mirrors are filled in gray, and parabolic mirrors are black. BaF2 windows separate chambers with different pressure levels. The components are not shown to scale and are simplified to ensure clarity.

Close modal

As mentioned above, a small illumination spot on the sample is essential to reflect all light from the sample surface and keep most of the intensity focused on the molecular beam spot. The presence of a focus (shaping focus) between the input mirror and the illumination mirror facilitates low-loss light transfer over a distance of almost 1 m. An aperture called the illumination-shaping slit is placed directly in the shaping focus; its task is to limit the illuminated area on the surface, especially at rather grazing incidence angles (see Sec. III B 2). In addition, the two angle-selection plates located between the shaping focus and the illumination mirror control the illumination angle range onto the sample. The specific function of the illumination-shaping slit and the angle-selection plates will be explained in Sec. III B.

The ideal sample illumination spot size is primarily defined by the adsorbate-covered area on the sample. Independent of the single crystal’s surface area, the infrared beam should be restricted to the adsorbate-covered area created by the molecular beam, i.e., a circular region with a diameter of 3.5 mm in the center of the sample surface. Consequently, the measurement should include only radiation reflected from this region. IR radiation reflected from the uncovered sample regions contributes only to the noise, not to the signal. This consideration is also valid for a grazing incidence of radiation. Since the principal ray of the IR beam is not perpendicular to the sample surface, the roughly circular focus gets distorted into an elliptical shape. The major axis of the ellipse is oriented along the incidence plane, and the minor axis remains unchanged and equals the diameter of the focal spot. Hence, for very grazing incidence, the spot size is larger than the sample and adsorbate area, resulting in decreased sensitivity.

The IR light reflected from the sample is collected in the beam path drawn with a red outline in Fig. 3. This path features two elliptical mirrors, one wedged BaF2 window in between, and an IR detector. The defining aspects of this part of the IRAS system are the complete collection of specularly reflected light and subsequent transfer to the detector with minimal losses. To collect all the light reflected from the sample, the collector mirror features a slightly shorter front focal length than the back focal length of the illumination mirror. This makes the setup insensitive to angular errors originating from the sample tilt. Since the sensitive detector element (1 × 1 mm2) is smaller than the illuminated area on the sample, the conservation of étendue requires a very large solid angle of the light entering the detector. The focal lengths of the collector mirror and the detector mirror are balanced to fit the detector light acceptance angle and the element size of the detector within the spatial restrictions of the measurement chamber to ensure maximum throughput. In addition, the intermediate focal point between the collector mirror and detector mirror minimizes losses compared to using a parallel beam, overcoming a distance of ∼0.6 m between the sample and detector.

Besides the conservation of étendue, the optical throughput is also determined by the transmissive and reflective losses of the optical components. Therefore, the correct selection of materials for optical components for the targeted IR wavenumber range and reducing the optical surfaces in the IRAS system is vital. To minimize reflective losses, all four off-axis elliptical mirrors (gray in Fig. 3) are manufactured from EN AW-6061-T6 aluminum, which has a reflectivity close to unity in the mid-IR region. These mirrors were fabricated according to custom specifications (for the focal lengths, see Table SI of the supplementary material), and the reflective surface is covered only by the native oxide layer of aluminum. The spectrometer-link mirror (black in Fig. 3) is a commercial gold-plated, off-axis parabolic mirror. Three BaF2 windows are used to separate the vacuum chambers (gray) with different pressure regimes: FTIR spectrometer (∼1 mbar), high-vacuum box (∼10−3 mbar), and the UHV surface chemistry chamber (∼5 × 10−11 mbar), where the sample is located. The BaF2 window thickness has to be chosen as a compromise between mechanical stability and optical losses near the distinct cutoff in the transmission of BaF2 at ∼1000 cm−1. All windows feature a half-degree wedge angle to avoid interference peaks. In addition, the linear holographic polarizer uses BaF2 as a substrate material. Our system utilizes a highly sensitive liquid-nitrogen-cooled mercury cadmium telluride (MCT) detector from InfraRed Associates with a 1 × 1 mm2 detector element and a field of view of 60°. This detector features a BaF2 window, a spectral range of 850–12 000 cm−1, a specific detectivity bigger than 4 × 1010 cm Hz1/2 W−1, and a liquid nitrogen dewar lasting 12 h.

In addition to maximizing throughput, optimizing the peak heights was a central goal of our design strategy. Careful balancing of these measures is crucial, as optimizing peak heights at the cost of decreased intensity can improve the SNR. Maximizing peak heights requires restricting the illumination to the adsorbate-covered area, clean polarization of the IR beam, and a precise adjustment of the incidence angle range by the angle-selection plates (see Fig. 4). The importance of selecting the incidence angles is described in Sec. II and is an innovation in the surface science community.

FIG. 4.

Schematic function of the angle-selection plates (ASP) in three main measurement configurations. The plates are located before the illumination mirror, reducing the incidence angle range. Without using the ASP, the incidence angles are between θmin = 48° and θmax = 87°. (a) Setting for the full angle range, (b) selection of grazing incidence, and (c) selection of non-grazing incidence angles.

FIG. 4.

Schematic function of the angle-selection plates (ASP) in three main measurement configurations. The plates are located before the illumination mirror, reducing the incidence angle range. Without using the ASP, the incidence angles are between θmin = 48° and θmax = 87°. (a) Setting for the full angle range, (b) selection of grazing incidence, and (c) selection of non-grazing incidence angles.

Close modal

1. Effect of the incidence angle θ

The choice of the incidence angle range depends on the polarization, the orientation of the dipole moment of the adsorbate, and the Brewster angle of the investigated dielectric.

The IRAS system described here offers a wide range of incidence angles from 48° to 87°. We have implemented angle-selection plates that can limit the range of incidence angles at either side (see Figs. 3 and 4). The main application of the angle-selection plates is for p-polarized light, where the opposite sign of the peaks above and below the Brewster angle θB would lead to a cancellation of the total signal (see Fig. 2); this can be avoided by selecting only angles at one side of θB. Since θB depends on the refractive index of the substrate in use, our setup allows variation of the cutoff angle θL to optimize the IR measurement for different samples. Depending on the measurement system, the variation of θmin and θmax additionally to θL can lead to SNR improvements. For s polarization, the complete throughput of the system can be utilized without restricting the incidence angle range.

In our IRAS setup, the angle-selection plates are placed in the UHV chamber before the illumination mirror (see Fig. 3). The two plates can be moved independently, which makes it possible to select both the minimum and maximum incidence angles. In most cases, only one of the plates will be used at a time. Section V B provides measured data for the non-grazing, grazing, and full angle range.

2. Effects on peak height in the real system

A realistic view of an IRAS system must take into account that, for example, imperfect optical components or a finite source will alter the IR beam characteristics compared to the ideal case. Effects such as the degree of light polarization,47 the incidence angle spread,26,38 or the light reflected from the adsorbate-free area affect the maximum achievable peak height.

The degree of light polarization is limited by the polarizer itself and polarization aberrations caused by rays that are not in the plane of incidence of the principal ray and reflected from the mirror surfaces (skew aberration).41,47,48 The depolarization effect on the peak heights can be estimated from a weighted sum of the reflectivities of the sample47 for the two polarization directions, p and s polarization. Depending on the degree of depolarization, signals from p polarization may leak into the s polarization when the incidence angle range is sufficiently far from the Brewster angle. Usually, the opposite effect, leakage of s-polarized light into measurements with p polarization, is more relevant: The zero of reflectivity at the Brewster angle will disappear when a small s-component is present, and thus, the singularity in Fig. 2 will disappear. A distribution of incidence angles26,38 will also smear the singularity.

Because our setup features a small adsorbate-covered area, the magnitude of the ΔR/R0 ratio49 will also be reduced by rays reflected from uncovered areas. We deposit molecules with a molecular beam,29 creating a 3.5 mm circular adsorbate-covered region in the sample center, the molecular beam spot. Figure 5(a) shows a photograph of the sample holder, including a mounted sample. The dark circular area is the molecular beam spot. At grazing incidence, the illumination spot (and the collected light at the detector) includes regions outside the molecular beam spot (vicinity), as shown by the ray tracing simulation in Fig. 5(b). For a given angle of incidence, the reflected light from the vicinity reduces the relative peak height. To reduce the intensity of the vicinity illumination, we implemented an illumination-shaping slit in our system (see Fig. 3). It consists of two fixed aperture blades that modify the illumination on the sample to almost rectangular (see Sec. V A). For incidence angles below 65°, also almost all light at the detector that comes from within the measurement area is concentrated at the MB spot. At more grazing angles (>65°), the vicinity is illuminated; up to ≈48% of the detected radiation stems from the adsorbate-free region. In case all the sample area is covered by the adsorbate (background dosing), a large illuminated area poses no limitation, and one may also omit the illumination-shaping slit shown in Fig. 3.

FIG. 5.

Molecular beam (MB) spot on the sample. (a) A photograph of the sample holder, including the sample (6 × 6 mm2 surface area). (b) Simulated intensity at the detector, with the contributions reflected from the different areas of the sample surface, as a function of the incidence angle on the sample surface. The red curve represents the total intensity on the detector reflected from the sample. The blue dotted curve shows the intensity reflected inside the molecular beam spot. The green dotted curve represents the intensity reflected from the vicinity (region not covered by adsorbates) around the molecular beam spot.

FIG. 5.

Molecular beam (MB) spot on the sample. (a) A photograph of the sample holder, including the sample (6 × 6 mm2 surface area). (b) Simulated intensity at the detector, with the contributions reflected from the different areas of the sample surface, as a function of the incidence angle on the sample surface. The red curve represents the total intensity on the detector reflected from the sample. The blue dotted curve shows the intensity reflected inside the molecular beam spot. The green dotted curve represents the intensity reflected from the vicinity (region not covered by adsorbates) around the molecular beam spot.

Close modal

Figure 6 displays the calculated p-polarized ΔR/R0 for TiO2 at ν̃= 2178 cm−1 as a function of the incidence angle θ on the sample, including the effects of incidence angle spread, incomplete polarization, and a partly adsorbate-free sample. The calculation was performed for a model adsorbate with a dipole moment perpendicular to the surface, and the collection efficiency was taken into account. Curve (4) includes a ±2° angular spread in the angle of incidence, curve (3) shows the effect of the finite efficiency of the polarizer (99.5%), and curve (2) demonstrates the effect of the light that gets reflected from the adsorbate-free area of the sample in our system. The effects of the angular spread perpendicular to the angle of incidence and depolarization by the mirrors (skew aberration) are not included in this figure. The strongest effect near the Brewster angle comes from the 0.5% leakage of s-polarized light, whereas the deterioration of the ΔR/R0 ratio at grazing incidence is dominated by light reflected outside the adsorbate-covered area. This is expected since it becomes increasingly difficult to focus all light into the adsorbate-covered area at very grazing angles. The same considerations are also valid for p polarization and in-plane dipole moments. For IRAS measurements with s polarization, the leakage of p-polarized light is irrelevant in the region around the Brewster angle. At grazing angles, the relative peak height will be reduced by reflection from outside the adsorbate-covered area for p and s polarization [see Fig. 5(b)].

FIG. 6.

Different effects influencing the calculated normalized reflectivity difference ΔR/R0 for a model adsorbate with vertical dipole moment on TiO2.49 The calculation was performed for p polarization. Curve (1) shows the uncorrected case, and curves (2)–(4) depict the separate effects of (2) illumination of sample areas not covered by the adsorbate, (3) a polarization degree of 99.5%, and (4) a ±2° spread of the incidence angle. Curve (5) shows a calculation including all effects (2)–(4).

FIG. 6.

Different effects influencing the calculated normalized reflectivity difference ΔR/R0 for a model adsorbate with vertical dipole moment on TiO2.49 The calculation was performed for p polarization. Curve (1) shows the uncorrected case, and curves (2)–(4) depict the separate effects of (2) illumination of sample areas not covered by the adsorbate, (3) a polarization degree of 99.5%, and (4) a ±2° spread of the incidence angle. Curve (5) shows a calculation including all effects (2)–(4).

Close modal

Due to the higher reflectivity of the sample for s-polarized light, compared to p-polarized light, especially around the Brewster angle (see Fig. 1), it is primarily p-polarized spectra that suffer from the depolarization effect. The corrected curves for p polarization and s polarization with parallel dipole moment orientations can be seen in Fig. S2 of the supplementary material. For p polarization, depolarization and IR beam divergence significantly decrease the normalized reflectivity difference around θB, resulting in lower peak heights in the final spectrum.

In addition to optimization of the optical design, it is also essential to achieve mechanical stability of the setup. This reduces the sensitivity to environmental vibrations (e.g., vacuum pumps) and is also required to maintain the alignment of the optics and ensure precise and reproducible sample positioning. Factors such as temperature stability of the laboratory, sample vibrations, or tilt errors50 of the sample mounted on a cryostat influence the spectral stability, i.e., equal baselines for consecutive measurements. The setup is designed to fit an already existing UHV measurement chamber29,51 and takes these considerations into account. It is characterized by the precise positioning of the optical components, minimization of adjustable components (kinematic mounts), and a highly stable optical platform. All mirror holders feature precise pinned fittings and spring-loaded screws to keep them securely in place.

Figure 7 shows a partial section view of the IRAS setup with its primary components. The IRAS setup connects a commercial FTIR spectrometer (Bruker VERTEX 80v; exit window on the left side of Fig. 7) and a UHV surface chemistry chamber29,51 (black chamber on the right side). The detection platform with the transfer optics, polarizer, and detector are inside the high-vacuum box, which is pumped to a pressure of 1.5 × 10−3 mbar. The high-vacuum box is machined from EN AW-6061-T6 aluminum; two covers with Viton gaskets seal its top and bottom. The top cover includes a liquid nitrogen filling port for the IR MCT detector, and the bottom cover has feedthroughs for electrical connections. The right side of the HV box is connected with a DN 200 ISO-K tube to the central flange of the IRAS system.

FIG. 7.

Partial section view of the key components of the IRAS setup. The IR beam path is visualized in red. The arrows in the beam indicate the direction of the IR light. The UHV surface chemistry chamber is shown in black. The detection platform is in the high-vacuum box. The sample-focus platform and the angle-selection plates are inside the UHV chamber. The sample is mounted on the bottom of a helium flow cryostat and located in the IRAS measurement position, but it is hidden behind the sample-focus platform.

FIG. 7.

Partial section view of the key components of the IRAS setup. The IR beam path is visualized in red. The arrows in the beam indicate the direction of the IR light. The UHV surface chemistry chamber is shown in black. The detection platform is in the high-vacuum box. The sample-focus platform and the angle-selection plates are inside the UHV chamber. The sample is mounted on the bottom of a helium flow cryostat and located in the IRAS measurement position, but it is hidden behind the sample-focus platform.

Close modal

All optical components in the high-vacuum box and the UHV chamber are supported by this central flange, which provides precise positioning. The stability of the system depends sensitively on this link. Milled corner joints are used to ensure accurate and reproducible connections between the central flange of the IR setup and rigid customized stainless steel supports that hold the detection platform in the high-vacuum box and the sample-focus platform on the UHV side. The central flange includes two wedged BaF2 windows for the entry and exit of the IR beam into the UHV chamber. These are each fixed on a custom CF-40 flange to facilitate exchange if necessary. Thermocouple feedthroughs are mounted on the flange for temperature monitoring during bakeout.

The optical components outside the FTIR spectrometer are located on the two platforms. The detection platform on the high-vacuum side has two 2-axis motorized kinematic mounts (New Focus 8822-AC-UHV) for IR beam alignment on the illumination side (adjustable spectrometer-link mirror) and detector side (adjustable detector mirror). The input mirror is attached rigidly to the detection platform. The polarizer is mounted on a motorized rotation stage (SR-5714, SmarAct GmbH). With a precision of 0.1° or better, the rotation angle uncertainty of the stage does not contribute to the depolarization of the beam.

The MCT detector, the red cylinder in Fig. 7, is located on a homemade mount and is manually adjustable in the x, y, and z directions. It allows for slight tilting adjustments around all axes. After the initial alignment, the position of the detector is determined by a milled corner joint, ensuring reproducible remounting. The detector electronics supplied by Bruker was removed from the bottom of the detector and placed outside the high-vacuum box, with the wiring (coaxial cable) passing via a current feedthrough to the detector. The detection platform itself is manufactured from aluminum EN AW-6061-T6 and mounted onto the stainless steel support using a milled corner joint and a spring-loaded screw connection.

The two sample-side mirrors in UHV are rigidly mounted with pins and spring-loaded screws on the sample-focus platform. A hole in the center of the sample-focus platform allows the sample and the cryostat to pass to the measurement position. It provides room for adjustments in the x-, y-, and z-directions and rotation around the z-axis through a sample manipulator. The hole diameter (60 mm) is small enough to avoid the deposition of sputter debris on the mirrors when the sample is in the sputtering position (200 mm above the IRAS and molecular beam position). Cu shields (not shown in Fig. 7) protect the mirrors and the IR windows from material deposited with the electron-beam evaporators below the IRAS components in the UHV system.

The angle-selection plates are located in UHV on the side of the illumination beam. The rotation mechanism of each angle-selection plate is operated by a linear motion feedthrough (angle-selection manipulators in Fig. 7). The position of each plate can be read from a scale engraved in the rotation mechanism; these values directly translate into the cutoff angle for the incident beam.

A sample-position finder (see Fig. 8) mounted on the lower part of the sample-focus platform (visible in Fig. 7 behind the IR beam) enables the user to see the sample from the bottom, overlaid with a scale grid. This facilitates quick and reproducible sample positioning.

FIG. 8.

Function of the sample-position finder. (a) Two schematic optical paths, (1) and (2), are separated by the 50:50 beam splitter (BS) from Thorlabs (BSW04). Both paths have the same optical path length. Path (1) reflects on the BS to the sample. Path (2) is guided to the scale grid through the BS and two 90° fused silica. (b) The resulting image that is visible to the user. The sample bottom and the scale grid with a 1 mm grid spacing can be seen simultaneously.

FIG. 8.

Function of the sample-position finder. (a) Two schematic optical paths, (1) and (2), are separated by the 50:50 beam splitter (BS) from Thorlabs (BSW04). Both paths have the same optical path length. Path (1) reflects on the BS to the sample. Path (2) is guided to the scale grid through the BS and two 90° fused silica. (b) The resulting image that is visible to the user. The sample bottom and the scale grid with a 1 mm grid spacing can be seen simultaneously.

Close modal

During the bakeout of the UHV chamber, the IRAS system remains connected to the chamber. To avoid overheating of the components, the stainless steel beam holding the detection platform is connected via copper braids to an aluminum flange (on the bottom cover), where a water-cooling unit is attached. Therefore, no realignment is necessary after the bakeout. If optical realignment is required, the motorized kinematic mounts facilitate alignment without venting the high-vacuum box. These measures result in a mechanically highly stable system that achieves sensitive measurements with a flat baseline over long periods and benefit user-friendliness.

An initial performance characterization of our IRAS system was first carried out outside our UHV surface analysis chamber. The intensity distribution was measured with a camera at critical focal points of the system and compared to simulation data. Next, we tested the performance of the setup in UHV in comparison with IRAS results published in the literature.13,15,22,24,52,53 The FTIR spectrometer Bruker VERTEX 80v with the standard mid-IR source (glow bar) was used for all the characterizations.

To measure the intensity distribution in the critical focal points, the spectrometer was connected to a home-built optical table, with the IRAS setup mounted and exposed to atmospheric conditions. The beam-defining aperture (J-stop) inside the spectrometer was 6 mm during these measurements. The moving interferometer mirror of the FTIR spectrometer was stopped (usually, it is in motion during IR measurements), and the interferometer laser was switched off to prevent saturating the camera sensor by the laser beam.

The angle-selection plates were completely open, not obstructing the IR beam. The intensity characterization was performed without the BaF2 windows mounted on the central flange and without the polarizer. Custom-made holders positioned the camera sensor in the sample or detector focal point, with the position according to the simulation model. A flat aluminum mirror (Ø 12.6 mm) replaced the sample for the measurements in the detector position.

To measure the intensity maps, we utilized the CMOS camera DMM 37UX178-ML from the Imaging Source. The sensitivity of its sensor chip (Sony IMX 178, back-illuminated) extends into the near-infrared region. A thin, framed glass plate covers the sensor, whose pixels exhibit a limited light acceptance angle. Therefore, the intensity maps do not include light hitting the sensor at grazing angles.

Figure 9 shows the measured and simulated intensity distribution at the sample and detector focal points. The two intensity maps in the sample focus [Figs. 9(a) and 9(c)] show good agreement in the central region (red and yellow). The experimental image indicates slightly better focusing than in the simulation (smaller FWHM values). We attribute this to the limited acceptance angle of the camera, which suppresses light at grazing incidence, where the focus area is larger.

FIG. 9.

Simulated (a) and (b) and measured (c) and (d) intensity distributions in two positions of the IRAS setup: The sample position (left) and the position of the detector (right). The white lines in the maps cross the intensity-weighted centroid (“center of mass”) and indicate the position of the line profiles displayed at the periphery of the figure. The black double arrows in the line profile plots indicate the full width at half maximum (FWHM). The red arrow marks an additional faint feature originating from the mirror mount, replacing the sample in these tests.

FIG. 9.

Simulated (a) and (b) and measured (c) and (d) intensity distributions in two positions of the IRAS setup: The sample position (left) and the position of the detector (right). The white lines in the maps cross the intensity-weighted centroid (“center of mass”) and indicate the position of the line profiles displayed at the periphery of the figure. The black double arrows in the line profile plots indicate the full width at half maximum (FWHM). The red arrow marks an additional faint feature originating from the mirror mount, replacing the sample in these tests.

Close modal

In the detector position [Figs. 9(b) and 9(d)], we find excellent agreement between the measured and simulated intensity distributions in the high-intensity region; also, the FWHM values agree very well. The low-intensity tails of the distribution are more pronounced in the experiment than in the simulation. In summary, the intensity maps verify the correct simulations and design of the spectrometry setup.

The IRAS setup was tested using a synthetic rutile TiO2(110) single crystal (CrysTec GmbH, miscut < 0.05°) mounted on a Ta plate (with Ta clips) and cooled with a helium flow cryostat (base temperature of 37 K) of the UHV setup.29,51 The UHV chamber features a base pressure of 5 × 10−11 mbar. For good thermal contact, a gold foil (0.025 mm thickness) was placed between the TiO2 crystal and the Ta plate. The [001] direction of the TiO2(110) sample was in the incidence plane of the IR beam.

The sample was first annealed to 950 K. After that, the sample was prepared by cycles of sputtering (1 keV Ne+, 15 min, at 300 K) and UHV annealing to 900 K for 20 min. From time to time, the sample was reoxidized by annealing at 900 K in 5 × 10−7 mbar O2, with subsequent UHV annealing to 900 K for 10 min.43,54,55 The reduction state of the sample is not necessarily the same in all measurements presented here; it could have changed because of the repeated preparation steps during the lengthy test measurements.

IR spectra were recorded with a mirror speed of 19 mm/s (laser interferometer frequency 60 kHz), a zero filling factor of one, and a Happ-Genzel apodization. For p-polarized spectra, the spectrometer was set to average 1000 scans with a resolution of 4 cm−1 and a J-stop of 6 mm unless mentioned otherwise. Thus, an IRAS measurement took about 5 min in total for the reference and sample spectrum. S-polarized measurements required a smaller J-stop setting of 3 mm to avoid saturation of the MCT detector. Again, a resolution of 4 cm−1 was used; averaging was done over 4000 scans (≈20 min).

IRAS spectra require recording two spectra—the reference spectrum (R0) and the sample spectrum (R)—which are then used to calculate the normalized reflectivity difference according to Eq. (1). Here, the IR spectra were measured using two different approaches.

The first approach is to measure the clean sample for a reference spectrum. Then, the sample is moved to the molecular beam position for dosing. After moving the sample back to the IR position, the measurement from the adsorbate-covered surface is used as a sample spectrum. This method will be referred to as the adsorption spectrum and describes the standard procedure for IRAS measurements.

For a smooth workflow, the reference spectrum of the CO measurements was sometimes recorded after measuring the adsorbate-covered surface. Then, the adsorbate was desorbed at 350 K, and the reference spectrum (after the base temperature of 43.5 K was reached) was acquired. Spectra acquired in this manner will be referred to as desorption spectra in the following. This procedure has the advantage that no sample movement is necessary between the acquisition of the sample and reference spectra, which results in a better match between the sample positions for these two spectra.

In our UHV system, the reproducibility of the sample position is better than ±0.05 mm. Nevertheless, the small changes of the sample position lead to baseline shifts and decreased peak heights. Using the sample position finder shown in Fig. 8, these effects can be mitigated.

The SNR of the normalized reflectivity difference spectra was evaluated by determining the peak heights and the root mean square (rms) noise of the spectra in the range of 1900–2100 cm−1 using the OPUS software56 and a parabolic fit.

1. CO on rutile TiO2(110)

Low coverages of CO adsorbed on rutile TiO2(110) provide a good model system for checking the functionality of our IRAS setup for vibrational modes perpendicular to the surface because previous studies15,22,24,52,53 provided comprehensive IR measurements for this system.

a. Angle range optimization.

To optimize our IRAS measurements, we need to know how the SNR depends on the incidence angle range. Adjusting the angle-selection plates (see Figs. 3 and 4) allows us to change the incidence angles in the non-grazing and grazing range. For the measurements presented here, the absolute errors of the incidence angle are estimated to be less than ±3.5°. (The main reason for this uncertainty comes from the rather long levers holding the angle-selection plates. Small deviations of the angle around the rotation axis result in large position deviations.) The reproducibility of the incidence angle range is ±1°. All data were obtained with a coverage of 1 monolayer (ML) CO, which corresponds to a gas dose of 1.36 L (Langmuir; 1 L = 10−6 Torr s) adsorbed at a sample temperature of 43.5 K. Since the angle-selection plates were moved in each step, each measurement also includes a reference spectrum for this range of incidence angles (taken after desorption).

Comparing the non-grazing and grazing sides of the Brewster angle, we find the optimum SNR on the non-grazing side, which can be seen by directly comparing the spectra in Figs. 10(b) and 10(c). Finding the optimum SNR on the non-grazing side is expected. The widths of the incidence angle ranges between the limits and the Brewster angle for the two cases are comparable (18° and 21°). However, the non-grazing side provides two advantages: (i) a higher average signal intensity and (ii) in this range of incidence angles, all the beam gets reflected from the adsorbate-covered area of the sample in our setup (Fig. 5). The situation would be different for a material with a substantially lower index of refraction n, where the Brewster angle is lower (e.g., θB = 56° for n = 1.5). In such a case, the angle range at the non-grazing side will be very small, and the best SNR will be obtained at the grazing side of θB.

FIG. 10.

Experimental spectra for 1 ML CO/TiO2(110) obtained for different incidence angle ranges. Frame (a) shows the calculated angle dependence of ΔR/R0 for comparison. Note the low peak height of spectrum (d), which results from partial cancellation of the positive and negative peak heights on the right and left side of the Brewster angle. Thus, despite the high intensity leading to low noise, the SNR of this measurement is only 24.4, much less than in (b) or (c) (SNR = 69.4 and 51.3, respectively) where only peaks with the same orientation contribute to the signal. The spectra were measured with a resolution of 4 cm−1 and averaged over 1000 scans. No baseline correction was applied, and all spectra are plotted with the same scale.

FIG. 10.

Experimental spectra for 1 ML CO/TiO2(110) obtained for different incidence angle ranges. Frame (a) shows the calculated angle dependence of ΔR/R0 for comparison. Note the low peak height of spectrum (d), which results from partial cancellation of the positive and negative peak heights on the right and left side of the Brewster angle. Thus, despite the high intensity leading to low noise, the SNR of this measurement is only 24.4, much less than in (b) or (c) (SNR = 69.4 and 51.3, respectively) where only peaks with the same orientation contribute to the signal. The spectra were measured with a resolution of 4 cm−1 and averaged over 1000 scans. No baseline correction was applied, and all spectra are plotted with the same scale.

Close modal

Figure 11 demonstrates the impact of the incidence angle range on the SNR. Plot (a) depicts the SNR measurements with non-grazing angles between 48° and θL for different θL values. With increasing θL, i.e., increasing slit size (indicated by the black arrows), more light passes through, and the SNR increases. The maximum SNR of 69.4 is reached at θL = 65° [spectrum in Fig. 10(a)]. This is close to the Brewster angle of 66.2°. The SNR decreases when the plate is opened more toward the grazing side. This can be directly related to the band inversion above the Brewster angle: When including angles at both sides of θB, positive and negative peaks superimpose and partly cancel out, leading to a decrease in the SNR although the intensity increases (see also Fig. 2).

FIG. 11.

SNR for CO/TiO2(110) (vertical dipole moment) and different incidence angle ranges. In (a), the incidence angle range is between 48° and θL (non-grazing side), while (b) includes angles between θL and 87° (grazing side). A gray circle marks the maximum SNR. For each plot, one angle-selection plate is fixed (indicated by the vertical black line) not to obstruct the IR beam, and the second plate is moved to limit the incidence angle by θL (dashed lines). See Fig. 4 for the measurement configuration. The schematic peaks on the zero line indicate the orientation of the peaks in the ΔR/R0 spectrum at the given angle range.

FIG. 11.

SNR for CO/TiO2(110) (vertical dipole moment) and different incidence angle ranges. In (a), the incidence angle range is between 48° and θL (non-grazing side), while (b) includes angles between θL and 87° (grazing side). A gray circle marks the maximum SNR. For each plot, one angle-selection plate is fixed (indicated by the vertical black line) not to obstruct the IR beam, and the second plate is moved to limit the incidence angle by θL (dashed lines). See Fig. 4 for the measurement configuration. The schematic peaks on the zero line indicate the orientation of the peaks in the ΔR/R0 spectrum at the given angle range.

Close modal

In Fig. 11(b), the SNR for the grazing range between θL and θmax = 87° is plotted. Again, the SNR increases with increasing intensity as the plate at the non-grazing side opens. The maximum SNR is reached at θL = 67.3° [see Fig. 10(c)], again close to the Brewster angle. In the spectra recorded, the peak orientation changes at a cutoff angle of θL = 56°. Here, the positive and negative contributions cancel out; thus, the SNR is zero.

Finally, we should mention that the best SNR is obtained for incidence angle ranges not reaching to the Brewster angle θB. Although the difference between the optimal cutoff θL and the Brewster angle is within the error bars, we consider this a valid result. In a real system, the ΔR/R0 ratio close to the Brewster angle is low (see Fig. 6). The reflected light close to the Brewster angle will be dominated by s polarization (leaking through the polarizer or caused by polarization aberration); this light carries no signal, but it can contribute to the noise and decrease the peak height in the ΔR/R0 spectrum.

2. Spectra at high resolution and low coverage

Figure 12 shows two spectra acquired with a high resolution of 1 cm−1 and the optimized incidence angle range (48°–67°). Plot (a) shows 1 ML CO adsorbed on the surface. The CO stretch again appears at 2181 cm−1; the SNR obtained with 1000 repetitions is 43. Figure 12(b) shows the spectrum obtained for 0.1 ML CO acquired with 4000 scans. The main peak appears at 2187 cm−1 and has a SNR of 11.6. A side peak is visible at 2193 cm−1; but no peak is seen at the position of the 1 ML peak (2181 cm−1). The height of the main peak for 0.1 ML CO is seven times smaller than for 1 ML CO.

FIG. 12.

Raw spectra for (a) 1 ML and (b) 0.1 ML CO/TiO2(110) acquired with 1 cm−1 resolution and an incidence angle range of 48°–67°. Averaging was done over (a) 1000 and (b) 4000 repetitions.

FIG. 12.

Raw spectra for (a) 1 ML and (b) 0.1 ML CO/TiO2(110) acquired with 1 cm−1 resolution and an incidence angle range of 48°–67°. Averaging was done over (a) 1000 and (b) 4000 repetitions.

Close modal

The peak positions and shapes generally agree with the work presented previously.15,22 The small differences in the peak position (2181 cm−1 in our work vs 2178 cm−1 in the literature for 1 ML CO) may be due to a different reduction state of the TiO2(110) sample. The detailed parameters of the shown spectra can be seen in Table I.

TABLE I.

Root-mean-square (rms) noise values, peak heights (both given as a fraction of the intensity), and signal-to-noise ratios (SNRs) in the different spectra. All measurements were performed on TiO2(110) and are shown in Figs. 10, 12, and 13. The noise values for ΔR/R0 were evaluated in the wavenumber range of 1900–2100 cm−1 with the OPUS software56 from Bruker and the use of a parabolic fit.

Resolution (cm−1)No. scansAdsorbateAngle rangepolrms noisePeak heightSNR
1000 1 ML CO 48°–87° 1.23 × 10−5 3.00 × 10−4 24.4 
1000 1 ML CO 48°–65° 2.63 × 10−5 1.83 × 10−3 69.4 
1000 1 ML CO 67°–87° 2.02 × 10−5 1.04 × 10−3 51.3 
4000 0.1 ML CO 48°–65° 3.01 × 10−5 3.50 × 10−4 11.6 
1000 1 ML CO 48°–65° 5.74 × 10−5 2.50 × 10−3 43.6 
4000 1 ML D248°–87° 2.17 × 10−6 1.40 × 10−4 64.5 
Resolution (cm−1)No. scansAdsorbateAngle rangepolrms noisePeak heightSNR
1000 1 ML CO 48°–87° 1.23 × 10−5 3.00 × 10−4 24.4 
1000 1 ML CO 48°–65° 2.63 × 10−5 1.83 × 10−3 69.4 
1000 1 ML CO 67°–87° 2.02 × 10−5 1.04 × 10−3 51.3 
4000 0.1 ML CO 48°–65° 3.01 × 10−5 3.50 × 10−4 11.6 
1000 1 ML CO 48°–65° 5.74 × 10−5 2.50 × 10−3 43.6 
4000 1 ML D248°–87° 2.17 × 10−6 1.40 × 10−4 64.5 

3. 1 ML D2O on rutile TiO2(110)

The performance in s-polarized IRAS measurements was assessed by adsorbing 1 ML (1.15 L) D2O on TiO2(110) and utilizing the full incidence angle range. The vibrational modes in this system are mainly oriented parallel to the surface. This system has also been well studied13,24 and provides an excellent foundation for benchmarking our system.

The sample was prepared using the standard procedure described in Sec. V B. After flashing it to 950 K, it was cooled to 45.9 K, and a reference spectrum was acquired. After hydroxylation (through reaction with the oxygen vacancies) at 315.4 K with 1.5 L of D2O, the reduced TiO2(110) sample was further treated by adsorbing 1 ML D2O at 186.4 K. Following this, a sample spectrum was recorded at 45.9 K with a resolution of 4 cm−1 and 4000 repetitions (Fig. 13). It shows two peaks. The first is located at 2602 cm−1, and the second, broader feature with the maximum at 2328 cm−1, has a SNR of 64.5. The spectrum shape generally agrees with the spectra reported in previous work24 and was measured in ∼20 min.

FIG. 13.

S-polarized spectra of 1 ML D2O on the hydroxylated TiO2(110) surface with the [001] direction in the incidence plane. The spectrum was acquired with 4 cm−1 resolution, and the averaging was done over 4000 repetitions. A baseline correction was applied to the spectrum.

FIG. 13.

S-polarized spectra of 1 ML D2O on the hydroxylated TiO2(110) surface with the [001] direction in the incidence plane. The spectrum was acquired with 4 cm−1 resolution, and the averaging was done over 4000 repetitions. A baseline correction was applied to the spectrum.

Close modal

Table I summarizes the analysis results of the spectra shown above. The rms noise, the peak height in the spectrum, and the SNR are shown for different coverages of CO and D2O adsorbed on the reduced TiO2(110) surface. The best result for p polarization was achieved in the angle range of 48°–65° for 1 ML CO. The measurements of D2O with s polarization show a good SNR for the full incidence angle range of 48°–87°.

This work introduces a novel IRAS setup tailored to study adsorbates on dielectrics such as metal oxide single crystals under UHV conditions. The most important features for achieving a high signal-to-noise ratio in measurements of a small sample area are (i) high throughput based on a careful design of the optics and a large numerical aperture obtained by using mirrors with short focal lengths, (ii) the selection of the optimal incidence angle range to only acquire signals with the same orientation of the peak, resulting in maximized peak heights, and (iii) the precise integration of the optical platforms and components in the IRAS system ensuring long-term stability of the optical alignment and flat baselines. By considering these features, we achieved a signal-to-noise ratio close to 70 at 4 cm−1 resolution acquired in a time of 5 min for p-polarized IRAS of one monolayer CO on TiO2(110) and sub-monolayer sensitivity, with minutes of measurement time. Moreover, the instrument was designed to be user-friendly and compact, thereby expanding its application range from any UHV system where a free DN 150 CF port is available up to IRAS operando studies of model catalysts with a low density of reaction sites.

The supplementary material includes the formulas used to calculate the reflectivities of the clean and adsorbate-covered surfaces, as well as calculated curves for the s and p polarization of the incidence angle dependent ΔR/R0, taking into account non-ideal polarization, angle spread, and adsorbate illumination. In addition, simulated bandshapes are shown and the focal lengths for the mirrors used in the IRAS setup are given.

This research was funded in part by the Austrian Science Fund (FWF) (Grant Nos. 10.55776/Y847 and 10.55776/F81). For the purpose of open access, the author has applied a CC BY public copyright license to any Author Accepted Manuscript version arising from this submission. The authors acknowledge the support by the European Research Council (ERC) (Consolidator Grant “E-SAC,” Grant Agreement No. 864628 and Advanced Research Grant “WatFun,” Grant Agreement No. 883395). M.E. acknowledges the support by the Marie Skłodowska-Curie Actions under the Horizon Europe Framework Programme for action Grant No. 101103731—SCI-PHI. The authors acknowledge TU Wien Bibliothek for the financial support through its Open Access Funding Programme. The authors thank Rainer Gärtner and Herbert Schmidt for the excellent fabrication of the custom-designed components.

The authors have no conflicts to disclose.

David Rath: Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Validation (lead); Visualization (lead); Writing – original draft (lead); Writing – review & editing (equal). Vojtěch Mikerásek: Investigation (supporting); Methodology (supporting). Chunlei Wang: Investigation (supporting); Writing – review & editing (supporting). Moritz Eder: Investigation (supporting); Writing – review & editing (supporting). Michael Schmid: Conceptualization (equal); Formal analysis (equal); Investigation (supporting); Methodology (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal). Ulrike Diebold: Conceptualization (equal); Investigation (supporting); Methodology (supporting); Resources (equal); Supervision (equal); Validation (supporting); Writing – review & editing (equal). Gareth S. Parkinson: Conceptualization (lead); Investigation (supporting); Methodology (supporting); Project administration (lead); Resources (lead); Supervision (supporting); Validation (supporting); Writing – review & editing (equal). Jiří Pavelec: Conceptualization (equal); Investigation (equal); Methodology (equal); Project administration (equal); Supervision (lead); Validation (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
F. M.
Hoffmann
and
M. D.
Weisel
, “
Fourier transform infrared reflection absorption spectroscopy studies of adsorbates and surface reactions: Bridging the pressure gap between surface science and catalysis
,”
J. Vac. Sci. Technol., A
11
,
1957
(
1993
).
2.
G.
Rupprechter
, “
Sum frequency generation and polarization–modulation infrared reflection absorption spectroscopy of functioning model catalysts from ultrahigh vacuum to ambient pressure
,”
Adv. Catal.
51
,
133
263
(
2007
).
3.
F.
Zaera
, “
New advances in the use of infrared absorption spectroscopy for the characterization of heterogeneous catalytic reactions
,”
Chem. Soc. Rev.
43
,
7624
(
2014
).
4.
J.
Kattner
and
H.
Hoffmann
, in
Handbook of Vibrational Spectroscopy
, edited by
P.
Griffiths
and
J. M.
Chalmers
(
John Wiley and Son Ltd.
,
Chichester
,
2002
), pp.
1
19
.
5.
F.
Hoffmann
, “
Infrared reflection-absorption spectroscopy of adsorbed molecules
,”
Surf. Sci. Rep.
3
,
107
(
1983
).
6.
S. A.
Francis
and
A. H.
Ellison
, “
Infrared spectra of monolayers on metal mirrors
,”
J. Opt. Soc. Am.
49
,
131
(
1959
).
7.
R. G.
Greenler
, “
Infrared study of adsorbed molecules on metal surfaces by reflection techniques
,”
J. Chem. Phys.
44
,
310
(
1966
).
8.
R. G.
Greenler
, “
Reflection method for obtaining the infrared spectrum of a thin layer on a metal surface
,”
J. Chem. Phys.
50
,
1963
(
1969
).
9.
J. A.
Mielczarski
, “
External reflection infrared spectroscopy at metallic, semiconductor, and nonmetallic substrates. 1. Monolayer films
,”
J. Phys. Chem.
97
,
2649
(
1993
).
10.
H. A.
Pearce
and
N.
Sheppard
, “
Possible importance of a ‘metal-surface selection rule’ in the interpretation of the infrared spectra of molecules adsorbed on particulate metals; infrared spectra from ethylene chemisorbed on silica-supported metal catalysts
,”
Surf. Sci.
59
,
205
(
1976
).
11.
M.
Moskovits
, “
Surface selection rules
,”
J. Chem. Phys.
77
,
4408
(
1982
).
12.
R. A.
Dluhy
, “
Quantitative external reflection infrared spectroscopic analysis of insoluble monolayers spread at the air-water interface
,”
J. Phys. Chem.
90
,
1373
(
1986
).
13.
G. A.
Kimmel
,
M.
Baer
,
N. G.
Petrik
,
J.
VandeVondele
,
R.
Rousseau
, and
C. J.
Mundy
, “
Polarization- and azimuth-resolved infrared spectroscopy of water on TiO2(110): Anisotropy and the hydrogen-bonding network
,”
J. Phys. Chem. Lett.
3
,
778
(
2012
).
14.
M.
Creutzburg
,
K.
Sellschopp
,
R.
Gleißner
,
B.
Arndt
,
G. B.
Vonbun-Feldbauer
,
V.
Vonk
,
H.
Noei
, and
A.
Stierle
, “
Surface structure of magnetite (111) under oxidizing and reducing conditions
,”
J. Phys. Condens. Matter
34
,
164003
(
2022
).
15.
M.
Xu
,
H.
Noei
,
K.
Fink
,
M.
Muhler
,
Y.
Wang
, and
C.
Wöll
, “
The surface science approach for understanding reactions on oxide powders: The importance of IR spectroscopy
,”
Angew. Chem., Int. Ed.
51
,
4731
(
2012
).
16.
N. A.
Khan
,
A.
Uhl
,
S.
Shaikhutdinov
, and
H.-J.
Freund
, “
Alumina supported model Pd–Ag catalysts: A combined STM, XPS, TPD and IRAS study
,”
Surf. Sci.
600
,
1849
(
2006
).
17.
M.
Yulikov
,
M.
Sterrer
,
T.
Risse
, and
H.-J.
Freund
, “
Gold atoms and clusters on MgO(100) films; an EPR and IRAS study
,”
Surf. Sci.
603
,
1622
(
2009
).
18.
J.
Hoffmann
,
S.
Schauermann
,
V.
Johánek
,
J.
Hartmann
, and
J.
Libuda
, “
The kinetics of methanol oxidation on a supported Pd model catalyst: Molecular beam and TR-IRAS experiments
,”
J. Catal.
213
,
176
(
2003
).
19.
M.
Schwarz
,
F.
Faisal
,
S.
Mohr
,
C.
Hohner
,
K.
Werner
,
T.
Xu
,
T.
Skála
,
N.
Tsud
,
K. C.
Prince
,
V.
Matolín
,
Y.
Lykhach
, and
J.
Libuda
, “
Structure-dependent dissociation of water on cobalt oxide
,”
J. Phys. Chem. Lett.
9
,
2763
(
2018
).
20.
D.
Stacchiola
,
F.
Calaza
,
L.
Burkholder
, and
W. T.
Tysoe
, “
Vinyl acetate formation by the reaction of ethylene with acetate species on oxygen-covered Pd(111)
,”
J. Am. Chem. Soc.
126
,
15384
(
2004
).
21.
L. R.
Merte
,
C. J.
Heard
,
F.
Zhang
,
J.
Choi
,
M.
Shipilin
,
J.
Gustafson
,
J. F.
Weaver
,
H.
Grönbeck
, and
E.
Lundgren
, “
Tuning the reactivity of ultrathin oxides: NO adsorption on monolayer FeO(111)
,”
Angew. Chem., Int. Ed.
55
,
9267
(
2016
).
22.
N. G.
Petrik
and
G. A.
Kimmel
, “
Adsorption geometry of CO versus coverage on TiO2(110) from s- and p-polarized infrared spectroscopy
,”
J. Phys. Chem. Lett.
3
,
3425
(
2012
).
23.
Y.
Wang
and
C.
Wöll
, “
IR spectroscopic investigations of chemical and photochemical reactions on metal oxides: Bridging the materials gap
,”
Chem. Soc. Rev.
46
,
1875
(
2017
).
24.
N. G.
Petrik
,
M. D.
Baer
,
C. J.
Mundy
, and
G. A.
Kimmel
, “
Mixed molecular and dissociative water adsorption on hydroxylated TiO2(110): An infrared spectroscopy and ab initio molecular dynamics study
,”
J. Phys. Chem. C
126
,
21616
(
2022
).
25.
Y.
Kobayashi
and
T.
Ogino
, “
Angle-variable infrared external reflection spectroscopy in UHV and its application to the observation of SiH vibrations on Si(001) substrates with a buried metal layer
,”
Surf. Sci.
368
,
102
(
1996
).
26.
D.
Blaudez
,
T.
Buffeteau
,
B.
Desbat
,
P.
Fournier
,
A.-M.
Ritcey
, and
M.
Pézolet
, “
Infrared reflection—Absorption spectroscopy of thin organic films on nonmetallic substrates: Optimal angle of incidence
,”
J. Phys. Chem. B
102
,
99
(
1998
).
27.
A.
Udagawa
,
T.
Matsui
, and
S.
Tanaka
, “
Fourier transform infrared reflection spectroscopy studies of organic films formed on inorganic substrates
,”
Appl. Spectrosc.
40
,
794
(
1986
).
28.
J. S.
Wong
and
Y.-S.
Yen
, “
Intriguing absorption band behavior of IR reflectance spectra of silicon dioxide on silicon
,”
Appl. Spectrosc.
42
,
598
(
1988
).
29.
J.
Pavelec
,
J.
Hulva
,
D.
Halwidl
,
R.
Bliem
,
O.
Gamba
,
Z.
Jakub
,
F.
Brunbauer
,
M.
Schmid
,
U.
Diebold
, and
G. S.
Parkinson
, “
A multi-technique study of CO2 adsorption on Fe3O4 magnetite
,”
J. Chem. Phys.
146
,
014701
(
2017
).
30.
D.
Halwidl
,
Development of an Effusive Molecular Beam Apparatus
(
Springer Fachmedien Wiesbaden
,
Wiesbaden
,
2016
).
31.
M. N.
Polyanskiy
, “
Refractiveindex.info database of optical constants
,”
Sci. Data
11
,
94
(
2024
).
32.
J.
Kischkat
,
S.
Peters
,
B.
Gruska
,
M.
Semtsiv
,
M.
Chashnikova
,
M.
Klinkmüller
,
O.
Fedosenko
,
S.
Machulik
,
A.
Aleksandrova
,
G.
Monastyrskyi
,
Y.
Flores
, and
W.
Ted Masselink
, “
Mid-infrared optical properties of thin films of aluminum oxide, titanium dioxide, silicon dioxide, aluminum nitride, and silicon nitride
,”
Appl. Opt.
51
,
6789
(
2012
).
33.
D. C.
Cronemeyer
, “
Electrical and optical properties of rutile single crystals
,”
Phys. Rev.
87
,
876
(
1952
).
34.
D. C.
Langreth
, “
Macroscopic approach to the theory of reflectivity
,”
Phys. Rev. B
39
,
10020
(
1989
).
35.
R. G.
Tobin
, “
Asymmetric lines and background shifts in reflection-absorption infrared spectroscopy
,”
Phys. Rev. B
45
,
12110
(
1992
).
36.
J. D. E.
McIntyre
and
D. E.
Aspnes
, “
Differential reflection spectroscopy of very thin surface films
,”
Surf. Sci.
24
,
417
(
1971
).
37.
D. L.
Allara
,
A.
Baca
, and
C. A.
Pryde
, “
Distortions of band shapes in external reflection infrared spectra of thin polymer films on metal substrates
,”
Macromolecules
11
,
1215
(
1978
).
38.
H.
Brunner
,
U.
Mayer
, and
H.
Hoffmann
, “
External reflection infrared spectroscopy of anisotropic adsorbate layers on dielectric substrates
,”
Appl. Spectrosc.
51
,
209
(
1997
).
39.
O.
Gamba
,
H.
Noei
,
J.
Pavelec
,
R.
Bliem
,
M.
Schmid
,
U.
Diebold
,
A.
Stierle
, and
G. S.
Parkinson
, “
Adsorption of formic acid on the Fe3O4(001) surface
,”
J. Phys. Chem. C
119
,
20459
(
2015
).
40.
J.
Chaves
,
Introduction to Nonimaging Optics
, 2nd ed. (
CRC Press
,
Boca Raton
,
2016
).
41.
R. A.
Chipman
,
W. S. T.
Lam
, and
G.
Young
,
Polarized Light and Optical Systems
(
CRC Press
,
Boca Raton
,
2018
).
42.
Y.
Wang
,
A.
Glenz
,
M.
Muhler
, and
C.
Wöll
, “
A new dual-purpose ultrahigh vacuum infrared spectroscopy apparatus optimized for grazing-incidence reflection as well as for transmission geometries
,”
Rev. Sci. Instrum.
80
,
113108
(
2009
).
43.
J.
Kräuter
and
K.
Al-Shamery
, “
Bulk defect-dependent initial steps of acetone oxidation on rutile TiO2(110)
,”
Mol. Phys.
119
,
e1963870
(
2021
).
44.
Z.
Ren
,
S.
Qi
,
H.
Zhou
,
S.
Yan
,
M.
Xu
, and
S.
Hu
, “
Tracking the atomic structure evolution of ZnO(000-1) polar surface via carbon dioxide infrared probe
,”
Appl. Surf. Sci.
635
,
157730
(
2023
).
45.
J.
Libuda
,
I.
Meusel
,
J.
Hartmann
, and
H.-J.
Freund
, “
A molecular beam/surface spectroscopy apparatus for the study of reactions on complex model catalysts
,”
Rev. Sci. Instrum.
71
,
4395
(
2000
).
46.
S.
Attia
,
E. J.
Spadafora
,
J.
Hartmann
,
H.-J.
Freund
, and
S.
Schauermann
, “
Molecular beam/infrared reflection-absorption spectroscopy apparatus for probing heterogeneously catalyzed reactions on functionalized and nanostructured model surfaces
,”
Rev. Sci. Instrum.
90
,
053903
(
2019
).
47.
A.
Gericke
,
A. V.
Michailov
, and
H.
Hühnerfuss
, “
Polarized external infrared reflection-absorption spectrometry at the air/water interface: Comparison of experimental and theoretical results for different angles of incidence
,”
Vib. Spectrosc.
4
,
335
(
1993
).
48.
G.
Yun
,
K.
Crabtree
, and
R. A.
Chipman
, “
Skew aberration: A form of polarization aberration
,”
Opt. Lett.
36
,
4062
(
2011
).
49.

Strictly speaking, the quantity discussed in this section is the normalized intensity change ΔI/I0, but for simplicity, we refer to it as ΔR/R0.

50.
D. A.
Rath
, M.Sc. thesis,
Technische Universität Wien
,
Vienna
,
2019
.
51.
J.
Pavelec
, Ph.D. thesis,
Technische Universität Wien
, Vienna,
2019
.
52.
S.
Hu
,
Z.
Wang
,
A.
Mattsson
,
L.
Österlund
, and
K.
Hermansson
, “
Simulation of IRRAS spectra for molecules on oxide surfaces: CO on TiO2(110)
,”
J. Phys. Chem. C
119
,
5403
(
2015
).
53.
M.
Kunat
,
F.
Traeger
,
D.
Silber
,
H.
Qiu
,
Y.
Wang
,
A. C.
Van Veen
,
C.
Wöll
,
P. M.
Kowalski
,
B.
Meyer
,
C.
Hättig
, and
D.
Marx
, “
Formation of weakly bound, ordered adlayers of CO on rutile TiO2(110): A combined experimental and theoretical study
,”
J. Chem. Phys.
130
,
144703
(
2009
).
54.
P.
Scheiber
,
A.
Riss
,
M.
Schmid
,
P.
Varga
, and
U.
Diebold
, “
Observation and destruction of an elusive adsorbate with STM: O2/TiO2(110)
,”
Phys. Rev. Lett.
105
,
216101
(
2010
).
55.
A.
Linsebigler
,
G.
Lu
, and
J. T.
Yates
, “
CO photooxidation on TiO2(110)
,”
J. Phys. Chem.
100
,
6631
(
1996
).
56.
Bruker
, OPUS base package (
2024
).