Polarization-dependent differential reflectance spectroscopy for real-time monitoring of organic thin film growth

Optical spectroscopy is a non-destructive, powerful technique which is commonly applied to monitor physical processes occurring at surfaces and interfaces in situ and in real-time. It has proven to be particularly suitable to study the structural, electronic, and optical properties of molecular thin films and their interfaces with inorganic materials.^1–4 Due to the fact that light has a relatively large mean free path in matter, it is necessary to eliminate the optical response from the bulk in order to obtain the optical response of a surface. This can be achieved by applying differential methods, namely, differential reflectance spectroscopy (DRS)⁵ or reflectance difference spectroscopy (RDS).⁶ In contrast to RDS, for which a surface anisotropy is required, DRS can be applied to anisotropic as well as isotropic materials due to the fact that it measures the change in reflectance of a surface upon physical or chemical modification. With DRS, one records continuously the incremental evolution of the reflectance during material deposition onto a surface, thus allowing real-time monitoring of growth processes. Furthermore, for very thin films (d ≪ λ), the real and imaginary parts of the dielectric function can be directly obtained from DRS measurements.^5,7

Two fundamentally different types of configurations can be used for DRS: (i) close to normal incidence (θ = 0°), and (ii) oblique incidence (where the angle of incidence θ is significantly different from zero). For oblique incidence, the reflectance will depend on the linear polarization of the light (perpendicular (s) or parallel (p) to the plane of incidence),^5,8 even for optically isotropic samples. As a result, DRS will produce different signals for s- and p-polarizations, which can be measured independently. Therefore, polarization-dependent DRS (pol-DRS) measurements are expected to deliver complementary information on the optical transitions, electronic states, and the anisotropy of a surface. Moreover, depending on the polarization of light, the different excited optical transitions can be correlated with the arrangement of molecules on a surface, providing additional structural information.⁹ 

While DRS has been widely applied to successfully investigate the growth of organic thin films on metals and semiconductor surfaces,^1,7,10,11 pol-DRS has been so far only applied in the context of the optical anisotropy of metallic nanoparticles.^12–14 Proehl et al. pointed out the possibility to use polarized light while monitoring the reflectance during deposition of PTCDA on different substrates. However, in their analysis, they found it not necessary to consider the exact experimental geometry at θ = 20° and discriminated between the different polarizations states (s and p), because they found no significant influence on the DRS signal for the entire absorption range of their specific organic molecule.¹⁰ Therefore, no polarization dependent DRS study on the growth of organic thin films is available up to now.

In this work, we present a pol-DRS setup operated at off-normal incidence (θ = 65°) and its application to the in situ and in real-time study of the growth of organic thin films. As an example, we investigate the deposition of perflouropentacene (PFP) thin films on a Ag(110) surface. We have chosen Ag(110) as substrate because its surface optical properties are well-known.^15,16 PFP is a n-type organic semiconductor with a high charge carrier mobility, which has been employed for organic field-effect transistors.^17,18 We observe significant polarization-dependent changes in the reflectance during the deposition, related to the different optical transitions excited by the two linear polarization states of light. From these results, we can obtain important information on the alignment of the molecules on the Ag(110) surface. The present study demonstrates the advantages of employing pol-DRS as a tool for in situ monitoring organic thin film growth.

The DRS signal is defined in terms of the reflectance of a surface as follows:

Here, R(E, t) and R(E, 0) denote the reflectance of the surface at a photon energy E after a time t (after deposition) and at t = 0 (pristine surface), respectively. The second part of the equation defines the DRS signal in terms of the measured intensities (I(E, t) and I(E, 0)) of the reflected beam at times t and 0, respectively. The relation R(E, t) ∝ I(E, t) only holds for stable light sources and a robust mechanical setup. A DR spectrum may exhibit positive and negative features since the reflected light intensity can either increase or decrease upon deposition of a molecular film. If light is reflected from a surface at oblique incidence (or if an anisotropic surface is used), the reflectance will depend on the two linear polarization states of light, which are defined with respect to the plane of incidence as follows: The electric field of s-polarized light oscillates in a plane orthogonal to the plane of incidence; the $E →$-vector of p-polarized light is parallel to the plane of incidence. The Fresnel equations for the reflection coefficients of s- and p-polarized light incident from a material 1 to an absorbing material 2 are given by¹⁹ 

where N₂₁ = N₂/N₁ is the ratio of the complex refractive index of both materials and θ is the angle of incidence measured with respect to the surface normal. Considering that $R= r 2$⁠, it can be seen from Equations (2) and (3) that for θ ≈ 0 (normal incidence), the reflectance for both states of polarization is identical. Only for an anisotropic sample, the measured intensities of the reflected light will differ for the two linear polarization states. If the anisotropy of the sample varies in time, e.g., due to deposition of a thin film or a phase transition, one can expect that this difference also evolves in time. Likewise, the evolution of the DRS signal derived from the reflectances R_s and R_p will evolve in a different fashion for s and p-polarized light.

In order to obtain the DR spectra for a certain polarization state of light in a conventional setup, it is necessary to use linearly polarized light and perform two independent experiments for each polarization state. However, even though our deposition system has shown high reproducibility, small variations of the substrate temperature and of the deposition rate of the molecules for each experiment cannot be ruled out, so that measuring the DRS signal simultaneously for both polarizations during the same experiment is extremely advantageous. Taking this into consideration, we have developed a DRS setup which allows splitting the reflected beam into its two states of linear polarization (s and p, respectively) and collecting them simultaneously, in a way similar to the one proposed by Grachev et al.¹² and Lazzari et al.¹³ Our setup differs from the one described in Refs. 12 and 13 by: (i) the splitting angle for the two polarization states is larger, and (ii) we do not employ any optical fibers but use free beam optics to ensure the highest possible signal intensity to improve the overall signal-to-noise ratio (SNR).

The polarization dependent DRS setup is depicted in Fig. 1. It consists of a modular system mounted directly on the view ports of an ultra-high vacuum chamber housing a focus photo-electron emission microscope (PEEM). This allows in situ monitoring of the deposition of the organic molecules on surfaces. The angle of incidence of θ = 65° with respect to the surface normal is fixed and determined by the requirement to mount the PEEM directly in front of the sample as can be seen in Figs. 1 and 2(b). As a light source, we use a super-quiet Hamamatsu Xe-lamp (peak to peak stability ≤0.2%),²⁰ which is mounted into a home-built chassis which is flooded with nitrogen. The constant gas flow yields a constant temperature of the lamp and, thus, lowers the intensity fluctuations. In addition, it ensures high intensity in the UV spectral range (λ ≤ 250 nm) by avoiding light absorption in the ambient air. The incoming light beam is collimated by an achromatic lens and passes through a Glan-Thompson calcite rotating polarizer, whose optical axis is set to 45° with respect to the scattering plane, to ensure the same intensity for s and p polarized light on the sample. Employing the rotating polarizer is optional. However, the advantage of using the rotating polarizer is that one can tune the s-to-p ratio of the incoming beam, if desired.

A Glan-Thompson UV-transparent barium borate (α-BBO) polarizing prism splits the reflected light into its s- and p-components. The p-polarized light passes straight through the prism while the s-polarized light is split 60° away from the transmitted beam. Before passing through the polarizing prism, the reflected light is collected by a lens. It focuses the light into two separate spectrometers, each of which just detects the light of one polarization state. For the signal detection, either STS-VIS or STS-UV spectrometers (from Ocean Optics) are used. The STS-VIS spectrometers operate in the spectral range between 350 nm and 800 nm (3.54 eV–1.55 eV) with a spectral resolution of about 1.5 nm. The STS-UV spectrometers span the range from 190 nm to 650 nm (6.52 eV–1.91 eV). Both types of spectrometers have a digital resolution of 14 bit²¹ and can be triggered by an Arduino board in a home-built control box connected to a PC via USB. The spectrometers (one for s- and one for p-polarized light) are triggered every 20 ms. This interval time is a good compromise between the actual exposure time (10 ms) and the time needed by the internal electronics for the read-out of the CMOS detectors (≲10 ms). In order to reduce the traffic on the USB and increase the SNR, 10 spectra are averaged online on each spectrometer. Additional averaging is applied to the collected data, so that typically 10 transferred spectra are used to obtain a single DR spectrum. Therefore, the final time resolution for the data shown is 2 s, which is sufficient for most real-time studies of adsorption and growth.

All optical components on the side of the reflected beam are mounted on a home-built compact optomechanical cage system (see Fig. 2(a)), ensuring a high mechanical stability. The cage system is mounted directly on the view port of the vacuum chamber as can be seen in Fig. 2(b). The distance between the rods holding the cage is compatible with the commercially available cage systems (60 and 30 mm, respectively). The compact spectrometers are directly mounted on commercial XYZ-stages, used for their alignment.

The base pressure in the chamber is 3 × 10⁻¹⁰ mbar. During the deposition experiments the sample is kept at room temperature. Simultaneously with the DRS measurements, PEEM can be performed to correlate the optical data with the morphological information provided by the local changes in the work function of the surface.^22,23 Prior to the deposition experiment, the Ag(110) surface is cleaned in situ by repeated sputtering and annealing cycles as described elsewhere.²⁴ The sample is mounted onto an Omicron tantalum sample holder with its crystallographic [1$1 ̄$0] axis oriented parallel to the optical scattering plane and the [001] axis normal to it. Therefore, the s and p polarized light will excite optical transitions with transition dipole moments oriented along the [001] and [1$1 ̄$0] crystallographic directions, respectively. The evaporation temperature for the PFP molecules is 185 °C, which yields a growth rate of ≈0.5 monolayers (ML)/min.

In Fig. 3, we present a detailed analysis of the reflected intensities and the signal-to-noise ratio (SNR) after data processing (off line averaging and binning into energy intervals of 0.02 eV). The data were recorded from the pristine Ag(110) surface in intervals of 2 s over a total time span of 20 min, which corresponds to the duration of a typical growth experiment. Because there was no deposition in the experiment reported in Figs. 3 and 4, the reflectance of the sample should not change in time and the DRS signal should be constant over time and equal to zero. The mean value μ of the acquired signals for s and p polarizations on the two spectrometers for 600 processed spectra is depicted in Fig. 3(a). For both polarization states, the mean value μ, as well as its standard deviation σ, is of the same order of magnitude across the entire spectral range, leading to a SNR greater than 1000 for both spectrometers over almost the entire spectral range.

In order to verify the stability of the acquired signal over time, we present in Fig. 4 the signal transients for both polarization states of light at 2.1 eV. In both cases, p we obtain an overall mean value μ of around 12 500 counts. The standard deviation is less than 7 counts taking all 600 spectra into account. The so determined standard deviation takes into account the short term intensity fluctuations of the lamp due to instabilities of the arc discharge, as well as long term intensity drifts. The overall peak to peak difference (maximum - minimum) for the 600 s amounts to 40 counts. We can thus conclude that the long term drift is of the same order of magnitude as the short term fluctuations. This is corroborated by the average over a moving window of 60 data points (μ_av) in Fig. 4. The standard deviation for the moving average (σ_av) is found to be 2 counts for the s polarized, and 4 counts for the p polarized signal, while the long term drift of the mean signal lies between 4 and 10 counts for each polarization, respectively (by taking the difference between the maximum and minimum values of the moving average). Interestingly, the long term drift for both signals is not correlated at all. We can hence conclude that it is not simply the long term stability of the Xe lamp that is limiting the signal quality. One could improve the short term quality of the DRS signal by increasing the signal-to-noise ratio of the spectrometers. Here, the compact miniature spectrometers provide a compromise between cost of investment and signal-to-noise. In fact, more sophisticated spectrometers using cooled detectors exhibit a better signal quality albeit at a price ten times higher.

The stability of the calculated DRS signal is shown in the insets of Fig. 4 for each polarization state, respectively. The mean value of both DRS signals is 1 × 10⁻³, while the standard deviation is only 6 × 10⁻⁴. As can be seen from the DRS transients, the peak to peak fluctuations in the reflectance are of the order of 0.2%. As the detectors are based on CMOS and not on CCD technology, the DRS signal stability is practically independent of the signal intensity. The peak to peak fluctuations and standard deviation over the entire energy range are of the same order of magnitude as the ones presented for 2.1 eV for both polarization states.

The polarization dependent DRS signals acquired during the deposition of PFP on Ag(110) are presented in Fig. 5. The two different polarization states, s (Fig. 5(a)) and p (Fig. 5(a)), show clear and distinctive features evidencing that the reflectance is strongly dependent on the polarization of light. While for p-polarized light the DRS signal reaches a maximum amplitude of 6.5% at an energy of 3.3 eV after 20 min of deposition (about 10 ML of PFP), the reflectance of the s-polarized light changes by almost 18% with respect to that of the pristine silver surface.

The negative change in reflectance can be easily understood: Upon deposition of the PFP on the metallic surface, the reflectance of the sample decreases due to the absorption of light by the PFP molecules. In fact, most of the detected features in both DR spectra can be identified as absorption lines previously observed for PFP thin films on SiO₂.^26,27 The optical transitions at around 2.7 and 2.8 eV observed in the s-polarized DR spectra are a fingerprint of lying molecules excited along the long molecular axis (L) (see inset in Fig. 5(b)).²⁵ Taking into consideration that s polarized light is parallel to the [001] crystallographic direction of the Ag(110) surface, we can conclude that the molecules are flat lying and preferentially aligned with their long axis (L) parallel to the [001] direction, i.e., parallel to the closely packed atomic rows of the Ag(110) surface. Indeed, both features are practically missing in the spectra obtained for p-polarized light.

Additional information on the growth process can be obtained from the DRS transients at those energies for which pronounced changes in the DR spectra are observed. In Fig. 6, the DRS transients at 2.71 eV and 3.61 eV for s and 1.71 eV and 3.31 eV for p polarized lights are depicted.

Prior to the deposition (t < 0), all DRS transients are constant at around 0. After opening the shutter of the PFP evaporator (first dashed line in Fig. 6 at t = 0), the DRS signal starts to decrease, exhibiting a dip after approximately 105 s ± 5 s of deposition. The dip is most pronounced in the DRS signal for p-polarized light, and we assign it tentatively to the completion of the first monolayer. We further observe two changes in slope for all DRS transients at t = 210 s and t = 320 s (marked with dashed lines), which we assign to the completion of the second and third monolayers in agreement with the growth rate of about 0.5 ML/min suggested by the simultaneous PEEM experiment. After t = 320 s the slope of all transients does not present any further abrupt changes until the shutter of the evaporator is closed at t = 1260 s. The change in slope at t = 320 s can be explained as a transition from a 2D layer by layer growth to the formation of 3D islands, similar to that observed in the deposition of α-sexithiophene on Ag(111).²³ 

We have presented a polarization-dependent DRS (pol-DRS) setup. The first measurements on the deposition of PFP on a Ag(110) surface demonstrate that pol-DRS is a powerful tool for monitoring the growth of organic thin films by simultaneously probing the two states of linearly polarized light. We find different optical absorption features in the DR spectra for s and p polarized lights of PFP thin films grown on Ag(110), which are related to the specific absorption of the molecules and their alignment on the metal surface. Moreover, by analyzing the DRS transients at selected absorption energies of the PFP, additional information of the growth process is obtained. The experimental setup is sensitive enough to monitor changes of organic thin films with submonolayer resolution and time intervals in the range of 1 s.

One of the main advantages of the pol-DRS setup presented here is its modular character which allows to add or remove the different modules (such as, the rotating polarizer, beam-splitter, lenses). Our setup could be extended to acquire fixed-angle ellipsometric measurements by including a rotating polarizer on the reflected beam arm. The STS-VIS spectrometers can easily be replaced by STS-UV or STS-IR spectrometers to shift the spectral range more to the ultra-violet (190 nm–650 nm; 6.52 eV–1.91 eV) or infrared (650 nm–1100 nm, 1.91 eV–1.13 eV). In order to increase the intensity in the UV spectral range, one could exchange the Xe-lamp by a D₂-lamp. If required, a beam splitter can be mounted just after the rotating polarizer to couple out a fraction of the incident beam as a reference signal to correct the DR spectra for instabilities of the light source.²⁸ In the results presented here, no optical referencing was required, as we have demonstrated a high signal stability with good signal-to-noise ratio.

Our results are expected to encourage the application of pol-DRS to technologically relevant organic/inorganic systems in order to explore the fundamental physical processes occurring at surfaces and interfaces.

We gratefully acknowledge the financial support by the Austrian Science Fund (FWF): No. P24528-N20, “Real-time observation of growing organic nanostructures.” The authors would like to acknowledge Ewald Fink and Robert Leimlehner for the construction of the pol-DRS cage system, Stefan Buchgeher for the DRS-control box and Konrad Fragner for technical support.