Ellipsometry is an advanced polarization-sensitive measurement technique suitable for the determination of a large variety of sample properties. Its principle is based on analyzing the change in polarization caused by light–matter interaction. The polarization change is determined by evaluating intensity ratios and typically described by the ellipsometric parameters $\Psi$ and $\Delta$ referring to the amplitude ratio and the phase difference between two orthogonal polarization components, respectively. This measurement scheme results in several important benefits: on the one hand, ellipsometry is a non-destructive, contactless and label-free method for absolute measurements (no reference measurements required); on the other hand, since polarization changes can arise from manifold different processes, $\Psi$ and $\Delta$ data can be translated to numerous sample properties of interest, e.g., layer thickness, complex refractive index, anisotropy, roughness, defects, etc. [1]. The conversion of ellipsometric quantities into meaningful material properties is realized by fitting theoretical models to measured data—thus wavelength-dependent spectroscopic ellipsometry (SE) provides maximum information.

SE in the infrared (IR) spectral range additionally provides access to chemical information. Especially the fingerprint region of the mid-infrared (MIR) is of particular interest, as it is sensitive to an abundance of fundamental molecular vibrations. Thus, for anisotropic samples, SE in the MIR simultaneously enables chemical analysis and identification of anisotropy axes that are due to a predominant alignment of specific bonds or molecular chains. However, despite the fact that IR SE is able to provide unique sample information, potential applications have often been restricted by the limited performance of IR instrumentation. To date, commercial IR ellipsometers rely on Fourier-transform IR (FTIR) spectrometers and thermal IR emitters that are strongly limited in spectral brightness. Further, their omni-directional emission leads to limitations in signal-to-noise ratios (SNR), measurement times, and spot sizes [2].

Apart from immovable, large and extremely complex synchrotrons, modern MIR laser sources—namely quantum cascade lasers (QCL) or more recently MIR supercontinuum laser sources (SCL)—are the dominating high-brightness IR sources that have already found application in numerous fields [3–8]. In contrast to emerging MIR SCLs, the well-established QCLs in external cavity configuration provide monochromatic and broadband tunable emission, thus eliminating the need for a spectrometer. Since QCLs offer at least 10⁴ times higher brightness compared with thermal IR emitters as well as a fast spectral tunability of several 1000 cm⁻¹ s⁻¹, sub-second acquisition times for broadband IR ellipsometry spectra are feasible.

In 2019, we introduced broadband QCL-based spectroscopic ellipsometry, outperforming the SNR of a conventional FTIR ellipsometer by a factor of 290 [9]. This exceptional noise performance found expression in a measurement time of only 890 ms for broadband (900 cm⁻¹–1204 cm⁻¹) $\Psi$, $\Delta$-spectra. A comparison with its FTIR based state-of-the-art counterpart revealed that more than 16 hours of acquisition are necessary to achieve a similar noise performance—thus the QCL ellipsometer reduced the measurement time by a factor of 6×10⁴. From then on QCLs were employed in related polarization-sensitive setups including a balanced detection scheme for investigations on vibrational circular dichroism [10,11]. A first approach for hyperspectral polarimetry (i.e., spatially resolved spectroscopic polarimetry) was reported for measurements on polymer and fatty-acid films, with a stated spatial resolution of ${250}\times {125}{\mathrm{\mu} {\rm m}^2}$ [12].

In this contribution we present an enhanced hyperspectral phase-modulated QCL micro-ellipsometer with diffraction-limited spatial resolution better than 13.3 µm. The spatial resolution is characterized by knife-edge measurements and a USAF resolution test target. Furthermore, the benefits of the developed instrument are demonstrated by examining the portrait window of a 200-euro bank note and the cross section of a polypropylene/ethylene vinyl alcohol/polypropylene (PP/EVOH/PP) multilayer film.

The configuration of the QCL micro-ellipsometer is schematically illustrated in Fig. 1(a). The applied QCL (Hedgehog, DRS Daylight Solutions) emits a linearly polarized high-brightness MIR laser beam with a diameter of approximately 5 mm in TEM₀₀ mode and is tunable from 900 cm⁻¹ to 1204 cm⁻¹. In order to minimize the potential of sample destruction, the QCL was driven in pulsed mode with a repetition rate of 1 MHz and a pulse length of 20 ns resulting in a duty cycle of 2 %. As indicated, the radiation is redirected by two uncoated gold mirrors used for alignment. Two consecutive wire grid polarizers on Si substrates—one rotatable, one fixed—are used to adjust the intensity and ensure vertical polarization, respectively. The following telescope system of two ZnSe lenses with focal lengths of 75 mm and 15 mm reduces the beam waist by a theoretical factor of five at cost of increasing beam divergence. This configuration allows us to spatially separate multiple reflections occurring inside the following tilted ZnSe photo-elastic modulator (PEM) and block them with the subsequent razor blade (similar to Refs. [9,10]). The applied PEM (PEM-90, Hinds Instruments) modulates the polarization at a frequency of 37 kHz before the radiation is focused onto the sample by means of an infinity corrected $\times 40$ reflective objective with a numerical aperture $\text {NA} = {0.5}$ (LMM40X-P01, Thorlabs). Consequently, the reduced Gaussian beam waist leads to a distribution of incidence angles dominated by angles between 14°–19°, as estimated with geometrical optics. The radiation transmitted through the sample is collimated with a 6 mm focal length ZnSe objective (#88-446, Edmund Optics) with an $\text{NA} = {0.25}$ and a clear aperture of 5 mm before passing another wire grid polarizer which acts as analyzer. Finally, an achromatic ZnSe/ZnS lens doublet with a focal length of 50 mm focuses the radiation on a thermoelectrically cooled mercury-cadmium-telluride (MCT) detector (PCI-4TE-12, Vigo System).

 

Fig. 1. (a) Experimental setup. Red and yellow beam path for QCL micro-ellipsometer and VIS microscope, respectively. (b) Knife-edge measurements in (top plot) the focal plane and (bottom plot) derived beam profiles for $x$ and $y$ axes.

Download Full Size | PDF

The pulsed detector signal is demodulated with a boxcar integrator (UHFLI, Zurich Instruments) before a signal extraction unit consisting of a 10 kHz high-pass filter and a 2 kHz low-pass filter splits it into AC and DC parts, respectively. While the DC part ($I_{DC}$) is fed directly to the applied data acquisition card (USB-231, Measurement Computing), the AC part is preprocessed beforehand by a lock-in amplifier (eLockIn 204, Anfatec Instruments), which extracts the amplitudes of the first ($I_{\omega }$) and second harmonic ($I_{2\omega }$) of the polarization modulation. As reported in Ref. [9], from these three signals the desired ellipsometric parameters are obtained as follows:

In addition to the QCL micro-ellipsometer, the optical setup comprises the visible (VIS) microscope depicted in Fig. 1(a) (yellow beam path)—a flip mirror allows to switch between these two modalities. Whenever the flip mirror is inserted, a halogen lamp illuminates the sample through the system of a 150 mm focal length glass lens and the reflective objective of the QCL micro-ellipsometer. The sample is imaged through the same components. The image is then redirected by a pellicle beam splitter and projected on a VIS microscope camera (CS165MU1/M, Thorlabs) by means of an additional glass lens with a focal length of 75 mm. Importantly, the focal planes of both the QCL micro-ellipsometer and the VIS microscope coincide. Thus, the VIS microscope drastically simplifies navigation to sample features of interest and sample alignment in the focal plane of the QCL micro-ellipsometer—both are realized by mounting the sample on a motorized two-dimensional (2D) microscope stage (ASR050B050B-T3, Zaber Technologies), which in turn is mounted on a motorized linear stage (076514, Standa) for $z$ axis adjustments.

The focal plane of the micro-ellipsometer was determined by means of a knife-edge method, which additionally revealed the QCL beam profile at the sample position. Therefore, a razor blade was successively moved across the beam path along the $x$ and $y$ axes at various $z$ positions, while simultaneously recording the induced intensity variations. The first derivative of the resulting intensity curves recorded at each $z$ position is directly related to the respective beam profile at this position. Figure 1(b) depicts both the intensity curves in the focal plane for the $x$ and $y$ axes recorded with 0.63 µm steps at a wavelength $\lambda = {10}{\mathrm{\mu} {\rm m} }$ as well as the beam profiles derived from them. As illustrated, similar curves for both axes indicate a highly symmetrical beam profile. The resulting Airy pattern features a narrow maximum with a full width at half maximum (FWHM) of approximately 8 µm as well as—through the application of a reflective objective—rather prominent side maxima.

The actual spatial resolution enabled by the narrow beam profile was evaluated by examining a clear optical path USAF resolution test target and the portrait window of a 200-euro bank note at close to normal incidence. Figure 2(a) illustrates the intensity ($I_0$) transmitted through element 6 of group 3 of the investigated self-standing substrateless aluminum test target (#58-402, Edmund Optics) and the normalized intensity profile across the center of the element (red). The 2D image was acquired by scanning the target in 7.81 µm steps at a wavelength $\lambda$ of 10 µm (1000 cm⁻¹) with an integration time of 20 ms per step. The illustrated element—which is the smallest available on the target—features a linewidth of 35.08 µm. According to the depicted profile, this element is well resolved as the intensity drops approximately 77 % between the lines of the element, well beyond the Rayleigh criterion indicated in blue.

 

Fig. 2. (a) Two-dimensional intensity scan and normalized profile (red) of element 6 of group 3 of a USAF resolution test target acquired at a wavelength of 10 µm (1000 cm⁻¹). The Rayleigh criterion is indicated (blue). (b)(i), (b)(ii) VIS microscope images of the portrait window of a 200-euro bank note and (b)(iii) normalized intensity image acquired with the QCL micro-ellipsometer at 908.4 cm⁻¹. The profile along the position where the two dots of metallization are closest is indicated (red). (c) Hyperspectral ellipsometry data. The $\Psi$ and $\Delta$ images result from averaging the shaded spectral ranges. Similar $\Psi$, $\Delta$-spectra at the position between the two metallization dots and at an open area prove the feasibility of MIR micro-ellipsometry at diffraction-limited spatial resolution.

Download Full Size | PDF

To prove the functionality of the instrument at diffraction-limited spatial resolution, the illustration of princess Europa from the Greek mythology on the portrait window of a 200-euro bank note was investigated. As depicted in the VIS microscope images in Figs. 2(b)(i) and 2(b)(ii), the portrait is composed of numerous small metallizations on an anisotropic polymer window. Of particular interest for this study is the location near Europa’s left eye, highlighted with a red box in the centered microscope image. This location features two dots of metallization that are very close together compared with other features and was thus examined with the QCL micro-ellipsometer. Therefore, the sample was scanned in $64\times 64$ steps of 3.13 µm each. During each step, the QCL was tuned over its full spectrum from 900 cm⁻¹ to 1204 cm⁻¹ leading to three hyperspectral data cubes ($I_{DC}$, $I_{\omega }$, $I_{2\omega }$) acquired simultaneously in approximately 75 minutes in total (for comparison, a full image at a specific wavelength only would result in a measurement time of just 80 s). The intensity image in Fig. 2(b)(iii) shows the normalized slice of the hyperspectral $I_0$ cube at 908.4 cm⁻¹ (11 µm) calculated according to Eq. (3). The two examined dots of metallization are clearly visible and well resolved. The profile along the position where the two dots are closest is depicted in red and shows a bell-shaped curve with a peak at approximately 42 % of the maximum intensity—which again considerably exceeds the Rayleigh criterion and in turn indicates that even smaller distances are resolvable. A Gaussian fit of the profile reveals a FWHM of 13.3 µm, which is in good agreement with the distance between the dots estimated with the VIS microscope and near Abbe’s diffraction limit $d = \frac {\lambda }{2 NA} = {11}{\mathrm{\mu} \rm{m} }$.

In addition to intensity measurements, the developed QCL micro-ellipsometer provides hyperspectral MIR ellipsometry data with diffraction-limited resolution. Figure 2(c) shows hyperspectral $\Psi$, $\Delta$-data calculated with Eqs. (1) and (2), respectively. The depicted $\Psi$ and $\Delta$ images result from averaging the spectral ranges from 969 cm⁻¹ to 988 cm⁻¹ and 962 cm⁻¹ to 981 cm⁻¹, respectively—highlighted as shaded area in the illustrated $\Psi$, $\Delta$-spectra. These spectra were extracted from two specific spatial channels labeled Positions 1 and 2 and plotted as red and blue graphs, respectively. While Position 1 is directly located at the narrow gap between the two metallization dots, Position 2 is located at an open area of the polymer window. The spectra at both positions reveal a prominent absorption band at approximately 979 cm⁻¹, which corresponds to a ${{\rm C}={\rm C}}$ bending vibration. Considering the similarity of the depicted spectra and the small structures investigated, this experiment proves the feasibility of MIR micro-ellipsometry at diffraction-limited spatial resolution.

A comparison of the intensity image in Fig. 2(b) and the $\Psi$, $\Delta$-images of Fig. 2(c) reveals the complementary information provided by the latter—prominent diffraction effects influencing the polarization state can be observed at the edges of the metallization [13]. To demonstrate the benefits, specifically provided by diffraction-limited MIR micro-ellipsometry, Fig. 3 shows a detailed investigation of a PP/EVOH/PP multilayer film with a total thickness of 1.12 mm (the investigation of a similar film with a conventional QCL microscope has been reported in Ref. [14]). Among others, PP is used extensively as a packaging material, with EVOH often serving as an oxygen barrier interlayer, e.g., in shelf-stable food packaging. For this study, a cross section of the examined multilayer with a thickness of approximately 6 µm was prepared with a microtome. Similar to the previous experiment, the QCL was tuned from 900 cm⁻¹ to 1204 cm⁻¹ during scanning step of 3.13 µm. In approximately 75 minutes $64\times 64$ spatial positions were scanned at close to normal incidence, resulting in simultaneously acquired hyperspectral $I_{DC}$, $I_{\omega }$, and $I_{2\omega }$ cubes, which were further processed to hyperspectral $\Psi$, $\Delta$, and $I_0$ cubes according to Eqs. (1)–(3). The latter was further processed to a hyperspectral absorbance cube, using a reference spectrum acquired in 887 ms while no sample was present in the beam path. In addition to a VIS microscope image of the investigated sample, the resulting absorbance data is illustrated in Fig. 3(a). At the bottom of the image, approximately 70 µm from the left edge, the sample features an inclusion of unknown origin, which was used for orientation. The depicted absorbance image results from averaging the spectral range from 1116 cm⁻¹ to 1147 cm⁻¹, where EVOH features strong absorption. As illustrated, the absorbance spectra extracted from Position 1 and 2—which are located directly in high and low absorbance regimes, respectively—adequately fit to the dashed reference spectra of pure PP and EVOH films recorded in transmission measurements with a conventional FTIR spectrometer (VERTEX 70, Bruker).

 

Fig. 3. (a) VIS microscope image and hyperspectral absorbance data of the cross section of a PP/EVOH/PP film recorded with the QCL micro-ellipsometer. Reference spectra of pure PP and EVOH films are illustrated as dashed graphs. (b) Hyperspectral ellipsometry data: $\Psi$ and $\Delta$ images result from averaging the shaded spectral ranges. Decreasing peak amplitudes of PP bands with distance to the EVOH layer indicate a higher degree of anisotropy in proximity to the EVOH.

Download Full Size | PDF

In contrast to the absorbance spectra, which only provide chemical information, the acquired ellipsometry data additionally provides structural information. Figure 3(b) illustrates $\Psi$ and $\Delta$ images and $\Psi$, $\Delta$-spectra extracted from the respective hyperspectral cubes at the four indicated positions on the sample. The depicted images result from averaging the shaded spectral ranges from 967 cm⁻¹–1004 cm⁻¹ and 934 cm⁻¹–974 cm⁻¹ for $\Psi$ and $\Delta$, respectively. Unlike the absorbance image in Fig. 3(a), these images reveal significant variations in proximity to the EVOH interlayer, which can be directly assigned to variations of the anisotropy within the PP layer. This consideration is verified by evaluating the depicted spectra. The fact that Position 3 reveals hardly any ellipsometric features within the EVOH layer despite the presence of strong absorption bands, indicates uniformly distributed EVOH chains with respect to the sample plane. At Positions 4–6, however, the absorption bands characteristic for PP are clearly present indicating anisotropic absorption that can be attributed to a predominant alignment of the PP chains. Positive peaks with amplitudes above 45° indicate an alignment of the molecular chains parallel to the EVOH interlayer. The fact, that the peak amplitudes decrease with distance to the EVOH layer implies a higher degree of anisotropy in proximity to the EVOH—as also expected from the extracted $\Psi$ and $\Delta$ images.

In summary, a new level of spatial resolution for SE in the MIR has been achieved by demonstrating the first diffraction-limited micro-ellisometry setup in this spectral range. The outstanding advantages of a QCL as MIR source for SE enable sub-second acquisition times, exceptional noise performance, and—as reported in the present study—diffraction limited spot sizes. A knife-edge method in two dimensions, revealed both a narrow maximum of the airy pattern with a FWHM of approximately 8 µm and a highly symmetrical beam profile. The actual spatial resolution was found to be diffraction limited by investigating a USAF resolution test target and the portrait window of a 200-euro bank note. Thereby, two dots of metallization on the portrait window approximately 13.3 µm apart could be easily resolved. Furthermore, valid ellipsometry spectra could be acquired directly within the narrow gap between the dots, which is why the QCL micro-ellipsometer can be considered diffraction limited. Finally, the benefits of the developed instrument were highlighted by examining the orientation of molecular chains within a PP/EVOH/PP multilayer. Inhomogeneous anisotropy of the PP layers could thereby be detected at the microscopic level. With diffraction-limited spatial resolution and sub-second acquisition times the QCL micro-ellipsometer allows insight into complex sample properties and parameters in an unprecedented manner.