Grazing incidence scattering

In order to probe the inner morphology, grazing incidence scattering (GIS) methods are used. Unlike in standard scattering experiments a reflection geometry is employed in GIS measurements instead of transmission geometry, which is schematically shown in figure 3.3.

Figure 3.3: Schematic presentation of a grazing incidence X-ray scattering setup (GISAXS and GIWAXS). The X-ray beam impinges on the sample with the incident angleα_i and with a wavevector Ki (red arrow). It is scattered under an angleαf. 2D detectors with short and long sample-detector distances record wide angle and small angle X-ray scattering signals, respectively.

The use of a reflection geometry is particularly suitable for characterizing thin films as grazing incidence conditions give a large footprint of the x-ray beam on the samples. Due to the size of the detector, the sample-detector distance (SDD) determines the accessible scattering angles. Hence, various structural length scales of the sample can be detected

by varying the SDD. For example, very small distance below 1 nm can be probed by setting a short SDD (≈ 10 cm). This measuring technique is named as grazing incidence wide angle x-ray scattering (GIWAXS). A longer SDD (≈ 2-4 m) usually allows for ac-cessing structural distances ranging from approximately 1 nm up to 1 µm. It is called grazing incidence small angle x-ray scattering (GISAXS). For both measurements, it is very important to set a proper incident angle α_i of the x-ray beam which determines that the measurements are surface or volume sensitive. To probe the volume of the film in the present thesis, all the incident angles are set above the critical angles of the probed materials.

The GIWAXS measurements were performed at the Austrian SAXS beamline of the Elettra synchrotron source (Trieste, Italy). The x-ray wavelength was 1.54 ˚A, corre-sponding to an x-ray energy of 8 keV. The GISAXS measurements were conducted at the Elettra synchrotron source and the P03/MiNaXS beamline of the PETRA III storage ring at DESY (Hamburg, Germany). The difference between the two beamlines is that the x-ray energy was higher at P03, which provided an energy of 13 keV, corresponding to a x-ray wavelength of 0.957 ˚A. Moreover, the intensity of x-ray flux was also higher at the P03 beamline. The smaller wavelength of x-ray results in smaller critical angles of the probed materials, which should be taken into consideration when the value of the incident angle is determined for the measurements.

GISAXS

The basic setup of GISAXS measurements is schematically displayed in figure 3.3. Due to the long distance between the sample and the detector, a vacuum flight tube is necessary in the x-ray pathway to minimize scattering from air. At the Elettra beamline, a SDD of about 2 m is used, whereas a typical SDD in the range of 3 m to 4 m is chosen at the P03 beamline. Different SDDs meet the demand for different resolvable structural length scales. A Pilatus 1 M (Dectris) detector is used to measure the scattered signal. It consists of ten modules (981 by 1043 pixel array) and each pixel has a size of (172×172) µm². The detector has a total area of (169 × 179) mm² with blind areas of intermodule gap accounting for 8.4% of the total area. The horizontal black stripe in the 2D GISAXS images shown in figure 3.3 is caused by the intermodule gap. Moreover, the detector has a readout time of 3.6 ms, which allows kinetic measurements with high temporal resolution.

To protect the detector from oversaturation, the transmitted and the specularly reflected beams are both blocked with beamstops.

Vertical line cuts of the 2D GISAXS data provide information about the film perpen-dicular to the substrate (in the vertical direction), whereas horizontal line cuts contain information about lateral structures that are parallel to the substrate. For a quantita-tive analysis, such line cuts are performed on the 2D GISAXS data with the aid of the software Fit2D (Andy Hammersley, 1987–2005, ESRF, Grenoble) or DPDAK (Gunthard Benecke, DESY Hamburg & MPIKG Potsdam). Afterwards, the line cuts are fitted with a custom-made program, which is modeling the data in the framework of the distorted wave Born approximation (DWBA) using the local monodisperse approximation (LMA).

Within this model, the scattering objects, described by their form factor, are assumed to have a certain shape (cylinder or sphere) with a Gaussian size distribution. The distance between two neighboring objects, named as the structure factor, are modeled with an one-dimensional paracrystal approach [91]. From fitting, the lateral structure sizes and their corresponding center-to-center distances can be obtained.

GIWAXS

The setup of the GIWAXS measurements is similar to that of the GISAXS measurements except that the SDD is much shorter. With a SDD in the order of 10 cm, information about the molecular stacking, crystal orientation and crystal size can be achieved with the GIWAXS measurements.

The GIWAXS measurements are conducted at the Elettra beamline in the present thesis. The beam source, detector and experimental environment are the same as for GISAXS measurements except the absence of a flight tube. Due to the very short SDD, the scattering from air is negligible. Before analyzing the 2D GIWAXS data, solid angle correction, q-reshaping and conversion, efficiency correction (air path and pixel sensitiv-ity under oblique angles) and polarization correction on the raw 2D GIWAXS data are required in order to retrieve the corrected reciprocal space patterns [90]. 2D GIWAXS data before and after correction are shown in figure 2.17. All corrections are done with the aid of the software Grazingincidence x-ray Scattering Graphical User Interface (GIXS-GUI), which is developed by the Advanced Photon Source, Argonne National Laboratory, USA [89]. Afterwards, the radial integrals are taken in vertical or horizontal directions.

The vertical integrals offer crystalline information along the direction perpendicular to the substrate, whereas the horizontal integrals provide crystalline information along the direction parallele to the substrate. The obtained curves can be fitted with Gaussian functions. From fitting, the lattice constants can be revealed from the peak positions, the crystal sizes can be estimated from the full-width-at-half-maximum (FWHM) values, and relative crystallinity between different samples can be compared by a comparison of peak

intensities. Moreover, information about crystal orientations can be extracted with tube integrals.