Transmission zone plates as analyzers for efficient parallel 2D RIXS-mapping

We have implemented and successfully tested an off-axis transmission Fresnel zone plate as spectral analyzer for resonant inelastic X-ray scattering (RIXS). The imaging capabilities of zone plates allow for advanced two-dimensional (2D) mapping applications. By varying the photon energy along a line focus on the sample, we were able to simultaneously record the emission spectra over a range of excitation energies. Moreover, by scanning a line focus across the sample in one dimension, we efficiently recorded RIXS spectra spatially resolved in 2D, increasing the throughput by two orders of magnitude. The presented scheme opens up a variety of novel measurements and efficient, ultra-fast time resolved investigations at X-ray Free-Electron Laser sources.

. Using an imaging FZP, the second direction is free for acquiring additional information. For instance, it can be used for hν 2 -spectroscopy, where the incoming energy is varied along one axis and the outgoing energy along the other axis (see. Fig. 1c).

Experimental setup
Here, we present the results from our soft X-ray RIXS spectroscopy measurements using an off-axis transmission Fresnel zone plate spectrometer setup. The experiments were performed at P04 beam line of the PETRA III synchrotron in Hamburg, Germany 38 , utilizing the ChemRIXS end station 39 . The off-axis FZP (3 mm × 3 mm) had 70 nm outermost line period (14285 lines/mm) and 7.53 mm outermost radius. The 370 nm high FZP structures consisted of SiO 2 , supported by a 250 nm thick Si 3 N 4 -membrane. They were fabricated by exposure of an HSQ (Hydrogen silsesquioxane) resist layer by electron-beam lithography. The FZP was mounted to be energy-dispersive in the horizontal direction.
The FZP had a focal length of 170 mm at 400 eV photon energy and was placed at a distance of 200 mm from the sample to focus onto a CCD-detector at 1.13 m distance from the FZP, which results in a magnification of 5.6. Unlike reflecting VLS analyzer gratings, the transmission zone plates are largely insensitive to tilt, therefore alignment of our transmission FZP took only a few minutes.
As detector we used an ANDOR iKon-M 934 series camera with 1024 × 1024 pixels and a pixel size of 13 μm × 13 μm. The smallest achieved signal spot size on the detector was 6 pixel (FWHM) in the dispersive direction and the dispersion of the spectrometer was 134 meV per pixel. Thus, the best achievable resolution was 804 meV at 400 eV (resolving power E/dE ~ 500), which is in good agreement with the predicted resolution calculated using the equations shown in a recent publication 37 .
To produce a line focus of 10 μm(h) × 1000 μm(v) on the sample, the vertical monochromator exit slits were fully opened. The mean photon energy was 399.4 eV and varies along the vertical line focus by about 0.7%, roughly a factor of two smaller than the FWHM bandwidth emitted by the 72 period undulator. The fact that we can accept such a wide bandwidth increases the incident flux at least by an order of magnitude compared to the narrowband flux required by a conventional RIXS configuration.

RIXS mapping
In our experiment we used a round liquid jet with a diameter of ~20 μm of an acetonitrile-ethanol mixture as sample. In the acquired signal shown in Fig. 2, the incident energy is varied along the vertical axis and the outgoing energy is varied along the horizontal axis. As the incoming energy is varied along the line focus, the elastic line is oriented in diagonal direction.
The focal length of the FZP changes with the photon energy. Thus, one image is only focused for one specific energy. We recorded two images, which were acquired at slightly different detector positions, one focused at the elastic line at ~399.4 eV (see Fig. 2b) and the other at the nitrogen emission lines of 7a 1 and 2e transitions 17 at ~392 eV (see Fig. 2a). The combination of both images is shown in Fig. 2c. By tilting the detector with respect to the optical axis according to the image positions of the different photon energies, this effect can be partially compensated in the future and it will be possible to obtain a good overall resolution by one exposure only.
We extracted horizontal intensity profiles from Fig. 2c, resulting in the emission lines shown in Fig. 3. In both images one can identify the elastic line, the electronic transitions and the vibrational modes. The results are in excellent agreement with recent measurements 17,39,40 . The major advantage of our method based on an off-axis transmission FZP is, that entire RIXS maps can be acquired with single images. Taking the spectra shown in Fig. 3 with a conventional RIXS setup would require 27 individual measurements.

Imaging RIXS
The imaging capabilities of the zone plate analyzer can be also used to efficiently collect spatially resolved RIXS spectra (see. Fig. 1d). RIXS using a conventional analyzer requires illuminating the sample with a tight point focus, while raster-scanning across the sample surface and acquiring RIXS spectra at each point. In our experiment, the sample is illuminated by a line focus and a scan is performed only in the dimension perpendicular to the focus line. In our experiment the line focus had a length of 1000 μm and a width of 10 μm. Therefore scanning the sample with a line focus increases the throughput by two orders of magnitude compared to a 2D scan with a 10 μm × 10 μm spot in a conventional RIXS geometry.
The sample consisted of a silicon wafer coated with 250 nm Si 3 N 4 . On top of the Si 3 N 4 layer, we patterned 20 nm thick Au structures forming Siemens stars with 500 m diameter and "RIXS" letters with 250 m height. The sample was tilted by 60deg with respect to the incident beam and 30deg with respect to the optical axis of the spectrometer.
The sample surface was scanned horizontally in 101 steps at a mean incident energy of 401 eV. For each step, the signal of the CCD-detector (Fig. 4a) was recorded. The outgoing energy increases along the horizontal axis from the left to the right. One can identify the electronic transitions, the elastically scattered signal, as well as the band-gap from Si 3 N 4 in between. From the series of RIXS patterns, we reconstructed an image of the sample from the resonant nitrogen emission at around 392eV (see. Fig. 4b) as well as from the elastically scattered signal at around 401 eV (see Fig. 4c).
In the two-dimensional image displaying emission energy and one spatial dimension at the same time (Fig. 4a), we now use the different emission characteristics of the structured sample to derive two complementary contrast mechanisms that appear with inverted contrast to each other: For an incident energy of 401 eV, Au provides stronger elastic scattering than the Si 3 N 4 layer. Au appears bright, when selecting 401 eV emission energy, while the regions that lack the Au layer appear darker (Fig. 4b). At the same incident energy, the nitrogen atoms in Si 3 N 4 are resonantly excited and provide a strong inelastic emission signal at photon energies around 392 eV. Selecting this emission energy, the regions not covered with Au thus show up bright, while the Au cover reduces the nitrogen excitation and inelastic emission, leading to ~80% lower signals at this emission energy (Fig. 4c).
In the inelastically scattered image (Fig. 4b) the spatial resolution is about 10 m horizontally. It is limited by the horizontal size of the spot and the scanning step size. In the vertical direction the spatial resolution is 5 m vertically, being only limited by the imaging properties of the spectrometer as a result of magnification and detector pixel size. Here, with a magnification of 5.6 and detector pixels of 13.5 μm, 2-fold sampling leads to a theoretical resolution limit of 4.6 μm. The intrinsic diffraction limit of the FZP depending on the numerical aperture NA and the wavelength λ is λ/NA ~ 400 nm and thus not limiting the resolution here.
In the image reconstructed form the elastic line (Fig. 4c), the spatial resolution is worse, due to the chromatic aberration, as the system was focused for the inelastic emission signal at 392 eV.

Discussion
The opportunity to collect spatial and electronic information simultaneously is highly interesting as it enables to efficiently record local changes of the sample's electronic states 41 . These can be caused by artificial structures like in our experiment, but one can also investigate the local changes of the native characteristics of the material itself, revealing the materials properties at defects and interfaces. In the future the spatial resolution as well as the energy resolution can be increased by a reduction of the horizontal spot size on the sample and by a reduction of the effective detector pixel size.
Due to the chromatic aberration of FZPs, their focal length is changing linearly with the photon energy. As the depth of field is limited, the spectrometer can only be perfectly focused for one photon energy at the same time. We compensated for this effect by taking two images at different detector positions. This effect can likely be reduced by tilting the detector in future experiments. Tilting the detector would also reduce the effective pixel size in the dispersive direction leading to increasing energy resolution.
In summary, the use of an off-axis transmission FZP for RIXS enables exploiting the second axis of the detector order to obtain additional information, which opens up a variety of possible applications. In our experiments, we varied the incident energy across a homogeneous sample as well as the local material properties of a microstructured sample. In the future one could also vary properties like time, temperature, chemical composition, and others on the second axis, making this RIXS analyzer scheme very versatile.