Information on 3D elemental distribution is critically important in studies of catalysis, life sciences, materials research, toxicology, geology, and many other fields. To access this information X-rays are advantageous compared to electron microscopy and atom probe tomography techniques because of their short wavelengths, high penetrability and possibility to image samples non-destructively. X-ray fluorescence (XRF) tomography stands out among other synchrotron-based techniques for its high sensitivity to low elemental concentrations, simultaneous access to many elements in a single measurement, and for its compatibility with other simultaneous imaging modalities. Recent developments of nano-focussing optics for hard X-rays have advanced XRF tomography towards submicron resolution.
Although higher resolution can be easily achieved in 2D, practical limitations due to long scanning times and instrument instability and positioning accuracy have limited the 3D resolution to a few hundred nanometers [1, 2]. With expected improvements of brilliance at upgraded diffraction-limited storage rings, this limit will be pushed through the reduction of acquisition time. However, resolving the sample stability and positioning accuracy issue will immediately benefit existing XRF imaging stations (both for 2D and 3D case) and will have an even greater impact on the imaging quality of upgraded storage rings.
In this talk, we will present current developments in combining XRF tomography and ptychographic X-ray computed tomography (PXCT) at the cSAXS beamline of the Swiss Light Source. In this approach, we use flOMNI (flexible tOMography Nano Imaging, see Figure 1) [3], a unique instrument that allows accurate sample positioning with respect to the focusing optics by means of differential laser interferometry. The setup is now equipped with a fluorescence detector. Overcoming the stability issue gives us an opportunity to achieve resolution in XRF tomography on the level of the probe size with minimal data manipulation. While combining the electron density contrast from PXCT and the element specificity of fluorescence allows us to exercise various data analysis strategies that will be covered in this talk. We expect that this development will find interest from synchrotron users in fields ranging from life sciences to materials science.

[1] T. Victor et al., X-ray Fluorescence Nanotomography of Single Bacteria with a Sub-15 nm Beam, Sci. Rep. 8, 13415 (2018).
[2] G. Martínez-Criado et al., ID16B: a hard X-ray nanoprobe beamline at the ESRF for nano-analysis, J. Synchrotron Rad. 23, 344-352 (2016).
[3] M. Holler et al., X-ray ptychographic computed tomography at 16 nm isotropic 3D resolution, Sci. Rep. 4, 3857 (2014).

High-resolution imaging of 3D structure and elemental composition is critical for studies ranging from biology to materials science. ID16A is well up to the challenge with its established record in hard X-ray phase and fluorescence imaging. Two phase imaging modalities are routinely offered at the beamline: holography that explores longitudinal diversity and near-field ptychography that explores the transverse diversity. The recent upgrade of ESRF further improves the beamline performance through increased flux and coherence and reduced spectral bandwidth.

For phase imaging the improvements of the source can lead to higher resolution and better data quality (particularly in the near-field ptychography regime). It will also directly benefit X-ray fluorescence imaging where higher flux translates directly into increased elemental sensitivity and reduced acquisition times, allowing to survey more samples resulting in the higher statistical significance of the results.

In this poster, we will discuss the status of the beamline after the transition to the ESRF-EBS and its exciting implications for the user community. We will also discuss immediate plans in the beamline development where new collaborations with the users will be mutually beneficial.

Coherent imaging methods such as ptychography have a growing interest for multidimensional imaging applications due to the high spatial resolution at which quantitative information is obtained. In the application of X-ray imaging to the study of dynamic processes [1], the achievable temporal resolution is limited by detector performance. However, for ptychography this is also determined by the degree of redundancy in the diffraction data that is needed to reliably reconstruct real-space images. The success of ptychography lies in the incorporation of redundancy in the spatial dimensions from transverse translation diversity of the object, which allows for relaxation of constraints on the illumination and other experimental factors. Other schemes such as longitudinal translation (phase) diversity [2], and probe diversity [3], have been reported.

Using the time dimension redundancy for the introduction of a constraint has recently been proposed in coherent diffractive imaging (CDI) [4,5], exploiting “overlap” between successive images to achieve similar advantages to ptychography. However, these methods impose limits on the object or illumination and require a priori knowledge of the location of time-independent regions of the object, making them incompatible with scanning methods such as ptychography. We have developed an algorithm that removes these limitations and introduces redundancy without guidance and without requiring reference object regions by exploiting intentional or incidental temporal diversity in the diffraction data [6]. Our approach automatically defines a spatiotemporal constraint through automatically segmenting time-dependent and time-independent regions within the image field, dependent on the detected sample dynamics, and is able to suppress ambiguity and artefacts in the reconstructions.

Spatiotemporal redundancy in time-series coherent diffraction data provides a viable path toward studying nanoscale dynamics by X-ray imaging. We demonstrate this potential through CDI simulations of different dynamic phenomena under realistic conditions modelled on the XFM beamline at the Australian Synchrotron, the motion of nanoparticles, as well as the oscillatory behaviour of a 2-dimensional chemical reaction. An extension to dynamic ptychography that allows high quality reconstructions to be achieved with a relaxed spatial overlap constraint and faster scanning is then examined through simulation. Preliminary experimental investigation of crack propagation within thin metallic films, obtained using the EIGER2 detector at the XFM beamline, will be presented.

This work was performed in part at the Melbourne Centre for Nanofabrication (MCN) in the Victorian Node of the Australian National Fabrication Facility (ANFF). This research was undertaken on the XFM beamline at the Australian Synchrotron, part of ANSTO.

[1] J. Lim et al., “Origin and hysteresis of lithium compositional spatiodynamics within battery primary particles,” Science, 353, 566 (2016).

[2] C. T. Putkunz, et al., “Phase-Diverse Coherent Diffractive Imaging: High Sensitivity with Low Dose,” Phys. Rev. Lett. 106, 013903 (2011).

[3] I. Peterson, et al., “Probe-diverse ptychography”, Ultramicroscopy 171, 77 (2016).

[4] Y. H. Lo, et al., “In situ coherent diffractive imaging,” Nat. Commun. 9, 1826 (2018)

[5] X. Tao, et al., “Spatially correlated coherent diffractive imaging method,” Appl. Opt. 57, 6527 (2018).

[6] G. N. Hinsley, et al., “Dynamic Coherent Diffractive Imaging Using Unsupervised Identification Of Spatiotemporal Constraints,” Opt. Express 28, 36862 (2020).

Coherent X-ray diffraction imaging (CXDI) is a powerful lensless imaging technique utilizing X-rays with a high degree of coherence [1]. Conceptually, CXDI can image isolated microscopic objects with high resolution and since its first demonstration in 1999 [2] it has laid the foundation for the development of other methodologies such as ptychography and Bragg CXDI [1]. With the new low-emittance storage ring [3] at ESRF combined with a state-of-the-art Eiger 4M detector and efficient iterative algorithms, the ID10 beamline is optimised for CXDI experiments. Three-dimensional imaging of crystalline and amorphous particles at ~14 nm resolution [4] has recently been demonstrated. In contrast to conventional characterisation methodologies such as electron microscopies, CXDI is ideally suited to study the surface morphology and interior of such microscopic particles without sectioning or ion-milling.

We used CXDI to image in 3D a series of CaCO₃ microparticles prepared under different crystallization and growth conditions, revealing that the microparticles systematically assume a wide range of morphologies as function of temperature [5]. In this presentation, we demonstrate 3D CXDI imaging of CaCO₃ particles 3-6 µm in diameter, with 16 nm voxel size. Wide-angle X-ray diffraction (WAXD) patterns [6] were recorded in combination with the CXDI datasets to identify the crystalline phase of the CaCO₃ microparticles and obtain information of characteristic crystal planes. Figure 1 shows the evolution of CaCO₃ particle morphology as a function of precipitation temperature: from the nested hexagonal morphology at T = 25˚ C to an appearance of spikes at T = 35˚, and finally transforming to an extended rod-shape morphology at T = 45˚ C. In addition, the internal structures of the particles and density variations within the particles can also be appreciated in 3D. Finally, we discuss the challenges arising due to radiation damage and how to resolve them, and also the future prospects of dynamic or serial CXDI experiments.

[1] Chapman, H. N. & Nugent, K. A. (2010). Nat. Photon. 4, 833.

[2] Miao, J., Charalambous, P., Kirz, J. & Sayre, D. (1999). Nature 400, 342.

[3] ESRF News. (2017). No. 77, December 2017.

[4] Cherkas, O., Beuvier, T., Breiby, D. W., Chuskhin, Y., Zontone, F. & Gibaud, A. (2017). Cryst. Growth Des. 17, 4183.

[5] Oral, M. Ç. & Ercan, B. (2018). Powder Technology. 339, 781.

[6] Chushkin, Y., Zontone, F., Cherkas, O. & Gibaud, A. (2019). J. Appl. Cryst. 52, 571.

For non-destructive imaging of three-dimensional (3D) materials and biological specimen, hard X-ray in-line holography is particularly suitable, since it offers a phase-sensitive imaging scheme which can cover large specimen in a full-field approach without the need for scanning. Unfortunately, the resolution of holographic imaging is limited by the source size of the cone-beam illumination, and does not reach values in the sub-20 nm regime which are routinely achieved by ptychography or CDI.

We have implemented holographic X-ray imaging based on cone-beam illumination, beyond the resolution limit given by the cone-beam numerical aperture. In this new single-shot approach [1], image information encoded in the far-field diffraction and in the holographic self-interference is treated in a common reconstruction scheme, without the usual empty beam correction step of in-line holography. An illumination profile tailored by waveguide optics and exactly known by prior ptychographic probe retrieval is shown to be sufficient for solving the phase problem. We demonstrate the improved experimental capability by reconstruction of a test pattern with a field of view of 5×5µm² and a resolution of 11 nm, using a waveguide exit source size of about 30 nm (FWHM).

Figure 1 shows the divergent illumination (a) from the waveguide exit (b) to the sample plane (c). To quantify the resolution we have analyzed the reconstruction of a pattern with 50 nm (half-period) lines and spaces (d,e). The reconstruction by the presented method shows higher resolution and image quality compared to conventional single-shot reconstruction by the contrast-transfer-function approach after empty-beam division. The resolution of the new reconstruction approach was determined by Fourier ring correlation (FRC) indicating a resolution (half-period) of Δ = 11.2 nm (f).

The approach paves the way towards high resolution and dose-efficient X-ray tomography, well suited for the current upgrades of synchrotron radiation sources to diffraction limited storage rings.

The three-dimensional X-ray diffraction (3dxrd) technique provides a useful tool to investigate polycrystalline materials, grain-by-grain, in a non-destructive way. The approach of the scanning 3dxrd microscopy is to probe the sample by moving a pencil beam horizontally across it (y direction) with a resolution dependent on the beam size. For each step, the sample is rotated of 180° (or 360°, ω angle) in order to collect the diffraction spots of all the grains in the sample [1].

We used a combined approach of scanning 3dxrd and Phase Contrast Tomography (PCT) to investigate the hydration of a widespread hydraulic binder material, namely gypsum plaster. This material forms when the bassanite (calcium sulfate hemihydrate) reacts with water. In-situ 3dxrd measurements allowed to understand the crystallographic lattice, orientation and position of each grain in the sample during the hydration reaction (Figure 1 a,b).

The PCT reconstructions, instead, allowed the visualization of the shape of the crystals in the sample over time and a quantification of density and porosity (Figure 1 c,d).

Monitoring the evolution of the hydration reaction of gypsum plaster with both these techniques appears to be a promising tool to gain insights about the kinetics of the hydration reaction, the crystallization and growth of the hydrated phase and the shape of the final gypsum crystals that build the interlocked and porous gypsum plaster hardened mass.

X-Ray Photon Correlation Spectroscopy (XPCS) is a flexible tool to quantify dynamics on the nanometer to micrometer scale in bulk samples and was used in the recent years in grazing incidence (GI) geometry for application to thin film samples, such as quantifying thin film growth [1]. Measurements in GI geometry introduce distortions of the detected signal. These distortions are due to refraction and reflection and known from the Distorted Wave Born Approximation (DWBA), which leads to superpositions of signal within detector areas [2]. Zhang et al. [3] showed that these superpositions also influence GI-XPCS measurements and can alter observation quantities like decorrelation times and stretching exponents. We present an approach to quantify the influence these refraction and reflection effects due to the DWBA have on decorrelation analysis by conducting grazing incidence transmission (GT) XPCS and Gi-XPCS simultaneously for a thin film sample of Methyl Ammonium Leadiodide, showing non-equilibrium dynamics. A combination of the GI- and GT XPCS results with calculations of Fresnel reflectivities and transmissivities within the simplified DWBA allows to determine the origin of scattering contributions for GT and GI regions. Considering calculations of the non-linear effect of refraction in GISAXS and GTSAXS, comparable regions to XPCS experiments in transmission are identified and differences for phenomena like altered decorrelation times and decay stretching are elucidated. This allows the use of this technique to analyze dynamics in thin films for certain experimental conditions. [4]
[1] Headrick, R. L., Ulbrandt, J. G., Myint, P., Wan, J., Li, Y., Fluerasu, A., Zhang, Y., Wiegart, L. & Ludwig, J. K. F. (2019). Nature Communications. 10, 1–9.
[2] Lu, X., Yager, K. G., Johnston, D., Black, C. T. & Ocko, B. M. (2013). Journal of Applied Crystallography. 46, 165–172.
[3] Zhang, Z., Ding, J., Ocko, B. M., Fluerasu, A., Wiegart, L., Zhang, Y., Kobrak, M., Tian, Y., Zhang, H., Lhermitte, J., et al. (2019). Physical Review E. 100, 1–8.
[4] Greve, C. R., Kuhn, M., Eller, F., Buchhorn, M. A., Kumar, D., Hexemer, A., Freychet, G., Wiegart, L. & Herzig, E. M., manuscript in preparation.

Functional properties such as dielectric constant [1] and hydrogen storage [2] in fine crystalline materials often exhibit particle size effects. Understanding the phenomena with size effects and utilizing their functions require observing the structure in conjunction with shape, size, and heterogeneity information.

We report the development and improvement of an apparatus for Bragg coherent x-ray diffraction imaging (Bragg-CDI) [3] at BL22XU in SPring-8. The achieved observable particle size was 40–500 nm and Pd (~40 nm) and ferroelectric barium titanate (BaTiO₃, BTO, 200~500 nm) fine crystals were investigated.

This study aims to achieve two primary goals. (1) The first is to reduce background noise due to x-ray scattering by air. To this end, we newly prepared a vacuum chamber for samples, enabling us to obtain high-contrast x-ray diffraction pattern for a shorter time. (2) The second is to optimize a real-space constraint; our modified phase-retrieval algorithm can use appropriate real-space constraints with shrinking [4] support to refine the phase distribution.

We succeeded in expanding the observable particle-size range from 100–300 [3] to 40–500 nm [5] for the Bragg-CDI at BL22XU in SPring-8. The reconstructed three-dimensional image showed the outer shape, size, and internal phase (strain) for a single particle. A single 500-nm BTO particle showed a straight and sharp antiphase-boundary shape, whereas smaller BTO particles showed different phase boundary shapes. The present Bragg-CDI thus allows the observation of the outer shape, size, and inner phase distribution for a single particle with a size of tens to hundreds of nanometres, which may lead to a simple understanding of mesoscale ferroelectricity.

This work was partly supported by JSPS Grant-in-Aid for Scientific Research (Grant Nos. JP19H02618, JP18H03850, JP18H05518, JP19H05819, JP19H05625) and The Murata Science Foundation.

Full author list: N. Oshime, K. Ohwada, K. Sugawara, T. Abe, R. Yamauchi, T. Ueno, A. Machida, T. Watanuki, S. Ueno, I. Fujii, S. Wada, R. Sato, T. Teranishi, M. Yamauchi, K. Ishii, H. Toyokawa, K. Momma and Y. Kuroiwa.

[1] T. Hoshina, J. Ceram. Soc. Jpn. 121, 156 (2013). [2] M. Yamauchi, R. Ikeda, H. Kitagawa, and M. Takata, J Phys C, 112, 3294 (2008). [3] K. Ohwada, K. Sugawara, T. Abe, T. Ueno, A. Machida, T. Watanuki, S. Ueno, I. Fujii, S. Wada, and Y. Kuroiwa, Jpn. J. Appl. Phys. 58, SLLA05 (2019). [4] S. Marchesini, H. He, H. Chapman, S. Hau-Riege, A. Noy, M. Howells, U. Weierstall, and J. Spence, Phys. Rev. B 68, 140101 (2003). [5] N. Oshime, K. Ohwada, K. Sugawara, T. Abe, R. Yamauchi, T. Ueno, A. Machida, T. Watanuki, S. Ueno, I. Fujii, S. Wada, R. Sato, T. Teranishi, M. Yamauchi, K. Ishii, H. Toyokawa, K. Momma and Y. Kuroiwa, Jpn. J. Appl. Phys. (in press). https://doi.org/10.35848/1347-4065/ac148b

Interpretation and analysis of XFEL data can depend critically on a fundamental understanding of the characteristics of the XFEL pulses. To exploit the unique repetition rate of the EuXFEL requires understanding of both the inter- and intra-train fluctuations in pulse fluence, spatial energy distribution, coherence and wavefront, and beam pointing, which are frequently implicated in the loss of information in XFEL single particle imaging (SPI) and other classes of coherent diffraction experiment. Failure to account for fluctuations of the electron bunch phase-space and/or trajectory within a pulse train can result in deviations of the recorded wavefront, intensity statistics and intensity integrals from theoretical behaviour.
Preliminary X-ray optical data collected at the SPB-SFX instrument of the European XFEL demonstrates a sensitivity of inter- and intra-train variations in beam pointing to different beam delivery parameters (pulses-per train). We present this data alongside a model of the SASE1 photon beam. A partially coherent wave optical simulation of the beam propagated from the undulator exit to the SPB-SFX instrument hutch is compared to experimental data collected in the same plane. Also discussed will be the design of wavefront measurement methods that can be made for comparison with theory. Moving forward, we outline a novel method for investigating the relationships between the statistical behaviour of the XFEL source (including inter- and intra-train jitters) and the optical (wavefront) properties observed at the instrument to extend these observations.

Coherent diffraction imaging (CDI) allows for sub-wavelength spatial resolution by reconstructing an image from recorded diffraction patterns using a phase-retrieval algorithm [1, 2]. CDI is e particularly advantageous in the extreme ultraviolet (EUV) and X-ray ranges, where optics manufacturing is difficult and expensive [2]. Ptychography, the scanning version of CDI, has several benefits, such as large field of view imaging and robustness. The sample (object) is scanned by moving a spatially confined illumination (probe) while ensuring overlap in the illuminated regions [3]. The complex object function is typically retrieved by using an iterative algorithm that relies on two constraints [4]. First, the real space (or overlap) constraint assumes that the exit wave leaving the sample is formed by the probe function multiplied with the object function, i.e. the thin object approximation [3]. Second, the Fourier constraint enforces the estimated diffraction pattern intensity to match the measured diffraction data.

To optimize the reconstruction procedure, additional constraints have been suggested, based on a priori knowledge of the object and the measurement system. For example, Guizar-Sicairos et al. [5] have proposed a statistical optimal reconstruction procedure that finds the solution to the ptychographic problem by a least-squares approximation of the maximum likelihood function. Alternatively, an approach by Katkovnik et al. [6] that uses a sparse approximation of the probe and object function additionally to a maximum likelihood technique, to improve the reconstruction quality compared to a non-optimized algorithm. Recently, Ansuinelli et al. [7] have directly used the sample’s layout information to build an optimal reconstruction algorithm for imaging a photolithography mask, by penalizing the deviation of the reconstructed mask image to a full mask model.

We present here a phase-retrieval algorithm similar to Chang et al. [8] and Enfedaque et al. [9] that solves the blind ptychography problem (retrieving the probe and object) using total variation regularization (TV) as an additional constraint on the object function. TV promotes a sparse object gradient and is therefore preferential for (quasi) binary structures, removing noise and image artefacts [10]. We will discuss the total variation based algorithm for EUV photolithography mask inspection and show the impact of the algorithm for reconstruction of simulated and experimental data.

[1] Chapman, H. N. & Nugent, K. A. (2010). Nature Photonics, 12, pp. 833 – 830.

[2] Gardner, D., Tanksalvala, M., Shanblatt, E.R., et al. (2017). Nature Photonics, 11, pp. 259 – 263.

[3] Rodenburg, J. & Faulkner, H. (2004). Appl. Phys. Lett., 85, pp. 4795 – 4797.

[4] Thibault, P., Dierolf, M., Bunk, O., et al. (2009). Ultramicroscopy, 109, pp. 338 – 343.

[5] Guizar-Sicairos, M. & Fienup, J. (2008). Optics Express, 16, pp. 7264 – 7282.

[6] Katkovnik, V. & Astola, J. (2012). J. Opt. Soc. Am. A, 30, pp. 367 – 379.

[7] Ansuinelli, P., Coene, W. M. J. & Urbach, H. P. (2020). Applied Optics, 59, pp. 5937 – 5947.

[8] Chang, H., Enfedaque, P. & Marchesini, S. (2019). 2019 IEEE International Conference on Image Processing, pp. 2931 – 2935.

[9] Enfedaque, P., Chang, H., Krishnan, H., et al. (2018). Computational Science – ICCS 2018, pp. 540 – 553.

[10] Rudin, L. I., Osher, S. & Fatemi, E. (1992). Physica D, 60, pp. 259 – 268.