Date Published: March 01, 2019
Publisher: International Union of Crystallography
Author(s): Xiaojing Huang, Hanfei Yan, Yan He, Mingyuan Ge, Hande Öztürk, Yao-Lung L. Fang, Sungsoo Ha, Meifeng Lin, Ming Lu, Evgeny Nazaretski, Ian K. Robinson, Yong S. Chu.
Combining multi-slice ptychography with multi-modality scanning probe microscopy reconstructs two planes of nanostructures separated by 500 nm with sub-20 nm lateral resolution, assisted by simultaneously measured fluorescence maps for decoupling low-spatial-frequency features.
Diffraction-based imaging techniques such as coherent diffraction imaging (Miao et al., 1999 ▸) and its scanning variant ptychography (Rodenburg et al., 2007 ▸) are being pursued to achieve the diffraction-limited spatial resolution beyond the focus size provided by X-ray optics. With low numerical aperture (NA) imaging systems, the achievable resolution is primarily limited by the X-ray wavelength and the maximum diffraction angle (Sayre et al., 1998 ▸). For achieving higher resolutions with high-NA imaging systems, the thickness effect starts to make an impact, since the corresponding depth of field decreases dramatically with the enlarged NA as λ/NA2. As a result, the sample thickness is required to be thinner than the depth of field in order to obtain diffraction-limited resolution, δr, defined by the Rayleigh criterion as 0.61λ/NA. Otherwise, the wavefront propagation effect becomes non-negligible inside the sample, and the interaction with the illumination cannot be modeled by a simple multiplicative operation, which makes the projection approximation invalid. The thickness limitation T has been analyzed to satisfy T ≤ A(δr)2/λ, where the scaling factor A varies from 0.5 to 4.88 in different theoretical estimations (Rodenburg & Bates, 1992 ▸; Chapman et al., 2006 ▸; Thibault et al., 2008 ▸; Jacobsen, 2018 ▸) and a recent empirical observation suggests A = 5.2 (Holler et al., 2014 ▸; Tsai et al., 2016 ▸).
The multi-slice X-ray ptychography measurement was carried out at the Hard X-ray Nanoprobe beamline, National Synchrotron Light Source II (Chu et al., 2015 ▸). The schematic of the MLL-based scanning probe microscope setup (Nazaretski et al., 2015 ▸, 2017 ▸) with an off-axis focusing geometry (Yan & Chu, 2013 ▸) is illustrated in Fig. 1 ▸(a). The coherent portion of the incident 12 keV X-ray beam was selected by a µm secondary source aperture. The filtered illumination propagated 15 m downstream and was focused by a crossed pair of MLLs with a 4 nm outermost zone width. A 53 × 43 µm aperture size covering of a full MLL aperture delivered ∼ photons s−1 into a nm spot (Yan et al., 2018 ▸). The sample was a zone plate structure fabricated with ∼450 nm-thick gold on a 500 nm-thick silicon nitride membrane. Nickel oxide particles were drop-casted on the rear surface of the membrane. The nickel oxide particles overlapped with the zone plate pattern on the front surface and formed a layered structure with 500 nm separation. Before the multi-slice ptychography measurement, a single-slice ptychography dataset was collected at a sample area with only the gold zone plate feature inside the field of view. The reconstructed wavefront was propagated to locate the focal plane, which is shown in Fig. 1 ▸(b). The sample was then moved to a location with both the gold zone plate feature and nickel oxide particles. The sample was also translated along the axial direction, and the front surface was 10 µm downstream from the focal plane to improve the overlapping condition with an enlarged beam (Bunk et al., 2008 ▸; Huang et al., 2017 ▸). The probe functions at the front and rear surfaces of the sample are shown in Figs. 1 ▸(c) and 1 ▸(d). Considering the depth of focus of the used MLLs is 3.9 µm, the divergence of the X-ray propagation over the sample thickness is negligible. After the X-ray measurement, the scanned area was surveyed with scanning electron microscopy (SEM). Figs. 1 ▸(e) and 1 ▸(f) are SEM images of the zone plate structure and nickel oxide particles. A noticeable pile of carbon accumulated over the X-ray scanned area on both surfaces. A 30 keV electron beam was used to penetrate the carbon layer to observe the gold feature underneath.
We present our work on multi-slice X-ray ptychography, resolving two slices of nanostructures axially separated by 500 nm, with sub-10 nm and sub-20 nm lateral resolutions from a single 2D scan with 5 s dwell time per point. The highly convergent X-ray beam focused by large-NA MLLs provides a fast-changing wavefront, which facilitates the depth resolution enhancement. The high focusing efficiency of MLLs, which delivers ∼1 × 109 photons s−1 onto the focal spot, is beneficial for collecting a high-spatial-frequency signal to further improve depth resolution. The synergy between multi-slice ptychography and multi-modality scanning probe microscopy offers opportunities to utilize information obtained from other imaging channels in order to assist in the isolation of low-spatial-frequency features. In this work, we demonstrated the use of simultaneously measured fluorescence maps to decouple two separated slices with different elemental compositions. We showed that having additional information or constraints on the sample can aid the depth resolution of the multi-slice ptychography method. The generalization of this approach in practical sample systems would rely on choosing adequate imaging channels to properly represent the structural properties of the specimen. With the potential of improving depth resolution in the multi-modality measurement scheme, multi-slice X-ray ptychography as a technique is expected to be a powerful tool for high-resolution 3D imaging of specimens with extended dimensions.