A. The monochromator

Spatially coherent X-ray is produced by an elliptical polarized undulator (EPU), and monochromatized by a plane grating monochromator (PGM), as shown in Fig. 1. The sagittal cylinder mirror, M1, is to collimate the photon beam in the vertical direction, absorb the heat load, suppress the high order harmonics, and deflect the beam in the horizontal direction. To protect its surface from the high heat load, M1 is designed with an internal water-cooling scheme. The core instrument is a vertically dispersing PGM equipped with a double-ruling-area grating to guarantee mechanical accuracy during grating switching. Two Au-coated gratings are used: 800 l/mm (optimized for 250–750 eV) and 1200 l/mm (optimized for 275–2000 eV). The last optical element is the toroidal mirror, M3, which refocuses the beam simultaneously in horizontal and vertical directions at the exit slit. After the exit slit, the STXM is setup to refocus the monochromized beam to 30 nm with a zone plate.

Fig. 1.Full-screen View

(Color online) The layout of the BL08U1A beamline at SSRF.

The energy resolution was evaluated by measuring the shell excitation spectra of Ar and N₂. As shown in Fig. 2, the fine structure of 7d peaks can be seen clearly. By fitting the peaks with Voigt profile, the energy resolution is 10000 at Ar edge and 7400 at N edge. Also, the spatial resolution was measured by imaging a standard test sample, which has a fine pattern of 30 nm. As shown in Fig. 3, the innermost 30 nm lines can be seen clearly, demonstrating that the spatial resolution is better than 30 nm. The beamline specifications were listed in Table 1. The details of these measurements were described in Ref. [8].

Fig. 2.Full-screen View

The shell excitation spectra of (a) Ar and (b) N₂ gases, using the first grating (800 lines/mm groove density) with cff=1/2.5.

Fig. 3.Full-screen View

STXM images of a test pattern sample, showing 30 nm spatial resolution with a scanning range of 5 μm×3 μm in 15 nm steps. The innermost circle demonstrates a spatial pattern of 30 nm.

B. The STXM endstation

The STXM endstation, purchased from Xradia, consists of a beam focusing system, a sample scanning system, a high-speed proportional detector, a sample tank, a control system and a graphical user interface. The focusing system, sample tank and detector are installed in a vacuum chamber. The working condition of STXM is either vacuum of 1.33×10^-3 Pa (d-5 Torr) or helium environment. The vacuum chamber is separated from the beamline vacuum tube by a 100 nm thick silicon nitride window. As shown in Fig. 4, a Fresnel zone plate (FZP) is used to focus the spatially coherent soft X-rays into a nano-probe, which scans the sample point by point to form a complete image. In most cases, the image resolution depends on the focus spot size.

Fig. 4.Full-screen View

(Color online) Schematics of the STXM with a divergent beam downstream of the exit slit. The FZP produces a diffraction-limited focus, the OSA selects only the first-order focus, and the detector measures intensity of the photons transmitted from the sample. The picture is from CLS SM manual (http://www.lightsource.ca/sm/Pages/SM-Home.aspxhttp://www.lightsource.ca/sm/Pages/SM-Home.aspx).

The sample is placed at the first-order focalspot of the zone plate, which is equivalent to a convergent lens. The minimum size of the focal spot depends on the outermost width of the zone plate. An FZP with a central stop in combination with an order-sorting-aperture (OSA) is designed to produce the 1st order focal spot on the sample, and minimize the higher order diffraction and direct lights on the sample. The sample is moved by the sample stage (Sample X and Sample Y) and Piezo stage (Piezo X and Piezo Y), and rotated by a rotation stage in sample plane. Transmitted X-rays from the sample are detected by a photomultiplier (PMT) detector. A sample can be scanned point by point by its two-dimensional movement (X and Y directions). The scanning precision of piezo stages is about 5 nm, and the scanning range can be selected from several microns to several hundred microns. A laser interferometer system is used for high-precision position measurement of zone plate and sample in X and Y directions. The piezo stages are scanned / positioned based on the differential position of sample and zone plate, to guarantee the highest possible X-ray-optical resolution. Figure 5 shows the STXM chamber and components inside.

Fig. 5.Full-screen View

(Color online) The STXM chamber and the components inside.

A. Point spectrum scanning

The sample location is fixed. By scanning the photon energy within a certain range near the absorption edge of a specific element in the sample, one obtains the characteristic absorption fine structure of the element. And chemical state information of the elements can be obtained by comparing the sample spectra with standard spectra of various compounds or single elements with fitting analysis. There are two ways for point spectrum scanning—focus scanning and large spot scanning.

B. Energy stack scanning

Two-dimensional scanning transmission imaging is performed at each energy point for a series of energies near the absorption edge of the interested element, obtaining N absorption contrast images. Post-analysis and processing are performed on the data of the N images by using principal component analysis and cluster analysis. The information of chemical constituents of the element is acquired, including composition of the near edge absorption spectra, and 2D spatial distribution of the constituents. The stack data processing software can be obtained free online (http://xray1.physics.sunysb.edu/data/idl programs/stackanalyze.savhttp://xray1.physics.sunysb.edu/data/idl programs/stackanalyze.sav; http://xray1.physics.sunysb.edu/data/idl programs/pcagui.savhttp://xray1.physics.sunysb.edu/data/idl programs/pcagui.sav). In stack data analysis and processing, the drift of images along the X/Y directions is very common, hence the use of image correlation maximization method and the multi-window correlation maximization method to perform image registration and trimming.

C. Dual-energy contrast imaging

This experiment method for STXM was developed after SSRF’s opening to users. The main steps are as followed. For an element of interest, select the energy at the absorption edge and another energy just above the edge, and perform two-dimensional scanning transmission imaging of the sample, respectively. Comparing and processing the two images, 2D spatial distribution of the element can be determined semi-quantitatively. Details about this method were described in Ref. [9].

D. TEY method

TEY is an indirect method to measure X-ray absorption near-edge spectra by detecting the electric current caused by secondary electrons [10]. The advantage of this method is independent of sample thickness and its signal-to-noise ratio is much higher than the transmission mode. TEY detects interfacial properties of the sample, rather than bulk information using the transmission mode. TEY requires that the sample be conductive or prepared on conductive substrates. In addition, the magnetic field can be easily added to study the magnetic samples using this method.

E. Nano-CT technique

Three-dimensional microscopic imaging is of importance in such research areas as cell biology and environmental science, where 3D sample imaging is needed to know more detailed structural information. For example, the ferroferric oxide distribution in human cervical carcinoma samples can be detected with STXM, but this does not tell where it is located (on the surface, or in the cell) and how it is distributed along the depths [11]. We combined STXM and computed tomography technique to develop the Nano-CT technique, which obtains, non-destructively, both 3D elemental mapping and chemical information. The Nano-CT technique in identification capability of 50 nm 3D structures has been opened to users.

F. SXEOL technique

SXEOL monitors optical luminescence (UV to near IR) emitted from the sample excited by soft X-ray photons in energies across the absorption edge of interest. Combining with NEXAFS spectra, it can provide elemental, chemical and site-specified information under favorable conditions. The total (zero order) or partial (selected wavelength) photoluminescence yield (PLY) is often used to determine the origin of luminescence. BL08U1A operates in the soft X-ray regime (250–2000 eV), which covers the so-called "water window". This offers an opportunity of checking the contributions from low atomic number elements, such as carbon, nitrogen and oxygen, which are not accessible with hard X-rays. Most importantly, at these energies the X-ray attenuation length is sufficiently short, allowing probing surface, interfacial, and/or bulk regions of the sample. At present, a concurrent collection of SXEOL and TEY NEXAFS spectra has been realized at SSRF. Figure 6 shows the schematic diagram and photo of the SXEOL setup (top view) in the vacuum chamber. The monochromatized X-ray photons impinge on a sample at an angle of 45. The lights emitted by the sample are collected by a collimating lens mounted at the end of an optical fiber, and transferred to the spectrometer by the optical fiber. A software triggers the spectrometer to collect a SXEOL spectrum for every energy in the energy scanning.

Fig. 6.Full-screen View

(Color online) Schematics diagram (a) and photo of the SXEOL (b).

The spectrometer has replaceable slits, a composite grating HC1 with a bandwidth of 725 nm, fairly enough for the synchrotron radiation application purposes. A commissioning of the system was carried out, and measurements on single crystal ZnO were taken at oxygen K edge. Two main features were observed from the SXEOL spectra, i.e., a broad intense defect emission band at 531.3 nm and a weak band-gap peak at 380 nm. Wavelength selected (531.3 nm and 380 nm) and zero order PLY spectra were shown in Fig. 7. The PLY spectrum at 380 nm was averaged between 377 nm and 383 nm due to very weak luminescence. A TEY spectrum is shown for comparison. All the spectra were normalized to the incident photon flux collected on a gold-coated silicon wafer. The zero order PLY is the same as the PLY spectrum taken at 531.3 nm. It is interesting to note that the PLY spectra, being similar to the TEY, exhibit a decrease of the feature at 539.4 eV relative to that at 535.8 eV, agreeing well with the literatures [12].

Fig. 7.Full-screen View

PLY spectra recorded at 380 nm and 531.3 nm, together with the zero order PLY and TEY spectra [13].

G. X-ray CDI methodology

The CDI is a novel and powerful lensless imaging method for imaging materials and biological specimens [14, 15], with spatial resolution achieving 3 nm [16]. Since its first demonstration in 1999 [17], this method has been advancing rapidly through the use of synchrotron radiation, X-ray free-electron lasers, and tabletop soft X-ray lasers. In CDI, the diffraction pattern of a noncrystalline specimen is measured in the far field, and its phase is then directly retrieved to obtain an image of the specimen using iterative algorithms in combination with the oversampling method [18]. This method overcomes the shortcoming of traditional X-ray microscopes, where the resolution is limited by the X-ray focusing element. Theoretically, resolution of this method is limited by just the X-ray wavelength and the attainable maximal diffraction angle, hence the potential possibility of reaching atom level resolution and realizing molecule imaging [19]. We have carried out the algorithmic and experimental study of CDI methodology, and achieved some good results in plane-wave CDI, scanning CDI and broadband CDI [20-22]. These pave the ways for our implementation and application of this newly emerging ultra-resolution imaging method.

The newly emerging scanning CDI method (also called ptychography) is the most suitable CDI method for STXM setup [23-26]. It allows samples of unlimited lateral extent to be imaged by scanning a localized illuminating probe wavefront across the transmissive sample and recording the multiple diffraction patterns at a set of partly overlapping illuminated regions, overcoming the disadvantages of the classic plane-wave CDI such as the rigid size limitation and isolation requirement on samples, slow convergence, stagnation, and the nonuniqueness of the solution. Therefore we have been developing the scanning CDI as our emphasis.

A scanning CDI experimental rig was installed at the STXM endstation (Fig. 8), and primary scanning CDI methodology experiments using a star pattern as the sample were carried out. A number of pretty good diffraction datasets were obtained, and the scanning CDI images (Fig. 9) were reconstructed using scanning CDI reconstruction software we developed. In the reconstructions, the spokes of star pattern can be distinguished distinctly, in both absorption (amplitude) and phase-shift images. The 30 nm spokes can be seen clearly. The probe (illumination) amplitude and phase were also reconstructed successfully. The highest resolution is about 10 nm, derived by a power spectral density analysis of the reconstructed image (Fig. 9(b)).

Fig. 8.Full-screen View

(Color online) The CDI experiment platform (a) in the STXM chamber, and the square cone mask covering the CCD camera (b) which reduces the noise-to-signal ratio significantly.

Fig. 9.Full-screen View

(Color online) Scanning CDI images using a star pattern as specimen. The illumination spot is 3 μm in diameter and the scanning step is 800 nm. (a) a typical diffraction pattern recorded by CCD; (b) reconstructed amplitude image, (c) reconstructed phase image, (d) reconstructed amplitude image of the illumination probe, (e) STXM image of the star pattern, the red square marks the region scanned in the ptychographic CDI experiment.

In developing the CDI reconstruction code, we implemented a number of image reconstructing algorithms of multiple CDI methods using Matlab language, including [(a)] t t

(a) Plane-wave CDI: Hybrid input-output (HIO), Error reduction, Difference Map, Relax averaged alternate reverse (RAAR), Guided-HIO, Oversampling smoothness, Shrinkwrap, Noise-robust HIO, etc., and the source codes were perfected gradually to improve the efficiency and accuracy;

(b) Scanning CDI: Ptychographic iterative engine (PIE), extended PIE, parallel PIE, position-correction PIE, Nonlinear optimization (NLO), etc., and tested by simulation and experiment data, in which the position-correction PIE (pcPIE) and NLO are the newest algorithms that could correct the position errors;

(c) Broadband CDI algorithm was implemented and tested by simulation and experiment;

(d) Fresnel CDI reconstruction algorithm was realized, and the Ewald sphere space transformation was used for large-angle CDI;

(e) Data preprocessing programs were finished and tested, such as background subtraction, diffraction stitching, diffraction center finding, beam stop area determination, etc., which are necessary for disposing experimental data before CDI reconstruction;

(f) Dual energy CDI ratio contrast element mapping method was implemented. Except that the detector is a CCD camera and the sample is not necessary to be placed at the zone plate focus, the setup of ptychography is almost the same as the STXM. Therefore, the sample preparation scheme and the sample thickness range are fully equal to those of STXM. This newly developed technique will be opened to the users soon after an improvement of the automatic control system of the CCD synchronized triggering and an optimization of the experimental parameters and operations.

A. Nanomaterials 2D imaging and absorptions

Using STXM technique, the sample can be imaged, while element absorption spectra can be obtained by stack analysis, i.e., analyses of a series of images at different energies near the edge of certain element. As shown in Fig. 10, Peng et al. studied graphene-templated 2D single crystal lepidocrocite nanosheets (γ-FeOOH), an efficient catalyst to treat phenol in wastewater, and confirmed the formation of γ-FeOOH2D nanosheets [33].

Fig. 10.Full-screen View

(Color online) Stack analysis (a) for the stack-scanned image at energies around Fe L edges of γ-FeOOH nanosheets (yellow area) and nanoparticles (blue area) on rGO; and the Fe³⁺ contents in nanosheets (b) obtained by stack analysis of the Fe L edge absorption [33].

N-doping in CNTs has shown excellent performance as electrocatalyst support or even directly as metal-free electrocatalyst for oxygen reduction, which can be used in fuel cells or biosensors [34, 35]. However, due to the complicated situation in CNTs bundles, such as variations in helicity, defects, diameter, residues, and surface modification, further explorations are needed to understand key factors affecting performance of CNTs in different applications. Using XANES spectroscopy and STXM Xie et al. investigated N-doped carbon nanotubes (NCNTs) and their electronic structures, and state of iron atoms in the catalysis. The results are shown in Figs. 11 and 12. The iron exists mainly in the form of ferrocene in the carbon nanotube, but it is partly transferred into oxidation state in NCNTs. Encapsulated ferrocene residues in CNTs was revealed by STXM, which may help for the N₂ sealing [36].

Fig. 11.Full-screen View

(Color online) Fe L-edge XANES spectra (a) of NCNTs, CNTs and ferrocene, and N K-edge XANES spectra (b) of NCNTs in TEY and TFY modes. The inset shows the magnified spectrum at TFY mode [36].

Fig. 12.Full-screen View

(Color online) SEM image of CNTs (a); STXM image at the Fe L-edge (b); averaged Fe L-edge XANES spectrum of the sample (c); and color composite mapping of the residues (d), red: Fe residues, yellow, CNTs [36].

B. Biological sample imaging

STXM, of high spatial resolution (30 nm) and high sensitivity of chemical identification, is an excellent method to study biological samples, such as cell, virus and tissues [37, 38]. Zhang et al. studied the biotransformation of ceria nanoparticles (NPs) in cucumber plants, and found that CeO₂ NPs were stable under environmental and biological conditions [39]. NEXAFS showed that the root Ce contents are CeO₂ and CePO₄ while the shoots Ce contents are CeO₂ and cerium carboxylates. As shown in Fig. 13, some CeO₂ NPs were transformed to CePO₄ on surface of cucumber roots, which might be attributed to the enhanced dissolution of NPs at the nano-bio interface.

Fig. 13.Full-screen View

(Color online) TEM images (A and E) of the root cells; Ce mappings (B and F) of rectangle area in Panels A and E obtained by a ratio of 886 eV and 888 eV images; XAFS spectra (D and H) extracted from Panels C and G images, respectively. The black spectra are of the standard compounds and the colored spectra are the root samples. The vertical red dotted lines indicate the characteristic peaks of CePO₄ and the dash lines indicate the characteristic peaks of CeO₂ NPs [39].

In 2014, Chen et al. studied immunological modulation of nanomaterials [40]. The endohedral metallofullerenol, Gd@C₈₂(OH)₂₂, serves as a potential nanomedicine that can efficiently inhibit the growth and invasion of tumors with low toxicity. The immunological modulation is involved in the processes but little has been revealed about the mechanism. Figure 14 shows the STXM-revealed uptake process of Gd@C₈₂(OH)₂₂ and the distribution inside the macrophages, an important immune cell, and the key signal pathways involved in the immune activation and modulation by Gd@C₈₂(OH)₂₂. In details, macrophages can internalize a large number of Gd@C₈₂(OH)₂₂ and their immune activity is induced by activating the pathways of TLRs//MyD88/NF-B and NLRP3 inflammasomes that promotes the secretion of IL-1β and regulates efficiently the innate and acquired immunological responses.

Fig. 14.Full-screen View

(Color online) Internalization of metallofullerenol by macrophages in vivo and in vitro. (A) A Gd M-edge NEXAFS spectrum obtained from Gd@C₈₂(OH)₂₂ nanoparticles. (B) Soft X-ray STXM dual energy contrast images of Gd@C₈₂(OH)₂₂ internalized by a primary mouse peritoneal macrophage in vivo via analyzing two images obtained at 1189 eV and 1185 eV, which are above and below the absorption edge of Gd, respectively. (C) Soft X-ray STXM dual energy contrast images of time-dependent uptake of Gd@C₈₂(OH)₂₂ by primary mouse peritoneal macrophages and RAW 264.7 cell line in vitro. (D) ICP-MS quantification of time-dependent uptake in macrophages of primary mouse peritoneal macrophages [40].

C. Magnetic imaging and spectroscopy

BL08U1A provides circularly polarized or linear light for X-ray magnetic circular dichroism in X-ray absorption (XMCD) and X-ray magnetic linear dichroism in X-ray absorption (XLCD) experiments [28, 29]. The interaction between lattice, charge, spin, and orbital degrees of freedom is related directly and crucially to performance of electronic materials, displaying a rich spectrum of exotic phenomena. Cui et al. used a gate voltage to manipulate the orbital occupancy of (La, Sr) MnO₃ films in a reversible and quantitative manner [41]. The variation of orbital occupancy is effectively proved by X-ray linear dichroism (XLD) (Fig. 15). Positive gate voltage increases the orbital occupancy proportion and magnetic anisotropy that were initially favored by strain (irrespective of tensile and compressive), while negative gate voltage reduces the concomitant preferential orbital occupancy and magnetic anisotropy.

Fig. 15.Full-screen View

(Color online) Self-assembled structures in LSMO films (a), and soft X-ray absorption and linear dichroism (b) [41].

Also, using soft X-ray spectroscopy combining with X-ray magnetic circular dichroism, Sun et al. obtained STXM images of micro magnetic structures of Ni₈₀Fe₂₀ film, in circular, triangular and vortex shapes [42]. As shown in Fig. 16, the quantitative analysis results of magnetic vortex of STXM agree well with the simulation results.

Fig. 16.Full-screen View

(Color online) Comparison between experimental and simulated vortex magnetization. (A, B, C) STXM images of magnetic vortex state for perm alloy circle, square and triangle geometries, respectively, scan size: 2.0 μm; (D, E, F) Micro magnetic simulations of the magnetic vortex state for circular, square and triangle geometries, respectively. The color and the arrows in D-F indicate the direction of the in-plane magnetization component; (G, H, I) Mx profiles along the red circular in (A, B, C) (red hollow circle) and (D, E, F) (blue solid square dots) [42].

D. Liquid samples

Liquid samples can be investigated by STXM. The liquid sample is confined between two silicon nitride windows. Zhang et al. imaged interfacial micro- and nano-bubbles with the scanning transmission soft X-ray microscopy [43, 44]. They found that nanoscopic gas bubbles in heights of <100 nm and lateral size of several tens of nm to micrometers can exist on various surfaces immersed in water, with a lifetime of hours and even days. The astonishing stability of the nanobubbles contradicts classical thermodynamics. STXM showed the first images of submicro-bubbles at solid-water interfaces (Fig. 17). The length scale for stable bubbles is about 2.5 μm.

Fig. 17.Full-screen View

(Color online) A typical STXM image (a) and diameter distribution (b) of nanobubbles (b). The number of stable bubbles observed by the STXM experiments in α-CD, urea or carbon solution as a function of their diameters as shown in inserted picture [43].