Four-dimensional (4D) STEM-EELS

Chapter/Index: Introduction | A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | Appendix

Four-dimensional (4D) STEM-EELS

A technique known as 4D STEM-EELS, reported by Konrad Jarausch et al.[1], employed a rotation holder and pillar-shaped samples to capture high-angle annular dark-field (HAADF) and EELS images over a complete 180° rotation in STEM mode, effectively minimizing artifacts. This method produces a four-dimensional (4D) data set comprising two spatial dimensions, rotation angle, and energy-loss information I(x, y, θ, ΔE). The resulting data can be processed to extract EELS signals as rotation or “tilt-series” or “rotation-series” maps. When these extracted properties meet the linear projection criteria, they can be utilized for tomographic reconstruction, yielding volumetric maps of the corresponding properties. By integrating STEM HAADF with energy-loss data from such a series of spectrum images, it becomes possible to map not only the microstructure but also the elemental, physical, and chemical state information of a material in three dimensions.

This technique is demonstrated through two examples:

4D STEM-EELS was applied to chemically tomograph the 3D electronic structure of a W-to-Si contact in a semiconductor device. Core-loss data facilitated the three-dimensional reconstruction and rendering of the W-to-Si contact’s composition, with MLLS fitting of the fine structure at the 99 eV Si edge used to map variations in Si bonding in 3D.
The second example highlights the technique's ability to directly probe intrinsic material anisotropy, as 4D STEM-EELS was used on a ZnO thin film. Subtle, systematic changes in low-loss structure were detected as the electron beam orientation varied relative to the ZnO crystallographic axes.

The 4D STEM-EELS technique was made possible by integrating a high-brightness FE-STEM equipped with a rotation holder and a high-acquisition-rate EELS spectrometer. The use of pillar-shaped samples in conjunction with the 360° rotation holder provided two key advantages:

The ability to perform a full rotation, thereby eliminating missing wedge artifacts.
Maintaining a constant projection thickness at all rotation angles, which is particularly beneficial for EELS studies.

The acquisition rates exceeded 100 spectra per second, and it was feasible to capture large spectrum images (>200 × 200 pixels) within minutes. This efficiency allowed for the practical recording of spectrum images at regular angular intervals, with the total 4D data acquisition for each sample requiring only a few hours.

Note that in research settings, experiments can often take days or even weeks.

Sample Preparation for Tomography in TEM are described at page0004. For acquiring data at the detector's maximum speed, the STEM beam current can be selected to ensure that the intensity of the EELS spectrum would nearly saturate the detector in the most intense spectral region. [1]

In 4D STEM-EELS measurements, reported by Konrad Jarausch et al. [1], the EELS SI image recording in the 4D-STEM EELS was done at at 10° rotation intervals from 10° to 360°. The parameters of the EELS condition are listed at page4717. Once acquired, STEM-EELS data sets were combined, and acquisition artifacts were removed in a three-step process:

The individual SIs were merged into a single four-dimensional rotation-series data set I(x, y, θ, ∆E).
Remove the energy shift from the data.
Remove spatial drift between the successive spectrum-image acquisitions (page0005).

Figure 0001a shows the RGB composite image of the W-to-Si contact displays volumetric elemental distribution maps for Ti, N, and Co, which were obtained through tomographic reconstruction from the corresponding elemental map rotation series. The rotation series of elemental maps was derived from 4D STEM-EELS core-loss data sets using established EELS analysis techniques. The resulting segmented volumetric map from EELS can be queried locally to investigate site-specific property variations or analyzed for various 3D properties, including volume fraction and surface area.

RGB composite image of the W-to-Si contact displays volumetric elemental distribution maps for Ti, N, and Co, which were obtained through tomographic reconstruction from the corresponding elemental map rotation series

Figure 0001a. RGB composite image of the W-to-Si contact displays volumetric elemental distribution maps for Ti, N, and Co, which were obtained through tomographic reconstruction from the corresponding elemental map rotation series. [1]

Figure 0001b shows the volumetric chemical-state mapping of Si from the 4D STEM-EELS spectrum image rotation series and the reconstructed 3D image.

Figure 0001b. Volumetric chemical-state mapping of Si from the 4D STEM-EELS spectrum image rotation series. (a) Spectra extracted from the EELS core-loss data reveal variations in the Si L_3,2 edge shape specific to different chemical states. (b) MLLS fingerprint mapping over the 90–130 eV loss range generates a distribution map rotation series for each silicon phase (maps are shown with 601 steps for clarity). (c) Tomographic reconstruction of each map rotation series results in a 3D chemical state map, presented as an RGB composite using the color key from (a). [1]

[1] Konrad Jarausch, Paul Thomas, Donovan N. Leonard, Ray Twesten, Christopher R. Booth, Four-dimensional STEM-EELS: Enabling nano-scale chemical tomography, Ultramicroscopy 109 (2009) 326–337.

Practical Electron Microscopy and Database

An Online Book, Second Edition by Dr. Yougui Liao (2006)

Four-dimensional (4D) STEM-EELS