Banner for QUANTAX EDS on TEM

Large Collection Angle EDS Detectors for High-End STEM in Materials Research

Example: Semiconductor Structure

Single EDS detectors for transmission electron microscopes must meet very specific geometric constraints. A large solid angle for X-ray collection is not enough. High take-off angle, a suitable collimator shape or pole piece cover and adapted specimen holder geometry are vital for successful EDS analysis and help keeping shadowing and system peaks in check. A large oval detector area (e.g. 100 mm2) can support the optimization of all these parameters for certain situations. For other TEM geometries though, constraints posed by the cold trap and pole piece shape may only be possible to address by smaller detector areas (30 mm2 and 60 mm2). Only careful evaluation of each individual microscope geometry will ensure the correct detector choice. Bruker’s detector slim-line design helps in all cases described above.

The XFlash® 6T-100 oval, installed on an aberration corrected STEM, was used to evaluate the element distribution in a cross-section specimen prepared from a FinFET structure. (Data Courtesy: ACE, Fig. 1). For the correct evaluation of the chemical composition of this semiconductor structure it is particularly important to separate overlapping element lines, such as from Ni and Ti in the low energy range,  Hf, Si and W between 1.4 keV and 2 keV, as well as overlapping lines from Ge and W around 10 keV. Furthermore, residues from Ga ion milling and secondary fluorescence from the microscope interior and specimen holder can complicate the analysis. Bruker’s ESPRIT software provides the means to comfortably tackle these issues (Fig. 2). Overlapping element lines can easily be identified. Clear handling of background models and background subtraction helps to consider and identify each line present in a spectrum. System peaks can be identified routinely and excluded from quantification while correcting for their impact on quantitative results. The relative Cliff-Lorimer-factor and the absolute Zeta-factor method are available for quantification.

Here the Cliff-Lorimer method was applied to quantify the element distribution in the specimen. The theoretical Cliff-Lorimer factors used were calculated from ESPRIT's constantly updated atomic data base and the known detector-specimen geometry and detector material composition. The distribution of all elements, including nitrogen as well as hafnium and titanium, which are, per device design, present in thin layers, were analyzed quantitatively. With the resulting data, the quality of the structure can be assessed with nm precision and potential causes of device failure can be found (Fig. 1-3).

Fig. 1: Fin FET detail: Left: HAADF; middle: EDS raw data (for clarity not all elements are shown). Colors from overlapping elements add to white. The quantitative display on the right using pseudo colors, e.g. for Hf, is more informative. Data Courtesy: ACE.
Deconvolution all
Fig. 2: Separation of overlapping element lines in Esprit. Black: measured spectrum; colors: counts assigned; light gray: sum of overlapping counts, dark gray: background. Data courtesy: ACE.
Detail at overlay
Fig. 3: Element map of a device detail. Qualitative (for clarity, not all elements are shown) and quantitative element display of N and Ti overlayed onto the HAADF image. Data courtesy: ACE.

More information

Please consider the following peer reviewed publications for further applications of the high-solid angle oval detector in other areas of research:

[1] Direct atomic scale determination of magnetic ion partition in a room temperature multiferroic material L. Keeney et al., Scientific Reports 7, Article number: 1737 (2017), open access

[2] Individual heteroatom identification with X-ray spectroscopy R. M. Stroud et al., APL 108, 163101 (2016), open access