X-ray Nano-Analysis of Sub-100nm Particles Using EDS in Conjunction with SEM

Abstract

Practical considerations for the detection, characterisation and analysis of sub-100nm particles in the field emission SEM have been investigated using a sample consisting of Ag nano-particles on a PZT thin film substrate.The principle for lowering beam accelerating voltage is well established and can be used to constrain the electron-sample interaction and X-ray generation to a small volume. In dense materials, interaction volumes of less than 100nm are practical at less than 5kV [1]. To achieve good lateral resolution in the FEGSEM at these voltages, the probe diameter must not significantly degrade the analytical resolution so beam current can only be increased to a point where the probe diameter is still much smaller than the feature being analysed. Furthermore, X-ray yield drops off with reduced beam voltage so there is a limit to what count rate can be generated. Therefore, high detector solid angle is an important parameter. Even with the benefit of a 30mm 2 detector, in our experiments on Ag nanoparticles we achieved count rates of typically 2-3kcps and no change in detector type or electronics would have improved the count rate achieved. An SDD or Si(Li) detector with area 10mm 2 in the same position would detect less than 1kcps.Ag nano-particles in the size range 30-80nm have been analysed (Fig. 1). It has been found that drift of the sample can be substantial and exceed the size of the particle in a time much shorter than is practicable for analysis (<<60s). Therefore we used software that effectively tracked the movement of the particles and locked the position of analysis by periodic correction of the scanning coordinates. This allowed us to collect EDS spectra from the particles as well as building up a full drift-corrected spectrum image of the sample.Monte carlo simulations at 4kV (Fig. 3) showed that substrate X-rays would be expected for a 60nm thick particle. However, even spectra collected from the 80nm particles (Fig. 2) showed a substrate contribution whereas Fig. 3 suggests this would be unlikely. We therefore used a thin film analysis method [2] to measure particle thickness and found it to be in the range 4-20nm for particles of 40-125nm diameter (Fig. 4). Although these measurements are known to slightly underestimate the true thickness, it is clear that these nano-particles have shapes closer to squashed domes rather than spheres. Clearly the visible diameter of nano-particles does not necessarily provide an accurate indication of actual particle size and whether the interaction volume will be fully contained within the particle. However, electron penetration into the substrate can be used to obtain useful information about the morphology of particles.

Optimizing the SEM specimen preparation method for accurate microanalysis of carbon nanotube/nanocluster hybrids

“…Although EDS maps can be provided under this mode, it was performed at an incident electron beam of 5 kV to increase the counts ratio of C/Si as the thickness of CNTs is too small. 23 Because of the lowvoltage mapping and the low content of other elements in CNTs, the EDS mapping cannot finger out other elements. As the CNTs are only tens of nanometres, far less than the X-ray penetration depth, it is not suitable for elemental quantitative analysis of the sample.…”

“…In order to improve the map identification, the distribution of silicon elements is characterised by using the L lines. Although EDS maps can be provided under this mode, it was performed at an incident electron beam of 5 kV to increase the counts ratio of C/Si as the thickness of CNTs is too small 23 . Because of the low‐voltage mapping and the low content of other elements in CNTs, the EDS mapping cannot finger out other elements.…”