Symeon Aris Paysanidis Your challenges with EDS analysis on small Nb/V-rich precipitates are quite common in microanalysis, especially when dealing with nanoscale features. At 20 kV, the interaction volume may include surrounding matrix. Lower the voltage (5-10 kV), shorten the working distance, and consider elemental mapping instead of point analysis.

Thanks for your answer! The working distance was 7 mm and the accelerating voltage 20 kV, as we were advised to increase it. We’ve also done some measurements using 15 kV and an 11 mm working distance — I’m attaching those below.

From what I observed, increasing the voltage from 15 to 20 kV and reducing the working distance from 11 mm to 7 mm resulted in more EDS analyses showing clear presence of niobium and vanadium in each precipitate.

Hello,

as mentioned before the (electron) interaction volume and the X-ray fluorescence range may be larger than the size of the precipitates.

For a discussion see here: Maximizing Spatial Resolution

https://smf.probesoftware.com/index.php?topic=1354.msg9720#new

Lowering the beam energy will (i) reduce the size of the electron interaction volume, but (ii) will affect the X-ray generation because for lower beam voltages the generation of K-series would be impossible and you will have to use L-series peaks with lower accuracy.

The working distance should be set to maximum X-ray intensity for a main peak.

There are some analytical expressions for the electron range:

Luke, Keung, L., Choice of a range-energy relationship for the analysis of electron-beam-induced-current line scans, Journal of Applied Physics 76, 1081 (1994)

https://doi.org/10.1063/1.357827

K Kanaya and S Okayama, Penetration and energy-loss theory of electrons in solid targets, J. Phys. D: Appl. Phys. 5(1972) 43-58

https://10.1088/0022-3727/5/1/308

and

http://www.globalsino.com/EM/page4795.html

You can try to do some Monte Carlo simulation to learn the size of the excitation volume, e.g. using Casino

https://www.memrg.com/publications

https://www.memrg.com/programs-download

See the answer of @Warren Straszheim from 2025-04-01 here (for a layered system):

https://www.researchgate.net/post/EDX_EDS_detector_difference_in_quantification_using_10_kV_vs_5_kV_Why

EDS Simulation for an inclusion could be done with the 3D version.

The DTSA-II software has a simulation mode where inclusion in a bulk substrate can be simulated

https://www.cstl.nist.gov/div837/837.02/epq/dtsa2/index.html

Hello Aris,

The link below discusses using WD to vary image contrast. In this case they can use a shorter WD for spectrum acquisition as they have a new style of detector orientation:

https://www.oxinst.com/blogs/eds-at-the-imaging-sweet-spot

Hello Aris,

Probably covered by Stefan:

A moderately low beam current (0.22 nA) was used in order to reduce the beam spreading and maximize the spatial resolution.

I found this in my reading just now by chance:

axia-bend-failure-mech-an164.pdf

I think there are three issues, the first of which has already been covered.

i. The lateral beam spread (or drift) may be of the same size as your particle size, which seems ~0.5 µm. You can check with MC simulations and improve using either low kV or mapping with drift correction.

2. The depth penetration of your electrons can be larger than your particles so that a lot of the signal comes from the matrix BELOW the particles.

3. In same spectra you seem to have a strong Mo signal not included in your quantification. If that comes from an aperture strip, change to a Pt-Ir aperture instead, as low amounts of Nb (Z=41) are very difficult to detect when there is a strong Mo (Z=41) signal superimposed (both are near 2.2keV while Pt-Ir gives M-lines at lower 1.9-2.0keV).