Summary

In this presentation I present the design, production, and characterization of 
multilayer mirrors intended as wavelength dispersive elements for the soft 
x-ray range. The central theme is the analysis and control of the different 
processes occurring at the materials interfaces between the layers. These 
interfaces largely determine the multilayer optical characteristics and form a 
continuous challenge to improve the multilayer performance, i.e. its optical 
reflectivity and wavelength selectivity. Both properties are of relevance for 
the main application motivating this particular work, namely X-Ray Fluorescence 
spectroscopy (XRF). XRF is an elemental analysis technique for which multilayer 
dispersive elements are employed. The results collected during this project, 
however, also deal with more general thin film phenomena of relevance for other 
thin single and multilayer systems in microelectronics and elsewhere.

The ability to freely design and synthesize multilayer mirrors has in the past 
led to experimental and theoretical approaches to increase the multilayer 
wavelength selectivity. This was done by modifying the ideal, block-shape, 
density profile of a multilayer of two pure materials with perfectly sharp 
interfaces into a profile with a certain gradient within the materials density. 
However, so far, no general theory has existed that includes the effect on the 
reflectivity of such modifications. I will show a theoretical proof, namely 
stating that the ideal, classical design in all cases leads to the highest 
reflectivity for any given selectivity. For this purpose, a Fourier theory 
about reflection in multilayer optics was used, taking into account the effect 
of absorption on the penetration depth of the radiation. However, the use of 
graded density multilayer systems does have the advantage that a certain target 
selectivity can be reached at slightly higher metal fractions in the bi-layer, 
which is favorable for the practical fabrication.

Obviously, in the experimental programme of fabricating these and other density 
profiles, it is imperative to avail oneself of a clear method to determine the 
actual profiles created in the samples. Although numerous analysis methods do 
exist, they either alter the structure prior to the analysis or require 
extensive knowledge of the structure beforehand. In this work a new technique 
has been developed, which allows extracting quantitative information on 
materials densities from cross-section transmission electron microscopy (TEM) 
images. Using TEM it was already possible to obtain a qualitative in-depth 
density profile of the structure. By combining this technique with x-ray 
reflectometry, this intensity profile can now be converted into a calibrated 
density profile. The technique is demonstrated for W/Si and Mo/Si multilayers 
with sharp interfaces as well as multilayers of which the interfaces were 
deliberately intermixed. The in-depth density profile, the layer thickness, 
the layer roughness and even the stoichiometry for each individual layer in 
the structure could accurately be determined.


The use of these and other analysis methods have revealed clearly the effects 
of different forms of ion treatment, commonly applied to improve the 
reflectivity of multilayer mirrors. Dependencies of film properties were 
investigated for respectively low and high energies of the ions impacting 
during or after deposition of W layers in W/Si multilayer mirrors. 
Analysis revealed that the addition of energy to the W layer resulted in a 
smoother surface of the W layer, and thus a sharper interface with the next 
deposited Si layer. This positive effect was (partly) counter balanced by 
ion-induced intermixing of the W-on-Si interlayer below, due to the penetration 
of ions through the thin (2 nm) top W layer. For higher ion energies it was 
observed that the intermixing at the latter interface increased, ultimately 
leading to a complete intermixing of W into the Si layer. Combining the effects 
at the Si-on-W and the W-on-Si interface, an optimal energy for the ion 
treatment was determined having a net positive effect on the reflectivity. 
Chemical affinity between Si and W was identified as a main reason for the 
intermixing.

The studies on this and other types of layer intermixing have led to the 
formulation of a method to control interface intermixing phenomena. It is based 
on H-passivation of the Si layer and the subsequent deposition of an atomically 
thin Si layer acting as an adhesion layer. Upon W-overcoat, the latter is 
converted into silicide of which the thickness is limited due to the finite 
amount of unpassivated Si. Using planar TEM analysis it was confirmed that 
deposited W chemically hardly interacted with the Si:H:Si layer underneath. 
The Si:H/Si/W multilayer mirror showed an 18 rel.% higher reflectivity than 
the conventional Si/W mirror.