TEMMeasurement has revealed for the twostudied glass samples the
NPs diameter D. The size is 25bDb40 nm for the copper glass, The average
size population has been represented in Fig. 13 for the silver NPs
glass.
We deduce an average diameter Dav=4.3 nm determined by a
Lorentzian fit of a monosize population of NPs, the size dispersion is
low (w≈1 nm). This evaluation is not sufficient to describe the nature
of the NPs because of the small size and the copper nature of
the microscope probe. The structure is identified by the fits of both
X-rays diffractograms for the more significant peaks. Table 2 presents
the parameters obtained by Scherrer's formula (Eq. (6)) allowing the
silver crystallites size determination.
The typical size of the crystallites forming the silver NPs is very
small, in the range of 2 to 4 nm which is lower but in correlation
with the TEM determination. Table 3 gives the parameters for the
copper crystallites identification.
The copper crystallites size estimate is in the range of 9 to 21 nm
which is lower than the TEM determination. We suggest that NPs are
compound of a core made of copper or silver crystallites and a shell
made of another material. In this last case, the structure of one compound
cannot be identified by the XRD method because of its too low
rate or its disordered structure composition. We assume that silver
and copper NPs are in a nanometer range. The staining of the glass
has direct consequence on oxygen diffusion too and can influence
thus the NPs composition by oxidizing.
The infrared measurement gives the refractive index value n
obtained by the fit of the imaginary part and evaluated in the range
of 1500 to 2500 cm-1 (Fig. 14) [21,26].
The n value is higher for the silver Glass than in the REF glass case,
the concentration of silver exchanged is about 1.27% at the surface.
The concentration of exchanged copper is about 3.46% at the surface
and the index value is higher than in the REF glass case too. With
this infrared method the NPs polarizability is not evaluated. The
ionic radius for copper is lower than the sodium one and this last
one is lower than the silver one (rCu+=73 pm ; rNa+=102 pm ;
rAg+=115 pm). It means that either the copper ion polarizability is
lower than the silver, or the rate of metallic NPs on total metal is
higher for copper. As a consequence, the sodium/copper substitution
during staining does not favor an increase of the polarizability
[27,28]. The interpretation of the silicate network evolution of the
glass is very sensitive because it may involve various SiO4 tetrahedral
species (Fig. 15a).
The frequency evolution of the various νi bonds indicate a good
definition of each of the vibration modes because of their stability
whatever the process. Frequency of the Na-O-Na bonds does not
change either but Na-O-Si bonds decrease under the metal ion introduction
effect, and a M-O-Si bond is created as a consequence around
250 cm-1 where M+ is associated to Ag+ or Cu+ ions. A checking of
the FWHM stabilization associated to each νi vibration frequency confirms
the good choice of the model (Fig. 15b).
Each contribution is width-stable except for Cu-O and the corresponding
Na-O links. A decrease is observed in the Cu+ introduction