Atomic charge distribution in sodosilicate glasses from terahertz time-domain spectroscopy

Terahertz time-domain spectroscopy has been used to extract the light-to-vibration coupling coefficient for sodosilicate glasses, from which it was possible to calculate the variance in the distribution of uncorrelated charges. It was found that increasing the sodium content of the glasses increased the standard deviation of the uncorrelated charge distribution in a linear fashion, and was almost an order of magnitude higher when compared to the charge distribution for pure silica reported previously, in agreement with previously published simulation data.

Figure 1

(Color online) (a) Absorption coefficient data for the three NS glass samples between 0.1 and 3 THz. The inset shows the NS2 glass data at low frequencies where the black dashed line represents the smoothed absorption data used for the later analysis of the coupling coefficient. (b) Refractive index data for the three NS glasses between 0.1 and 3 THz.

Figure 2

(Color online) A plot of the coupling coefficient $C\left(\omega \right)$ extracted for the three glasses as a function of frequency between 0.1 and 1.2 THz, along with the coupling coefficient data for ${\text{SiO}}_2$ published previously (Ref. 5). The dashed lines represent the fits to Eq. (1) calculated from these data using ${\beta }_0=3.4$. Only the data below the boson peak were included in the fit (NS2 and $\text{NS}3\lesssim 0.9\text{THz}$, $\text{NS}4\lesssim 0.6\text{THz}$).

Figure 3

The values of ${ϵ}_{\infty }$, $A$ and the standard deviation of the uncorrelated charge distribution, ${\sigma }_1$, as a function of Na concentration. The error bars represent the uncertainty in ${\sigma }_1$ as a result of the experimental uncertainty in the value of $A$ and ${ϵ}_{\infty }$. Filled circles (triangles in middle and right panels) are the values calculated in this paper for ${\beta }_0=3.4$ $\left({\beta }_0=2\right)$, and the open circle is the value calculated for ${\text{SiO}}_2$ previously (Ref. 5). The straight lines through the points represent best linear fits to the data.