Eagle XG® glass: Optical constants from 196 to 1688 nm (0.735–6.33 eV) by spectroscopic ellipsometry

Abstract

Corning® Eagle XG® glass is widely used in the manufacture of electronic displays. In a previous submission, the authors reported its optical constants from 230 to 1690 nm (0.73–5.39). This spectral range reflected the fact that at a thickness of 0.5 mm, the material had zero transmission below 230 nm. These previously reported optical constants relied on a simple curve fitting approach that consisted of low- and high-energy poles, Tauc-Lorentz (T-L), and Gaussian oscillators. While the model agreed with the experimental data over the selected spectral range, it showed a sharp decrease in absorption when extrapolated to shorter wavelengths, which seemed physically unreasonable. To improve upon this earlier model, the authors have now obtained transmission data down to 196 nm using thin (100 and 200 μm) samples of the glass. Using this data, the optical constants were modeled using a T-L oscillator, 2–3 Gaussian oscillators, and a pole in the infrared. The center energies of the Gaussian oscillators corresponded approximately to previously reported absorptions from iron and tin in silicate glasses, and the band gap of the T-L oscillator was fixed at 5.8 eV, corresponding approximately to previously reported values of the onset of intrinsic absorptions. The resonance energy of the T-L oscillator was initially fixed at 12 eV following published values for fused silica, after which it was allowed to vary. The optical constants for the 100 and 200 μm samples could be fit simultaneously using the same model, but slight adjustments to the amplitude and broadening of the Gaussian oscillators were required to fit the transmission data of standard 0.5 mm (500 μm) Eagle XG. The authors attribute this apparent difference in the optical properties of these nominally identical materials, which were manufactured at different sites, to differences in the concentrations of iron and tin. Iron is a trace impurity in Eagle XG, and tin is a minor constituent. These have a profound effect on the ultraviolet absorption of glass. For the simultaneous fit of the 100 and 200 μm thick samples, our approach gave an unweighted mean squared error (MSE) of 1.19. The same model adapted to fit the 0.5 mm sample gave an unweighted MSE of 0.97. These models yielded small, nonzero values of k at shorter wavelengths. In situations where the glass was not analyzed with the inclusion of transmittance spectra, the refractive index can be reasonably modeled over this entire spectral range using only a Sellmeier (pole–pole) model. This model showed an MSE value of 1.09, where, of course, k was null throughout it. This reflection approach may be useful where transmission data is absent.