We performed measurements of SF5, SF6 and SF56A specimens using the Perkin-Elmer lambda9 system at the National Optical Institute of Florence. The resulting external transmission is plotted in the bottom panels of Fig. 6 where one can see that both glasses are basically transparent up to 1.35 m, i.e. up to the edge of the J atmospheric window. Absorption features appear at longer wavelengths, a relatively narrow and symmetric line centered at 1.43 m followed by much broader absorption bands which become particularly strong beyond 2.1 m.
Figure 6: Lower panels: measured external transmission of SF glass specimens, thickness t=21.7 mm. The dashed line represents the expected Fresnel's losses. Upper panels: derived absorption coefficients (internal transmission = ), the open circles are the values derived from the Schott catalog
The absorption coefficients are derived by normalizing the observed transmission to the level of the Fresnel's losses, and the results are shown in the upper panels of Fig. 6. These are in good agreement with the values derived from the internal transmission at laser wavelengths listed in the Schott catalog. The only significant discrepancy is for SF5 whose absorption coefficient at 1.53 m is similar to other glasses but a factor 2 larger than that reported by Schott.
Apart for the narrow absorption feature at 1.43 m (a wavelength where the atmosphere is basically opaque) the SF glasses considered here have negligible absorption, i.e. , up to 1.65 m (the center of the H atmospheric window) and up to 1.8 m, the long wavelength edge of the H band. In practice, this means that the fraction of light absorbed by the SF lenses in the cameras shown in Fig. 3 is 6% below 1.65 m and 12% between 1.65 and 1.8 m. These small losses are relatively unimportant in astronomical instruments such as fiber-fed spectrographs whose long wavelength transmission must be cut at about 1.7 m to avoid flooding the array with the thermal background from the warm parts of the instrument.
Beyond 2 m the SF glasses become quite opaque and cannot therefore be used in IR instruments extending to the K atmospheric window. However, the IR absorption bands are only due to water which is normally bound in the glass matrix, and can be eliminated by melting small cubicles of glass in vacuum environment, i.e. using the same fabrication process of FK54 and IR glasses (Knapp 1997). Clearly, this requires a special and expensive preparation which could be most conveniently supported by a large consortium of astronomers.
We are grateful to Konrad Knapp (Schott Glaswerke, Mainz) for many useful information on the SF and IRG glasses. We would like to thank A. van Dijsseldonk for helpful information, V. Castellini and D. Iafrancesco for their assistance with the Perkin-Elmer instrument, and A. Pecchioli (Gestione SILO, Florence) for preparing the SF specimens.