We present a laboratory study of silicate glasses of astrophysically relevant compositions including olivines, pyroxenes and melilites. With emphasis on the classic Si-O stretching feature near 10 microns, we compare infrared spectra of our new samples with laboratory spectra on ostensibly similar compositions, and also with synthetic silicate spectral data commonly used in dust modeling. Several different factors affect spectral features including sample chemistry (e.g., polymerization, Mg/Fe ratio, oxidation state and Al-content) and different sample preparation techniques lead to variations in porosity, density and water content. The convolution of chemical and physical effects makes it difficult to attribute changes in spectral parameters to any given variable. It is important that detailed chemical and structural characterization be provided along with laboratory spectra. In addition to composition and density, we measured the glass transition temperatures for the samples which place upper limits on the formation/processing temperatures of these solids in space. Popular synthetically-generated optical functions do not have spectral features that match any of our glass samples. However, the ~10 microns feature generated by the synthetic data rarely exactly matches the shape and peak position of astronomically observed silicate features. Our comparison with the synthetic spectra allows astronomers to determine likely candidates amongst our glass samples for matching astronomical observations.

Since their discovery, cosmic crystalline silicates have presented several challenges to understanding dust formation and evolution. The mid-infrared spectrum of IRAS 17495$-$2534, a highly obscured oxygen-rich asymptotic giant branch (AGB) star, is the only source observed to date which exhibits a clear crystalline silicate absorption feature. This provides an unprecedented opportunity to test competing hypotheses for dust formation. Observed spectral features suggest that both amorphous and crystalline dust is dominated by forsterite (Mg\_2 SiO\_4) rather than enstatite (MgSiO\_3) or other silicate compositions. We confirm that high mass-loss rates should produce more crystalline material, and show why this should be dominated by forsterite. The presence of Mg\_2 SiO\_4 glass suggests that another factor (possibly C/O) is critical in determining astromineralogy. Correlation between crystallinity, mass-loss rate and initial stellar mass suggests that only the most massive AGB stars contribute significant quantities of crystalline material to the interstellar medium, resolving the conundrum of its low crystallinity.