The silicates that produce the 10 and 18 micron features appear to be broadly similar to terrestrial minerals such as Forsterite and Enstatite. Forsterite is Mg2SiO4 and has two Mg atoms associated with SiO4 tetrahedrons. Enstatite is MgSiO3, differing from Forsterite in that there is an Si--O double-bond and two Si--O single bonds rather than three Si--O single bonds in the silicate part and just the one metal ion per silicate unit, and so it has a different lattice structure than that of Forsterite. In both cases there is a range of minerals with different metal ions in the lattice. On Earth the main "competition" to Mg for the lattice sites is Fe, so there is a continuous range of Olivines, (Fe/Mg)2SiO4, from the pure Mg form Forsterite to the pure Fe form Fayalite (Fe2SiO4). Similarly there is a continuous family of minerals from Enstatite (MgSiO3) to Ferrosilite (FeSiO3) for different fractions of Mg and Fe in the material. These two families are called the Olivine and Orthopyroxene silicate mineral series.
Under terrestrial conditions Forsterite can react with quartz (SiO2) to produce Enstatite. In space there could be a wide variety of silicates formed depending on how much of the constituent elements (Mg, Fe, Si, O, Al) are available. On the other hand, while we see differences in the silicate features from object to object in the main the features are surprisingly stable.
It is thought that the Mg-based forms of the silicates are strongly preferred to the Fe-based forms when the materials form. The bulk composition of the Earth's crust is heavily dominated by the Mg-bearing silicates. Where pre-solar silicate dust grains can be isolated from meteorite material, these also tend to be Mg-rich silicates rather than Fe-rich silicates. However as I understand things the Mg/Fe ratio in the pre-solar silicate grains is not as weighted to the Mg forms as is the case for the Earth's crust and mantle. The Mg/Fe ratio for silicates in circumstellar dust shells is not known for the amorphous silicate component. For the crystalline silicates the feature positions are thought to vary depending on the Fe/Mg ratio, and where observations have been made they appear to show a significant fraction of Fe-rich silicates.
The silicates in stars appear to be different from the terrestrial silicates in some way that is not fully understood. We see clear evidence that the stellar dust opacity in the 2 to 8 micron range is much larger than that of any known terrestrial silicate grains. This effect is seen in the Figure above where the blue curve starts to deviate from the black one at about 5 microns, due to excess emission from the stellar dust grains compared to what the Olivines would give. This is a well-known problem in understanding astronomical dust, and is generally handled by assuming an ad-hoc "astronomical silicate" dust type is present even though we do not understand what the physics behind the difference is (or should that be the chemistry behind the difference?). It has been suggested that a combination of small iron grains and Mg-type silicate grains can reproduce the properties of the grains seen in space. This is a matter that is still being debated. It has also been suggested that simply having the more Fe-rich silicates will produce the requisite infrared absorption.
Thousands of examples of these silicate spectra are known. There appear to be some subtle variations in the silicate feature from object to object: for example the 10 micron feature comes in "narrow" and "wide" forms that differ because of additional emission around 11 to 12 microns in the "wide" form. (See the plot below which shows the ISO spectra of three bright M-type stars with prominent silicate features, for which there are significant variations in the 10 micron feature shape.) The oddity is that there does not seem to be a continuous change from one form to the other, instead there are the two distinct groups and not many examples of intermediate profiles. As far as I know these differences in th silicate feature shape have not been satisfactorily explained.
The plot below shows that there is a well defined "narrow" 10 micron silicate feature shape. Scaling the ISO spectra of four bright M-type stars to the same peak value produces a good match in the shape. Note, though, that the 18 micron feature still shows some variations in shape from object to object.
All these subtle (and sometime not so subtle) differences in the feature shapes between different stars reflects variations in the dust properties, but this is difficult to quantify from object to object.
Some objects also show narrow crystalline silicate features along with the amorphous silicate features. This adds yet another level of complexity to the interpretation of the spectra. But for many objects there is no sign of crystalline silicate features in the spectra from 5 to 25 microns.
In the interstellar medium there are definitely silicate grains present which produce 10 and 18 micron absorption features in the spectra of distant objects. These dust grains are thought to be very cold and they don't radiate any energy at wavelengths of 10 to 20 microns. It has been suggested that the interstellar silicate grains have a different 10 micron feature profile than what is observed for the silicates around evolved stars. If this difference is real it could indicate that the silicate grains in space are affected by their environment over time, and possibly they slowly change character under exposure to cosmic rays and ultraviolet radiation in the diffuse interstellar medium.
up to the top page
"bare" stellar spectra in the mid-infrared
the most common features: silicates and SiC
less common features: extreme carbon stars, the AlO/silicate complex, the "unidentified infrared" features, unusual silicate features
dusty HII regions, planetary nebulae, and related objects
unusual features: mixed chemistry sources, Wolf-Rayet stars, ice bands, featureless carbon-star spectra