Vibrational spectroscopy is based on the principle that vibrational motions occur within a crystal lattice at frequencies that are directly related to crystal structure and elemental composition (i.e., mineralogy) [e.g., Wilson et al., 1955; Farmer, 1974]. The fundamental frequencies of geologic materials typically correspond to wavelengths greater than ∼5 μm, and provide a diagnostic tool for identifying virtually all minerals.
 An extensive suite of studies over the past 40 years has demonstrated the utility of vibrational spectroscopy for the quantitative determination of mineralogy and petrology [e.g., Lyon, 1962; Lazerev, 1972; Vincent and Thompson, 1972; Farmer, 1974; Hunt and Salisbury, 1976; Salisbury et al., 1987a, 1987b; Salisbury and Walter, 1989; Bartholomew et al., 1989; Salisbury et al., 1991; Salisbury, 1993; Christensen and Harrison, 1993; Lane and Christensen, 1997, 1998; Feely and Christensen, 1999; Christensen et al., 2000a; Hamilton, 2000; Hamilton and Christensen, 2000; Wyatt et al., 2001; Hamilton et al., 2001]. The fundamental vibrations within different anion groups, such as CO3, SO4, PO4, and SiO4, produce unique, well separated spectral bands that allow carbonates, sulfates, phosphates, silicates, and hydroxides to be identified (Figure 3). Additional stretching and bending modes involving major cations, such as Mg, Fe, Ca, and Na, allow identification of specific minerals within the broad mineral groups. Significant progress also has been made in the development of quantitative models to predict and interpret the vibrational spectra produced by emission of energy from complex, natural surfaces [e.g., Conel, 1969; Henderson et al., 1992; Hapke, 1993; Salisbury et al., 1994; Moersch and Christensen, 1995; Wald and Salisbury, 1995; Mustard and Hays, 1997].
Figure 3. Thermal infrared spectra of representative mineral groups. Note the significant offset between the fundamental absorptions of the major mineral groups.
Download figure to PowerPoint
3.2. Mineral Groups
 The primary silicate minerals associated with igneous rocks are the most abundant mineral class found on Mars [Christensen et al., 2000b; Bandfield et al., 2000; Christensen et al., 2001a; Bandfield, 2002], and the ability to distinguish and quantify volcanic minerals, such as feldspars, pyroxenes, and olivines, is crucial to describing the fundamental geological character of the MER landing sites. All silicates have Si-O stretching modes between 8 and 12 μm that vary in position with mineral structure (e.g., Figure 4). This absorption shifts to higher frequency (shorter wavelength) as bond strength increases for isolated, chain, sheet, and framework tetrahedron structure. These shifts allow for detailed identification of the igneous silicates, including variations within solid solution series.
Figure 4. Examples of individual minerals within the silicate mineral group, illustrating the unique nature of these minerals over the mid-infrared spectral range. The individual spectra have been scaled and offset for clarity. These spectra are representative of the spectral sampling and radiometric precision that will be obtained from the Mini-TES. The emissivity minima in the absorption bands in these minerals vary from 0.5 to 0.8.
Download figure to PowerPoint
 Oxides, in particular gray crystalline hematite, have been found to be an important minor component on Mars [Christensen et al., 2000c; Christensen et al., 2001b; Bell and Morris, 1999], and provided a major impetus for the selection of the Meridiani Planum landing site for the MER-B rover. The crystalline hematite at Meridiani has been suggested to form by aqueous processes [Christensen et al., 2001b; Hynek et al., 2002], with possible pathways from goethite, magnetite, or other iron-rich precursors [Glotch et al., 2003]. The search for associated oxides or other precursor minerals will provide significant clues to the origin and history of Meridiani Planum.
 The evaporite mineral group includes carbonates, sulfates, chlorides, and phosphates that are precipitated from marine or non-marine waters. As such, they provide direct mineralogical evidence for standing water. The abundance of these minerals in a sedimentary basin is a function of the dissolved chemical constituents contained in the water, as well as the history of the basin inundation/denudation, and the identification and quantification of the different evaporite minerals can yield information about the environments in which they were produced. Thermal-infrared spectra provide distinguishing characteristics for the different groups, with carbonates, sulfates (gypsum) and phosphates (apatite) having deep, well-defined features in the 8.3 to 10 μm region that vary with position on the basis of composition [Lane and Christensen, 1998; Farmer, 1974].
 Hydrothermal systems produce characteristic mineralization that is dominated by microcrystalline quartz (chert, chalcedony, opal, etc.) and carbonates. Carbonates can precipitate in thermal spring environments, and provide evidence for wetting episodes in several Martian meteorites [e.g., McSween, 1994]. The fundamental C-O absorption occurs near 6.7 μm [e.g., Farmer, 1974; Nash and Salisbury, 1991; Lane and Christensen, 1997] in a region that is distinct from other mineral classes (Figures 3 and 4), and varies with cation composition [Lane and Christensen, 1997] (Figure 5a). A suite of rocks formed by hydrothermal precipitation and alteration in the Castle Hot Springs Volcanic Field of central Arizona is shown in Figure 5b. The travertine samples are characterized by the absorption features typical of carbonates (calcite). The hydrothermal silica spectrum exhibits the major absorption features between 8 and 10 μm that are characteristic of crystalline quartz. The basalt sample is a potential analog for hydrothermal alteration on Mars. It contains small (<1 mm) calcite-bearing vesicles and veins, similar to those found in the SNC sample ALH84001 [McKay et al., 1996]. The spectral feature between 6.0 and 7.1 μm (1600 and 1400 cm−1) is due to carbonate, demonstrating that a small amount of carbonate (<5%) can be detected in volcanic rocks using thermal-IR spectra.
Figure 5. Individual minerals and rocks associated with aqueous formation and alteration. The individual spectra have been scaled and offset for clarity. These coarse particulate mineral samples have band depth (emissivity minima) relative to the nearby local emissivity maxima of ∼0.4 to 0.8. (a) Selected examples of minerals in the carbonate mineral group. (b) Thermal emission spectra of hydrothermal rocks. The travertine and hydrothermal quartz samples are from a hot-spring system within a volcanic environment. The hydrothermally altered basalt sample contains small (<1 mm) calcite-bearing vesicles, demonstrating that <5% carbonate can be detected in volcanic rocks. (c) Selected examples of minerals in the phyllosilicate mineral group.
Download figure to PowerPoint
 Minerals that incorporate hydroxyl (OH)- anions into their structure give clues about the availability of water during their formation. The majority of such minerals are silicates, and most of these are in the phyllosilicate group (Figure 5c). The hydrous silicates have characteristic mid-IR features due to fundamental bending modes of (OH)-attached to various metal ions, such as an AL-O-H bending mode near 11 μm in kaolinite clay [e.g., Farmer, 1974; Van der Marel and Beeutelspacher, 1976]. Though all the hydrous silicates have the hydroxyl anion, they range widely in their mode of occurrence. Some form as primary constituents of igneous rocks, giving clues about the magmatic conditions under which the rock was formed. Most hydrous silicates crystallize as secondary products of aqueous alteration and their composition provides insight into the pressure and temperature where they formed.
3.3. Quantitative Analysis of IR Spectra
 A key strength of mid-infrared spectroscopy for quantitative mineral mapping lies in the fact that mid-infrared spectra of mixtures typically are linear combinations of the individual components [Thomson and Salisbury, 1993; Ramsey and Christensen, 1998; Feely and Christensen, 1999; Hamilton and Christensen, 2000]. The mid-IR fundamental vibration bands have very high absorption coefficients and therefore much of the emitted energy only interacts with a single grain. When absorption coefficients are low, as is the case for overtone/combination bands, the energy is transmitted through numerous grains and the spectra become complex, non-linear combinations of the spectral properties of the mixture. The linear nature of the thermal spectral emission of mineral mixtures has been demonstrated experimentally in particulates >60 μm in size for mixtures of up to five components [Thomson and Salisbury, 1993; Ramsey and Christensen, 1998]. In these experiments the mineral abundance could be quantitatively retrieved using linear deconvolution techniques to within 5% on average.
 The linear mixing of mineral components in rock spectra has also been confirmed [Feely and Christensen, 1999; Hamilton and Christensen, 2000; Wyatt et al., 2001; Hamilton et al., 2001], with retrieved mineral abundances that are accurate to 5–10% in laboratory spectra. Mineral composition and abundance were determined both spectroscopically and using traditional thin-section techniques for a suite of 96 igneous and metamorphic rocks [Feely and Christensen, 1999]. The rocks were used in their original condition; no sample cutting, polishing, or powdering was performed, and weathered surfaces were observed where available to best simulate remote observations. Comparison of the mineral abundances determined spectroscopically with the petrographically estimated modes for each sample gave an excellent agreement using high-resolution data. The spectroscopically determined compositions matched the petrologic results to within 8–14% for quartz, carbonates, feldspar, pyroxene, hornblende, micas, olivine, and garnets. These values are comparable to the 5–15% errors typically quoted for traditional thin section estimates.
3.4. Environmental Effects
 Variations in particle size and porosity produce variations in the spectra of materials at all wavelengths. Numerous quantitative models have been developed to investigate these effects [Vincent and Hunt, 1968; Hunt and Vincent, 1968; Conel, 1969; Hunt and Logan, 1972; Hapke, 1981, 1993; Salisbury and Eastes, 1985; Salisbury and Wald, 1992; Salisbury et al., 1994; Moersch and Christensen, 1995; Wald and Salisbury, 1995; Mustard and Hays, 1997; Lane, 1999] and have demonstrated the importance of specular reflectance and scattering. Two basic behaviors are observed with decreasing grain size: (1) strong bands (high absorption) tend to get shallower; and (2) weak bands (low absorption) increase in contrast, but appear as emission maxima and reflectance minima [Vincent and Hunt, 1968].
 Dust coatings and weathering rinds present a potential problem for any optical remotely-sensed measurements of Mars. However, the thickness of material through which sub-surface energy can escape increases linearly with wavelength. Thermal IR spectral measurements through coatings have been studied using mechanically deposited dust [Ramsey and Christensen, 1992; Johnson et al., 2002] and terrestrial desert varnish [Christensen and Harrison, 1993] as analogs to Martian rock coatings. These results have shown that thermal-IR spectral observations can penetrate relatively thick (mean thickness up to ∼40–50 μm) layers of these materials to reveal the composition of the underlying rock.
 Downwelling radiation is reflected off of the surface materials and this reflected component will be included in the total radiance received by the Mini-TES. The downwelling radiance will be measured directly using Mini-TES sky observations. It will also be modeled using MGS TES downward-looking observations of atmospheric temperature and dust, water-ice, and water-vapor abundances acquired simultaneously or at identical seasons from similar atmospheric conditions during previous Mars years [Smith et al., 2001b].