Spectral and thermal properties of perchlorate salts and implications for Mars

Abstract

K⁺, Na⁺, Ca²⁺, Mg²⁺, Fe²⁺, Fe³⁺, and Al³⁺ perchlorate salts were studied to provide spectral and thermal data for detecting and characterizing their possible presence on Mars. Spectral and thermal analyses are coordinated with structural analyses to understand how different cations and different hydration levels affect the mineral system. Near-infrared (NIR) spectral features for perchlorates are dominated by H₂O bands that occur at 0.978–1.01, 1.17–1.19, 1.42–1.48, 1.93–1.99, and 2.40–2.45 μm. Mid-IR spectral features are observed for vibrations of the tetrahedral ${\mathrm{ClO}}_4^{-}$ ion and occur as reflectance peaks at 1105–1130 cm⁻¹ (~8.6–9 μm), 760–825 cm⁻¹ (~12–13 μm), 630 cm⁻¹ (~15.9 μm), 460–495 (~20–22 μm), and 130–215 (~50–75 μm). The spectral bands in both regions are sensitive to the type of cation present because the polarizing power is related to the band center for many of the spectral features. Band assignments were confirmed for many of the spectral features due to opposing trends in vibrational energies for the ${\mathrm{ClO}}_4^{-}$ and H₂O groups connected to different octahedral cations. Differential scanning calorimetry (DSC) data show variable patterns of water loss and thermal decomposition temperatures for perchlorates with different cations, consistent with changes in spectral features measured under varying hydration conditions. Results of the DSC analyses indicate that the bond energies of H₂O in perchlorates are different for each cation and hydration state. Structural parameters are available for Mg perchlorates (Robertson and Bish 2010) and the changes in structure due to hydration state are consistent with DSC parameters and spectral features. Analyses of changes in the Mg perchlorate structures with H₂O content inform our understanding of the effects of hydration on other perchlorates, for which the specific structures are less well defined. Spectra of the hydrated Fe²⁺ and Fe³⁺ perchlorates changed significantly upon heating to 100 °C or measurement under low-moisture conditions indicating that they are less stable than other perchlorates under dehydrated conditions. The perchlorate abundances observed by Phoenix and MSL are likely too low to be identified from orbit by CRISM, but may be sufficient to be identifiable by a VNIR imager on a future rover.

Introduction

Chlorine has long been known to be an important component of the martian regolith, but the molecular form of this element was unknown (e.g., Clark et al. 1977, 2005). One of the achievements of the Phoenix mission was to identify perchlorate as the predominate form of soluble chlorine in the regolith at the landing site (Hecht et al. 2009). Recent Phoenix Wet Chemistry Laboratory (WCL) measurements of soluble anions and cations indicate that the perchlorate anion is likely present at a level of 0.6 wt% as a combination of Ca(ClO₄)₂·nH₂O and Mg(ClO₄)₂·nH₂O in the soil (Kounaves et al. 2014). Remote sensing analyses of soil features in 12 trenches observed with Phoenix’s Stereo Surface Imager (SSI) instrument found a band at 0.967 μm consistent with hydrated salts such as perchlorate (Cull et al. 2010b). Remote sensing studies by the Compact Reconnaissance Imaging Spectrometer for Mars (CRISM) of the Phoenix site with 18 m/pixel ground sampling show water ice (Cull et al. 2010a), but no evidence of perchlorate. On the Curiosity rover, Sample Analysis at Mars (SAM) instrument results indicate the presence of Ca perchlorate at the Rocknest site in Gale Crater (Glavin et al. 2013; Leshin et al. 2013) and possibly Fe-perchlorates at the John Klein and Cumberland sites at Yellowknife Bay in Gale Crater (Ming et al. 2014).

At about the same time Phoenix identified perchlorates in the soil, orbital remote sensing analyses using thermal IR data led to the detection of a widespread region containing small chloride salt deposits with unusually high emissivity at shorter wavelengths in the southern highlands (Osterloo et al. 2008, 2010; Glotch et al. 2013). Most of these chloride-bearing outcrops occur in topographic lows and are attributed to brines formed in evaporitic environments. Anhydrous chloride salts are difficult to confirm on Mars using CRISM, but have been identified at sites such as Terra Sirenum by a weaker 3 μm band and an increasing slope from 1 to 2.6 μm (Murchie et al. 2009; Wray et al. 2009). Analysis of the stratigraphy at these sites indicates that chloride deposits overlay ancient phyllosilicate outcrops (Murchie et al. 2009; Wray et al. 2009; Glotch et al. 2010); in some cases there is evidence that the salt deposits were mobilized by aqueous processes (Glotch et al. 2010). Mixtures of halite (NaCl) and basalt minerals match the observed spectral slope characteristic of these chloride deposits, with the most favorable results for halite grains in the 63–180 μm range (Jensen and Glotch 2011). The relationship of chlorides, perchlorates, and other Cl compounds on Mars is unknown; however, they are observed together in cold desert sediments in the Antarctic Dry Valleys (Kounaves et al. 2010a) and thus could be related on Mars as well.

On Earth, perchlorate salts have been detected in the waters of ice-covered lakes of the Antarctic Dry Valleys, where they have been found to occur at 0.1–0.2 μg/L in many lakes but as high as 8 μg/L in one part of Lake Bonney (Jackson et al. 2012). Analyses of Atacama Desert samples revealed perchlorate in some regions up to levels as high as 42 μg/g (Parro et al. 2011). In rare cases that are limited to some Atacama nitrate deposits (Ericksen 1983), levels of naturally occurring perchlorate on Earth approach those found on Mars. Studies have shown that perchlorates on Earth are generally derived from reaction of atmospheric oxidants and Cl (Bao and Gu 2004; Michalski et al. 2004; Parker 2009; Kounaves et al. 2010b) and a similar process is thought to have occurred on Mars (Catling et al. 2010). However, photo-oxidation of Cl by oxide minerals in aqueous environments is another possible formation mechanism for perchlorate on early Mars (Schuttlefield et al. 2011).

Perchlorates have been identified on Mars by the WCL on Phoenix (Hecht et al. 2009; Kounaves et al. 2014) and SAM on Mars Science Laboratory rover (MSL) (Archer et al. 2013; Glavin et al. 2013; Leshin et al. 2013; Ming et al. 2014) and modeling of Viking, Thermal and Evolved Gas Analyzer (TEGA), and SAM results is consistent with perchlorate reacting with organics during thermal analysis (Navarro-Gonzalez et al. 2010; Glavin et al. 2013). However, the possibility of perchlorate-induced combustion of soil organics combined with the presence of terrestrial organic contamination in the Viking Gas Chromatograph/Mass Spectrometer (GC/MS) and MSL SAM instruments has so far thwarted definitive identification of soil organics on Mars (e.g., Biemann and Bada 2011; Ming et al. 2014). Additionally, lower oxidation state oxychlorine compounds related to perchlorate may directly react with organics present on the surface of Mars (Quinn et al. 2013).

Perchlorates and other chloride salts are highly deliquescent. The deliquescence of hygroscopic minerals such as chloride salts could allow for the occurrence of transient liquid water on Mars, even under the current arid climate (Renno et al. 2009; Davila et al. 2010; Gough et al. 2011) and perchlorates could represent a potential near-surface H₂O reservoir on Mars (Robertson and Bish 2011). The visible/near-infrared (VNIR) spectral properties of perchlorates are dominated by features due to H₂O, and differential scanning calorimetry (DSC) measures dehydration as well as phase changes. Because the number of H₂O molecules present in the unit formula for perchlorates determines the structure (Robertson and Bish 2010), the H₂O bond energies change as well as the spectral features and temperatures of dehydration. VNIR spectra are currently measured by CRISM in orbit around Mars and thermal properties were measured by TEGA on Phoenix. The purpose of this study is to present the spectral and thermal properties of several perchlorate salts to facilitate their detection on Mars.

Background

Structure and bonding of perchlorates

The perchlorate ion (${\mathrm{ClO}}_4^{-}$) forms stable complexes with several metals, especially in the absence of Cl⁻ or other halides; however, most are readily soluble in water (Michalski et al. 2013; Sowe et al. 2013). The metal perchlorate bond is largely covalent and usually weak (Michalski et al. 2013). Perchlorate ions can form structures with monodentate, bridging bidentate or chelating bidentate configurations (Olmstead et al. 1982; Horgan et al. 2013). The perchlorate ion is considered to be a strong oxidant; however, it is thermally stable at Mars temperatures. Although it is difficult to chemically reduce perchlorate in aqueous solution, it can be reduced under some conditions by Ru²⁺ to the chlorate ion (${\mathrm{ClO}}_3^{-}$) and by Ti³⁺ and a few other rare metals to Cl⁻ (Michalski et al. 2013; Sowe et al. 2013). Perchlorate is not readily formed from chlorate in solution because the thermodynamically favored disproportionation reaction of 4 ${\mathrm{ClO}}_3^{-}={\mathrm{Cl}}^{-}+3{\mathrm{ClO}}_4^{-}$ proceeds very slowly, even at 100 °C (Sowe et al. 2013). Typically, perchlorates are synthesized by electrolytic oxidation of chlorates. Perchlorate anions have tetrahedral coordination polyhedra and solid perchlorates are generally isomorphous with salts of sulfates and other tetrahedrally coordinated anions (Sowe et al. 2013).

Crystal structures of perchlorates vary depending on the number of H₂O molecules per unit cell (e.g., Robertson and Bish 2010, 2011). Each Mg²⁺ cation is octahedrally coordinated by some combination of bridging O²⁻ anions and H₂O. Through spectral analyses and symmetry considerations Ross (1962) determined that the ${\mathrm{ClO}}_4^{-}$ ions are connected to the metal cation via two O²⁻ anions. In the Mg(ClO₄)₂·6H₂O structure described by West (1935), every Mg²⁺ is bonded to six H₂O molecules. Robertson and Bish (2010) described the structures of multiple hydrated Mg perchlorates using a combination of an ab initio charge-flipping model based on X-ray powder diffraction data and subsequent fundamental-parameters Rietveld refinement. Analyses of Mg perchlorate at multiple hydration levels showed Mg²⁺ in octahedral coordination with H₂O molecules and ${\mathrm{ClO}}_4^{-}$ ions in different forms (Robertson and Bish 2010, 2011; Solovyov 2012).

(1) For the hexahydrate state: six H₂O molecules are connected to Mg²⁺ and the ${\mathrm{ClO}}_4^{-}$ ions are H-bonded to the H₂O molecules.

(2) For the tetrahydrate state: four H₂O molecules and two ${\mathrm{ClO}}_4^{-}$ ions are connected to the Mg cation with the large tetrahedrally coordinated perchloride ions positioned in a trans arrangement at opposite ends of the octahedron with space group C2.

(3) For the dihydrate state: two H₂O molecules and 4 ${\mathrm{ClO}}_4^{-}$ ions are connected to the Mg cation giving space group C2/m.

Models of the Mg(ClO₄)₂·4H₂O and Mg(ClO₄)₂·2H₂O structures from Robertson and Bish (2010) are shown in Figures 1 and 2. Skogareva et al. (2012) also published the crystal structure of the trivalent iron perchlorate, $\mathrm{Fe}({\mathrm{H}}_2\mathrm{O}{)}_6{]}^{3+}({\mathrm{ClO}}_4{)}_3^{-}\cdot 3{\mathrm{H}}_2\mathrm{O}$. That structure contains two Fe cation sites, both of which are bonded to six H₂O molecules each.

Figure 1.

Close-up of the Mg coordination polyhedra in Mg(ClO₄)₂·2H₂O and Mg(ClO₄)₂·4H₂O described by Robertson and Bish (2010) using a combination of an ab initio charge-flipping model (e.g., as used in Dyar and Gunter 2007) based on X-ray powder diffraction data and subsequent fundamental-parameters Rietveld refinement. Oxygen anions are red, Mg²⁺ is yellow, and H⁺ is purple. The O²⁻ anions that are not part of the water molecules are corners of perchlorate anions (see Fig. 2). (Color online.)

Figure 2.

Changes in the structure of Mg(ClO₄)₂·nH₂O perchlorate as thermal decomposition occurs from n = 4 to n = 2, demonstrating the fundamental structural rearrangements that occur. Mg(ClO₄)₂·4H₂O contains Mg octahedra, each with two perchlorate anions in the trans position. Structural rearrangement occurs in Mg(ClO₄)₂·2H₂O, which is composed of chains of Mg octahedra linked by perchlorate anions that share two of their four corners. (Color online.)

Low-temperature studies of Mg(ClO₄)₂·6H₂O found at least eight distinct types of H₂O and multiple types of H⁺ cations with variable charge densities below 108 K, where a phase transition occurs (White and Falk 1986). This indicates a complex low-temperature structure. Above this transition there are only two types of H⁺ cations that were assigned to inequivalent sites in the H₂O molecules (White and Falk 1986). All H⁺ cations became equivalent after a phase change at 273 K and then the perchlorate ions became increasingly disordered following another phase change at 325 K (White and Falk 1986). Experiments under low P_H2O showed that Mg(ClO₄)₂·6H₂O is the stable phase at 273 K and that it resisted dehydration at low temperatures and did not dehydrate until 323 K (Robertson and Bish 2010), consistent with the phase change observed at 325 K by White and Falk (1986). Robertson and Bish (2010) suggest that transitions occur with thermal decomposition in the Mg perchlorate hydrate system from a structure involving isolated polyhedra for Mg(ClO₄)₂·6H₂O, to isolated clusters for Mg(ClO₄)₂·4H₂O, to chains for Mg(ClO₄)₂·2H₂O, and an infinite framework for anhydrous Mg(ClO₄)₂.

Deliquescence and efflorescence of perchlorates

Perchlorates are highly hygroscopic and readily transform from a crystalline solid to aqueous state (deliquescence) under humid conditions. Recent experiments and modeling work provide constraints on the geochemical conditions governing deliquescence and efflorescence (transformation from a liquid state to a crystalline solid) of perchlorates (Chevrier et al. 2009; Davila et al. 2010; Marion et al. 2010; Gough et al. 2011). Modeling by Marion et al. (2010) indicates that Na and Mg perchlorate were more likely to have precipitated at the Phoenix landing site than Ca²⁺ perchlorate because Ca²⁺ would have preferentially formed calcite and gypsum. The eutectic points were modeled at ~239 K for NaClO₄·2H₂O, ~205 K for Mg(ClO₄)₂·6H₂O, and ~199 K for Ca(ClO₄)₂·8H₂O (Marion et al. 2010) using data from Pestova et al. (2005). However, re-analysis of the Wet Chemistry Lab data on Phoenix support Ca perchlorate as the dominant salt in the soil (Kounaves et al. 2012, 2014). Davila et al. (2010) modeled the temperatures and relative humidity (RH) on Mars and determined that many Cl⁻ salts would deliquesce on Mars, resulting in transient Cl⁻ brine solutions on Mars today.

Gough et al. (2011) studied the stability of Mg(ClO₄)₂·6H₂O, NaClO₄, and NaClO₄·H₂O over the range 223–273 K and variable RH conditions and found that their data were consistent with the modeling results of Chevrier et al. (2009). Mg(ClO₄)₂·6H₂O deliquesces at 55%RH at 223 K and at 42%RH at 273 K, while NaClO₄·H₂O deliquesces at 64%RH at 228 K and at 51%RH at 273 K. Similarly, Nuding et al. (2013) studied Ca(ClO₄)₂ and found that deliquescence occurred at 55%RH at 223 K and at 23%RH at 263 K. They also observed that higher hydration states of Ca²⁺ perchlorates are less deliquescent. Efflorescence occurred at much lower RH levels for perchlorates and was largely independent of temperature: ~19%RH for Mg(ClO₄)₂·6H₂O (Gough et al. 2011), ~13%RH for NaClO₄, (Gough et al. 2011), and ~16%RH for Ca(ClO₄)₂ (Nuding et al. 2013). This hysteresis effect could result in metastable, supersaturated perchlorate salt solutions on Mars today (Gough et al. 2011). The Na⁺ perchlorate solution only formed anhydrous NaClO₄ from solution and hydration of NaClO₄ is very slow. Thus, on Mars either anhydrous NaClO₄ salt or NaClO₄ solution is more likely to be present than a hydrated form of Na⁺ perchlorate (Gough et al. 2011). At ~250 K water activity levels increase for the Ca²⁺ perchlorate system from 0 for the anhydrous form to ~0.25 for Ca(ClO₄)₂·4H₂O and to ~0.4 for Ca(ClO₄)₂·8H₂O (Nuding et al. 2013).

Spectral properties of perchlorates

Perchlorate salts have been the focus of several transmittance IR studies (Miller and Wilkins 1952; Miller et al. 1960; Ross 1962; Nebgen et al. 1965; Gadsden 1975; Lewis et al. 1975; White and Falk 1986; Chen et al. 2004; Weitz et al. 2013). Hathaway and Underhill (1961) found that isolated ${\mathrm{ClO}}_4^{-}$ ions have Td symmetry (T2 modes are active), unidentate ${\mathrm{ClO}}_4^{-}$ coordination has C3ν symmetry (A1, E, and A1+E modes active), and bidentate ${\mathrm{ClO}}_4^{-}$ coordination has C2ν symmetry (A1, and A1+B1+B2 modes active). The perchlorate ion has bands near 1100 and 620 cm⁻¹ due to the T2, A1+E, and A1+B1+B2 modes (Ross 1962; Lewis et al. 1975; Weitz et al. 2013). Additional bands are observed near 930 cm⁻¹ for the A1 mode and near 500 cm⁻¹ for the E mode. Furthermore, the A1+E modes in unidentate perchlorates with C3ν symmetry show variation in the position of the 600–650 cm⁻¹ band depending on the type of cation (Lewis et al. 1975).

Mid-IR transmission spectra of ${\mathrm{ClO}}_4^{-}$ ions exhibit vibrational bands near 620 and 1100 cm⁻¹, while the chlorate ${\mathrm{ClO}}_3^{-}$ ions have bands at 1000–1100, 620, and 450–500 cm⁻¹ (Gadsden 1975). A nitronium perchlorate IR and Raman study assigned an asymmetric stretch for the ${\mathrm{ClO}}_4^{-}$ ion at 1082–1143 cm⁻¹, a symmetric stretch at 936 cm⁻¹, an asymmetric bend at 625 cm⁻¹, and a symmetric bend (Raman only) at 461 cm⁻¹ (Nebgen et al. 1965). The asymmetric stretching band shifts with cation type and was observed near 1100 cm⁻¹ for Na²⁺ perchlorate, as a doublet at 1075 and 1140 cm⁻¹ for K⁺ perchlorate and as a doublet at 1060 and 1130 cm⁻¹ for Mg²⁺ perchlorate (Miller and Wilkins 1952). The symmetric stretch has been observed at 940 cm⁻¹ for K⁺ perchlorate, and as a doublet at 945 and 962 cm⁻¹ for Mg²⁺ perchlorate (Miller and Wilkins 1952). The asymmetric bend was observed at 624 cm⁻¹ for Na⁺ perchlorate, 625 cm⁻¹ for K⁺ perchlorate and at 622 cm⁻¹ for Mg²⁺ perchlorate (Miller et al. 1960). Weak bands were also observed near 480–495 cm⁻¹ for a few perchlorates, which is consistent with some chlorate being present in the samples (Miller et al. 1960). Presumably, the symmetry change from ${\mathrm{ClO}}_4^{-}$ to ${\mathrm{ClO}}_3^{-}$ allows the symmetric bending vibration to become IR active. Chlorates have C3ν symmetry and exhibit bending vibrations at 480–510 and 615–630 cm⁻¹ (Miller et al. 1960), and stretching vibrations at 913–990 cm⁻¹ (Miller and Wilkins 1952). The vibrational bands measured in transmittance spectra of perchlorates and chlorates are summarized in Table 1. ATR studies of perchlorate solutions found that the asymmetric stretching band (ν₃) shifted toward lower frequencies (longer wavelengths) with increasing perchlorate concentration, while the position of the symmetric stretching band (ν₁) did not shift with perchlorate concentration (Chen et al. 2004). Kloprogge et al. (2002) noted similarities in the IR and Raman vibrations of ${\mathrm{SO}}_4^{2-}$ and ${\mathrm{ClO}}_4^{-}$ molecules present in the interlayer region of hydrotalcite due to similarities in the tetrahedral bonding of these molecules. These interlayer ${\mathrm{ClO}}_4^{-}$ molecules exhibited transmission IR absorptions at 626–635 cm⁻¹ for ν₄, 935 cm⁻¹ for ν₁, and 1096–1145 cm⁻¹ for ν₃ (Kloprogge et al. 2002). Free ${\mathrm{ClO}}_4^{-}$ molecules with Td symmetry only have spectral absorptions for the ν₄ and ν₃ vibrations at 645 and 1128 cm⁻¹ (Ross 1962, 1972). Peak positions in transmittance spectra of minerals differ from those observed in reflectance and emissivity spectra because transmittance spectra are a function of absorbance only, while reflectance and emissivity spectra depend on both absorbance and scattering properties (e.g., McMillan and Hofmeister 1988).

Table 1.

Summary of mid-IR absorptions in transmittance spectra of perchlorates (ClO₄) and chlorates (ClO₃) from the literature

Vibrational bands (cm−1)		KClO4	NaClO4·H2O	NaClO4·2H20	Ca(ClO4)2·4H2O	Mg(ClO4)2*	Mg(ClO4)2·6H2O	Fe2+(ClO4)2·6H2O	KClO3	NaClO3
asymmetric stretch	ν3	1075–1093		1081	1087	1060	sh 1087			990
		1111	1100	1111	1111	1130	1111	1111	962	965
		1140–1143		1124	1139		wsh 1143
symmetric stretch	ν1	940–941		939	939	945 962	940	943	938	935
asymmetric bend	ν4	625–627	624	625	627	622	627	628	623	627
symmetric bend	ν2		480			495 460			490	484

Notes: Band assignments from Miller and Wilkins (1952); Miller et al. (1960); Ross (1962). sh indicates shoulder and wsh indicates weak shoulder.

^*H₂O molecules were not indicated to be present in this sample, but a H₂O bending vibration is observed as in the spectra of the Na ClO₄·H₂O sample.

Additional IR bands are observed for perchlorate salts due to H₂O molecules. A band at 3580 cm⁻¹ (~2.79 μm) is assigned to H₂O H-bonded to ${\mathrm{ClO}}_4^{-}$ in an ATR study and does not vary with cation (Chen et al. 2004). Another band is observed at 3196 cm⁻¹ (~3.13 μm) for Na⁺ perchlorate and 3254 cm⁻¹ (~3.07 μm) for Mg²⁺ perchlorate that varies with differences in polarization of the cation (Chen et al. 2004). Near-IR reflectance spectra of Mg²⁺ perchlorates and other Cl⁻ salts show differences in the H₂O bands as a function of the hydration level of the salt (Hanley et al. 2010, 2011), which are likely due to structural changes as shown in Figures 1 to 2 and by Robertson and Bish (2010).

Low-temperature experiments were performed with Ca²⁺ perchlorate solutions to study spectral changes as the liquid changed to a glassy state (Kanno and Hiraishi 1981). These studies of H-bonding in Ca²⁺ perchlorates at 77 K show that additional H₂O stretching bands are present near 3545 cm⁻¹ (~2.82 μm) in Raman spectra of a hydrated glassy low-temperature perchlorate system in addition to the H₂O stretching band observed near 3585 cm⁻¹ (~2.79 μm) for hydrated perchlorate solution (Kanno and Hiraishi 1981). These were assigned to OH/H₂O groups weakly H-bonded to perchlorate ions (e.g., Adams et al. 1971).

Method

Perchlorate samples in this study were obtained from Sigma Aldrich and were reported to be 98–99% pure. The Na⁺ and K⁺ perchlorates were anhydrous. The Mg²⁺ perchlorate is not described as having any water, but H₂O is clearly present in the data and the sample must be hydrated. The Fe²⁺ and Fe³⁺ perchlorates are described as hydrated without providing information on how many waters are expected per formula unit. The Ca²⁺ perchlorate is described as Ca(ClO₄)₂·4H₂O and the Al³⁺ perchlorate is described as Al(ClO₄)₃·9H₂O. The Na⁺, K⁺, Ca²⁺, Mg²⁺, and Al³⁺ perchlorates are all white, the Fe²⁺ perchlorate is pale green, and the Fe³⁺ perchlorate is pale orange to yellow.

Measurement of reflectance spectra

Reflectance spectra were measured in a horizontal sample dish using an ASD spectrometer under ambient conditions at the SETI Institute, using a bidirectional VNIR spectrometer under ambient conditions at Brown University’s RELAB (Pieters 1983), and using a biconical Nicolet FTIR spectrometer in a controlled, dry environment at RELAB (e.g., Bishop et al. 1995). The SETI Institute measurements included acquiring spectra of the perchlorates as received and then again following short-term heat treatments at 100 °C. The samples in these experiments were heated in an oven in room air for 10–20 min intervals, removed from the oven for a few minutes for spectral measurements and then in some cases returned to the oven for continued heating at 100 °C.

Ambient spectra in both labs were measured relative to Halon from 0.3 to 2.5 μm. Infrared reflectance spectra were measured relative to a rough gold surface with 2 cm⁻¹ spectral sampling from 1–50 μm in an environment purged of H₂O and CO₂ for 10–12 h as in past studies (e.g., Bishop et al. 1995). Composite, absolute reflectance spectra were prepared by scaling the FTIR data to the bidirectional data near 1.2 μm. Additional FTIR spectra were collected following measurement of the IR spectra in three spectral regions (including some heating of the sample during measurement) and extended exposure to the dehydrated environment of the sample chamber. For some samples heating by the FTIR beam and/or longer exposure to the dehydrated conditions in the H₂O-purged environment resulted in changes to the spectral features.

Measurement of DSC data

Differential scanning calorimetry (DSC) scans were obtained using a Perkin-Elmer Pyris calorimeter. Perchlorate samples weighing from 5 to 10 mg were crimp sealed in standard 40 μL Perkin-Elmer aluminum pans with covers. Samples were analyzed over a temperature range of 20 °C up to a maximum of 600 °C at a scan rate of 20 °C/min. A nitrogen purge gas (99.999% purity) at 1 bar with a flow rate of 20 mL/min was used.

Results and discussion

VNIR reflectance spectra

Visible/near-infrared (VNIR) spectra of perchlorates are dominated by combinations and overtones of the H₂O vibrations near 1.42–1.48, 1.93–1.98, and 2.7–2.78 μm (Fig. 3; Table 2). Spectra of the K⁺ and Na⁺ perchlorates in our study do not exhibit these bands, confirming their anhydrous form. Iron perchlorates have additional absorptions due to electronic excitations (e.g., Burns 1993). These bands are broader than the vibrational features and occur near 0.767 μm for Fe³⁺ perchlorate and near 0.95–1.1 μm for Fe²⁺ perchlorate. Overtones and combinations of the ${\mathrm{ClO}}_4^{-}$ vibrations are observed near 3.2—4 μm and are most prominent in the spectra of the K⁺ and Na⁺ perchlorate (Fig. 3).

Figure 3.

VNIR reflectance spectra from 0.3 to 5 μm of several perchlorate salts, offset for clarity. (Color online.)

Table 2.

Summary of spectral features for perchlorates from reflectance spectra in this study

	Anhydrous KClO4	Anhydrous NaClO4	Ca(ClO4)2·4H2O	Mg(ClO4)2·nH2O	Fe2+(ClO4)2·nH2O	Fe3+(ClO4)3·nH2O	Al(ClO4)3·9H2O
Electronic excitations in μm (and cm−1)
					~0.95–1.1	0.767
					(~9000–10500)	(~13000)
NIR bands in μm (and cm−1)
			1.423 (7030)	1.435 (6970)	1.439 (6950)	1.472 (6790)	1.476 (6775)
			1.750(5715)
			1.803 (5550)
			1.93 (5180)	1.93 (5180)	1.94 (5155)	1.99 (5025)	1.98 (5050)
			2.15 (4650)	2.16 (4630)	2.16 (4630)	2.18 (4585)
			~2.4 (~4170)	2.41 (4150)	2.4 (~4170)	2.45 (4080)	2.39 (4185)
H2O stretch			2.74, 2.79 (3655,3585)	2.77 (3610)	2.77 (3615)	2.83 (3530)	2.84 (3540)
3× ν3	3.22 (3105)	3.17 (3150)
3× ν3	3.90 (2950)	3.34 (2995)
2× ν3+ ν4	3.79 (2640)	3.75 (2670)	3.64 (2750)	3.66 (2730)
2× ν3+ ν4	4.00 (2500)	3.98 (2515)	3.87 (2585)	3.88 (2575)
2× ν3	4.57, 4.95	4.47, 4.87		4.52, 4.96	4.52, 4.93	4.33, 4.93	4.15, 4.90
	(2190, 2020)	(2235, 2055)		(2210, 2015)	(2210, 2030)	(2310, 2030)	(2410, 2040)
	(2135, split)	(2160, split)		(2115, split)	(2115, split)	(2130, split)	(2140, split)
Mid-IR peaks in cm−1 (and μm)
ν3+ ν4	1870 (split)	1900
	(5.35)	(5.26)
	1750 (split)	1760 (split)
	(5.71)	(5.68)
H2O bend			1655 (6.04)	1650 (6.06)	1645 (6.07)	1635 (6.12)	1670 (5.99)
ν3+ ν4	1565 (split)	1570 (split)
	(6.39)	(6.37)
ν3	1185, 1125	1140 (split)	1160, 1095	1115 (8.97)	1105 (9.05)	1120 (8.93)	1130 (8.85)
	(8.44, 8.89)	(8.77)	(8.62, 9.13)
			(1120 split)
ν1	940d (10.6)			950d (10.5)
	885, 825 (11.3, 12.1)	825 (12.1)	980 (10.2)
ν4	630 (15.8)	644(15.5) (split)	630 (15.8) (split)	630 (15.8)	630 (15.8)	630 (15.8)	630 (15.8)
	465 (21.5) (split)	485 (20.6) (split)			350	500	460
		~115 (87) (split)	~215 (46) (split)	~210 (48) (split)	~125 (80) (split)		~110 (90) (split)
Radius (Å)*	1.52	1.16	1.14	0.860	0.920	0.785	0.675
Charge/radius*	0.66	0.86	1.75	2.33	2.17	3.82	4.44

Notes: Split indicates that the spectral feature is split into two or more bands; superscript d indicates downward band instead of upward peak at longer wavelegnths.

^*Effective ionic radii from Huheey et al. (1993) assuming octahedral coordination of the cation and high-spin state for Fe²⁺and Fe³⁺(Telser et al. 2005; Sowe et al. 2013).

The band centers near 1.4, 1.9, and 2.7 μm all shift in similar ways (Fig. 4) with the octahedral cation such that the vibrational energy depends on the polarizing power of the cation and structure of the solid. Polarizing power is the charge divided by the effective ionic radius and is also termed charge density or ionic potential (e.g., Huheey et al. 1993). Polarizing power was determined for the cations in the perchlorates of our study and follows the trends: Al³⁺ > Fe³⁺ > Mg²⁺ > Fe²⁺ > Ca²⁺ > Na⁺ > K⁺ (Table 2). These values were compared with the putative water vibrations to confirm their assignment (Fig. 5).

Figure 4.

NIR reflectance spectra from 1.2–2.7 μm illustrating variations in the perchlorate spectral features due to changes in the cation. The Mg perchlorate spectrum has a much brighter reflectance in this region and is divided by 2 for this plot. (Color online.)

Figure 5.

Comparison of vibrational energies (in cm⁻¹) of the H₂O bands with polarizing power of octahedral cations showing that the frequency of these vibrations decreases with increase charge/radius. (a) H₂O stretching vibration and its overtone, and (b) H₂O stretching vibration and H₂O combination (stretching plus bending) band. (Color online.)

For the polyhydrated perchlorate structure the H₂O stretching vibration changed from 3610 cm⁻¹ for Mg²⁺, to 3615 cm⁻¹ for Fe²⁺, to 3530 cm⁻¹ for Fe³⁺, to 3540 cm⁻¹ for Al³⁺ (Table 2), which compares well with the polarizing power of these cations. Similar trends were observed for the H₂O stretching overtone near 1.43–1.48 μm (6775–6970 cm⁻¹) and the H₂O combination stretching plus bending band near 1.93–1.99 μm (5025–5180 cm⁻¹). The Ca(ClO₄)₂·4H₂O spectrum exhibits two H₂O stretching vibrations at 3655 and 3585 cm⁻¹ due to different H₂O sites in the structure as shown for Mg²⁺ perchlorate in Figure 1. The Ca(ClO₄)₂·4H₂O spectrum also includes a doublet at 1.75 and 1.80 μm that is not observed for the other perchlorates in our study. These are attributed to overtones and combinations of H₂O vibrations in a constrained structure such as the isolated polyhedral in tetrahydrated perchlorates. Related features near 1.75–1.76 and 1.79–1.82 μm were observed in spectra of Mg(ClO₃)₂·2H₂O, MgCl₂·2H₂O, and CaCl₂·2H₂O (Hanley et al. 2010, 2011, in review), supporting the assignment of these features to H₂O in a constrained environment.

VNIR reflectance spectra under changing hydration environments

To evaluate changes in the spectral properties with hydration level, reflectance spectra of the samples were measured under diffeting environmental conditions. The K⁺, Na⁺, and Mg²⁺ perchlorate samples each converted to a liquid after heating for 10 min in an oven with room air at 100 °C, so reflectance spectra were not measured after heating for these samples. Because 100 °C is below their melting points, these samples may have gained H₂O and deliquesced. In contrast, the K⁺, Na⁺, and Mg²⁺ perchlorate samples did not liquefy after 10+ h of exposure to H₂O-purged conditions inside the Nicolet spectrometer. Reflectance spectra measured of these samples after purging out the H₂O for ~10 h showed virtually no changes relative to spectra measured again following an additional few hours in the chamber and heating by the FTIR beam. In different experiments by Morris et al. (2009) spectra of Mg(ClO₄)₂·6H₂O exhibited small reductions in hydration bands after heating at 210 °C in dry N₂ and large reductions in hydration bands after heating to 330 °C (Morris et al. 2009).

The spectrum of Ca²⁺ perchlorate exhibited almost no change after heating for 10 min in an oven with room air at 100 °C, but did show some decrease in intensity of the H₂O bands following dehydration and heating by the FTIR beam in the H₂O-purged environment of the Nicolet spectrometer. This suggests that either 10 min at 100 °C were insufficient to dehydrate the sample or it rehydrated within the few minutes required to run the spectra. The Fe²⁺, Fe³⁺, and Al perchlorates did change upon dehydration and heating and spectra of these samples under different conditions are shown in Figure 6.

Figure 6.

VNIR reflectance spectra measured of perchlorates under variable hydration conditions (ambient, dehydrated and heated): (a) Fe³⁺ perchlorate, (b) Fe²⁺ perchlorate, and (c) Al perchlorate. Note that the heated samples reacted during heating but spectra were measured under ambient conditions after cooling and in many cases they rehydrated. (Color online.)

Spectra of Fe³⁺(ClO₄)₃·nH₂O shown in Figure 6a exhibit the strongest H₂O band for the spectra measured under ambient conditions. This sample was pale orange in color, but changed to bright orange after heating 10 min at 100 °C. The reflectance is 10–20% brighter, but the H₂O overtone and combination bands near 1.46 and 1.98 μm appear about the same intensity and the broad water saturation past 1.5 μm is unchanged. This suggests as above that the sample either did not dehydrate after heating 10 min or it rehydrated within the few minutes required to run the spectra. After removing the sample from the oven for ~10 min and heating the sample for an additional 20 min at 100 °C, the crystals partially melted and turned white, which could indicate a change to a different structure. The spectrum of this sample heated for 10 and then 20 min is about twice as bright as the original spectrum and the H₂O bands are sharper and shifted to 1.45 and 1.96 μm. A new weak feature appeared in the spectrum as well at 2.18 μm. This is consistent with a phase change. Spectra measured after dehydrating the samples 10+ h in the spectrometer chamber (where H₂O is pumped out of the environment) resulted in bands near 1.46 and 1.98 μm, and some reduction in the broad water saturation past 1.5 μm. Subsequent measurements after heating the sample with the IR beam and continued exposure to the low-humidity environment produced a spectrum with bands again at 1.46 and 1.98 μm and further reduction in the broad water saturation past 1.5 μm. Thus, it appears that reducing the humidity level in the FTIR spectrometer experiments decreased the adsorbed water in the system, but did not cause a phase change. In contrast, heating the sample for 10+20 min at 100 °C did cause a structural change that produced a different color and shift in the wavelength of the H₂O bands. Neither the Fe³⁺ electronic excitation absorption at 0.76 μm, nor the reflectance maximum near 0.62 μm changed, but the shoulder near 0.45–0.5 μm changed shape slightly as the color changed.

Spectra of Fe²⁺(ClO₄)₂·nH₂O in Figure 6b show strong H₂O overtone and combination bands that are similar to those observed for the Fe³⁺ perchlorate. The H₂O stretching overtone occurs at 1.44 μm in these spectra and the H₂O stretch + bend combination band occurs at 1.95 μm; neither appeared to change wavelength as the sample was heated. A weak band is observed at 2.16 μm in all the spectra and is more clearly present in spectra of the heated samples. Spectra of the heated and dehydrated samples are less saturated from 1.5–2.5 μm and all of the bands in this region are better resolved. The shape of the strong crystal field transition absorption near 0.97 μm appears to change on the long wavelength side of the band as the sample is dehydrated, but this may be due to variations in the intensities of the H₂O overtones near 1.0–1.17 μm, ratherthan changes in the Fe²⁺ electronic excitation. However, the reflectance minimum near 0.55 μm in the ambient spectrum shifted toward longer wavelengths with heating as the color changed from pale green to orange and then dark orange. The spectrum of the sample heated for 10 min at 100 °C (changed from pale green to partly orange) has a reflectance maximum at 0.62 μm as does the Fe³⁺ perchlorate spectrum, but there is no Fe³⁺ electronic excitation absorption at 0.76 μm as would be expected if the whole sample had been oxidized. The reflectance maximum occurred at 0.70 μm for the Fe²⁺ perchlorate sample heated for 10+20+30 min.

Spectra of Al(ClO₄)₃·9H₂O in Figure 6c show strong H₂O overtone and combination bands and the spectrum measured under ambient conditions is largely saturated past 1.5 μm, similar to what was observed for Fe³⁺ perchlorate. The reflectance was brighter for the sample heated at 100 °C for 10 min and the saturation decreased slightly. Upon dehydration in the FTIR spectrometer, the saturation decreased further and the H₂O bands near 1.48 and 1.97 μm are readily observable.

Differential scanning calorimetry data

DSC data are shown for the K⁺, Na⁺, Mg²⁺, Fe²⁺, and Fe³⁺ perchlorates in Figure 7. The scan of the anhydrous KClO₄ (Fig. 7a) shows an endothermic peak with an onset of ~300 °C due to the transition from rhombic to cubic structure. This transition is followed by an endothermic peak due to the fusion of the salt (onset 585 °C) that occurs just prior to the onset of the exothermic decomposition at ~600 °C. Similar to KClO₄, the NaClO₄ DSC scan (Fig. 7b) shows an endothermic peak with an onset at 307 °C due to a solid-phase transition (orthorhombic to cubic), an endothermic salt fusion peak (melting) with an onset at 467 °C followed by a broad exothermic decomposition peak. Very small endotherms observed at 41 and 88 °C may indicate trace amounts of water in the KClO₄ sample. NaClO₄ monohydrate has been reported to dehydrate in two stages with the release of 0.2H₂O and 0.8H₂O (per mole) at 60 and 150 °C, respectively (Devlin and Herley 1987). These water endotherms are not observed in our NaClO₄ DSC scans.

Figure 7.

DSC scans of sodium (Na⁺) and potassium (K⁺) perchlorates (a), magnesium (Mg²⁺) and calcium (Ca²⁺) perchlorates (b), and ferrous (Fe²⁺) and ferric (Fe³⁺) perchlorates (c). Major transitions are indicated by: P = phase transformation; W = water loss; F = fusion (melting); D = decomposition. (Color online.)

The Mg(ClO₄)₂·6H₂O scan (Fig. 7c) shows three broad dehydration steps each comesponding to the loss of two H₂O (Devlin and Herley 1987). The onsets for the first two dehydration endotherms occur at ~135 and ~240 °C, followed by a complex final dehydration endotherm that is completed by ~370 °C. Dehydration is followed by the stepwise decomposition of Mg(ClO₄)₂·2H₂O. Ca(ClO₄)₂·4H₂O shows an endothermic loss of water over a broad temperature range with dehydration ending at ~320 °C. Dehydration of this sample is followed by a sharp exothermic peak at 348 °C, likely due to a phase transition. The onset of decomposition is seen at ~460 °C (Fig. 7d).

In comparison to K⁺, Na⁺, Ca²⁺, and Mg²⁺ perchlorates, the decomposition of Fe²⁺ and Fe³⁺ occurs at a low temperature (~125–150 °C) and is followed by a complex endotherm that is likely due to sample dehydration.

Mid-IR reflectance spectra

The mid-IR reflectance spectral properties of perchlorates in our study are related to the bands observed for transmittance IR spectra of perchlorates in past studies (Table 1), and to reflectance (Bishop and Murad 1996, 2005) and emission (Lane 2007) spectra of sulfates. The reflectance peaks described here would occur as emissivity minima. The v₃ asymmetric stretch near 1100–1200 cm⁻¹ (8–9 μm) and v₄ asymmetric bend near 630 cm⁻¹ (−16 μm) are observed for all our perchlorate spectra (Figs. 8-9, Table 2). The v₁ and v₂ vibrations were also observed for the anhydrous perchlorates, suggesting that their structures are disrupted enough to enable these modes to become IR active. The v₄ band was observed at 630 cm⁻¹ for all of the polyhydrated perchlorates in our study indicating that this vibration does not vary with cation as observed in past transmittance IR studies. However, this band was split into doublets for the anhydrous and tetrahydrated structures. The v₃ vibrational energies were observed to vary with the polarizing power of the cation such that the vibrational frequency increases with increasing charge/radius (Fig. 10). The v₃ bands in the anhydrous perchlorates and Ca(ClO₄)₂·4H₂O spectra were split into doublets, likely due to differences in the ${\mathrm{ClO}}_4^{-}$ sites in the structures. The v₃ band in spectra of the anhydrous perchlorates formed a trend of increasing vibrational energy with increasing polarizing power (charge/radius) that was distinct from the trend for the hydrated perchlorates. The Fe²⁺, Mg²⁺, Fe³⁺, and Al³⁺ data follow a fairly linear trend, while the Ca²⁺ data are offset a bit from the others (Fig. 10). The v₃ overtone followed this same trend, but was not observed in the Ca(ClO₄)₂·4H₂O spectra. These data confirm the assignment of the band near 2110–2160 cm⁻¹ as an overtone of the v₃ vibration.

Figure 8.

Mid-IR reflectance spectra from 200–2000 cm⁻¹ illustrating the dominant vibrational modes near 630 and 1130 cm⁻¹, as well as additional modes for some samples. The hydrated perchlorates with Mg, Fe²⁺, Fe³⁺, and Al have similar ${\mathrm{ClO}}_4^{-}$ features due to their related symmetries and an H₂O bending vibration near 1635–1670 cm⁻¹. The anhydrous Na and K perchlorates have multiple strong ${\mathrm{ClO}}_4^{-}$ bands, some of which are split. The mid-IR spectral features for Ca perchlorate exhibit some similarities with each group suggesting that this Ca perchlorate structure falls in between the structures of the other groups. (Color online.)

FIGURE 9.

Mid-IR reflectance spectra from 100–1300 cm⁻¹ illustrating variations in the longer wavelength features with type of cation. (Color online.)

Figure 10.

Comparison of ${\mathrm{ClO}}_4^{-}$ ν₃ and ν₃ overtone (OT) band positions (in cm⁻¹) with polarizing power of octahedral cations showing that the frequency of these vibrations increases with increasing charge/radius. (Color online.)

In addition to the ${\mathrm{ClO}}_4^{2-}$ vibrations, the hydrated perchlorate spectra (Fig. 8) exhibit H₂O bending virbations near 1650 cm⁻¹. This band spans the range of 1635 cm⁻¹ for Fe³⁺ perchlorate to 1670 cm⁻¹ for A1 perchlorate (Table 2) and depends on the polarizability of the cation and the degree of hydration of the salt.

The longer wavelength mid-IR perchlorate spectra in our study (Fig. 9) exhibited additional features that are not well understood. A doublet feature centered between ~110–215 cm⁻¹ (~45—90 μm) is observed for many but not all of the spectra. The position of this doublet does not appear to be correlated with structure or octahedral cation.

Relating spectral features to structure and composition

Perchlorates exhibit spectral features due to vibrations of the ${\mathrm{ClO}}_4^{-}$ ion and bound H₂O in the structure. The spectral features observed here are consistent with structural observations from previous studies (e.g., West 1935; Olmstead et al. 1982; Pascal et al. 1985; Robertson and Bish 2010). For the hexahydrated perchlorate the structure is based on isolated polyhedra of a metal cation octahedrally coordinated to 6 H₂O molecules and each of these is H-bonded to the ${\mathrm{ClO}}_4^{-}$ ions. The tetrahydrated perchlorate structure includes isolated clusters of the metal cation octahedrally coordinated to 4 H₂O molecules and 2 ${\mathrm{ClO}}_4^{-}$ ions. The dihydrate perchlorate structure forms chains of the metal cation bound to 2 H₂O molecules and 4 ${\mathrm{ClO}}_4^{-}$ ions. The anhydrous perchlorate structure forms an infinite framework of the metal cation coordinated to 6 ${\mathrm{ClO}}_4^{-}$ ions.

The ${\mathrm{ClO}}_4^{-}$ bands followed different trends with polarizing power than the H₂O bands. The H₂O stretching bands followed a pattern of decreasing band energy with increasing polarizing power (charge/radius) of the metal cation and the degree of hydration did not appear to greatly influence the vibrational energy of the bands. Thus, the H₂O stretching vibration occurs at different wavelengths for the different perchlorate structures. In contrast, the ${\mathrm{ClO}}_4^{-}$ stretching band energies increased with increasing polarizing power of the metal cation and the degree of hydration influenced these bands as well. Differences in mid-IR band v₃ stretching vibration energies for anhydrous K⁺ and Na⁺ perchlorates compared with those of hydrated Mg²⁺, Fe²⁺, Fe³⁺, and Al³⁺ perchlorates are consistent with structural differences. The v₃ vibrational energy for Ca(ClO₄)₂·4H₂O falls in between the other groups indicating this could be intermediate between the other structures.

Spectra of the hydrated Fe²⁺ and Fe³⁺ perchlorates changed significantly upon dehydration through exposure to a low humidity environment or heating to 100 °C. Both samples exhibited color changes as well as changes in the water bands. The Al³⁺ perchlorate and the Ca(ClO₄)₂·4H₂O exhibited some changes due to dehydration, while the anhydrous K⁺ and Na⁺ samples and hydrated Mg²⁺ perchlorate did not change under a low-humidity environment, but they liquified when heated to 100 °C and exposed to lab air, likely due to water adsorption.

Implications for Mars

The NIR spectral properties of hydrated perchlorates resemble the spectral properties of hydrated sulfates, hydrated phosphates, and zeolites making it difficult to uniquely identify some of these phases. Specifically, the band near 1.9–2.0 μm and the band or drop in reflectance near 2.4 μm found in hydrated perchlorate spectra are both characteristic of the spectra of some sulfates, phosphates, and zeolites (e.g., Bishop 2005; Bishop et al. 2012). Features near 1.9–2.0 and 2.4–2.5 μm in CRISM or Observatoire pour la Minéralogie, l’Eau, les Glaces, et l’Activité (OMEGA) spectra of Mars are generally attributed to hydrated sulfates, although some features are more consistent with selected zeolites (Bibring et al. 2005; Murchie et al. 2009). Characterization of the spectral properties of specific perchlorates in this study provides additional data to assist in identifying this mineral group using VNIR remote sensing on Mars. If perchlorates are sufficiently abundant on the surface, then they should be identifiable with CRISM or OMEGA, though they have not yet been identified at the Phoenix landing site (Cull et al. 2010b; Poulet et al. 2010). If perchlorates are present in the soil, near-surface, or inside rocks then these data could enable spectral detection of perchlorates using a VNIR imaging spectrometer on a future landed mission with some digging or scraping ability.

The mid-IR spectral features of perchlorates are related to those of sulfates and phosphates (e.g., Lane and Christensen 1998; Lane 2007; Lane et al. in review). Thus, these data could contribute to detection of salt minerals on Mars using the Thermal Emission Spectrometer (TES) data set (Christensen et al. 2001).

Analyses of TEGA data from Phoenix and laboratory experiments have shown that thermal decomposition products of perchlorates could react with inorganic carbonates in the soil to release CO₂ in the temperature range 400–680 °C (Camion et al. 2012). CO₂ release in this temperature range has also been attributed to the possible presence of organic C (Boynton et al. 2009) in soils at the Phoenix landing site. Reaction of perchlorate with organic C has been suggested as one possible explanation for the lack of organic signatures observed by TEGA (Ming et al. 2009).

Results from the SAM instrument on the MSL Curiosity rover indicate the presence of Ca²⁺ perchlorate in the soil at the Rocknest site at Gale Crater, although the hydration level of the perchlorate is not yet known (Archer et al. 2013; Glavin et al. 2013; Leshin et al. 2013). Fe-perchlorates are most consistent with observations at the John Klein and Cumberland sites, although mixtures of Ca- or Mg-perchlorates and Fe oxyhydroxides may also match the data (Ming et al. 2014). Based on laboratory experiments with perchlorate added to Atacama Desert soils, Navarro-Gonzalez et al. (2010) suggested that on Mars, perchlorate could have reacted with putative martian soil organics to produce some of the results of the Viking gas chromatograph mass spectrometer (Biemann et al. 1977). Other recent experiments have shown that radiation-damaged perchlorate salts can also explain the Viking gas exchange and labeled release experiments (Quinn et al. 2013).

The low-thermal decomposition temperatures of Fe³⁺ and Fe²⁺ perchlorate make them unlikely candidates for the type of perchlorate detected by TEGA at the Phoenix site (Hecht et al. 2009) and by SAM at the Rocknest site (Archer et al. 2013; Glavin et al. 2013). Additionally, soil oxidation-reduction potential measurements made using the WCL show that readily soluble iron salt concentrations do not exceed ~1 ppm (Quinn et al. 2011) in the soils measured at the Phoenix site. This low solubility of iron perchlorates precludes their presence above trace levels at the Phoenix landing site. However, while in situ measurements indicate that Mg²⁺ and/or Ca²⁺ perchlorate may be the primary form at the Phoenix (Hecht et al. 2009; Kounaves et al. 2014) and MSL Rocknest sites (Glavin et al. 2013), the presence of other types of perchlorates at other locations on Mars is likely, as indicated by the results of the SAM analyses of sedimentary deposits in Yellowknife Bay (Ming et al. 2014).

The DSC results show that the type of metal cation plays an important role in the patterns of water loss and thermal decomposition. Changes in temperature of the exotherms appear to be related to the perchlorate structure as well as the polarizing power of the cation. Endotherms for water release in the Fe²⁺ and Fe³⁺ perchlorates were both observed at 170 °C. The results of the DSC analyses show that H₂O bond energies in perchlorates depend on the type of cation and hydration state. The highly hygroscopic nature of Ca, Mg, Fe, and Al perchlorates observed during our experiments suggest that these salts may be in hydrated forms on the surface or in the near sub-surface of Mars.

The perchlorate abundances observed by Phoenix and MSL are likely too low to be identified from orbit by CRISM, but may be sufficient to be identifiable by a VNIR imager on a future rover. As chloride deposits have been observed from orbit by TES, there may also be regions of elevated perchlorate abundance on the surface that could be detected by TES or CRISM in the future. The data presented here will enable remote sensing searches for perchlorates on the surface of Mars.