Earth and Planetary Science Letters 476 (2017) 189–198

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

www.elsevier.com/locate/epsl

Brine formation via deliquescence by salts found near Don Juan Pond, Antarctica: Laboratory experiments and field observational results

R.V. Gough, J. Wong, J.L. Dickson, J.S. Levy, J.W. Head, D.R. Marchant, M.A. Tolbert

a r t i c l e i n f o a b s t r a c t

Article history:Received 1 March 2017Received in revised form 31 July 2017Accepted 2 August 2017Available online xxxxEditor: W.B. McKinnon

Keywords:Antarcticadry valleyssaltdeliquescencewaterMars

The observed darkening of water tracks near Don Juan Pond (DJP) as well as the formation of wet patches elsewhere in the McMurdo Dry Valleys is attributed at least partially to deliquescence, a process by which salts absorb atmospheric water vapor and form brine, coupled with liquid-phase growth when the atmospheric relative humidity exceeds the water activity. Here we perform laboratory experiments to investigate the temperature and relative humidity conditions necessary for deliquescence to occur in calcium chloride-rich sediments collected from the DJP watershed. We use a Raman microscope equipped with an environmental cell to study both deliquescence and efflorescence (recrystallization) of the soluble salt component of DJP soils between −30 and +15 ◦C. In this temperature range, we find that the soluble salt component of the DJP sediments begins to deliquesce between 19 and 46% RH, slightly higher than the deliquescence relative humidity of the primary pure component, calcium chloride. We find a limited hysteresis between deliquescence and efflorescence, but much greater supersaturation of the salt brine can occur at temperatures above 0 ◦C. The relative humidity conditions were varied either slowly (over ∼8 h) to observe near-equilibrium phases or rapidly (over <1 h) to better mimic Antarctic conditions and no differences in deliquescence relative humidity or efflorescence relative humidity were noted. The results of this work can help predict when deliquescence could be actively occurring in the soils near Don Juan Pond and explain darkening of the salt pan after a high humidity period. In tandem with field data, our experimental results suggest that brines can be generated near Don Juan Pond via deliquescence frequently during the southern summer and autumn. Additionally, the soluble salts may persist in the aqueous phase continuously for several months during the southern summer. This work also suggests that salt deliquescence could be impacting the year-round hydrological cycle of the DJP watershed. Steep-sloped water tracks found near DJP have been suggested as a terrestrial analog for recurring slope lineae on Mars, for which salt deliquescence is a proposed formation mechanism. Therefore, understanding the formation of deliquescent brines in a hyper-arid region on Earth may have relevance to Mars.

© 2017 Elsevier B.V. All rights reserved.

1. Introduction

The McMurdo Dry Valleys in Antarctica are a cold, dry desert (Doran et al., 2002; Marchant and Head, 2007; Fountain et al., 2009; Marchant et al., 2013; Head and Marchant, 2014). Limited precipitation in the region has led to the accumulation of high levels of soluble salts (chlorides, nitrates, sulfates) in the soils (Meyer et al., 1962; Claridge and Campbell, 1977; Wilson, 1979;Keys and Williams, 1981; Toner and Sletten, 2013; Bisson et al., 2015), as well as the existence of shallow, hyper-saline lakes such as Don Juan Pond (DJP) in the South Fork of the Upper Wright Valley. Despite the arid conditions, shallow groundwater is present in seasonally thawed active soil layers, expressed at

E-mail address: raina.gough@colorado.edu (R.V. Gough).

http://dx.doi.org/10.1016/j.epsl.2017.08.0030012-821X/© 2017 Elsevier B.V. All rights reserved.

the surface as wet patches and water tracks (Head et al., 2007;Levy et al., 2011, 2012) and potentially deeper within permafrost-affected soils (Dickinson and Rosen, 2003; Mikucki et al., 2015). These hydrological features are believed to form, at least partially, due to the deliquescence of the abundant salts in the shallow subsurface (Wilson, 1979; Levy et al., 2012, 2015; Dickson et al., 2013).

Water tracks near Don Juan Pond have been suggested as a possible terrestrial analog for recurring slope lineae (RSL) on Mars (Levy, 2012; Dickson et al., 2013; McEwen, 2014). RSL are widespread, narrow, low-albedo features that appear, grow and fade seasonally on steep slopes on Mars (McEwen et al., 2011;Ojha et al., 2014, 2015, Stillman et al., 2014, 2017). While the seasonality is consistent with RSL being triggered by an H2O phase change and the RSL growth patterns are compatible with downslope flow of a liquid through a soil matrix (Levy, 2012),

190 R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198

the source and composition of the material moving downslope is unknown and actively debated (Dundas et al., 2016). Because deli-quescence is a proposed mechanism for the formation of liquid in RSL (McEwen et al., 2011; Ojha et al., 2014, 2015; McEwen, 2014), it is important to understand the conditions under which deliques-cent brines play a role in terrestrial hyper-arid regions that lack pluvial (rainfall) activity.

Deliquescence, the phase transition from a crystalline solid to an aqueous solution, occurs when the relative humidity (RH) is equal to or greater than the deliquescence relative humidity (DRH) of a salt (Martin, 2000; Davila et al., 2010). At equilib-rium, the brine formed has a water activity equal to the rela-tive humidity (%) divided by 100. For each salt or salt mixture, the DRH varies as a function of temperature (Seinfeld and Pan-dis, 1998), with deliquescence occurring at temperatures as low as the eutectic temperature of the salt. Calcium chloride (CaCl2) is a highly soluble and deliquescent salt that is abundant in the MDV, having been found in shallow groundwater (Wilson, 1979)and lakes and ponds (Meyer et al., 1962; Torrii and Ossaka, 1965;Siegel et al., 1979). Calcium chloride has a DRH that can be as low as 12% RH but as high as 80% RH depending on the hydration state of the crystalline salt and on the temperature (Gough et al., 2016).

Deliquescence may play a role in the hydrological cycle of Don Juan Pond and the Dry Valleys in general, and possibly regional geochemistry as well. Wilson et al. (1979) first proposed that del-iquescence of soluble salts could be responsible for the fraction-ation of different salt species on the surface and in lakes of the Dry Valleys. Dickson et al. (2013) observed that water tracks near Don Juan Pond rapidly responded to high humidity air by ab-sorbing atmospheric water vapor and darkening. They suggested that the source of the salts in the pond and perhaps the source of some of the water in the pond is saline groundwater flowing above the permafrost. Some of this brine is formed from salt del-iquescence during periods of high humidity (Levy et al., 2015). CaCl2-rich solutions formed via salt deliquescence in the water tracks and flushed by streamwater and snowmelt could then per-colate through the soil matrix or along the top of the permafrost table, subsequently flowing into Don Juan Pond, impacting its geo-chemistry and area/volume. Shallow groundwater (water track) discharge from the eastern margin of Don Juan Pond, is consis-tent with this process (Kounaves et al., 2010; Dickson et al., 2013). However, brines formed this way are susceptible to efflorescence, the recrystallization of salt solution into a crystalline phase. Ef-florescence typically requires drier conditions than deliquescence for a given salt at constant temperature but has not been stud-ied in detail for MDV soils. The efflorescence relative humidity (ERH) is lower than the DRH for most inorganic salts (Tang, 1997;Martin, 2000) because crystallization is kinetically hindered. Super-saturated salt solutions are metastable and can exist for at least several hours at RH < DRH (Gough et al., 2011; Nuding et al., 2014; Davis et al., 2015; Martin, 2000). Laboratory experiments are needed to determine the extent and duration of this hysteresis be-cause it cannot easily be calculated thermodynamically.

Levy et al. (2015) performed laboratory studies to test the hy-pothesis that atmospheric water vapor can be concentrated in high salinity soils via deliquescence and subsequent solution growth. In-deed, the water content of salt-rich MDV soils from Taylor Valley was found to be as high as 7% by mass after dry soils were exposed to 75% RH in the laboratory and as high as 16% by mass after dry soils were exposed to 100% RH. In certain cases, the sediment vis-ibly darkened, became clumpy, and the presence of standing water was even observed as solution volume grew to exceed the pore space. This study was performed at room temperature; however, lower temperatures could impact the deliquescence of salts in the sediments by changing the DRH of the salts, as well as impact-ing the kinetics of water uptake. Accordingly, the investigation of

deliquescence of MDV salts under lower temperatures relevant to Antarctica is warranted, and essential for any potential application to Mars.

Here we report the results of laboratory studies of the deli-quescence and efflorescence of salts in soil collected from the salt pan around Don Juan Pond. Most of the dissolved salt is CaCl2(Torrii and Ossaka, 1965; Harris et al., 1979; Levy et al., 2011;Toner and Sletten, 2013) but sodium, magnesium and sulfate are also present (Torrii and Ossaka, 1965). We focus on the deliques-cence behavior of the soluble salt component of the DJP sediments. Once the environmental conditions needed for the deliquescence and efflorescence phase transitions are understood, estimates can be made regarding when and where deliquescence of CaCl2-rich salts near DJP can occur and how long surface or shallow subsur-face brines may persist until efflorescence occurs.

2. Materials and methods

2.1. Sample preparation

The sample used for this work was a green-brown, fine-grained silt collected from the salt pan around Don Juan Pond (77.564015◦ S, 161.189716◦ E) during 2012. Although a few exper-iments were performed on the entire sediment sample, for most experiments we focused on the soluble portion. These salts are likely controlling deliquescence. To extract the soluble portion of the sediment, 10.0 mL of high-purity water was added to 5.00 g of soil. After mixing well, the mixture sat for 24 h at room temper-ature. The slurry was then filtered through a 0.2 μm filter, which yielded a colorless solution. The mixture of soluble salts that re-sulted from this aqueous extraction will be called “DJP extract” in this paper. The DJP extract was analyzed via ion chromatography and inductively coupled plasma mass spectrometry to quantify the concentration of dissolved anions and cations, respectively.

To prepare a sample of DJP extract for a deliquescence exper-iment, the aqueous solution was nebulized onto a hydrophobic quartz disk. The DJP extract particles generated were crystalline and ranged from 5 to 30 μm in diameter. For the deliquescence experiments performed on the entire sediment sample, (not just the soluble extract), a thin layer was deposited onto the disk with no prior physical or chemical treatments.

2.2. Environmental cell and Raman microscope

A brief summary of the experimental setup and protocol is given here, with more details in the supplemental material. The Raman microscopy system (Fig. S1) has been previously described in detail (Baustian et al., 2010; Gough et al., 2011). Briefly, a Nico-let Almega XR Dispersive Raman spectrometer was outfitted with a Linkam environmental cell, Linkam automated temperature con-troller, and Buck Research chilled-mirror hygrometer. In this work, temperatures in the environmental cell were varied from −30 to 25 ◦C, and RH ranged from <1 to 100%. Changes in phase (solid vs. aqueous) or hydration state of individual salt particles were mon-itored visually with the optical microscope (Olympus BX51 with ×10, ×20, and ×50 magnification capabilities) or spectrally using Raman spectroscopy. A 532 nm laser was used to collect spectra between 400 and 4000 cm−1 with 2 cm−1 resolution. Spectra were always collected in the center of a particle. The Raman spectra obtained allow for molecular identification of individual particles as small as 1 μm. Raman spectroscopy is sensitive to the phase of water; water in the liquid, ice, and hydrated, crystalline phases exhibit different spectra, particularly in the O–H stretching region (∼3400 cm−1).

We performed experiments on two different timescales. Longer experiments (6–10 h) were used to probe equilibrium or near-

R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198 191

Fig. 1. The different Raman spectra (O–H stretching region only) of DJP extract that we observed over the course of our experiments are plotted in red. There is a clear difference between anhydrous salt (no O–H stretch), the four crystalline hydrates we observed (sharp O–H stretch features) and aqueous DJP extract brine solution (very broad O–H stretch). We have identified spectral matches for hydrate 1 (CaCl2 ·2H2O) and hydrate 4 (NaCl·2H2O). We have not determined spectral matches for hydrates 2 or 3, although some relevant reference spectra are included (Baumgartner and Bakker, 2009; Bakker, 2004; Gough et al., 2014, 2016). Spectra are offset for clarity. (For interpretation of the colors in this figure, the reader is referred to the web version of this article.)

equilibrium processes between salt and water vapor. Relative hu-midity was slowly increased with sufficient time between addi-tions (2–5% RH) to ensure there was no further change in the Ra-man spectrum, color, shape, or morphology of the particle. Shorter experiments (∼1 h) simulated the short time scales of the rapidly changing Antarctic isolation and wind conditions (e.g., shadowing, eddies, etc.).

2.3. Determination of DRH and ERH values for DJP extract

Because DJP extract contains multiple anions and cations, there is not likely to be a single RH value at which the deliquescence phase transition occurs completely. Even a mixture of two salts typically exhibits complex phase transition behavior, with the more soluble, hygroscopic salt component deliquescing at a lower RH value and complete conversion to an aqueous phase occurring at a higher RH dictated by the less soluble salt (Gough et al., 2014). In the present work, the DJP extract sample is more complex than a binary mixture due to the presence of multiple salts and differ-ent possible hydration states for each salt (each with different DRH values). Here we attempted to simplify the phase transition behav-ior of DJP extract by focusing solely on when deliquescence and efflorescence are spectrally observed to begin in the center of the particle.

Fig. 1 illustrates the Raman O–H stretch region for all spectrally unique phases observed during this study of DJP extract and also some likely spectral matches. All red traces in Fig. 1 are Raman spectra of DJP extract under the temperature and RH given. These spectra include an anhydrous DJP extract phase (bottom) with no visible O–H stretch, 4 different crystalline hydrates with a rela-tively sharp O–H stretch (numbered 1 through 4), and aqueous DJP extract (top), with a broad O–H stretch from 3000 to 3700 cm−1

indicating liquid water. The black, blue or green spectra in Fig. 1are spectra of pure hydrated chloride salts from the literature that were considered as spectral matches to the DJP hydrates. In all hy-drate spectra, the most intense peaks are in the O–H stretching region, although there are low intensity peaks below 1000 cm−1

(not shown). The specific locations and patterns of the peaks for each hydrate are unique.

Fig. 2. Microscope images (50x magnification) taken of a ∼20 μm particle of DJP sediment collected as the relative humidity was increased over 120 min while the sample was held constant at −30 ◦C. The particle is observed to become darker, larger and more spherical with increasing humidity; however, it is unclear when a brine phase has formed.

We are confident in our ability to spectrally differentiate be-tween an aqueous (liquid) phase, a hydrated crystalline phase, and an anhydrous crystalline phase, even if we cannot in some cases determine the exact mineralogy due to the complexity of the mix-ture and the lack of available reference spectra for all potential minerals. Because we are primarily interested in when a liquid phase can form, all DRH values reported in this paper indicate the RH at which a particle was spectrally observed to transition from an anhydrous phase or a crystalline hydrate into brine. All ERH values reported in this paper indicate the RH at which the reverse occurred: a particle was spectrally observed to transition from liq-uid brine to either an anhydrous phase or a crystalline hydrate. Although we focus on the phase transitions between a liquid and solid rather than solid-to-solid hydration (i.e., anhydrous NaCl (s) converting to NaCl·2H2O(s)), we did note the specific crystalline species that existed prior to deliquescence of the DJP extract (hy-drate 1, etc.) and also which specific solid phase was the product of efflorescence.

3. Results

3.1. Deliquescence of the full DJP sediment sample

Fig. 2 shows a series of microscope images collected as the rel-ative humidity was increased around a particle of DJP sediment at −30 ◦C over 120 min. This sample contained insoluble mineral grains and soluble salts. The particle grows darker due to water adsorption as humidity is increased. The most significant change in darkness occurs between 31 and 49% RH. The particle also be-comes more spherical with increasing humidity; the jagged edges visible at lower humidity become smoother, likely due to water adsorption, by 68% RH.

Although corresponding Raman spectra were collected, there was significant fluorescence that rendered them unusable (not shown). Such fluorescence is a known limitation when perform-ing Raman spectroscopy on geological samples (Marshall and Olcott Marshall, 2015). Because fluorescence obscures the O–H stretching region used here to determine DRH and ERH values, we were unable to obtain useful phase information from the Raman spectra of the DJP sediment. Therefore, in all subsequent experi-ments we focused on the soluble portion of the DJP sample, i.e., DJP extract, obtained as described in Section 2.1. These salts, the phase of which can be spectrally analyzed, are very likely control-ling hydration and deliquescence.

192 R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198

Table 1Measured concentration (in ppm by mass) of soluble anions and cations in the aqueous DJP extract solution prepared from the DJP sample.

Concentration (ppm)

15,049ChlorideSulfate 1147Nitrate 283Phosphate 12.2Total anions in extract 16,491

Calcium 5663Sodium 3255Magnesium 134Strontium 39.3Potassium 34.9Other (Mn, Ba, Si, Li) 32.5Total cations in extract 9159

Total soluble salt concentration in soil: 5.16%

3.2. Composition of the soluble salts in DJP

The results of the ion chromatography and inductively coupled plasma mass spectrometry analyses of DJP extract are shown in Table 1 as concentration of an ion in the soluble extract (ppm by mass). The most abundant cation was calcium (61.83% of total cation mass) followed by sodium (35.54%) and magnesium (1.46%). There were minor amounts (<1%) of strontium, potassium, man-ganese, barium, silicon and lithium. The most abundant anion was chloride (91.25% of total anion mass) followed by 6.96% sulfate, 1.72% nitrate and 0.07% phosphate. From these ion concentrations in the DJP extract and the relative amounts of sediment and water present during the extraction, we calculated that the original DJP sediment sample contained 5.16 wt% soluble salts. This reported salt concentration is likely a lower limit because some ions (like perchlorate) were not analyzed and others may not have been fully extracted due to low solubility.

Our findings that the soluble salts mostly consist of calcium and sodium chloride are generally consistent with the analysis of others, who report that soluble salts near DJP are primarily CaCl2with some sodium and magnesium chlorides (Meyer et al., 1962;Claridge and Campbell, 1977; Harris et al., 1979; Levy et al., 2012).

3.3. Deliquescence of DJP soluble salt extract

Fig. 3 shows a Raman spectral series and corresponding micro-scope images illustrating the deliquescence of a 15 μm particle of DJP extract as relative humidity was increased at constant tem-perature (0 ◦C) over 894 min. The Raman spectra (vertically offset for clarity) show that the DJP extract particle is initially anhydrous (2% RH) with no peak seen in the O–H stretching region, but tran-sitions to hydrate 1 by 4% RH and then hydrate 2 by 9% RH. The microscope images show that these hydrated crystalline particles are brighter and larger than the anhydrous particle. A likely major component of hydrate 1 is CaCl2 ·2H2O due to the spectral similari-ties, an assignment consistent with the compositional analysis and also with the stability region of CaCl2·2H2O (Gough et al., 2016). The primary phase present in hydrate 2 is not known. By 20% RH, the O–H stretch has broadened indicating the presence of liquid water. Additionally, the microscope images in Fig. 3 show an in-crease in size, darkness and sphericity between 14 and 20% RH, all consistent with the appearance of brine due to deliquescence. The DRH of DJP extract for this particular experiment was 20% RH. Af-ter deliquescing, the particle continues to grow in volume and the O–H stretch increases in size due to the increasing liquid water

Fig. 3. Spectra and images taken of a DJP extract particle while the relative humidity is increased at constant temperature (0 ◦C) from 0 to 100% RH over 894 min. The particle is 15 μm in diameter when dry. The particle is initially anhydrous (at 2% RH), then transitions to hydrate 1 (at 4 and 9% RH), then transitions to hydrate 2 (at 10% and 14% RH), and then deliquesces into a liquid brine by 20% RH. The DRH for this experiment was recorded as 20% RH. After deliquescence, the particle absorbs additional liquid water, as indicated by the increase in particle size and the increase in the size of the broad O–H stretch. (For interpretation of the colors in this figure, the reader is referred to the web version of this article.)

content. Even at 77% RH, however, the particle appears to contain small (∼1 μm) particulate matter.

Fig. 3 shows one type of deliquescence behavior; however, we observed different behavior in other experiments. For example, we often saw hydrate 1 deliquesce directly into brine (Fig. S2) or form hydrate 3 prior to deliquescence (Fig. S3). Hydrate 3, characterized by a sharp O–H stretch at 3463 cm−1, was observed to form under lower temperatures and higher RH values than hydrates 1 or 2, and so likely has additional water molecules relative to hydrates 1 or 2. Like hydrate 2, the O–H stretching region of hydrate 3 does not resemble known CaCl2 or NaCl hydrates. We have compared the spectra of DJP extract (particularly the hydrates 2 and 3) to likely hydrated sulfates such as the hydrates of MgSO4 (Wang et al., 2006), CaSO4 (Chang et al., 1999) and Na2SO4 (Hamilton and Menzies, 2010). None of these hydrated sulfates had features in the O–H stretching region that were consistent with hydrated DJP extract salts.

We performed over 20 deliquescence experiments on DJP ex-tract between −30 and +17 ◦C. The DRH values from all ex-periments are shown as colored symbols in Fig. 4, plotted on a graph of relative humidity vs. temperature. The color of the sym-bol indicates the last hydration state observed prior to deliques-cence (hydrate 1, 2 or 3). Hydrate 4 was not seen during deli-quescence experiments. Measured DRH values vary from 19% to 46% RH over the temperature range studied. The crystalline hy-drate present just prior to deliquescence was more likely to have a higher hydration state at lower temperatures, which is an ex-pected relationship. Additionally, the DRH value of the DJP extract increased with increased hydration and also with decreasing tem-perature. Both of these trends are consistent with the results of

R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198 193

Fig. 4. Summary of all DRH values measured for DJP extract (solid circles). Differ-ent colors indicate the different hydrated DJP phases observed immediately prior to deliquescence. DRH values for pure CaCl2 (two different hydration states) and pure NaCl are shown for comparison (open symbols); these two species compose ∼95% of the soluble salts in the DJP sample. (For interpretation of the colors in this figure, the reader is referred to the web version of this article.)

many other studies of soluble salt hydrate stability and deliques-cence (Chevrier et al., 2009; Gough et al., 2011; Wise et al., 2012;Hanley et al., 2012).

Approximately half of the deliquescence experiments were per-formed on long timescales (6–10 h) and the other half were per-formed on short, Antarctic-relevant timescales of approximately 1 h (see Section 2.2 and the supplementary material for the description of the experimental differences). During the short timescale experiments, the relative humidity was changed to match the conditions and timescale reported in Dickson et al.(2013). Specifically, we simulated the diurnal cycle that occurred near Don Juan Pond on December 6, 2010. The DRH values ob-tained from these two types of experiments were not noticeably different. Fig. S4 shows the same deliquescence data as in Fig. 4but as a function of experimental timescale. The lack of differ-ence in DRH for short vs. long timescale experiments suggests that equilibrium between the sample and the atmospheric water vapor was reached in both types of experiments and also that it is not necessary to closely monitor timescale in future experiments.

Also plotted in Fig. 4 are reference DRH values for the pure salts that are most abundant in DJP extract, specifically sodium chlo-ride (NaCl, diamond symbols) and two hydration states of calcium chloride (CaCl2·2H2O and CaCl2·6H2O, triangle symbols). Over the temperature range studied, the DRH values of DJP extract lie be-tween the DRH values of pure CaCl2·2H2O (13–21% RH) and pure CaCl2·6H2O (52–65% RH). The DJP extract deliquesced at much lower relative humidity values than NaCl, which has a DRH of ∼75% RH at a wide range of temperatures (Koop et al., 2000;Wise et al., 2012). The only hydrate with a component that could be spectrally determined was hydrate 1, which contained CaCl2·2H2O. The DRH of hydrate 1 (red symbols in Fig. 4) was 24% RH on average, which is close to the DRH of CaCl2·2H2O (13–21% RH). For a more complete comparison between the deliquescence of DJP extract and the major pure salt components (CaCl2 and NaCl), our experimental data are overlaid on the phase diagrams of the CaCl2 + water or NaCl + water systems (Figs S5 and S6 respectively).

Regardless of intermediate hydrates, when humidity is in-creased around crystalline DJP extract, deliquescence occurs at RH values between the DRH values of two pure calcium chloride hy-drates (CaCl2·2H2O and CaCl2·6H2O). We are not implying that the sample contains a mixture of these two specific hydrates, rather

Fig. 5. Typical Raman spectral series collected during an efflorescence experiment for DJP extract. During this experiment, the temperature was held constant at−30 ◦C and the relative humidity was decreased slowly from 100% to 1% RH over 6.5 h. The ERH recorded during this experiment was 40% RH and the solution ef-floresced into hydrate 4. For comparison, the Raman spectra of pure NaCl·2H2O (Bakker, 2004) is shown in black. Spectra are offset for clarity. (For interpretation of the colors in this figure, the reader is referred to the web version of this article.)

that the deliquescent behavior of DJP extract is, in general, much more similar to that of CaCl2 than of NaCl. The sulfate component (4% of total anion mass) is typically not as deliquescent, requiring high RH values and long time scales (Vaniman and Chipera, 2006;Linnow et al., 2014), but this did not inhibit the deliquescence of DJP extract as a whole.

3.4. Efflorescence of DJP extract

Next, we determined the ERH of the sample by lowering the relative humidity around a brine droplet until efflorescence oc-curred. Fig. 5 shows a typical spectral series and microscope im-ages illustrating efflorescence of a DJP extract particle (diam =7 μm, 6.5 h, −30 ◦C). The broad O–H peak is replaced by the sharp O–H stretch of a crystalline hydrate by 40% RH; therefore 40% is the ERH value recorded for this experiment. The efflores-cence product is hydrate 4, which has an intense, sharp peak at 3526 cm−1 with a minor peak at 3546 cm−1 and was a com-mon efflorescence product below 0 ◦C. Hydrate 4 likely contains dihydrated sodium chloride (NaCl·2H2O, hydrohalite). The Raman spectra of pure NaCl·2H2O (Bakker, 2004) is shown in Fig. 5. As RH is further lowered below the ERH, the sharp hydrate peaks de-crease and then disappear. The particle contains only anhydrous salt by 1% RH.

Not all efflorescence of DJP extract brine resulted in hydrate 4. Other outcomes were efflorescence into hydrate 1 (which contains CaCl2·2H2O), efflorescence into an anhydrous particle, or no efflo-rescence observed (Figs. S7, S8 and S9, respectively). Often, the initial efflorescence product was observed to transition to differ-ent crystalline hydrates or anhydrous salt as RH was lowered (Fig. S10). Although the DJP extract salt contained a mixture of multiple salts, we did not observe a step-wise efflorescence; rather, the salt particles visually appeared to completely recrystallize at a single RH value.

Plotted in Fig. 6 are all measured ERH values for DJP extract (solid circles). These ERH data points span the temperature range from −29 to +27 ◦C. The symbol color indicates the efflorescence product. Measured ERH values fall into two regimes: in the lower

194 R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198

Fig. 6. Summary of all ERH values measured for DJP extract. The different colored circles indicate the hydrate that was the product of efflorescence. In some cases (blue circles) efflorescence was not observed and the location of the symbol in-dicates the lowest RH at which a liquid brine was observed prior to ending the experiment. ERH values for pure CaCl2 (Gough et al., 2016) and pure NaCl (Wise et al., 2012 and Koop et al., 2000) are shown for comparison; these two species compose ∼95% of the soluble salts in the DJP sample. (For interpretation of the ref-erences to color in this figure legend, the reader is referred to the web version of this article.)

temperature regime (−30 to 0 ◦C), DJP extract effloresced into the NaCl·2H2O-containing hydrate 4 (purple symbols) between 34 and 50% RH with an average ERH of 42.6 (±6.1)%. In the higher tem-perature regime (−10 to +27 ◦C), very low ERH values (average of 5.7 (±4.5)% RH) were observed. In this warmer regime, multiple efflorescence products were observed: the CaCl2·2H2O-containing hydrate 1 (red symbols), the anhydrous salt (green symbols), or no efflorescence observed (blue symbols). In this last case, the ERH value plotted in Fig. 6 is a lower limit, representing the lowest rel-ative humidity reached before terminating the experiment.

Similar to the deliquescence experiments, efflorescence exper-iments were performed on either long (>5 h) or short (∼1 h) timescales. The shorter experiments simulated the rates of humid-ity changes measured by Dickson et al. (2013) near Don Juan Pond. Figure S11 shows the same DJP extract efflorescence data as in Fig. 6 but sorted by experimental timescale. As in the case of del-iquescence, ERH values do not vary with reaction time scale. This suggests that equilibrium between the sample and atmospheric water vapor was reached in all cases, and also that the rate of RH change does not affect the conditions needed for efflorescence.

Also plotted in Fig. 6 are ERH values of the major pure salts present in DJP extract, NaCl (diamond symbols) and CaCl2 (triangle symbols). We propose that there are two distinct regimes dictating the efflorescence phase transition for DJP extract: one controlled by the recrystallization of the NaCl component (lower T) and the other controlled by the recrystallization of the CaCl2 component (higher T).

Lower temperature efflorescence of DJP extract resulted in hy-drate 4 (purple symbols), which is spectrally similar to NaCl·2H2O. It is therefore consistent that the average ERH of 42.6% for DJP extract in the “cold temperature regime” is very close to the av-erage ERH value of 41 ± 3% RH for NaCl measured by Wise et al. (2012) and Koop et al. (2000) (diamond symbols). Wise et al.(2012) report that the low temperature efflorescence product of NaCl brine can be either anhydrous NaCl (halite) or NaCl·2H2O (hydrohalite), with a higher proportion of the latter at colder tem-peratures but with no noticeable difference in ERH value. Because anhydrous NaCl would have no O–H peak, we cannot confirm or exclude its presence as an efflorescence product. One discrepancy to note: Wise et al. (2012) never observed NaCl·2H2O as an efflo-

rescence product above −21 ◦C; however, we see formation of this NaCl·2H2O-containing hydrate up to 0 ◦C.

The average ERH for DJP extract in the “higher temperature regime” (between −10 and +27 ◦C) is 5.7% RH, with values rang-ing from 1 to 13% RH. These ERH values are very similar to those of pure CaCl2 (Gough et al., 2016), which was observed to effloresce between 2 and 9% RH. These low ERH values were observed for DJP extract regardless of whether or not the brine effloresced into hydrate 1 (a spectral match to CaCl2 ·2H2O) or an anhydrous salt. Additionally, experiments during which DJP extract failed to efflo-rescence (blue symbols) are consistent with observations of pure CaCl2 brines not efflorescing even under single-digit RH conditions (Gough et al., 2016).

In summary, we suggest there are two efflorescence regimes be-cause the less-soluble NaCl·2H2O that forms during low-tempera-ture efflorescence can only exist below 0 ◦C and therefore cannot initiate efflorescence at warmer temperatures. Instead, above 0 ◦C, the soluble CaCl2 component appeared to control the efflorescence of DJP extract by hindering recrystallization down to very low RH values.

Above 0 ◦C, there is a large amount of hysteresis: the average ERH of DJP extract is 5.7% RH, much lower than the average DRH of 22% RH. Below 0 ◦C, because of different salt components con-trolling the deliquescence and the efflorescence, there are actually temperatures for which the ERH is slightly higher than the DRH. To understand this unusual situation, we have performed several ad-ditional experiments to determine the implications of ERH > DRH.

First, we first probed what happens to the NaCl·2H2O-contain-ing, high-ERH efflorescence product prior to the next higher hu-midity period to determine if subsequent deliquescence will be affected. We simulated low temperature efflorescence followed an hour later by low temperature deliquescence (T < −15 ◦C the en-tire time). We found that after the low-T efflorescence product (the NaCl·2H2O-containing hydrate 4) was formed, continuing to lower the RH to 10% RH cause the salt to “reset” to a hydrate that sub-sequently deliquesced at the expected DRH value shown in Fig. 4next cycle. (An example spectral series is shown in Fig. S12.) There-fore, we find that even in the unusual case where ERH > DRH, the next experiment will still deliquescence at the expected DRH with no cycling effects.

Additionally, we probed the outcome if the relative humidity is raised above the DRH but not the ERH, in this special case of ERH > DRH occurring below −15 ◦C. We performed a series of experi-ments in which we raised the humidity to 30–35% RH, which was sufficient to deliquesce the DJP extract. We then lowered the RH to observe efflorescence (see Figure S13). Conditions were already below the expected “lower temperature regime” ERH, but the sam-ple did not immediately recrystallize. Rather, the sample remained in the brine phase until very low ERH values of 1–5% RH were reached. We concluded that the high ERH values (42.6% RH) ob-served at low temperatures occur because relatively insoluble salts (such as NaCl) become dissolved at high humidities (perhaps above 75% RH, the DRH of NaCl). If these salts are in the brine phase, they can control efflorescence when RH is lowered. If NaCl never enters solution, however, then these salts cannot initiate crystallization of the brine at the high ERH values of 40–45% RH and the brine can persist until the CaCl2 phase initiates efflorescence at low, single digit RH values.

4. Implications for soil microclimatology

The DRH and ERH results demonstrate that the temperature of the local environment will strongly affect the deliquescence and recrystallization of salts and the extent of hysteresis behav-ior near Don Juan Pond. Next, we use both summer and win-ter diurnal temperature and relative humidity changes that were

R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198 195

Fig. 7. Temperature (A) and relative humidity (B) data collected near Don Juan Pond every 15 min between Nov. 21, 2006 and May 31, 2007. Included in panel (B) are lines representing the experimentally determined DRH (blue) and ERH (orange) at the measured surface temperature. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

measured in the field near Don Juan Pond (Dickson et al., 2013, 2016) to demonstrate how our experimental data can help pre-dict the frequency and duration of deliquescence events near Don Juan Pond. It is known that humidity values in the this region range from saturation (100% RH) down to the low end of the sensor measurement capability (<18% RH), with an annual mean value of ∼55% RH (Doran et al., 2002). This same region expe-riences ground temperatures ranging from −52 to +18 ◦C with an annual mean temperature of −20 ◦C (Thompson et al., 1971;Doran et al., 2002). Here we use a dataset of temperature and rela-tive humidity values measured at the easternmost edge of the salt pan surrounding Don Juan Pond, near where the sediment sam-ple used in this work was collected. The data was collected from November 21, 2006 to May 31, 2007 (late southern spring to late southern fall). Temperature values were recorded at the ground surface using a sensor exposed to sunlight but covered with a thin dusting of soil. The relative humidity was collected at ∼2 m above the surface with a sensor freely exposed to the atmosphere but shielded from solar radiation. Measurements were made every 3 min and averaged/logged every 15 min during this entire period. Using measurements of surface temperature and ∼2 m relative hu-midity introduces some error because of the assumption that RH is constant in the lower 2 m. In reality the ground air mass is heated to some extent by the soil and thus the RH at the soil/air interface may be lower than at ∼2 m.

Fig. 7 shows (A) surface temperature and (B) relative humid-ity measured during this period. The temperature varies seasonally

and diurnally. The relative humidity shows a more subtle seasonal dependence and a less regular diurnal variation. There are high (>70% RH) values and low (∼10% RH) values that occur during all months as a result of variations in atmospheric circulation and the incidence of atmospheric drainage wind patterns in Wright Val-ley (Doran et al., 2002). However, the average RH in the first half of this dataset (austral spring and summer, from Nov. 21, 2006 to Feb. 25, 2007) is low (29.7%) and the average RH in the second half of this dataset (austral fall through May 31, 2007) is high (48.1%).

Fig. 7B also includes DRH and ERH values determined from temperature and our experimental data (colored lines). This will enable determination of when deliquescence is likely to occur and when the brines may persist. Information on how these DRH and ERH values were determined is detailed in the supplementary ma-terial. In the case of deliquescence, we applied a polynomial fit (Fig. S14) and used this DRH(T) relationship to interpolate the DRH at any temperature. For efflorescence, we simply assigned one of two ERH values based on temperature and corresponding efflorescence regime (Fig. S15). The resulting efflorescence line is therefore binary in value: for temperatures below −10 ◦C, the ERH was 42.6% and for temperatures above −10 ◦C, the ERH was 5.7%.

During the austral summer months of Jan. and Feb., the surface temperature primarily stays between −5 and +15 ◦C. Our experi-mental results suggest that at these temperatures the soluble salts near Don Juan Pond will deliquesce around 20% RH with a weak temperature dependence. There is thus little variation in the DRH (blue trace) during the summer. These summer temperatures are also solidly in the “higher temperature efflorescence regime”. Un-der these warm conditions, once brines are formed they will not effloresce until at or below approximately 6% RH (orange trace). Because the RH at Don Juan Pond is either above the DRH or be-tween the DRH and ERH during the summer months, salts may not recrystallize for several months at a time, even during the warmest and driest periods of the day. Instead, the soluble salts will exist in a liquid brine phase that may, at times, be supersaturated.

We have observed the darkening of the salt pan, a high con-centration of crystalline salts that surrounds Don Juan Pond and the area from which the sample studied here was collected. Fig. 8shows two frames from a time-lapse sequence of the salt pan on the southeastern portion of Don Juan Pond. The images were taken before (A) and after (B) a high humidity airmass passed through the valley. In panel C, data from a meteorological station on the edge of DJP shows that the air was saturated (100% RH) for 6 h. The most significant darkening was observed in the salt pan. Figure S16 is a difference image between A and B, highlight-ing regions that experienced reflectivity changes after the humidity pulse. The salt pan persists throughout the year and never entirely deliquesces, perhaps because some of these salts are less soluble (and thus less deliquescent) or perhaps because there is insuffi-cient atmospheric water vapor to deliquesce the entire salt pan.

Once fall arrives, the temperature rarely rises above −10 ◦C. In March, April and May, the temperature stays between −8◦C and −41 ◦C. At these temperatures, our measured DRH has a steeper temperature dependence, with DRH ranging from 18% RH at −10 ◦C to 46% RH at −30 ◦C (we did not perform experiments below −30 ◦C). Therefore, the blue DRH trace has much more variation in this season and the relative humidity at Don Juan Pond spends less time above the DRH. Additionally, these autumn temperatures lie almost entirely in the “lower temperature efflo-rescence regime”, and an ERH value of approximately 43% RH is expected. For several weeks during these colder months, the ERH is likely to be above the DRH due to NaCl·2H2O initiating efflo-rescence. These high DRH and high ERH values will shorten the window of time that liquid brine can exist. However, despite the unfavorable conditions there are still periods of time in the fall

196 R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198

Fig. 8. Frames from a time-lapse sequence showing salt pan darkening on the southeastern portion of Don Juan Pond. The images were taken before (A) and after (B) a moist front passed through the valley. In can be seen in panel C that the air was saturated (100% RH) for 6 h. The relative humidity measurement is from a meteorological station on the edge of DJP. The images were taken almost 24 h apart, ensuring that the darkening is not a photometric effect from variable solar incidence angle. The most significant reflectivity changes were observed in the salt pan, suggesting that water is retained. See Fig. S16 for a difference image between A and B, highlighting the regions that experienced darkening after the humidity pulse.

where a liquid brine phase formed by deliquescence is expected (see, for example, 4/1/07 to 4/21/07 and 5/13/07 to 5/22/07).

We do not present any winter-time data here; however, mea-surements taken during the 2015 austral winter found an average temperature of −33 ◦C, with daily temperature maxima never sur-passing −17 ◦C from May to August (Dickson et al., 2016). These conditions were not studied here; however, the eutectic tempera-ture of CaCl2 is −52 ◦C (Schiffries, 1990) and it is reasonable to predict deliquescence of CaCl2-rich DJP salts down to this temper-ature if high RH values are reached (exact DRH not known).

5. Implications for regional hydrology

It is tempting to hypothesize if brine formed via deliquescence could result in sufficient brine to cause downhill flow or the for-mation of water track features, either in Antarctica or on Mars. However, it is difficult to quantify the contribution of salt deli-quescence to the formation of water tracks relative to contribu-tions from snowmelt. Neither the quantification of deliquescence-derived water in the field or the transport of such brines were studied here; however, we can make some inferences about water content based on the salt content of the sample studied. The “wet patch” soils reported by Levy et al. (2012) have a soluble salt con-centration of ∼1.75% on average and maintain summertime water contents of ∼3% by mass. In contrast, the DJP basin soils analyzed here contain ∼5% soluble salts by mass. Levy et al. (2015) found a linear relationship between water content and the salt content of soil, so we infer that typical DJP fines can hold 8–9% water by mass, all of which could be atmospherically-emplaced and stored as soil pore water. These values suggest that the upper 1 cm of soil could contain ∼1.5 L of water per square meter, assuming a typical MDV soil bulk density of 1.8 g/cm3. Higher water contents would be expected for salt-encrusted soils closer to the pond margin. There is no evidence yet that these deliquescence-derived brines have caused downhill flow. Rather, although deliquescence-derived pore solutions may cause darkening, they are largely observed to be mostly bound by matric potential, and are removed from soils very slowly – if at all – unless flushed by snowmelt. Dickinson and Rosen (2003) suggest that high-elevation brines diffuse or slowly flow through MDV highland soils over timescales in excess of one summer, probably as thin films. Perhaps higher salt concentrations or a more deliquescent salt could create observable flow on shorter time scales. With snow-deposited salts representing the primary source of CaCl2 delivery to the MDV, a limited salt budget is likely preventing large scale, downhill brine flow on a diurnal time scale.

On Mars, the amount of salt on a slope as well as limited water budget may both be limiting factors in the formation and recur-rence of RSL.

6. Conclusions

We have performed laboratory experiments investigating the temperature and relative humidity conditions necessary for deli-quescence to occur in CaCl2-rich sediments collected from the Don Juan Pond watershed in Wright Valley, Antarctica. This is a location where the darkening of water tracks due to changes in relative hu-midity has drawn comparisons to recurring slope lineae behavior on Mars. Our experimental results show that the deliquescence of DJP extract, the soluble salt portion of the sediment, occurs be-tween 20 and 40% RH at temperatures between −30 and +15 ◦C. The increasing of the deliquescence relative humidity with de-creasing temperature is an expected relationship. In the laboratory, deliquescence appears to be controlled by the most abundant salt, CaCl2. Efflorescence, due to a kinetic barrier to nucleation, occurs at a very low humidity (5.7% RH on average) at temperatures above 0 ◦C, with CaCl2 controlling and effectively hindering efflorescence. Efflorescence occurred at much higher efflorescence relative hu-midity values at temperatures below 0 ◦C (42.6% on average), with NaCl·2H2O controlling and effectively encouraging efflorescence at these low temperatures. In our experiments, we observed no dif-ference in deliquescence relative humidity or efflorescence relative humidity values when relative humidity and temperature were in-creased or decreased at different rates. This suggests that the re-action kinetics observed in the laboratory are relevant to the DJP environment.

We used 6 months of field data collected near Don Juan Pond to assess the likelihood of deliquescence events occurring in sum-mer and fall. We find that deliquescence is very likely to occur in the summer. In fact, the soluble CaCl2-rich salts in the shal-low subsurface may be in the aqueous phase continuously (i.e., the soils may be “wet” at all times during the summer) solely due to deliquescence and the subsequent equilibrium with atmo-spheric water vapor. During fall and perhaps winter, a liquid brine phase is likely less abundant due to the higher deliquescence rel-ative humidity and higher efflorescence relative humidity values measured in our lab studies. Our experimental results suggest that the deliquescence of soluble salts near Don Juan Pond likely be-gins to occur at very low relative humidity values (20% RH) and could be contributing to the RSL-like slope streak darkening ob-served by others in the MDV. Analogous processes to those in the

R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198 197

Don Juan Pond area may be occurring on drier Mars, in areas with salty regolith and fluctuating relative humidity. The CaCl2 salt that is prevalent near Don Juan Pond, however, may not be analogous to Martian salts, although other highly deliquescent, soluble salts such as perchlorates are widespread on Mars (Hecht et al., 2009;Glavin et al., 2013; Navarro-González et al., 2010). The hysteresis between deliquescence and efflorescence, as well as the tempera-ture (and thus seasonal) dependence of the phase transitions, may be important when attempting to understand the seasonality and other behavior of recurring slope lineae on Mars.

Acknowledgements

RVG and MAT acknowledge NASA SSW Award # NNX14AJ96G. Part of this work (field reconnaissance, establishment of scientific setting, sample collection, instrument acquisition and deployment, MDV data analysis) was funded by the National Science Founda-tion Antarctic Science Division (Office of Polar Programs) through grants to James W. Head (ANT-0739702) and David R. Marchant (ANT-0944702), which are gratefully acknowledged. Logistical sup-port for this part of the project in Antarctica was provided by the U.S. National Science Foundation through the U.S. Antarctic Pro-gram. Sample collection in the McMurdo Dry Valleys was greatly helped by the Antarctic Support Contract, Raytheon Polar Services, Petroleum Helicopters International, and the staff of McMurdo Sta-tion.

Appendix A. Supplementary material

Supplementary material related to this article can be found on-line at http://dx.doi.org/10.1016/j.epsl.2017.08.003.

References

Bakker, R.J., 2004. Raman spectra of fluid and crystal mixtures in the systems H2O, H2O–NaCl and H2O–MgCl2 at low temperatures: applications to fluid-inclusion research. Can. Mineral. 42, 1283–1314.

Baumgartner, M., Bakker, R.J., 2009. CaCl2-hydrate nucleation in synthetic fluid in-clusions. Chem. Geol. 265, 335–344.

Baustian, K.J., Wise, M.E., Tolbert, M.A., 2010. Depositional ice nucleation on solid ammonium sulfate and glutaric acid particles. Atmos. Chem. Phys. 10 (5), 2307–2317.

Bisson, K.M., et al., 2015. Patterns and Processes of Salt Efflorescences in the McMurdo region. Arct. Antarct. Alp. Res. 47 (3), 407–425.

Chang, H., Huang, P.J., Hou, S.C., 1999. Application of thermo-Raman spectroscopy to study dehydration of CaSO4·2H2O and CaSO4·0.5H2O. Mater. Chem. Phys. 58, 12–19.

Chevrier, V.F., Hanley, J., Altheide, T.S., 2009. Stability of perchlorate hydrates and their liquid solutions at the Phoenix landing site, Mars. Geophys. Res. Lett. 36 (10), L10202.

Claridge, G., Campbell, I.B., 1977. The salts in Antarctic soils, their distribution and relationship to soil processes. Soil Sci. 123 (6), 1–9.

Davila, A.F., et al., 2010. Hygroscopic salts and the potential for life on Mars. Astro-biology 10 (6), 617–628.

Davis, R.D., et al., 2015. Contact efflorescence as a pathway for crystallization of atmospherically relevant particles. Proc. Natl. Acad. Sci. USA 112 (52), 15815–15820.

Dickinson, W.W., Rosen, M.R., 2003. Antarctic permafrost: an analogue for water and diagenetic minerals on Mars. Geology 31 (3), 199–202.

Dickson, J.L., et al., 2013. Don Juan Pond, Antarctica: near-surface CaCl2-brine feed-ing Earth’s most saline lake and implications for Mars. Sci. Rep. 3 (1166), 1–7.

Dickson, J.L., Head, J.W., Levy, J.S., 2016. Austral winter imaging of Don Juan Pond, Antarctica: polar desert brine flow at −40 ◦C on Earth and implications for Mars. In: 47th LPSC.

Doran, P.T., et al., 2002. Valley floor climate observations from the McMurdo dry valleys, Antarctica, 1986—2000. J. Geophys. Res., Planets 107 (D24), 4772.

Dundas, C.M., McEwen, A.S., Milazzo, M.P., 2016. How wet is recent Mars? Insights from gullies and RSL. In: 2016 LPSC.

Fountain, A.G., et al., 2009. Snow in the McMurdo Dry Valleys, Antarctica. Int. J. Climatol. 30 (5), 633–642.

Glavin, D.P., et al., 2013. Evidence for perchlorates and the origin of chlorinated hydrocarbons detected by SAM at the Rocknest aeolian deposit in Gale Crater. J. Geophys. Res., Planets 118 (10), 1955–1973.

Gough, R.V., et al., 2011. Laboratory studies of perchlorate phase transitions: sup-port for metastable aqueous perchlorate solutions on Mars. Earth Planet. Sci. Lett. 312 (3–4), 371–377.

Gough, R.V., Chevrier, V.F., Tolbert, M.A., 2014. Formation of aqueous solutions on Mars via deliquescence of chloride–perchlorate binary mixtures. Earth Planet. Sci. Lett. 393, 73–82.

Gough, R.V., Chevrier, V.F., Tolbert, M.A., 2016. Formation of liquid water at low tem-peratures via the deliquescence of calcium chloride: implications for Antarctica and Mars. Planet. Space Sci. 131, 79–87.

Hamilton, A., Menzies, R.I., 2010. Raman spectra of mirabilite, Na2SO4·10H2O and the rediscovered metastable heptahydrate, Na2SO4·7H2O. J. Raman Spectrosc. 41 (9), 1014–1020.

Hanley, J., et al., 2012. Chlorate salts and solutions on Mars. Geophys. Res. Lett. 39 (8).

Harris, H.J.H., Cartwright, K., Torii, T., 1979. Dynamic chemical equilibrium in a po-lar desert pond: a sensitive index of meteorological cycles. Science 204 (4390), 301–303.

Head, J.W., Marchant, D.R., 2014. The climate history of early Mars: insights from the Antarctic McMurdo Dry Valleys hydrologic system. Antarct. Sci. 26 (2), 774–800.

Head, J.W., et al., 2007. Slope streaks in the Antarctic Dry Valleys: characteristics, candidate formation mechanisms, and implications for slope streak formation in the Martian environment. In: Cooper, A.K., Raymond, C.R. (Eds.), Antarctica: A Keystone in a Changing World – Online Proceedings of the 10th ISAES. Santa Barbara, CA, pp. 1–5.

Hecht, M.H., et al., 2009. Detection of perchlorate and the soluble chemistry of Mar-tian soil at the Phoenix Lander site. Science 325 (5936), 64–67.

Keys, J.R., Williams, K., 1981. Origin of crystalline, cold desert salts in the McMurdo region, Antarctica. Geochim. Cosmochim. Acta 45, 2299–2309.

Koop, T., et al., 2000. Phase transitions of sea-salt/water mixtures at low tempera-tures: implications for ozone chemistry in the polar marine boundary layer. J. Geophys. Res. 105, 26393–26402. http://dx.doi.org/10.1029/2000JD900413.

Kounaves, S.P., et al., 2010. Discovery of natural perchlorate in the Antarctic Dry Valleys and its global implications. Environ. Sci. Technol. 44 (7), 2360–2364.

Levy, J., 2012. Hydrological characteristics of recurrent slope lineae on Mars: evi-dence for liquid flow through regolith and comparisons with Antarctic terrestrial analogs. Icarus 219 (1), 1–4.

Levy, J.S., et al., 2012. Hypersaline “wet patches” in Taylor Valley, Antarctica. Geo-phys. Res. Lett. 39 (L05402).

Levy, J.S., et al., 2011. Water tracks and permafrost in Taylor Valley, Antarctica: ex-tensive and shallow groundwater connectivity in a cold desert ecosystem. Geol. Soc. Am. Bull. 123 (11–12), 2295–2311.

Levy, J., et al., 2015. Experimental formation of pore fluids in McMurdo Dry Valleys soils. Antarct. Sci. 27 (2), 163–171.

Linnow, K., et al., 2014. Experimental studies of the mechanism and kinetics of hy-dration reactions. Energy Proc. 48, 394–404.

Marchant, D.R., Head, J.W., 2007. Antarctic dry valleys: microclimate zonation, vari-able geomorphic processes, and implications for assessing climate change on Mars. Icarus 192 (1), 187–222.

Marchant, D.R., et al., 2013. A review of geomorphic processes and landforms in the Dry Valleys of southern Victoria Land: implications for evaluating climate change and ice-sheet stability. In: Antarctic Palaeoenvironments and Earth-Surface Processes. Geol. Soc. (Lond.) Spec. Publ. 381 (1), 319–352.

Marshall, C.P., Olcott Marshall, A., 2015. Challenges analyzing gypsum on Mars by Raman spectroscopy. Astrobiology 15 (9), 761–769.

Martin, S.T., 2000. Phase Transitions of Aqueous Atmospheric Particles. Chem. Rev. 100 (9), 3403–3454.

McEwen, A.S., 2014. Recurring slope lineae in equatorial regions of Mars. Nat. Geosci. 7, 53–58.

McEwen, A.S., et al., 2011. Seasonal flows on warm Martian slopes. Science 333 (6043), 740–743.

Meyer, G.H., et al., 1962. Antarctica: the microbiology of an unfrozen saline pond. Science 138, 1103–1104.

Mikucki, J.A., et al., 2015. Deep groundwater and potential subsurface habitats be-neath an Antarctic dry valley. Nat. Commun. 6, 6831.

Navarro-González, R., et al., 2010. Reanalysis of the Viking results suggests per-chlorate and organics at midlatitudes on Mars. J. Geophys. Res., Planets 115 (E12010). http://dx.doi.org/10.1029/2010JE003599.

Nuding, D.L., et al., 2014. Deliquescence and efflorescence of calcium perchlorate: an investigation of stable aqueous solutions relevant to Mars. Icarus 243, 420–428.

Ojha, L., et al., 2014. HiRISE observations of Recurring Slope Lineae (RSL) during southern summer on Mars. Icarus 231, 365–376.

Ojha, L., et al., 2015. Spectral evidence for hydrated salts in recurring slope lineae on Mars. Nat. Geosci. 8 (10).

Schiffries, C.M., 1990. Liquid-absent aqueous fluid inclusions and phase equilibria in the system CaCl2–NaCl–H2O. Geochim. Cosmochim. Acta 54 (3), 611–619.

Seinfeld, J.H., Pandis, S.N., 1998. Atmospheric Chemistry and Physics: From Air Pol-lution to Climate Change, 2nd ed. John Wiley and Sons, Inc., New York, NY.

Siegel, B.Z., et al., 1979. Life in the calcium chloride environment of Don Juan Pond, Antarctica. Nature 280 (5725), 828–829.

198 R.V. Gough et al. / Earth and Planetary Science Letters 476 (2017) 189–198

Stillman, D.E., et al., 2014. New observations of martian southern mid-latitude recurring slope lineae (RSL) imply formation by freshwater subsurface flows. Icarus 233 (C), 328–341.

Stillman, D.E., Michaels, T.I., Grimm, R.E., 2017. Characteristics of the numerous and widespread recurring slope lineae (RSL) in Valles Marineris, Mars. Icarus 285, 195–210.

Tang, I., 1997. Thermodynamic and optical properties of mixed-salt aerosols of at-mospheric importance. J. Geophys. Res. 102 (D2), 1883–1893.

Thompson, D.C., Bromley, A.M., Craig, R.M.F., 1971. Ground temperatures in an Antarctic dry valley. N.Z. J. Geol. Geophys. 14 (3), 477–483.

Toner, J.D., Sletten, R.S., 2013. The formation of Ca-Cl-rich groundwaters in the Dry Valleys of Antarctica: field measurements and modeling of reactive transport. Geochim. Cosmochim. Acta 110 (C), 84–105.

Torrii, T., Ossaka, J., 1965. Antarcticite: a new mineral, calcium chloride hexahydrate, discovered in Antarctica. Science 149 (3687), 975–977.

Vaniman, D.T., Chipera, S.J., 2006. Transformations of Mg- and Ca-sulfate hydrates in Mars regolith. Am. Mineral. 91 (10), 1628–1642.

Wang, A., et al., 2006. Sulfates on Mars: a systematic Raman spectroscopic study of hydration states of magnesium sulfates. Geochim. Cosmochim. Acta 70 (24), 6118–6135.

Wilson, A.T., 1979. Geochemical problems of the Antarctic dry areas. Nature 280 (5719), 205–208.

Wise, M.E., et al., 2012. Depositional ice nucleation onto crystalline hydrated NaCl particles: a new mechanism for ice formation in the troposphere. Atmos. Chem. Phys. 12 (2), 1121–1134.