a b c d Figure 1. Laboratory set-up for Experiment #3. CaCl2 CO2 Sample cup CaCl2 Dry ice (195K) beneath the red foam-cover a CO2 H2O vapor supply b c Dry ice Raman objective d

Optical fiber feedthrough Raman probe Z stage Electric Feedthrough Rotation stage Motor-hub X stage cold plate Vacuum pump & pressure meter Laser Raman Spectrometer computer PEACh Optical fiber feedthrough feedthrough Quartz window RH meter Needle valve Extra-dry CO2 water bath at 30 C H2O CO2 flow control Extra dry CO2 mix CO2 & H2O supply T-sensor & control Cooling system LN2 Figure 2. Laboratory set-up for Experiment #4 Laboratory set-up for Experiment #4

Figure 3a of Wang et al., 2017 Figure 3b of Wang et al., 2017

Figure 3c of Wang et al., 2017 Figure 3d of Wang et al., 2017

green circle represent unknown solid phase with mass > 6w Figure 3e of Wang et al., 2017 Figure 3f of Wang et al., 2017 green circle represent unknown solid phase with mass > 6w

Figure 3g of Wang et al., 2017

Figure 3h of Wang et al., 2017 Figure 3i of Wang et al., 2017

a. AlCl3·6H2O Fe2+-sulfate·7H2O CaCl2·4H2O Mg-sulfates·7-6 H2O MgCl2·6H2O FeCl3·6H2O AlCl3·6H2O a. Fe3+-sulfates·9-20H2O Mg-sulfates·7-6 H2O Fe2+-sulfate·7H2O Figure 4a of Wang et al., 2017 Figure 4. Deliquescence phase boundaries of HyCOS compared with those of hydrous sulfates. The deliquescence boundary of each HyCOS was plotted, as a colored band, through the middle space between two sets of RH values in T-RH field, where deliquescence was seen at the higher RH values (right side of band) but not seen at the lower RH values. The width of colored bar reflects the uncertainty of phase boundary in RH. Filled orange data points indicate occurrence of deliquescence. Unfilled points indicate dehydration, or non-change, or rehydration, against the deliquescence boundary at the lowest RH (e.g., CaCl2·4H2O for chlorides). Similar deliquescence boundaries for hydrous sulfates were drawn based on data in Figure S1, S2, S3 and relevant publications in supporting document.

b. Mg(ClO4)2·6H2O Ca(ClO4)2·4H2O NaClO4·H2O Fe3+-sulfates·9-20H2O Mg-sulfates·7-6 H2O Fe2+-sulfates·7H2O Figure 4b of Wang et al., 2017

Tlow-end ~ 208K Figure 5a of Wang et al., 2017 T range of RSL sites RH range of 7 deliq-exp = 45-100% Average of the means of 7 t @ different RH 172 sols Deliquescence of MgCl2·6H2O Tlow-end ~ 208K 278K 294K 323K Figure 5a of Wang et al., 2017 Figure 5. Time range t (from t>30% to t100%) of the observed macroscopic manifestation of deliquescence of a HyCOS, as the function of T and RH. (a) in the case of MgCl2.6H2O, the time ranges t for its deliquescence at a specific T (e.g., 278K) but different RHs (e.g., 7 RH values from 45% to 100%) are heavily overlapped. For each t, a mean value was calculated, then an average over the means of all  t was taken, tave(T) (the red dot). It represents an average time for MgCl2.6H2O deliquescence to occur at a T (e.g., 278K) over a wide RH range (45-100%). An exponential regression line (black dotted line), in form of tave(T) = a*eb*T, pass through tave (323K), tave(294K) and tave (278K). The value of R2 reflects the quality of regression. We use 172(=688/4) sols (green dotted line) to set a virtual limit for a season on Mars, in order to have a time basis for comparison. This line intersects with the regression line and set the Tlow-end of a gray-shaded region. This region suggests that if the environmental T of a subsurface MgCl2.6H2O layer rises to be higher than 208 K, its deliquescence would occur in less than 172 sols.

Tlow-end ~ 238K Figure 5b of Wang et al., 2017 T range of RSL sites RH range of 8 deliq-exp = 31-100% Average of the means of 8 t @ different RH 172 sols Deliquescence of CaCl2·4H2O t100% t>30% t : Tlow-end ~ 238K 278K 294K 323K Figure 5b of Wang et al., 2017

Tlow-end ~ 231K Figure 5c of Wang et al., 2017 T range of RSL sites RH range of 7 deliq-exp = 45-100% Average of the means of 7 t @ different RH 172 sols Deliquescence of FeCl3·6H2O t100% t>30% t : Tlow-end ~ 231K 278K 294K Figure 5c of Wang et al., 2017

Tlow-end ~ 224K Figure 5d of Wang et al., 2017 T range of RSL sites 172 sols Deliquescence of AlCl3·6H2O RH range of 7 deliq-exp = 45-100% Average of the means of 7 t @ different RH t100% t>30% t : Tlow-end ~ 224K 278K 294K 323K Figure 5d of Wang et al., 2017

Tlow-end ~ 257K Figure 5e of Wang et al., 2017 T range of RSL sites RH range of 7 deliq-exp = 45-100% Average of the means of 7 t @ different RH 172 sols Deliquescence of NaClO4·H2O t100% t>30% t : Tlow-end ~ 257K 278K 294K 323K Figure 5e of Wang et al., 2017

Tlow-end ~ 232K Figure 5f of Wang et al., 2017 T range of RSL sites 172 sols Deliquescence of Ca(ClO4)2·4H2O RH range of 8 deliq-exp = 31-100% Average of the means of 8 t @ different RH t100% t>30% t : Tlow-end ~ 232K 278K 294K 323K Figure 5f of Wang et al., 2017

Tlow-end ~ 240K Figure 5g of Wang et al., 2017 T range of RSL sites RH range of 7 deliq-exp = 45-100% Average of the means of 7 t @ different RH 172 sols Deliquescence of Mg(ClO4)2·6H2O Tlow-end ~ 240K 278K 294K 323K Figure 5g of Wang et al., 2017

a CaCl2.2H2O CaCl2.4H2O + CaCl2.6H2O CaCl2.2H2O + 3800 3600 3400 3200 3000 Raman Shift (cm-1) MgCl2 MgCl2.xH2O (x=2, 4, 6, 8, 12) 2800 3550 3500 1000 980 960 940 NaClO4 NaClO4.xH2O (x=1, 2, more) b CaCl2 3600 3400 3200 3000 Raman Shift (cm-1) 1040 1020 1000 980 960 Ca(ClO4)2 Ca(ClO4)2 + Ca(ClO4)2.xH2O (x=2, 4) 940 920 3800 Mg(ClO4)2 + Mg(ClO4)2.2H2O + Mg(ClO4)2.6H2O Mg(ClO4)2 d e c Figure 6. Evidences of HyCOS rehydration at 195 K. (a). Rehydration of CaCl2 at 195 K, revealed by the appearance of H2O Raman peaks (marked by dotted lines); (b) Rehydration of MgCl2 at 195K, revealed by the appearance of H2O Raman peaks; (c). Rehydration of NaClO4 at 195 K, revealed by the appearance of multiple H2O Raman peaks, and by a new Raman peak at 942 cm-1, in addition to 952 cm-1 peak of NaClO4 (both are ClO4 vibrational modes); (d). Rehydration of Ca(ClO4)2 at 195 K, revealed by the appearance of multiple H2O Raman peaks, and by a new Raman peak in 900-1100 cm-1; (e). Rehydration of Mg(ClO4)2 at 195 K, revealed by the appearance of multiple H2O Raman peaks, and by two new Raman peaks in 900-1100 cm-1. Wang et al., 2017

3800 3600 3400 3200 Raman Shift (cm-1) Started as CaCl2 CaCl2·4H2O (4 min 50 s) a. CaCl2 1000 950 900 850 started as Ca(ClO4)2 Ca(ClO4)2 ·4H2O (14 min) Ca(ClO4)2 ·2H2O (10 min) b. Ca(ClO4)2 Wang et al., 2017 Figure 7. Evidence of extremely fast rehydration of chlorides and perchlorates measured at different T and RH in a Mars chamber in experiment #4. (a) rehydration of CaCl2 at 251 K, 23.6 %RH, and 16.4 mbar. The H2O peak (3445 cm-1) of CaCl2·2H2O appeared at 4 minutes 50 seconds after the starting of this experiment. (b). rehydration of Ca(ClO4)2 at 296 K, 4.8 %RH, and 19.7 mbar. The H2O peak (3491 cm-1) of Ca(ClO4)2 ·2H2O and the H2O peak (3550 cm-1) of Ca(ClO4)2·4H2O appeared at 10 and 14 minutes, respectively, after the starting of this experiment.

High obliquity period Current obliquity period 2H2O 4H2O 6H2O 10H2O 12H2O Figure 8. Calculated increase of hydration degree for seven anhydrous HyCOS in 172 sols, as the function of input H2O vapor volume. For current obliquity period, the assumed input H2O vapor volume range is 15-90 pr-µm (Jakosky, 1985). For high obliquity period, the assumed the input H2O vapor volume range is 100-500 pr-µm. Five dotted lines indicate the highest hydration degree of a HyCOS, with color code matches with the line. For example, the highest hydration degree of NaCl and NaClO4 (green color) is two H2O.