(a) Glass transition temperature, T_g, against the mass fraction of water, X_aqu, for aqueous ZnCl₂ solutions obtained under conventional cooling/heating process (circle). In those solutions with an , ice precipitation occurs first in the cooling process. The freeze-concentrated phases then vitrify at a constant and the corresponding can be directly read from the monotonous part of the T_g versus X_aqu curve at the point where . (b) Mass-normalized heat flow change at glass transition, δ_Tg, as a function of X_aqu for aqueous ZnCl₂ solutions. The red up triangle in panels (a) and (b) indicates the data point measured in the second round of protocol 2 for the sample with X_aqu = 0.61. (c) Raman spectra measured at 183 K for pure ice (blue dotted line), the solution with X_aqu = 0.49 (black dashed line), and the solution with X_aqu = 0.61 after recrystallization of water (red solid line). The experimental error, approximately 0.5% for T_g and approximately 10% for δ_Tg, arises mainly from in calibrating furnace temperature for the former and from extrapolating the heat flow curve at two sides of T_g for the letter. Error bars are omitted when they are smaller than the symbol size.