The nature of amorphous ices has been debated for more than 35 years. In essence, the question is whether they are related to ice polymorphs or to liquids. The fact that amorphous ices are traditionally prepared from crystalline ice via pressure-induced amorphization has made a clear distinction tricky. In this work, we vitrify liquid droplets through cooling at ≥106 K ⋅ s−1 and pressurize the glassy deposit. We observe a first order–like densification upon pressurization and recover a high-density glass. The two glasses resemble low- and high-density amorphous ice in terms of both structure and thermal properties. Vitrified water shows all features that have been reported for amorphous ices made from crystalline ice. The only difference is that the hyperquenched and pressurized deposit shows slightly different crystallization kinetics to ice I upon heating at ambient pressure. This implies a thermodynamically continuous connection of amorphous ices with liquids, not crystals.

In previous work, water's second glass transition was investigated based on an amorphous sample made from crystalline ice (Amann-Winkel et al., Proc. Natl. Acad. Sci. U.S.A. 110 (44) 17720-17725). In the present work, we investigate water's second glass transition based on the genuine glassy state of high-density water as prepared from micron-sized liquid water droplets, avoiding crystallinity at all stages. All the calorimetric features of water's second glass transition observed in the previous work are also observed here on the genuine glassy samples. This suggests that the glass transition indeed thermodynamically links amorphous ices continuously with deeply supercooled water. We proceed to extend the earlier study by investigating the effect of preparation history on the calorimetric glass transition temperature. The best samples prepared here feature both a lower glass transition temperature Tg,2 and a higher polyamorphic transition temperature Tons, thereby extending the range of thermal stability in which the deeply supercooled liquid can be observed by about 4 K. Just before the polyamorphic transition, we observe a spike-like increase of heat capacity that we interpret in terms of nucleation of low-density water. Without this spike, the width of water's second glass transition is 15 K, and the Δcp amounts to 3{plus minus}1 J K-1 mol-1, making the case for HDL being a strong liquid. We suggest that samples annealed at 1.9 GPa to 175 K and decompressed at 140 K to {greater than or equal to}0.10 GPa are free from such nuclei and represent the most ideal HDA glasses.

We here review mostly experimental and some computational work devoted to nucleation in amorphous ices. In fact, there are only a handful of studies in which nucleation and growth in amorphous ices are investigated as two separate processes. In most studies, crystallization temperatures T x or crystallization rates R JG are accessed for the combined process. Our Review deals with different amorphous ices, namely, vapor-deposited amorphous solid water (ASW) encountered in many astrophysical environments; hyperquenched glassy water (HGW) produced from μm-droplets of liquid water; and low density amorphous (LDA), high density amorphous (HDA), and very high density amorphous (VHDA) ices produced via pressure-induced amorphization of ice I or from high-pressure polymorphs. We cover the pressure range of up to about 6 GPa and the temperature range of up to 270 K, where only the presence of salts allows for the observation of amorphous ices at such high temperatures. In the case of ASW, its microporosity and very high internal surface to volume ratio are the key factors determining its crystallization kinetics. For HGW, the role of interfaces between individual glassy droplets is crucial but mostly neglected in nucleation or crystallization studies. In the case of LDA, HDA, and VHDA, parallel crystallization kinetics to different ice phases is observed, where the fraction of crystallized ices is controlled by the heating rate. A key aspect here is that in different experiments, amorphous ices of different “purities” are obtained, where “purity” here means the “absence of crystalline nuclei.” For this reason, “preseeded amorphous ice” and “nuclei-free amorphous ice” should be distinguished carefully, which has not been done properly in most studies. This makes a direct comparison of results obtained in different laboratories very hard, and even results obtained in the same laboratory are affected by very small changes in the preparation protocol. In terms of mechanism, the results are consistent with amorphous ices turning into an ultraviscous, deeply supercooled liquid prior to nucleation. However, especially in preseeded amorphous ices, crystallization from the preexisting nuclei takes place simultaneously. To separate the time scales of crystallization from the time scale of structure relaxation cleanly, the goal needs to be to produce amorphous ices free from crystalline ice nuclei. Such ices have only been produced in very few studies.

We investigate the glass polymorphism of dilute LiCl–H 2 O in the composition range of 0–5.8 mol % LiCl. The solutions are vitrified at ambient pressure (requires hyperquenching with ∼10 6 K s –1 ) and transformed to their high-density state using a special high-pressure annealing protocol. Ex situ characterization was performed via isobaric heating experiments using X-ray diffraction and differential scanning calorimetry. We observe signatures from a distinct high-density and a distinct low-density glass for all solutions with a mole fraction x LiCl of ≤ 4.3 mol %, where the most notable are (i) the jumplike polyamorphic transition from high-density to low-density glass and (ii) two well-separated glass-to-liquid transitions T g,1 and T g,2 , each pertaining to one glass polymorph. These features are absent for solutions with x LiCl ≥ 5.8 mol %, which show only continuous densification and relaxation behavior. That is, a switch from water-dominated to solute-dominated region occurs between 4.3 mol % LiCl and 5.8 mol % LiCl. For the water-dominated region, we find that LiCl has a huge impact only on the low-density form. This is manifested as a shift in halo peak position to denser local structures, a lowering of T g,1 , and a significant change in relaxation dynamics. These effects of LiCl are observed both for hyperquenched samples and low-density samples obtained via heating of the high-density glasses, suggesting path independence. Such behavior further necessitates that LiCl is distributed homogeneously in the low-density glass. This contrasts earlier studies in which structural heterogeneity is claimed: ions were believed to be surrounded by only high-density states, thereby enforcing a phase separation into ion-rich high-density and ion-poor low-density glasses. We speculate the difference arises from the difference in cooling rates, which are higher by at least 1 order of magnitude in our case.