Structure of bulk water from Raman measurements of supercooled pure liquid and LiCl solutions

In spite of decades of efforts, the puzzle of understanding the anomalies of liquid water remains unsolved. Several plausible theoretical scenarios have been proposed, but distinguishing among them requires accessing temperatures below the homogeneous nucleation temperature $T_h$, which for bulk water is an experimentally impossible task. A widely adopted way for bypassing such a difficulty consists in studying the behavior of samples confined in nanopores, supercooled aqueous salt solutions, or in investigating samples transiently heated or cooled. In this paper, we compare the results obtained from Raman experiments on supercooled bulk water and supercooled LiCl aqueous solutions. The obtained results indicate that while at relatively high temperatures water in ionic solution can appear indistinguishable from bulk water, large differences are detected after further cooling. The comparison with very recent experimental data on water confined in nanopores suggests generalizing our result for every kind of confinement. Even if an experiment carried out under space-time constraints can produce results that match those from bulk water (in the temperature range where bulk water data are available), we cannot assume that the matching will persist at lower temperatures. This implies that the interpretation of many literature experimental results should be reconsidered. A coherent picture of existing experimental data can be obtained accepting the idea that amorphous water is not a metastable state and that homogeneous nucleation temperature represents a metastability limit.

Figure 1

Bulk water. Left panel: Low-density (continuous line) water and high-density (dashed line) water normalized spectra, calculated through Eq. (1). Dash-dot line: HDA spectrum from Refs. 23 and 24. Right panel: Normalized spectra at three selected temperatures (dots); continuous lines represent the results of the fitting with Eq. (2). Inset: Typical difference spectrum (see text for details).

Figure 2

Semilog plot of the ratio between the weights of the MS and LS contributions. The fitting with Eq. (3) (confidence 0.5$\%$) has been obtained for $\mu =1.18$ and $T_S=229.8$ K (very close to $T_h$). Dashed straight line: Arrhenius fit of the data for $T>246$ K ($H$ $=$ activation enthalpy). Continuous line: Fit of the data with Eq. (3). Open circles: Bulk water. Triangles: $\mathrm{LiCl}+12{\mathrm{H}}_2\mathrm{O}$. Full circles: $\mathrm{LiCl}+40{\mathrm{H}}_2\mathrm{O}$.

Figure 3

LiCl aqueous solution. (a) Symbols: Spectra at two selected temperatures in the range $T\le 248$ K. Continuous lines: Results of the fitting with Eq. (4). (b) Symbols: Spectrum at 244.8 K for LiCl solution at $R=40$. Continuous line: Result of the fittings with Eq. (4). Inset: Spectrum of confined water, as obtained by inverting Eq. (4).

Figure 4

Values of the parameter as obtained through the fit of the experimental spectra with Eq. (4), as a function of the calculated volume fraction of water molecules free to self-interact. Inset (a): $R$ dependence of the minimum temperature of validity for Eq. (4). Inset (b): Values of the parameter as a function of the experimental densities of the solutions at 290 K.