Vitrification and Glass Transition of Water: Insights from Spin Probe ESR

Three long-standing problems related to the physics of water, viz., the possibility of vitrifying bulk water by rapid quenching, its glass transition, and the supposed impossibility of obtaining supercooled water between 150 and 233 K, the so-called “no man’s land” of its phase diagram, are studied using the highly sensitive technique of spin probe ESR. Our results suggest that water can indeed be vitrified by rapid quenching; it undergoes a glass transition at $\sim 135\mathrm{K}$, and the relaxation behavior studied using this method between 165 K and 233 K closely follows the predictions of the Adam-Gibbs model.

Figure 1

ESR spectra of TEMPOL in (a) fast quenched and (b) slowly cooled water and (c) fast quenched propylene glycol (PG) at a few representative temperatures. For the fast quenched samples, (a) and (c), the radicals are trapped in glassy matrices at low temperatures and give rise to broad “powder patterns” indicative of anisotropic $\mathbf{A}$ and $\mathbf{g}$ interactions. For the slowly cooled sample, (b), below the homogeneous nucleation temperature $\sim 232\mathrm{K}$, a single broad signal from aggregated TEMPOL, which is thrown out of the crystalline water, is observed. The method of estimating the extrema separation $2A_{zz}$ and the amplitudes and the linewidths used in Eq. (1) for calculating ${\tau }_c$ is also indicated.

Figure 2

Extrema separation $(2A_{zz})$ vs temperature for TEMPOL doped water (circles), propylene glycol (hollow squares), and glycerol (stars). All the three samples show sharp transitions of $2A_{zz}$. $T_{50\mathrm{G}}$, the temperatures at which $2A_{zz}=50\mathrm{\text{Gauss}}$, are indicated. The inset shows correlation plot for $T_{50\mathrm{G}}$ vs $T_g$ adapted from Ref. 22. The line and the numbered hollow circles are from the work of Kumler [Ref. [17] ]. The data points correspond to different polymer samples: (1) PDMS; (2) $0.42\mathrm{S}/\mathrm{B}$; (3) $0.62\mathrm{S}/\mathrm{B}$; (4) PS-706; (5) Polycarbonate. Reference 17 has many more data points which have been left out for clarity. Our results are shown in the same graph for glycerol (star), propylene glycol (hollow square), and water (filled square). Not only the $T_{50\mathrm{G}}$ for glycerol and PG lie exactly on the graph but the respective $T_g$’s estimated match with literature values. The value of $T_g$ for water is estimated to be 132 K from the graph.

Figure 3

The temperature dependence of the product ${\tau }_c\times T$ for propylene glycol (PG) (hollow squares) and water (filled squares) plotted as $\log ({\tau }_c\times T)$ vs $T_g/T$. The line through the data of PG is a least square fit to a straight line whereas the smooth line through the data of water is a guide to the eye. In the same figure $\log \eta$ vs $T_g/T$ for water obtained from Ref. 21 is also shown for comparison. The latter shows a few experimental points in a limited temperature range plotted over the theoretical curve determined assuming the validity of the Adam-Gibbs equation. It is seen that while the correlation time of propylene glycol follows the Arrhenius behavior excellently, ${\tau }_c$ of water shows qualitatively the anomalous behavior predicted by the theory.