Search

We study the temperature dependence of the structure and orientation of the first hydration layers of the protein lysozyme and compare it with the situation for a model homogeneous hydrophobic surface, a graphene sheet. We show that in both cases these layers are significantly better structured than bulk water. The geometrical constraint of the interface makes the water molecules adjacent to the surface lose one water-water hydrogen bond and expel the fourth neighbors away from the surface, lowering local density. We show that a decrease in temperature improves the ordering of the hydration water molecules, preserving such a geometrical effect. For the case of graphene, this favors an ice Ih-like local structuring, similar to the water-air interface but in the opposite way along the $c$ axis of the basal plane (while the vicinal water molecules of the air interface orient a hydrogen atom toward the surface, the oxygens of the water molecules close to the graphene plane orient a lone pair in such a direction). In turn, the case of the first hydration layers of the lysozyme molecule is shown to be more complicated, but still displaying signs of both kinds of behavior, together with a tendency of the proximal water molecules to hydrogen bond to the protein both as donors and as acceptors. Additionally, we make evident the existence of signatures of a liquid-liquid transition (Widom line crossing) in different structural parameters at the temperature corresponding to the dynamic transition incorrectly referred to as ``the protein glass transition.''

Recent investigations have demonstrated that the short-time fluctuations in a supercooled liquid can be used as predictors of the long-time dynamic propensity (that is, the regions of the sample with enhanced tendency to be mobile within time scales on the order of the $\ensuremath{\alpha}$-relaxation time). This could mean that the long-time dynamics (the actual mobility of the particles at such long times) would be implicit in the short-time dynamics or else, that the long-time dynamic propensity [as defined in A. Widmer-Cooper and P. Harrowell, Phys. Rev. Lett. 96, 185701 (2006)], while providing a measure of the degree of jamming of the local structure, would only be sensitive to the short time behavior. The first scenario is in clear disagreement with our recent finding that the influence of the local structure on dynamics (as determined by the propensity for motion) is only local in time, fading out at times close to the metabasin lifetime, much before the $\ensuremath{\alpha}$-relaxation time. Thus, in this work we show that the short-time fluctuations in supercooled liquids do in fact represent precursors to the dynamics at intermediate times commensurate with the metabasin lifetime (being thus able to predict the regions of the sample that will present high propensity for motion at such stage) but that the dynamical behavior at later times of the $\ensuremath{\alpha}$ relaxation is unpredictable, in agreement with a metabasin random walk scenario.