We know how unique water is and this question is based on a recent discovery that water can exist in two different liquid states. The density of the two states vary by almost 20 %.
Can someone explain me how a single fluid is being converted into two fluids of different densities at the atomic scale? Also why are the states observed at such a low temperature and not ordinary temperatures?
The mentioned recent discovery can be better described as a new experimental confirmation of a quite old conjecture. There is some reference to some hypothesis by Röntgen about the coexistence in low-temperature water of two different phases. Starting from the '90s of the last century, a combined theoretical, experimental, and computational effort has added increasing evidence in favor of the presence of a phase coexistence between a high-density liquid (HDL) and low-density liquid (LDL) in supercooled water.
Experimental evidence and computer simulations point towards a liquid-liquid first-order phase transition with a critical point well inside the supercooled region with an estimated critical temperature below 190 K.
Such a low temperature explains why experimental confirmations took so long and why part of the physical picture is still a work in progress.
The answer to the question why are the states observed at such a low temperature and not ordinary temperatures? is connected to the low critical temperature. It is a general rule that distinction between different fluid phases is possible only below the critical temperature. In turn, the precise location of the critical temperature is due to the interactions.
The question about what happens at the atomic scale is interesting but only part of the information can be considered as established. It is a general difficulty in the description of liquid phases, the lack of time persistence of any local structure. the characteristic diffusion process of a liquid always reshuffles atomic positions. It is not a case that experiments give easy access to average spatial distribution of atoms or molecules. For an example of the information that could be gathered about average positions see this paper, in particular fig.3.
What can be said for sure is that the LDL water should have a higher tetrahedral local order than HDL. Here, by tetrahedral order, one should not intend the presence of perfect tetrahedra in the liquid, but a well-defined shell of 4 first neighbors of a central water molecule (in average). HDL shows evidence of a less definite tetrahedral order that could be described as due to interstitial molecules, somewhat borrowing from solid-state physics language. Additional characterization of the two phases is possible, but hinging on computer simulations. In such a case, the role played by different models for the interactions is still a matter of debate.
