Water
Water was always there.
For years, it was simply the solvent surrounding the luminescent nanoparticles I was studying.
Then, while trying to understand their Brownian motion, I varied the initial temperature of the liquid in which they were dispersed.
Organic solvents behaved essentially as expected.
Water did not.
At a characteristic temperature, water's behaviour changed.
And suddenly, the solvent became the question.
Perhaps this is one of the dangers of familiarity. We stop looking carefully at what seems to be already understood.
Water is familiar enough to be taken for granted. Yet beneath that familiarity lies a liquid whose molecular organisation remains remarkably difficult to understand.
And I fell in love with the problem.
A hidden structural landscape
One useful picture of liquid water describes it as a dynamically fluctuating mixture of local environments with different local organisation and density.
Low-density (LD) environments are associated with more open, tetrahedrally organised hydrogen-bond networks. High-density (HD) environments are more compact and structurally disordered.
These are not separate liquid phases.
They are transient local environments that continually appear, disappear and interconvert.
At ambient conditions, HD environments predominate, while transient LD/HD structural fluctuations may persist over characteristic spatial and temporal scales.
We do not observe those fluctuations directly.
We observe how they affect something else.
Turning the experiment around
What had begun as a study of nanoparticle motion suggested a different possibility:
Experiments and molecular-dynamics simulations suggested a structural crossover involving changes in the organisation of LD and HD environments.
Particularly intriguing was the possibility that, around the crossover, the characteristic spatial extent of these fluctuations could become comparable to the size of the nanoparticles themselves.
Xiaogang Liu called those nanoparticles “nanorulers.”
But subsequent experiments made the ruler less innocent.
A crossover, but not a universal temperature
Experimentally, Tc is straightforward to define.
It is the temperature at which the measured response changes between two approximately linear regimes.
What that temperature means physically is much less straightforward.
Crossover temperatures have been observed in very different water-based systems, using different probes and different experimental observables.
They tend to occur within the same broad temperature range — roughly 40–55 °C, close to the minimum of water's isothermal compressibility — but they are not identical.
If Tc were simply an intrinsic temperature of bulk water revealed by a neutral observer, why should changing the probe change the temperature at which the crossover becomes visible?
Size mattered.
Surface chemistry mattered.
pH mattered.
And effective surface charge density increasingly emerged as a more useful descriptor of the crossover than either size or pH considered separately.
The crossover carried the fingerprint of the water–probe interface.
At first, this looked like a problem.
If the probe changes the water surrounding it, how can it tell us something about the water?
The nanoparticle had begun as a thermometer. It became a probe. Now I had to accept that the probe was also part of the physics.
The hydration layer could no longer simply be treated as bulk water observed locally.
The interaction between the surface and the surrounding water was itself part of what we were observing.
What first looked like a limitation of the probe was becoming the experiment itself.
The observable is in the interaction
This changed the way I began to think about Tc.
I now regard it less as a temperature belonging to water itself and more as an experimental signature of the conditions under which hydration-water structural heterogeneity remains observable on the characteristic spatial and temporal scales of a particular probe.
This does not make Tc less physical.
It makes the physics more interesting.
Different interfaces can affect the surrounding hydration water differently. Different probes can couple differently to these structural fluctuations.
The measured crossover may therefore reflect the conditions under which that heterogeneity remains capable of affecting a particular observable.
How could we test it?
When the crossover disappears
Pressure offered a different way of testing this picture.
In experiments developed by Ramon Filho in Dani Jaque's laboratory in Madrid, ultrasmall CdTe quantum dots were confined in a diamond anvil cell and their luminescence followed as a function of temperature and pressure.
At ambient pressure, they showed a clear optical crossover near 323 K.
At the higher pressures we explored — approximately 0.1 and 0.7 GPa — the bilinear signature was no longer detected, and the temperature dependence became linear within the investigated range.
Within the structural picture we are exploring, increasing pressure favours HD local environments and progressively reduces the population, spatial extent and lifetime of LD/HD structural fluctuations.
The disappearance of the crossover does not necessarily mean that every microscopic LD/HD fluctuation has disappeared.
It may mean that the fluctuations have become too limited in space, too short-lived in time, or both, to remain observable by that particular nanoprobe.
The fluctuations may still be there. The probe may simply no longer see them.
And that opens a door.
A different kind of nanoruler
A conventional ruler measures a length that exists independently of the ruler.
Our nanorulers may be doing something subtler.
Different probes may provide different windows onto the characteristic spatial and temporal scales of the underlying fluctuations.
The probe may help define the window through which this heterogeneity becomes observable.
We do not yet have direct access to those scales.
But the experiments suggest a way of asking for them.
Perhaps Tc reflects the conditions under which the spatial and temporal scales of hydration-water fluctuations become accessible to a particular probe.
That remains a hypothesis.
I am no longer particularly interested in adding another crossover temperature to the list.
I want to understand what makes the crossover observable, what makes Tc shift from one probe to another, and why it sometimes disappears altogether.
And ultimately I would like to know whether the dependence of Tc on the probe can be turned from a complication into a method.
I do not yet know.
That is where the question is now.