Skip to Content

Chemical News

Water Isn’t Water Down There

When I look at the chemical literature now versus what it was like when I was in graduate school, two things stand out: we’re gotten a lot better at looking at single-molecule events and characterizing behavior at a very small scale, and at the same time, we’ve gotten a lot better at detailed characterization of larger-scale structures and materials that used to be in “throw up your hands” territory. Here are a couple of examples of the former from the recent literature.

I’ve written before about how it’s become clear that our ideas about bulk water are just a crude approximation when you get down to the molecular scale. Surfaces and edges of water against some other substance have very different properties than the bulk phase. And as for single water molecules inside protein active sites and such places, forget it – those things are just plain weird, which makes life difficult for computational chemists, because they’re also very important in drug design.

One prediction about confined water molecules is that very thin layers of the solvent should have a wildly different dielectric constant than what we’re used to. That’s one measure of a solvent’s polarity, although the physicists would rather call it “relative permittivity“. Either way, it’s defined as how much the electric field between two charges is decreased in some medium relative to a straight vacuum (whose value is set to 1). The “dielectric constant” name came from the way that you can measure this in a capacitor, versus one that just has vacuum between its plates. To give you an idea, the value for hexane is 1.9, Teflon is 2.1, Pyrex is 4.7, acetone is 21, methanol is 30, DMF is 38, DMSO is 47, and water is 80 (these can vary with temperature). There’s more than one factor that goes into these numbers (iodinated compounds, for example, have rather high values because of the polarizability of the iodine atoms), but it large fits with chemists’ intuition of polarity.

Now this paper (a multinational collaboration between the University of Manchester and co-workers in Iran, Spain, and Japan) has actually measured values for water in extremely confined spaces (various-sized boron nitride channels). And lo and behold, the predictions are correct. Layers of water molecules up to about 3 atoms deep actually show a dielectric constant of about 2. The authors refer to this as an “electrically dead” layer, and I see why: polar bulk water has a thin skin of what might as well be Teflon around it. It’s very useful to have experimental verification of this, and the technique can doubtless be used on other solvents as well. Nanoscale workers will be very interested in such measurements, because (as you can see) assuming that “a substance is a substance no matter how small” will lead you wildly astray on such scales. When water starts acting like some weird form of cyclohexane, you need to know when and where that’s going to happen.

Here’s another nanoscale paper (Univ. of Twente in the Netherlands and Univ. Toronto) that gives you that odd feeling of peering into unseen realms as well: in this case, the authors are looking at what happens when you literally pull a single polymer chain off of a surface. They do that with an atomic force microscope tip, where the polymer initiator is actually on the tip, so the chain grows from that point, and they do this in an aqueous environment with two different polymers (one more water-soluble than the other) and three surfaces of varying polarity themselves.

What shows up is broadly similar across the range: when you’re pulling straight up, or close to it, there’s very little dependence on the force needed versus the angle you’re pulling at. But as you work your way down, things change abruptly at about a 50 degree angle. At that point, the force needed to pull off the polymer chain gets higher, and increases strongly the shallower you go from there. This is in line with at least one model of absorption/desorption behavior, and here’s another example of getting actual experimental proof of what we think is happening at the nanoscale. If I’m interpreting the paper’s Figure 7 correctly, though, the models predict even stronger angle dependence than what was observed, and it looks like they really start breaking down at the lowest angles to the surface, predicting much larger forces than really seem to occur. It’s interesting that this behavior is so similar across different surfaces and between two polymers that will have very different interactions with the water molecules as they come off the surface.

And that makes me wonder if there’s an intersection between these two papers: could the water molecule layers down around the substrate surface and around the polymer itself have a very different dielectric constant than what you might think? I don’t know to what extent this has been taken into account in the modeling of the polymer behavior – that may well be a step beyond what they’re looking at, but that first paper suggests that it’s going to be real-world effect.

Now, you may have read this far (has anyone?) and thought “What does this have to do with drug discovery, eh?” But consider the inside of a cell: protein surfaces sticking together in aqueous media, DNA and RNA polymers unwinding from their sites of production, proteins and small molecules coming on and off membrane surfaces, and so on. All our drug targets are down on this scale, and if they behave in ways that we don’t understand – and they sure do – it’s partly because we don’t understand even the basic chemistry and physics at this level. So the  more experimental data we can get, the better.

15 comments on “Water Isn’t Water Down There”

  1. Mutha Yucka says:


    1. Spider says:

      Gold titanium samarium

      1. jerzy popieluszko says:


      2. peptoid says:

        Oh deary me.

  2. Wallace Grommet says:

    And the first shall be last…

  3. Uncle Al says:

    Before one wonders about confined anomalous water, separate polarity from hydrogen bonding. Water is all about hydrogen bonding and surface tension. Squeeze two new microscope slides together under water. That “water” film freezes at -40° C. Acetonitrile and pivalonitrile have large dielectric constants (37, 21) and curiously high boiling points (82° C, 101° C; t-BuOH at 83° C).

    Acetonitrile freezes at -45° C, hinting that hydrogen bonding rather than dielectric constant is the hammer.

  4. Design Monkey says:

    Welll, IIRC usually water molecules inside receptors already are modelled with scaled down, wimpy, averaged dielectric constants, nothing like the one of bulk water. That probably was tweaked empirically, by observing behavior of models (less screwed, when scaled down).

  5. Peter Kenny says:

    I’m not sure what meaning a dielectric constant has for a cluster of water molecules which may have a dipole moment although these are really interesting measurements. The link between dielectric constant and molecular recognition is not as strong as one might think. For example, consideration of the dielectric constant (47) of DMSO might lead one to believe that it would interact equally well with cations and anions.

    1. Curious Wavefunction says:

      Exactly. It’s like asking what’s the temperature of a few dozen atoms of gold. But assuming the dielectric constant is what it is, it raises interesting questions about polarity. Consider that lipophilic sites inside proteins are usually assumed to have a dielectric constant of about 4, and that’s because they have essentially no polar electrostatic interactions and only non-polar VdW interactions. If small-scale water truly has a dielectric constant of 2, what’s happening to its electrostatics and its hydrogen bonds? Have these features disappeared?

      1. Peter Kenny says:

        Water molecules within a confined space will orient themselves to form hydrogen bonds with each other in the energetically most favorable manner. If they can’t ‘find’ energetically favorable orientations then they simply won’t allow themselves the be confined (see the Gerhard Klebe’s article that I’ve linked as the URL for this comment). It is worth remembering that contact between polar and non-polar molecular surfaces is not inherently repulsive even it may be destabilizing in the context of association of ligand with protein. The dielectric constant is a bulk (rather than a molecular) property and it quantifies the ability of an applied electric field to induce a dipole in the dielectric.

  6. Barry says:

    two molecules of water can’t form ice
    at best, they share a single hydrogen
    no less than five could possibly suffice
    to sketch the unit right tetrahedron
    thereafter, each incomer’s free to dock
    if its kinetic energy’s alright
    forsaking role as gas for role as rock
    to claim a free coordination site
    six vertices in six directions grow
    each independently hews to the norm
    as building blocks march outwards, row by row
    defining a new snowflake’s starry form
    reductionism picks a thing apart
    to grasp it by the method of Descartes

    1. Chris Phoenix says:


      Anyone who likes this should google “Lab Lemming geosonnet”.

    2. Rhenium says:

      Oh my goodness, beautiful chemistry poem!

  7. anon electrochemist says:

    This paper dismisses out of hand the entirety of electrochemistry and surface science which constantly involves measurements of electron transport through few-layer films of structured water at the interface. These water surface films have been well-studied since Helmholtz!

    I would have instantly rejected this one.

  8. GladToMoveToProcess says:

    Several years ago, I tried to help a colleague with criticisms of a paper dealing with solvent isotope effects in some biochemical system. The criticism centered on the different viscosities of H2O and D2O. I argued that the differences in bulk viscosities probably didn’t apply to the few-molecule layers involved in exposing the enzyme active site, but had no evidence for this. A local liquid expert wasn’t able to help us. Does anyone have any real evidence about this?

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.