Stockholms universitet

Group Members
Job Openings
How to find us

Water still holds many surprises.

As part of our long-term project to understand chemical bonding between molecules we have moved from strong (covalent and ionic) chemical bonds to weaker bonds such as the hydrogen bonding between amino acids and between water molecules. Applying our new synchrotron radiation and spectrum calculation techniques to the classical example of liquid water we found that the spectra did not at all look as one would expect. Current understanding is that water should be nearly four-fold coordinated, i.e. each molecule would bind to close to four other molecules. The spectrum should thus look like ice (where the molecules are truly four-coordinated), but it is very different. This had to be investigated!
From calculations of spectra using structures from molecular dynamics (MD) simulations of the liquid a picture began to emerge where the specific features in the liquid spectrum are due uniquely to molecules that have one donating hydrogen bond weaker than the other (one of the two hydrogens is less strongly interacting with its neighboring water molecule). This asymmetry between the hydrogens is the origin of the extra pre-edge feature (at 535 eV) and strong main-edge (at 537 eV) compared to ice. Through careful model experiments on the bulk and surface of ice (where most molecules have one of their hydrogens uncoordinated, i.e. without a partner molecule to bind to) this suggestion was experimentally confirmed. Having these experimental model spectra we could experimentally determine how many molecules in the liquid are in an asymmetric situation. The surprising result was around 80% at room temperature and not very different near the boiling point!
Going back to theory we could make a simple model of the ice using 11 molecules. By distorting this model (elongating or bending bonds) we could interpolate between the two experimental models through calculated spectra. This gave us geometrical criteria for how large the distortions must be to cause the observed features. In the process we generated a library of spectra that we could use to independently analyze the experimental spectrum. The result was again around 80%.
From the computed spectra we found that elongation or bending of the bonds could not be distinguished; the spectra are very similar and only sensitive to the asymmetry at the hydrogens, that is if one is bound to another molecule while the other isn't. A comparison with neutron diffraction data, which measures distances but is less sensitive to angles, solved the problem and showed that the bond is weakened through a combination of bending and elongation.
We thus arrive at a picture of the liquid where the molecules on average only use two of the possible four bonds (we can only measure the situation on the hydrogen side, not if there are hydrogens from other molecules attached to the oxygen, but a broken donating bond implies that a bond has been lost on an oxygen as well). There seems to be two stronger and shorter bonds and two weaker ones which are distorted off the preferred straight line. If we speculate about what this means we arrive at chain- or ring-like structures in the liquid. This is very exciting and highly controversial since it goes strongly against the current picture. Note, however, that it is in agreement with presently available experimental data, but not with theoretical simulations of the liquid.
The work, published in Science (available online here), was selected as one of the ten most important scientific breakthroughs in Science Top Ten 2004.
More information and a high-resolution image of water molecules is available at the SSRL group web site. For further reading, see the Local structure of water page under the Research section. A recent (March, 2006) and well-written news story about the research was published in Symmetry magazine from SLAC and Fermilab, USA, which can be found here.
Stockholm University |AlbaNova University Center| Fysikum |Updated 14 June 2010 | back to top |