Chemical physics

Ultrafast relaxation in water

Article metrics

When a molecule is excited, where and how fast does its excess energy go? The answer to this question is a prerequisite for understanding and predicting the course of many chemical and biological processes. Chemical reactions can take place only when enough internal energy has accumulated in the molecule. The study of chemical reactions is therefore intimately connected with the study of energy-relaxation processes that compete with the chemical reaction channel. Intermolecular excitation transfer is one such process that has been studied for more than half a century. A paper by Woutersen and Bakker on page 507 of this issue1 may force us to re-examine some of our notions about this important relaxation pathway.

When an isolated atom is optically excited it can relax to the ground state only by emitting radiation. In large molecular systems many degrees of freedom compete for the excitation energy and the winner rarely takes all. After a molecule is excited in solution, much of its energy usually ends up as increased solvent thermal motion, and can be regarded as wasted. Chemical interest often lies in other relaxation channels, for example electron transfer or chemical bond breaking. But even pathways that eventually lead to wasted thermal energy, such as intermolecular excitation transfer, can be of great interest at intermediate timescales. Knowing how energy flows between different molecular modes en route to complete relaxation can suggest ways to direct it to more useful channels, similar to harnessing the water flow in a river for useful work.

Intermolecular excitation energy transfer is the process by which one excited molecule, a donor, transfers its excess energy to another, an acceptor, leaving the latter in an excited state. This process continues until terminated by photon emission, chemical reaction or thermal relaxation. The intermolecular excitation pathway can be desirable or not, depending on your objective: it may obstruct an attempt to bring a molecule into higher excited states and it will destroy coherence that may otherwise help control a photochemical reaction. On the other hand it can provide ways to ‘conduct’ energy to where it is needed; an example is the use of sensitizers in photographic films in order to activate photoreactions in species that do not absorb natural light. Nature has also learned how to use these processes, for example in the light-harvesting complexes of photosynthetic systems2.

The theory of such energy-transfer processes goes back to the well-known works of Förster3 and Dexter4. The simplest Förster transfer mechanism is similar to the interaction between two electric dipoles. The rate of energy transfer, k, is described by k=T−1(ro/r)6, where T is the lifetime of the excited state, r is the distance between the donor and acceptor, and ro is a parameter called the Förster radius. This equation tells us that the rate of dipolar energy transfer behaves like r−6. With increasing r, higher-order interactions (such as dipole–quadrupole, quadrupole–quadrupole and exchange interactions) decay much more rapidly than the dipole–dipole interaction, and are effective only at very small intermolecular distances.

How important is this mode of energy flow? It is significant only when its rate is comparable to or faster than other relaxation processes. The most important competing processes are intramolecular vibrational relaxation, where vibrationally excited molecules relax by transfer of energy within the molecule itself, and vibrational energy relaxation, where vibrational energy is transformed into solvent thermal energy. These processes are fast; relaxation of polyatomic molecules in condensed phases at room temperature occurs over a few picoseconds or less.

In contrast, vibrational energy transfer between molecules is generally believed to be too slow to be important. It is only expected to play a significant role for diatomic molecules or at cryogenic temperatures, where vibrational relaxation is relatively slow. Indeed, these are the conditions under which such processes have been observed in the past5,6. Contrary to such expectations, the experiment by Woutersen and Bakker1 shows that resonant intermolecular energy transfer between OH bonds in liquid water is extremely fast. Moreover, it appears to be much faster than the vibrational energy relaxation of the OH group, which has recently been shown to have a short lifetime of 740 fs (femtoseconds)7.

In their experiment, Woutersen and Bakker use two 200-fs infrared pulses: one relatively strong, linearly polarized ‘pump’ pulse to excite the OH groups and another, low-intensity pulse to probe this excitation. They use thin-layer samples of either pure water (liquid H2O), or a mixture of HDO and D2O (D is deuterium, a heavy isotope of hydrogen). The mixed samples make it possible to measure the dependence of the energy transfer rate on the OH concentration. Woutersen and Bakker measure the rotational anisotropy of the molecules as a function of the time delay between pump and probe. Rotational anisotropy is induced by the pump pulse, which excites molecules with specific orientations. This vibrational excitation can then relax either by rotational motion of the excited molecules or by energy transfer between molecules of different orientations (Fig. 1).

Figure 1: Intermolecular energy transfer.

Woutersen and Bakker1 measure energy transfer in pure H2O and in mixtures of HDO dissolved in D2O using two infrared pulses. A pump pulse, polarized as shown in a, excites the OH bond of the HDO molecule, b. The induced rotational anisotropy can relax either by rotation of this molecule to configuration c, or by energy transfer to another molecule, d. Averaging over all final orientations amounts to loss of anisotropy, which is monitored by the probe pulse (not shown).

These two relaxation modes can be distinguished from each other for the low OH mixtures (that is, HDO dissolved in D2O): energy transfer between HDO molecules depends on the concentration of this species, whereas their rotation does not. Therefore measuring the rotational anisotropy at different delay times as a function of HDO concentration yields the characteristic time for molecular rotation (four picoseconds), and more importantly the rate of intermolecular vibrational energy transfer between the OH groups. These results show that the Förster theory accounts well for the observed intermolecular vibrational energy transfer in HDO–D2O mixtures and that the corresponding transfer rate is quite fast — in the range of a few picoseconds for molar concentrations of OH. With Woutersen and Bakker's technique, the transfer rate is measurable even though the competing processes of energy relaxation are very fast.

The real surprise comes from similar measurements in pure H2O. Using the Förster results from mixtures of HDO in D2O (ro=2.1 Å) to extrapolate to the intermolecular distance in pure water (2.8 Å) predicts an energy-transfer time in the range of a few hundred femtoseconds. But the observed intermolecular energy transfer in pure water takes place even faster than the experimental time resolution of 100 fs. This is considerably faster than the 740-fs lifetime of the excited OH population, and makes intermolecular vibrational energy transfer one of the fastest relaxation processes ever recorded in water. This means that vibrational energy cannot be localized in water long enough to affect most chemical reactions. On the other hand it implies that water is an extremely good conductor of vibrational energy through its OH groups. It is even possible that this energy-transfer process could involve other molecules containing OH groups, so water may play an important role in protein dynamics when energy is transported between different molecules.

The failure of the dipolar Förster theory in H2O is not unexpected because the OH groups are so close to each other that higher-order interactions come into play. But the observation that intermolecular vibrational energy transfer in water is so amazingly fast calls for a reassessment of vibrational energy transfer and relaxation in condensed phases. Many questions remain. For example, what is the actual rate of vibrational energy transfer in water? Is this behaviour peculiar to this liquid (perhaps it is associated with its special structural properties)? Previous studies from the same laboratory7 have suggested that the fast vibrational energy relaxation of the OH bond in water is possibly associated with its coupling to the nearest hydrogen bond. Is there a link between that and the present observations? If not, what is the origin of this extremely fast energy-transfer process? The need to answer these questions is our next challenge.


  1. 1

    Woutersen, S. & Bakker, H. J. Nature 402, 507–509 (1999).

  2. 2

    van Oijen, A. M., Ketelaars, M., Köheler, J., Aartsma, T. J. & Schmidt, J. Science 285, 400–402 (1999).

  3. 3

    Förster, Th. in Modern Quantum Chemistry Vol. III (ed. Sinanoglu, O.) 93–137 (Academic, New York, 1965).

  4. 4

    Dexter, D. L. J. Chem. Phys. 21, 836–850 (1953).

  5. 5

    Legay, F. in Chemical and Biochemical Applications of Lasers Vol. II 43–86 (Academic, New York, 1977).

  6. 6

    Apkarian, V. A., Wiedeman, L., Janiesch, W. & Weitz, E. J. Chem. Phys. 85, 5593–5610 (1985).

  7. 7

    Woutersen, S., Emmerichs, U., Nienhuys, H.-K. & Bakker, H. J. Phys. Rev. Lett. 81, 1106–1109 (1998).

Download references

Author information

Correspondence to Abraham Nitzan.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Nitzan, A. Ultrafast relaxation in water. Nature 402, 473–475 (1999) doi:10.1038/44976

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.