PERSPECTIVES ical (and, possibly in the future, chemical and biological) grounds. There is a sense of urgency that paleoclimate reconstructions must continue apace because key potential proxies are threatened, for example, through the melting of glaciers with global warming (13), harvesting of trees, and degradation of coral reefs. But it is also important to discuss the results and place them in context. This ought not be done with mere speculation, such as extrapolating broad conclusions about global mean temperatures or changes in El Niño from a single reconstruction. Conclusions about changes in ENSO from records that do not resolve interannual variability (7, 8) are especially fraught with difficulty. There is often confusion on what constitutes ENSO. In particular, ENSO involves both a combination of changes in the background climate state (such as whether it becomes more “El Niño–like” or not) and changes in the in-
terannual variability. What does the term “permanent El Niño” (14) mean? Certainly, SSTs alone do not define ENSO. It is clear that the expression of ENSO—including changes in the patterns of the Southern Oscillation and atmospheric circulation, changes in associated precipitation patterns, and changes in SST patterns—in models and in the past vary from that of today. Speculations depend critically on a conceptual framework, and it is desirable to use physically based arguments wherever possible. Climate system models can help enormously in this endeavor. At the same time, the improved knowledge of the climate record can help to evaluate and improve climate models, which have their own flaws. Climate dynamics and paleoclimate experts should challenge each other’s interpretations and assumptions. The mix is much more powerful than either community going it alone.
References and Notes 1. M. E. Mann et al., Geophys. Res. Lett. 26, 759 (1999). 2. R. G. Fairbanks, M. Sverdlove, R. Free, P. H. Wiebe, A.W. H. Bé, Nature 298, 841 (1982). 3. M. Werner, U. Mikolajewicz, M. Heimann, G. Hoffmann, Geophys. Res. Lett. 27, 723 (2000). 4. K. E. Trenberth et al., J. Geophys. Res. 103, 14291 (1998). 5. M. Blackmon et al., Bull. Am. Meteorol. Soc. 82, 2357 (2001). 6. B. L. Otto-Bliesner, E. C. Brady, J. Clim. 14, 3587 (2001). 7. A. Koutavas, J. Lynch-Stieglitz, T. M. Marchitto Jr., J. P. Sachs, Science 297, 226 (2002). 8. L. Stott, L. C. Paulsen, S. Lund, R. Thunell, Science 297, 222 (2002). 9. R. S. Bradley, M. Vuille, D. Hardy, L. G. Thompson, Geophys. Res. Lett. 30, 10.1029/2002GL016546 (2003). 10. M. Vuille, R. S. Bradley, M. Werner, R. Healy, F. Keimig, J. Geophys. Res. 108, 2003 (10.1029/2001JD002038 (2003). 11. J. E. Cole, D. Rind, R. S. Webb, J. Jouzel, R. Healy, J. Geophys. Res. 104, 14223 (1999). 12. M. Werner, M. Heimann, J Geophys. Res. 107, 10.1029/2001JD900253 (2002). 13. L. G. Thompson, Science 298, 589 (2002). 14. M. Huber, R. Caballero, Science 299, 877 (2003). 15. We thank J. Cole, D. Schrag, C. Ammann, and H. Cullen for comments and suggestions. NCAR is sponsored by the National Science Foundation.
M AT E R I A L S S C I E N C E
Creating Transient Crystal Structures with Light Craig W. Siders and Andrea Cavalleri
short-lived (millisecond) ionic state with long-range ferroelectric ordering. One of the most intriguing aspects of this experiment is that the product phase is not spontaneously formed by adding energy to the system. Rather, it usually appears only after the temperature is reduced below ~81 K. The experiment thus goes well beyond previous demonstrations of impulsive triggering of a phase transition (4–6). It shows that short light pulses can generate new product phases, directing the system toward otherwise unexplored local minima of the potential-energy landscape. This approach may be used in a way analogous to the use of doping, temperature, and pressure to control materials properties. Several aspects of this photoinduced transition remain mysterious and will stimulate new work. The response of molecular
ptical control of phase transitions (1) and chemical reactions—the use of light to drive matter through otherwise unexplored pathways—has long been a guiding vision for optical scientists (2), with potential applications ranging from new photonic materials to macromolecular chemistry. However, it is difficult to test the evolution of these photoexcited systems because the transition states between parent and product phases are short-lived (10−14 to 10− 12 s). Ultrafast spectroscopy in the visible and near-infrared provides appropriate time resolution but only indirect information about transient structures. The development of ultrafast x-ray probes now promises direct access to the dynamics of atomic, electronic, and magnetic structure. On page 612 of this issue, Collet et al. use such time-resolved x-ray diffraction to demonstrate optical control of a phase transition in a molecular charge transfer crystal (3). They provide convincing evidence of a
photoinduced structural transition in tetrathiafulvalene-p-chloranil (TTF-CA). Excitation of the sample with near-infrared light, combined with measurements of xray diffraction patterns comprising more than 800 spots, results in the unique determination of a three-dimensional (3D) transient structure. The phase transition described by the authors entails light-activated symmetry breaking of a stable, paraelectric structural phase to a metastable,
C. W. Siders is at the School of Optics/CREOL and the Florida Photonics Center of Excellence, University of Central Florida, Orlando, FL 32816, USA. E-mail:
[email protected] A. Cavalleri is in the Materials Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA. Email:
[email protected]
Concerted excitation. Schematic drawing of the mechanism leading to the macroscopic ordered phase. Step 1: an optical pulse excites a molecule; step 2: an intermolecular charge transfer occurs accompanied by a lattice relaxation, that is, a dimerization process, trapping the excitation; step 3: cooperative phenomena take place with the self-multiplication of the excited molecules (red) inside the stack; step 4: interstack interactions lead to the 3D ordering of the dimers, with a photon efficiency so high that one photon transforms a few hundred molecules.
O
Step 1
Step 2
Step 3
+
+
–
–
Step 4
e– + –
Photoinduced cooperative molecular switching
www.sciencemag.org
SCIENCE
VOL 300
25 APRIL 2003
3D ordering of ionic dimers
591
PERSPECTIVES crystals to intense photoexcitation appears to differ from that of inorganic solids, in which electronic excitations are delocalized and tend to trigger long-range lattice instabilities. Instead, a multistep process is the most likely interpretation for the phase transition observed by Collet et al. (see the figure). In the first step, localized absorption of a photon at a single molecular site produces a charge transfer transition state (step 1 in the figure) with an energy that greatly exceeds that characteristic of thermal vibrations. At a later time, structural motion ensues; the energy is shared with neighboring molecules and converted into a distributed rearrangement of the stack (steps 2 through 4). Although these early events could not be resolved because of insufficient time resolution, switching of long chains is thought to act as precursors for the formation of the observed macroscopic 3D domains (step 4). The system then seems to experience a domino-like effect, with domains of hundreds of unit cells switched by a single photon-absorption event. Highly cooperative interactions are thought to dominate domain formation, with domain growth dynamics that differ markedly from conventional temperature- and pressure-induced first-order phase transitions. The low dimensionality of the initial structural instability (1D chains in this case) may also affect the formation of product-phase domains. The experiment by Collet et al. adds a fine example to the short-pulse x-ray scat-
tering investigations that have appeared in the past few years. Some studies have used very fast (
