either single crystals or crystalline powders. Diffraction techniques are not applicable to amorphous phases at present, but the exciting possibility looms on the horizon to use ultra-fast techniques to obtain diffraction patterns of possibly as little as one biological molecule (see below).

The intensity and tunability of synchrotron radiation have revolutionized the application of x-rays for studying the structure of macromolecules, enabling much higher resolution information to be obtained on increasingly large and complex molecular systems. The average bulk crystalline structure determined from diffraction studies is expressed as simple, small, symmetric arrangements of atoms in a unit cell. However, local deviations from this average structure are often the driving force behind the collective behavior of a crystalline compound. Neutron crystallography has the unique advantage of high contrast for the location of hydrogen atoms, so it affords information complementary to that normally obtained from x-ray crystallography.

As conventionally applied today, x-ray methods give rise to “time-averaged” structural information. Since many chemical processes, including the making and breaking of chemical bonds, occur in the subpicosecond time domain, time-resolved structural information has been limited and only indirectly available. Recent developments in electron diffraction and soon-to-be-available x-ray laser sources could dramatically improve the investigation of structural dynamics. Compressed electron pulses can be produced with reasonable intensities and widths of a few picoseconds; these are being used to study relatively simple molecular reactions. X-ray free-electron lasers, based on using high energy linear accelerators providing beams to long undulators, have the promise of easily reaching pulse lengths of only a few hundred femtoseconds and, with additional magnetic and optical compression schemes, likely the regime of only a few femtoseconds. Such x-ray free-electron lasers might have sufficient photons in a single pulse to record an entire diffraction pattern, hence bringing the most powerful tools used today for structural determination to bear on understanding chemical and biological reactions.

A major limitation of diffraction techniques has been the need to obtain crystalline samples. If scientists could learn how to crystallize large molecules in a routine manner, a breakthrough would result. In the biological area, this limitation is keenly experienced for membrane-bound proteins, which are important in many biological functions. Scientists are now devising techniques and strategies to crystallize these proteins—if not in three-dimensional, then in two-dimensional lattices.

Future development of spectroscopic structure-determination methods will depend on the availability of more powerful photon and particle sources as well as advances in photon and particle detectors. Impressive progress has been made in molecular structure determinations based on advances in computation power and in computational algorithms, such as fast Fourier-transform techniques, for nearly every form of spectroscopy and diffraction analysis. Hajdu and co-work-