The temporal structure and high brilliance of the X-ray beams produced by third-generation synchrotrons open up new possibilities in time-dependent diffraction and spectroscopy, where timescales down to the sub-nanosecond regime can now be accessed. Moreover, the properties of these beams are such that one can envisage the development of the X-ray equivalent to various optical components, such as photon delay lines and resonators. Optical resonators, like those used in lasers, are available at wavelengths from the visible spectrum to soft X-ray energies. Equivalent components for hard X-rays have been discussed for more than thirty years, but have yet to be realised. Here we report the storage of hard X-ray photons in a crystal cavity. The photons are stored for as many as 14 back-and-forth reflections, each cycle separated by one nanosecond [1].

The cavity consists of a pair of vertical plates separated by 150 mm and cut into a monolithic silicon crystal. The 111 orientation is along their surface normals as sketched in Figure 131. The plates are slightly wedge-shaped in order to vary the effective crystal thickness between 50 µm and 500 µm by a horizontal translation perpendicular to the axis of the beam. The experiment was performed at the inelastic scattering beamline ID28. The X-rays were monochromatised by a Si 888 reflection at a Bragg angle of 89.865°, providing a beam of 15.817 keV, with an energy resolution of 3.7 meV and a divergence of about 10 µrad. Photons are Bragg reflected back exactly within the divergence into the axis of the incident beam. A fast avalanche diode detector was placed behind the cavity, to measure the transmitted intensity leaving the cavity.

The Bragg condition for the cavity was determined by an energy variation of the incoming photons by controlled thermal expansion of the monochromator lattice spacing. The result of such a scan is shown in Figure 132a. The exact Bragg condition is fulfilled when the transmitted intensity has a minimum. The observed FWHM on the relative energy scale is E/E = 7.4x10­7 and the minimum transmission through the two 292 µm thick silicon plates is 17% after consideration of normal absorption.

The time dependence of the transmitted beam at the centre of the Bragg reflection is shown in Figure 132b for various crystal thicknesses. The data was obtained in the ESRF's 16 bunch mode where X-ray flashes of 100 ps duration, separated by 176 ns, allow stroboscopic timing experiments [2]. The first curve shows the time structure measured by the detector without the cavity in the beam. There are no delayed photons and the transmitted intensity maximum defines the zero time. The time patterns with the cavity in Bragg position differ qualitatively from the first curve and show a series of sequential maxima separated by 1.0 ns with an exponential intensity decay towards longer times. The successive maxima correspond to photons trapped within the cavity for 1, 2, 3 ... N reflections from both crystal plates. When the beam impinges onto the first crystal slice, there is a probability for transmission, producing the forward diffracted beam. The same holds for the second plate and therefore yields a maximum signal at t = 0. But there is also a probability for reflection at each slice, permitting part of the beam to propagate back and forth and thus travelling several times the cavity length, i. e. multiples of 30 cm corresponding to a pulse separation of 1.0 ns. We observe up to 14 delayed peaks up to 14.0 ns, and even intensity beyond this for the thinnest crystals. The delayed maxima are less intense for thicker crystal slices because the transmission probability for entering and leaving the cavity goes down and one has to compromise between transmission and reflectivity.

We have demonstrated for the first time the storage of X-ray photons, with the highest energies so far of 15.817 keV, for up to 14 ns by multiple bounces in a crystal cavity. Highest ratios of 50% between neighbouring intensity peaks have been observed. Higher number of reflections were observed for thinner crystals. Discrepancies between the observed time patterns of transmitted intensity and calculations based on a simplified dynamical theory of diffraction are most likely due to a slightly distorted crystal cavity.

References
[1] K.-D. Liss, R. Hock, M. Gomm, B. Waibel, A. Magerl, M. Krisch, R. Tucoulou, Nature, 404, 371-373 (2000).
[2] K.-D. Liss, A. Magerl, R. Hock, B. Waibel, A. Remhof, Proceedings of SPIE, 3451, 117-127 (1998).

Authors
K.-D. Liss (a), R. Hock (b), M. Gomm (b), B. Waibel (c), A. Magerl (b), M. Krisch (a), R. Tucoulou (a).

(a) ESRF
(b) Lehrstuhl für Kristallographie und Strukturphysik, Erlangen (Germany)
(c) MTU Motoren- und Turbinen-Union GmbH, München (Germany)