A functional test for a pulse picker for synchrotron radiation was performed at Diamond Light Source. The purpose of a pulse picker is to select which pulse from the synchrotron hybrid-mode bunch pattern reaches the experiment. In the present work, the Bragg reflection on a multilayer was modified using surface acoustic wave (SAW) trains. Diffraction on the SAW alters the direction of the x rays and it can be used to modulate the intensity of the x rays that reach the experimental chamber. Using electronic modulation of the SAW amplitude, it is possible to obtain different scattering conditions for different x-ray pulses. To isolate the single bunch, the state of the SAW must be changed in the short time gap between the pulses. To achieve the necessary time resolution, the measurements have been performed in conical diffraction geometry. The achieved time resolution was 120 ns.
© 2017 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
X-ray time-resolved experiments allow the investigation of the dynamics of chemical reactions or physical phenomena [1–3]. Various pump-probe experiments can be conceived, from measuring the fast structural changes (for an overview of recent x-ray diffraction experiments with time resolutions down to , see ) in dependence on external excitation to spectroscopic in situ monitoring of chemical reactions using x-ray absorption spectroscopy (XAS) (for an overview about picosecond and femtosecond XAS applied to molecular systems in solution, see ). The studied processes span different time scales, from picosecond to millisecond. In pump-probe experiments, a fast change in the sample is triggered by external activation (pumping). A short time after the activation, an x-ray pulse is used to measure the state of the studied sample (probe). Ideally, no additional x-ray pulses should reach the sample, especially if the measurement hardware is not able discriminate the unwanted photons. To be able to serve the experiments probing different time scales at the same time, a versatile modulator for the x-ray pulse time structure is needed. Since at synchrotron light sources many experiments are served simultaneously, such a device is preferably implemented in each experimental station. Synchrotron radiation facilities provide strong and stable x-ray beam pulses that can be used for time-resolved measurements. The pulse sequence is the direct consequence of the filling pattern of electron bunches in the storage ring. Due to the physics of electron acceleration, the electrons are distributed in bins (buckets), which can be filled or not. The repeating frequency of the pattern is, therefore, given by the circumference of the storage ring. Often synchrotrons run in the so-called Hybrid mode fill pattern. At the Diamond Light Source (DLS) this consists of 686 contiguous bunches, 2 ns apart, each with a charge of 0.81 nC (0.43 mA), plus a single 3 nC (1.6 mA) bunch in the middle of the gap in the bunch train of 500 ns; . The power that is stored in the multibunch is much higher than the one stored in the single bunch, and it must be effectively blocked by the pulse picker. Modern pulse pickers are mechanical choppers rotating at high velocities. This comprises a wide range of different technical implementation, such as rotating crystals or mirrors, triangular-shaped metal plates, and the classic design consisting of a rotating disc with its rotational axis parallel to the propagation direction of the light beam [7–17]. The open time window for each bunch picker depends on the design of the bunch picker itself. Mechanical bunch pickers are limited by the rotation speed and by the shape of the aperture, in the order of a few mircoseconds, while rotating crystals bunch pickers are limited by the rotation speed and by the distance of the mirrors to a defining aperture, in the order of a few hundred nanoseconds. Another limitation common to most of the pulse pickers actually in use is that they operate at low repetition rate, in the kilohertz region. Nevertheless, synchrotron sources commonly operate with a repetition rates in the megahertz region. To overcome these problems, Tucoulou et al. presented a different kind of pulse picker based on the x-ray diffraction on a multilayer modulated by a surface acoustic wave (SAW) . The SAW induces a sinusoidal deformation on the multilayer surface, which acts as a grating when illuminated by x rays producing diffraction satellites [19–21]. This device can be used for temporal modulation of x rays by switching the grating structure on and off. If SAWs are pulsed, the diffraction satellites appear only when the SAW train goes through the irradiated area of the crystal surface. SAWs move with a constant velocity, usually between 2000–5000 m/s . The time resolution is defined as the time that the SAW train needs to cross and leave the x-ray beam footprint. Tucoulou et al. performed an experiment at the European Synchrotron Radiation Facility (ESRF) synchrotron facility, and managed to pick a single pulse in the middle of a gap of 1.8 μs using the grating produced by a SAW in meridional geometry . ESRF has a much larger diameter than DLS, and the ion clearing gap is much larger. The time resolution that Tucoulou et al. obtained at ESRF is too low to select the single bunch in DLS. To be of practical value for DLS, the time resolution has to be smaller than the 400 ns wide ion clearing gap. In this Letter we report about an x-ray pulse picker built taking advantage of diffraction of a SAW in sagittal geometry [22,23]. In this geometry, the wavefront is parallel to the scattering plane. This has the advantage of diminishing the time resolution to values that are at least one order of magnitude lower, compared to meridional geometry. The reason is that the SAW is traveling across and not along the footprint, which normally has much smaller transverse size than longitudinal. A similar optical layout was used by Roshchupkin et al. to visualize a SAW in conical geometry, taking advantage of the Talbot effect . In this Letter, unlike in Refs. [22,23], the authors propose to use a geometry close to sagittal, with a small inclination with respect to the SAW wavefront. This geometry takes advantage of the time resolution in sagittal geometry, and at the same time it allows for a good separation of the diffraction satellites, as in meridional geometry. We demonstrate that we achieved the necessary time resolution to isolate the single bunch.
The excitation frequency () of a SAW depends on the velocity () of the SAW on the exploited crystal, and on the period () of the SAW via the simple expression . The vertical displacement of the crystal surface in first approximation can be written as21]. The substrate was a cut of Lithium Niobate (), a piezoelectric crystal. The surface roughness did not exceed 5 Å. For our sample, the resonance frequency was 289 MHz, the SAW wavelength was , and the SAW aperture was 0.3 mm. The propagation velocity was . To excite the SAW, an aluminum interdigital transducer (IDT) was deposited on the surface of the crystal. The IDT converts the high-frequency signal into acoustic oscillations of the crystal lattice that propagate along the crystal surface . The amplitude of the SAW depends linearly on the voltage supplied to the IDT, and it can easily be varied from zero to several angstroms. In the present work, a multilayer with period (0.9 nm W, 1.8 nm , 200 bilayers) was deposited on the surface of a crystal; see Fig. 1. This extends the validity of the experimental results to the soft x-ray range, where the Bragg reflection from is not possible. Previous experiments show that the SAW effectively propagates to a multilayer deposited on the surface of a crystal [26–28]. Since the velocity of light is five orders of magnitude higher than the velocity of a SAW, the acoustic deformation can be considered to be quasi-static and characterized only by its wavelength and amplitude. Since the x rays are being scattered by the multilayer, they do not penetrate deeper into the crystal. Therefore, the kinematical theory of diffraction can be used to describe the x ray–SAW interaction. The interaction of x rays with the SAW in the multilayer can be divided in two distinct diffraction processes that can be treated separately. One can distinguish the multilayer Bragg diffraction from the diffraction due to the SAW acting as a grating. The Bragg angle depends only on the wavelength of the incident radiation , and on the period and the critical angle of the multilayer via the Bragg law:1. The diffraction takes place perpendicularly to the scattering plane; therefore, the Bragg-diffracted and the SAW-diffracted satellites appear simultaneously. The diffraction satellites lie on the surface of a cone. The angular separation of the -th order of diffraction can be written in the small angle approximation as
X-ray diffraction from a SAW-modulated Si/W multilayer was studied in a four-circle diffractometer at the B16 beamline  at DLS. The x-ray energy of 8 keV was selected by a double Si crystal monochromator, with a resolving power of . The beam was focused on the detector using 10 refractive Be lenses. The focal spot on the detector was 6 μm. This allows us to resolve the Bragg peak from the diffraction satellites due to SAW, and the footprint of the beam on the sample is big enough to have a good scattering of the beam by the SAW. A charge-coupled device (CCD) camera detector (Photonic Science Ltd.) was used to measure the diffracted intensity, with a pixel size of 6.5 μm. This corresponds to an angular resolution of 0.1 arcsec. This was enough to resolve the satellite peaks with a CCD sample distance of 1.5 m. The SAW were excited using a high-frequency generator (Hameg, HM8134/5), and a wideband radio frequency (RF) amplifier with 5 W power (AR, KAW1020). The Bragg angle at the multilayer was for the energy 8 keV, and the sample was tilted about the axis by . Surface acoustic waves were emitted in trains of 100 ns duration. The emission of the SAW was correlated with the DLS storage ring in order to scatter the x rays emitted from the selected electron bunch. Delay, and consequently the selected bunch, were varied by a delay generator (DG645, Stanford Research System).
In the CCD camera images, the peaks are rather well separated (see Fig. 2), even though they are not entirely decoupled from the halo of the main beam. The very high intensity of the main beam originates from the crystal areas not affected by the SAW. The sample used for the measurements had a rather narrow SAW path, 0.3 mm, while the beam footprint on the sample was in the order of 15 mm. Thus, a large portion of the incoming beam was hitting sample areas which were not covered by the SAW. Calculated from experimental data, including the scattered intensity from the non-disturbed area, the rejection ratio is about 50%. However this does not reflect the real rejection ratio of a SAW pulse picker, which is proven in static experiments to be around 0.1% .
The choice of represents a balance between the peak separation and the time resolution. To calculate the time resolution of the pulse picker, the integrated intensity from the satellites was correlated to the delay time. The integration area in the CCD image was automatically selected by numerical algorithm. Due to different time dependencies, the algorithm is able to distinguish between the multilayer Brag diffraction and the SAW diffraction satellites. The result is shown in Fig. 3. For a given delay between the storage ring signal and the SAW trigger, the SAW is crossing the footprint of the x-ray beam in the moment when a specific bunch is being scattered. When the SAW train reaches the beam footprint and interacts with the multibunch, the intensity is maximal, normalized to 1 in the plot. The intensity decreases when the delay is such that the SAW train reaches the footprint during the ion gap, and it has a relative maximum in the middle of the gap due to the interaction with the single bunch. We estimated the time resolution at three different positions: at the single bunch position (), and at the falling and rising edge, at 1.4 μs and 1.8 μs, respectively. The full width at half-maximum (FWHM) of the peak at the single bunch position corresponds to the time resolution of our device; in this case . To obtain the time resolution at the falling and rising edge, we differentiate the spectra and fit the two peaks. The FWHM is for the left slope and for the right slope. The fits were carried out assuming a Gaussian distribution. It was shown that the rotation of the SAW device about the axis of optimizes the spatial resolution without spoiling the time resolution. The value of the angle is a balance between the peak separation and the time resolution: rotating the angle to lower values would allow for a better separation of the diffraction satellites, but at the same time it would spoil the time resolution. This allowed us to resolve the diffraction satellites on the detector. The angular separation of the diffraction satellites, as calculated in Eq. (3), would have not been enough to distinguish the diffraction satellites from the Bragg peak with a sample detector distance of 1.5 m in sagittal geometry. The filling pattern at DLS during the experiment had the same time structure as in the standard DLS Hybrid mode described in the Introduction. The only difference was the intensity of the isolated bunch at 1.6 μs, which had the same intensity as the other bunches, Fig. 4. The SAW train generated by our device has a duration of 118 ns, as discussed above. This means that when the SAW train interacts with the multibunch, it interacts with 59 bunches (each bunch is 2 ns apart). Calculation of the intensity shows that the peak intensity of the isolated pulse is about 3% of the maximal intensity. The calculation is the convolution of the time synchrotron time pattern with the Gaussian curve having the width of the measured time resolution (green curve in Fig. 3). It matches the measured intensity.
The achieved time resolution confirms the feasibility of a pulse picker for pulses separated by at least 120 ns.
In the future we intend to implement such a pulse picker also at the BESSY II facility. Being a 1.7 GeV synchrotron storage ring, most beamlines operate in the soft x-ray or in the UV energy range. SAW operates like a grating and generates satellites in the reflected beam regardless of the actual reflection process. To optimize the pulse picker for different energy ranges, the most effective reflection process is selected. The devices based on Bragg diffraction on piezoelectric crystals can be used in the energy range of 3–30 keV, where the efficiency of the Bragg diffraction is close to 100%. The energy range of 1–3 keV can be covered with a multilayer coating of a piezo crystal surface with proper periods of optically contrast materials. And, finally, the energy range below 1000 eV can be covered with total external reflection from super-polished surfaces of metals. In all cases, the design of IDT devices for the SAW excitation remains the same and uses the piezoelectric properties of the substrate. The achieved time resolution is enough pick the single bunch out of the fill pattern at BESSY II. The standard fill pattern at BESSY II differs from the one at DLS due to the different diameter of the storage ring. In the so called Multibunch Hybrid mode, a single bunch at 4 mA is in the center of the 200 ns wide ion clearing gap followed by another isolated bunch of variable transverse excitation at 3 mA and 84 ns later. Together with the multibunch filling and the three slicing bunches on top of the multibunch train, 302 out of 400 possible buckets in the storage ring are filled and topped up .
The authors thank A. Firsov from Helmholtz Zentrum Berlin for his support and his efforts in producing high-quality samples, and A. Malandain and his team from Diamond Light Source for technical support. This work was carried out with the support of the Diamond Light Source.
1. C. Rischel, A. Rousse, I. Uschmann, P. Albouy, J. Geindre, P. Audebert, J. Gauthier, E. Foerster, J. Martin, and A. Antonetti, Nature 390, 490 (1997). [CrossRef]
2. V. Srajer, T. Teng, T. Ursby, C. Pradervand, Z. Ren, S. Adachi, W. Schildkamp, D. Bourgeois, M. Wulff, and K. Moffat, Science 274, 1726 (1997). [CrossRef]
3. R. Schoenlein, W. Leemans, A. Chin, P. Volfbeyn, T. Glover, P. Balling, M. Zolotorev, K.-J. Kim, S. Chattopadhyay, and C. V. Shank, Science 274, 236 (1996). [CrossRef]
4. A. Rousse, C. Rischel, and J.-C. Gauthier, Rev. Mod. Phys. 73, 17 (2001). [CrossRef]
5. C. Bressler and M. Chergui, Annu. Rev. Phys. Chem. 61, 263 (2010). [CrossRef]
6. D. L. Source, Machine Status (2016), http://www.diamond.ac.uk/Science/Machine.html.
7. D. F. Förster, B. Lindenau, M. Leyendecker, F. Janssen, C. Winkler, F. O. Schumann, J. Kirschner, K. Holldack, and A. Föhlisch, Opt. Lett. 40, 2265 (2015). [CrossRef]
8. A. Meents, B. Reime, M. Kaiser, X. Wang, R. Abela, E. Weckert, and C. Schulze-Briese, J. Appl. Crystallogr. 42, 901 (2009). [CrossRef]
9. M. Cammarata, L. Eybert, F. Ewald, W. Reichenbach, M. Wulff, P. Anfinrud, F. Schotte, A. Plech, Q. Kong, M. Lorenc, B. Lindenau, J. Raebiger, and S. Polachowski, Rev. Sci. Instrum. 80, 015101 (2009). [CrossRef]
10. S. Plogmaker, P. Linusson, J. Eland, N. Baker, E. Johansson, H. Rensmo, R. Feifel, and H. Siegbahn, Rev. Sci. Instrum. 83, 013115 (2012). [CrossRef]
11. A. D. LeGrand, W. Schildkamp, and B. Blank, Nucl. Instrum. Methods Phys. Res. A 275, 442 (2002). [CrossRef]
12. A. McPherson, J. Wang, P. Lee, and D. M. Mills, J. Synchrotron Radiat. 7, 1 (1999). [CrossRef]
13. A. McPherson, W. Lee, and D. M. Mills, Rev. Sci. Instrum. 73, 2852 (2002). [CrossRef]
14. M. Gembicky, D. Oss, R. Fuchs, and P. Coppens, J. Synchrotron Radiat. 12, 665 (2005). [CrossRef]
15. M. Gembicky and P. Coppens, J. Synchrotron Radiat. 14, 133 (2007). [CrossRef]
16. K. Ito, F. Penent, Y. Hikosaka, E. Shigemasa, I. H. Suzuki, J. H. D. Eland, and P. Lablanquie, Rev. Sci. Instrum. 80, 123101 (2009). [CrossRef]
17. S. Husheer, J. M. Cole, T. d’Almeida, and S. J. Teat, Rev. Sci. Instrum. 81, 043905 (2010). [CrossRef]
18. R. Tucoulou, D. Roschupkin, I. Schelokov, M. Brunel, L. Ortega, E. Ziegler, M. Lingham, C. Mouget, and S. Douillet, Nucl. Instrum. Methods Phys. Res. B 132, 207 (1997). [CrossRef]
19. D. V. Roschupkin, D. V. Irzhak, R. Tucoulou, and O. A. Buzanov, J. Appl. Phys. 94, 6692 (1995). [CrossRef]
20. D. Irzhak and D. Roschupkin, J. Appl. Phys. 115, 244903 (2014). [CrossRef]
21. A. Erko, M. Idir, T. Krist, and A. G. Michette, Modern Developments in X-Ray and Neutron Optics (Springer, 2008).
22. D. Roshchupkin, D. Irzhak, A. Snigirev, I. Snigireva, L. Ortega, and A. Sergeev, J. Appl. Phys. 110, 124902 (2011). [CrossRef]
23. S. Vadilonga, I. Zizak, D. Roshchupkin, E. Evgenii, A. Pestsiuk, W. Leitenberger, and A. Erko, J. Appl. Crystallogr. 50, 525 (2017).
24. D. Roshchupkin, L. Ortega, A. Snigirev, and I. Snigireva, Appl. Phys. Lett. 103, 154101 (2013). [CrossRef]
25. C. Campbell, Surface Acoustic Wave Devices and Their Signal Processing Applications (Academic, 1989).
26. D. Roshchupkin, I. A. Schelokov, R. Tucoulou, and M. Brunel, Nucl. Instrum. Methods Phys. Res. B 129, 414 (1997). [CrossRef]
27. J. Dong, J. Qi, and R. Miao, Appl. Opt. 49, 2054 (2010). [CrossRef]
28. D. Roshchupkin, R. Tucoulou, and M. Brunel, Appl. Phys. Lett. 75, 639 (1999). [CrossRef]
29. K. J. S. Sawhney, I. P. Dolbnya, M. K. Tiwari, L. Alianelli, S. M. Scott, G. M. Preece, U. K. Pedersen, and R. D. Walton, AIP Conf. Proc. 1234, 397 (2010).
30. H. Z. Berlin, “Machine Status (2016),” https://www.helmholtz-berlin.de/quellen/bessy/betrieb-beschleuniger/betriebsmodi_en.html.