Nuclear Instruments and Methods in Physics Research A 467–468 (2001) 1125–1128

Observation of hard X-ray pulses with a highly sensitive streakcamera

T. Hara*, Y. Tanaka, H. Kitamura, T. Ishikawa

Institute of Physical and Chemical Research, Harima Institute, SPring-8/RIKEN, 1-1-1 Kouto, Mikazuki-cho,

Sayo-gun, Hyogo 679-5148, Japan

Abstract

We have developed a highly sensitive X-ray streak camera system, which synchronously operates with the RF signal

of the SPring-8 storage ring. The streak camera was installed at an undulator beamline of SPring-8, and the beamloading effect for various electron bunch structures (filling pattern) has been observed. The camera has also beenoperated as a timing monitor for a synchronization system of synchrotron radiation and a Ti:sapphire laser. The highly

sensitive X-ray streak camera system can be used not only as an accelerator diagnosis, but also as a fast temporaldetector for beamline applications, such as the observation of fast temporal transitions of diffraction images andrelaxation process. # 2001 Elsevier Science B.V. All rights reserved.

PACS: 07.85.Qe; 07.85.Fv; 07.68.+m; 79.60.�i

Keywords: X-ray streak camera; X-ray detector; Streak camera; CsI

1. Introduction

Streak cameras are one of the most powerfultools for the observation of fast temporal phe-nomena and they have been widely used for visibleand soft X-ray light detection. In third generationsynchrotron radiation (SR) sources, the most ofbeamlines are designed and optimized for softX-ray or X-ray utilization, thus installation ofvisible streak cameras requires a specially designedlight port or a beamline for electron bunchobservation. On the other hand, X-ray streakcameras are compatible with existing beamlines

and they can be set up in an experimental hutchafter a monochromator. In addition, the X-raystreak cameras can be used not only as an electronbeam diagnostic but also as a detector for beam-line experiments. One defect of the X-ray streakcameras is relatively low sensitivity of photo-cathode compared with that of visible streakcameras. An Au cathode is widely used in X-rayregion for its material stability, but accumulativemeasurements are necessary due to its lowquantum efficiency and it may lead to largetemporal resolution. One way to improve thetemporal resolution is to develop a jitter freeaccumulative system [1], and the other way is toimprove cathode sensitivity. We have modified aphotocathode disc of a commercially availablecamera (Hamamatsu Photonics, C5680-06) and

*Corresponding author. Tel.: +81-791-58-2809; fax: +81-

791-58-2810.

E-mail address: toru@spring8.or.jp (T. Hara).

0168-9002/01/$ - see front matter # 2001 Elsevier Science B.V. All rights reserved.

PII: S 0 1 6 8 - 9 0 0 2 ( 0 1 ) 0 0 6 1 1 - 8

installed a CsI photocathode (100 nm thickness).As a result, 20–100 times higher sensitivity andbetter time response were obtained compared witha conventional Au cathode (30 nm thickness) inthe energy range of 8–46 keV [1,2]. Using thishighly sensitive X-ray streak camera, we observedbunch loading effect in SPring-8.

2. Instrumentation

The X-ray streak camera is installed at anundulator beamline (BL29XU) of SPring-8 andFig. 1 shows a schematic view of the camera setup.BL29XU is equipped with a double crystalmonochromator, so monochromatized SRphotons are observed by the streak camera [3].

The camera has two pairs of deflection electro-des. Fast deflection electrodes implement a psorder vertical scan at 84.76MHz, and slowdeflection electrodes displace vertical ps imageshorizontally on a phospher screen. This doublescan makes a ms or ms order record of fastrepetitive signals possible. The ring RF frequencyis 508.58MHz, so one of the six bunches aredetected on the camera when all RF buckets arefilled with electron bunches (fully filled mode). Thecamera scan trigger is generated from 508.58MHzRF signal of SPring-8 after the frequency beingdown converted to 84.76MHz by a 1

6 counter. Dueto low photocathode quantum efficiency to X-ray,

measurements are made in an accumulativemanner, so pulse to pulse jitter on the scan triggeraffects the camera temporal resolution. In order toreduce jitter on the scan trigger, we first operate asynchronized Ti::sapphire mode locked laser [4]with a 1

6 counter output, and we detect the laserpulses with a fast PIN detector and use it as thecamera scan trigger. In this configuration, the laseroscillator works as a frequency filter on the84.76MHz RF signal and the jitter reducesfrom 16 to 5 ps [2]. The intrinsic temporalresolution of the camera is about 4 ps (FWHM)measured by a single shot image of a 2 ps laserpulse (266 nm), and the overall temporal resolu-tion for accumulative measurements is estimatedto be about 6 ps.

A 50 mm Ta slit is inserted in front of thephotocathode and it limits vertical photon beamsize to avoid degradation of the temporal resolu-tion of the camera. The whole camera is tiltedby 58 from the SR beam axis to prevent thetransmitted X-ray directly hitting multi-channelplates (MCP).

3. Observation of X-ray SR pulses

Fig. 2 is a comparison of the observed SR pulsesat 46.5 keV using Au and CsI photocathodes. Bothplots are accumulations of 8� 105 SR pulses, and20 times higher MCP gain is applied for Au. The

Fig. 1. Instruments setup at BL29XU of SPring-8.

T. Hara et al. / Nuclear Instruments and Methods in Physics Research A 467–468 (2001) 1125–11281126

pulse profile obtained using the CsI cathode isclose to Gaussian, whereas that of Au is statisti-cally less accurate. This is due to the small numberof secondary electron emission from the Aucathode compared with the CsI [1].

Fig. 3 shows the beam loading effect measuredby the double scanning mode of the streak camera.Full scale of the horizontal axis corresponds to oneround trip of the SPring-8 storage ring and thevertical position of the image shows longitudinaldisplacement of the bunches. When an electronbunch passes through the ring RF cavity, the fieldinduced by the bunch reduces effective accelerationfield of the cavity. This will be compensated by RFpower supplied by klystrons, but it takes timebecause of a high Q value of the cavity. As a result,following electron bunches gain less energy and

Fig. 3. Longitudinal bunch positions when the ring is operated in (a) 24/29 filled mode and (b) 1/12 filled+10 equally spaced bunches

mode.

Fig. 2. SR pulses observed using CsI (solid line) and Au (dotted

line) cathodes. Figure shows 8� 105 accumulative shots. The

Au plot is measured with a 20 times higher MCP gain.

T. Hara et al. / Nuclear Instruments and Methods in Physics Research A 467–468 (2001) 1125–1128 1127

are shifted longitudinally in order to sit on the RFphase where the bunch can gain energy equivalentto radiation loss. The deformation of the RF fieldis significant when beam current per bunch islarge. In Fig. 3 (a), the ring is filled with 2026bunches (24/29 RF buckets are filled) with0.05mA/bunch. The RF cavity regains accelera-tion voltage during a train of empty buckets(5/29), so the longitudinal bunch positions at thetop and tail of 2026 bunch train are shifted byabout 20 ps. This is more significant in the caseof 203 bunch train (1/12 filled)+10 bunches(Fig. 3 (b)). In this case, the current per bunch isten times larger, and the positions at the top andtail of 203 bunch train shift by about 90 ps.Insertion devices also change the longitudinalbunch position due to the variation of radiationloss caused by gap movement. At SPring-8, weobserved that the bunch moves 42 ps when 14insertion devices close the gap [4].

In the beamline experiment, we have used theX-ray streak camera to observe time delay on the

wave front of SR pulses caused by asymmetriccrystal reflection [5].

Acknowledgements

The authors would like to acknowledgeH.Yamazaki, K.Tamasaku and M.Yabashi fortheir assistance in this work.

References

[1] K. Scheidt, G. Naylor, Proceedings of the 4th European

Workshop on Diagnostics and Beam Instrumentation for

Particle Accelerators, Chester, UK, 16–18 May 1999, p.51.

[2] T. Hara, Y. Tanaka, H. Kitamura, T. Ishikawa, Rev. Sci.

Instrum. 71 (2000) 3624.

[3] K. Tamasaku, Y. Tanaka, M. Yabashi, H. Yamazaki,

N. Kawamura, M. Suzuki, T. Ishikawa, Nucl. Instr. and

Meth. A 467–468 (2001), in this proceedings.

[4] Y. Tanaka, T. Hara, H. Kitamura, T. Ishikawa, Rev. Sci.

Instrum. 71 (2000) 1268.

[5] Y. Tanaka et al., to be submitted.

T. Hara et al. / Nuclear Instruments and Methods in Physics Research A 467–468 (2001) 1125–11281128