Measurement and Timing-Control Techniques of Femtosecond Electron Pulse Takahiro Watanabe, Kei Nakamura, Hokuto lijima, Yusa Muroya, Tomonao Hosokai, Kenichi Kinoshita, Koji Yoshii, Toru Ueda and Mitsuru Uesaka Nuclear Engineering Research Laboratory, the University of Tokyo, 319-1188, Ibaraki, Japan Abstract. Updated techniques and results on the measurement and timing-control of femtosecond electron pulses are presented. Radiation emitted by an electron pulse was measured by a femtosecond streak camera, a Michelson interferometer, a 10-channel polychromator and a fluctuation method in order to estimate a longitudinal pulse shape of the electron pulse. Measurements by the streak camera, the interferometer and the polychromator agree with one another within the error of 20 %, while that by the fluctuation method was different. The numerical simulation explained the reason for it that the transverse emittance of the electron pulse affects the fluctuation of incoherent Cherenkov radiation. The synchronization of the electron pulse with the femtosecond laser pulse was also carried out. The timing jitter was 330 fs in rms and the hours-long drift was more than 1 ps. The suppression of the drift is under way by introducing a stable water cooler (within 0.01 °C) for the accelerator tubes and RF gun, and an air-conditioner (within 2 °C). INTRODUCTION Generation, measurement and control of ultra-short electron pulses are continuing challenges in accelerators [1], Femtosecond pulses have already been available via a high-quality beam generation from a photo-injector [2], A brand-new technique, i.e., a plasma cathode, is expected to reduce the bunch length another order of magnitude [3,4]. Once such an ultra-short pulse has been generated, one has to verify the success of the generation by a reliable measurement way. The conventional timedomain technique, such as a streak camera, may not be able to catch up with the reduction of the pulse duration into 10-femtosecond regime. It is therefore necessary to develop an alternative measurement scheme. Further, even for the picosecond pulse that can be measured by the streak camera, it is beneficial to construct a simple and inexpensive measurement instrument. One more basic technique, i.e. timing-control of electron pulses, is equally important. The time resolution of the pump-probe experiment is dominated not by the speed of the mechanical shutter and the detector, but by the pulse duration and the precision of the timing-control of pulses. In the paper, the measurement and the timing-control of femtosecond electron pulses are studied. Incoherent and coherent radiations emitted from an electron pulse are measured by a several independent techniques and their results are compared with one another [5. 6]. The characteristics and error sources of the four methods are qualitatively discussed. The timing-control systems for femtosecond laser and electron pulses have been modified and the improvement is shown. MEASUREMENT THEORY Properties of electromagnetic radiation from charged particles depend greatly on the actual particle density distribution. The spectrum of the radiation from the pulse has a CP647, Advanced Accelerator Concepts: Tenth Workshop, edited by C. E. Clayton and P. Muggli © 2002 American Institute of Physics 0-7354-0102-0/02/$19.00 869 cutoff frequency determined by the inverse of the bunch duration. The degree of coherence changes dramatically across the boundary. The frequency-domain diagnostics of electron pulses, which are so-called coherent radiation technique, are based on the finding of the boundary [5-12]. Once one can observe the enhancement of the degree of coherence, it enables to estimate the bunch duration from the relation between the wavelength of the cutoff and the bunch duration. Incoherent radiation, which has much higher frequency of the cutoff, has been used for time-domain diagnostics such as a streak camera. Such a short wavelength radiation does not provide the information on the cutoff. However, as M.S. Zolotorev et al. proposed [13, 14], incoherent radiation is also useful for the frequency-domain measurement. They proposed to observe the relation between the bunch duration and the spectral width, instead of the frequency itself. Further, the relation is obtained not directly from the spectrum but from the statistical analysis of the fluctuation in the spectrum. Here we synthesize the measurement theories for coherent and incoherent radiations. The electric field of the radiation pulse E(CD) can be described as the superposition of radiation emitted from each electron [15]: p(icott), (1) where e(co) is the spectrum from a single electron and N is the number of electrons. The difference of position of each electron numbered k gives rise to the phase difference cotk. The first-order spectral correlation function between different energies can be written as, where the angular brackets denotes the ensemble average over bunches. By expanding Eq.(2), (E(co)E(co')) = =l N(N - l}F(co)F*(co'}}. (3) Here Aa> = CD - CD' and F is the Fourier transform of the bunch distribution. The function is connected to the bunch form factor / by \F\2 = f . The first term consists of a single N and is known to be the incoherent term. The second term is proportional to the square of N and known to be the coherent term. Another important feature of Eq.(3) is related to the function F. In the first term, there exists a single F, which is a function of Ao>. The second term includes two Fs, which is a function of co. These characteristic features of incoherent and coherent terms reflect the measurement principles for incoherent and coherent radiation techniques as discussed in the following. By substituting o) = a)f in Eq.(3), the important equation for the diagnostics using coherent radiation can be derived as, The bunch distribution can then be obtained from the bunch form factor f(co) . Therefore the relation between the bunch duration and the frequency co is the key to the measurement for coherent radiation. On the other hand, the second term vanishes for the incoherent radiation. In order to obtain the bunch form factor from the first term in Eq.(3), one has to introduce the second-order correlation function. By using the Siegert relation, the second-order correlation function g2(co,co') can be described by the bunch 870 form factor as, (5) Eq.(5) gives the understanding of the diagnostic method using incoherent radiation. Not the power spectrum of incoherent radiation but its correlation or fluctuation carries the information on the bunch distribution. Further, the important key is the relation between the bunch duration and the spectral width LCD . MEASUREMENS USING COHERENT RADIATION We introduced a Michelson interferometer and a 10-channel polychromator to resolve coherent radiation. The experimental setup is shown in Fig. 1. Transition radiation is sent to the two apparatus from the thin-Al radiator. Cherenkov radiation emitted in air is sent to the femtosecond streak camera by the retractable Al mirror. Figure 2 indicates the typical output of the two apparatus; (a) the interferogram obtained by the interferometer (b) output of the polychromator. The reconstruction of the bunch shape is calculated from the experimental results by using Eq.(4). As one can see in Fig. 3, three bunch distributions obtained from the streak camera, the Michelson interferometer and the polychromator were compared with each other. As a result, good agreements within 20 % were obtained. Streak Camera =1——— N^ ———————— 1 II Alfoil 6-stage x-stage rn—— 4—————— &± \ \ -II LINAC 35L '\\ 1 \^ i *^°^-^i > /"^A-^j v Oi/ ^^^^\ \ \/ 'J Michelson Interferometer r^ W Figure 1. Experimental setup for measurements of electron pulses. (b) 2.0 0.07 1.5 r 1.0 0.5 0 0.01 -0.5 10 Optical Path Difference (mm) 15 20 25 Wavenumber (cirr Figure 2. Interferogram and output of the polychromator. 871 30 The good agreement can be obtained only with careful thought of error sources [4], When the longitudinal bunch form factor fL(o)) is evaluated from the experimental data, the spectrum from the electron pulse 1(0)), the number of electrons N9 the transverse bunch form factor / r (o>), the divergence angle of the electron pulse x> and the spectrum from an electron I0(co) have to be considered as follows: fT((o)X(N-l) l0(a>)N -1 (6) ~ /r(fl>)ArVoM ' All quantities on RHS in Eq.(6) can be an error source. When it is assumed that each error is much smaller than the actual value, the total error is estimated by, A/ L (ftj)^ A/(ftj) A/ 0 (ftj) ^AJV AX A/ r (ftj) (7) fL(co) * I(co] I(co) IIn0(co] (co) N The misalignment of the interferometer or the polychromator yields the error of the spectrum. The transverse emittance of the electron pulse should be taken into consideration for the estimation of fT(co) and x- The current error is enlarged twice to the error of the longitudinal bunch form factor under the assumption of small error. The spectrum contributed by a single electron, I0(co), is calculated under several assumptions, such that the radiator is an ideal conductor without dispersion and has a finite dimension. In consequence, one has to be careful about these factors. It is worth noticing that all the error factors discussed above does not effect on the measurement limit. Even though the total error A/ r (o>) exists, the short pulse can be estimated if the bunch form factor fL(co) is acquired within an appropriate range. The appropriate range for the bunch form factor is around 0.1 as shown in Ref. [4], Streak camera (FWHM = 1.0ps) Interferometer (FWHM=1.2ps) Polychromator (FWHM=1.0ps) - 2 0 2 4 Time (ps) Figure 3. Bunch distributions by three methods. MEASUREMENTS USING INCOHERENT RADIATION There have been two remarkable achievements of the experimental results using incoherent radiation [16,17]. Both of them observed the spectral fluctuation of undulator radiation. P. Catravas et al., reported first measurements of the full single shot spectra with 100 % fluctuations and the extraction of the bunch duration [16]. V. Sajaev 872 reported the reconstruction of the bunch distribution from the shot noise [17]. In our case, the time-integrated power fluctuation of Cherenkov radiation was observed. The radiation is limited by the band-pass filter and detected by the photo-diode shot-by-shot. The fluctuations were measured as a function of the bandwidth. The experimental results are shown in Fig. 4. There were big discrepancies of the bunch duration between the streak camera and the fluctuation method. The 1.0 ps pulse from the 18 MeV linac was measured to be 4.5 ps. The 1.5 ps pulse from the 35 MeV linac was measured as 30 ps. In order to investigate the reason for the discrepancies, we developed the threedimensional numerical simulation. The electrons are placed at the three-dimensional position in the pulse and emit Cherenkov radiation. The radiation can be imaged onto the focal plane as a ring. The three-dimensional code gives the intensity profile on the focal plane, from which the intensity fluctuation can be estimated. Here the pulse duration is 300 fs and the energy is 20 MeV. The total fluctuation as a function of transverse beam emittance is shown in Fig. 5. One can see in Fig. 5 that the fluctuation is suppressed as the emittance becomes larger. The enhancement of transverse modes by the emittance is summarized in Fig. 6. The horizontal line in Fig. 6 indicates the threshold of the measurable range under the assumption that 10 modes are the largest transverse size. Consequently, typical emittances of our linacs are found to be out of the measurable range. The dependence of the transverse emittance upon the transverse modes of Cherenkov radiation is quite large compared with that of other radiations. It comes from the fact that Cherenkov radiation is not a spherical wave but a shock wave. Fluctuations 40 10 100 1000 vi \ 1 5ps - ^3Qps ® Experiirrental data \ \ 10 "^ 0 10 "> 100 1000 Band width [nm] Band width [nm] Figure 4. Power fluctuation of Cherenkov radiation. Another important feature in Cherenkov radiation is pointed out by M.S. Zolotorev [18]. When Cherenkov radiation is used in the experiment, the radiator length is usually much larger than the formation zone in order to make the intensity enough large. Then the electron emits Cherenkov radiation at each formation zone independently, which suppresses the fluctuation. The radiation lengths of undulator radiation and transition radiation are equal or smaller than the formation zone. Hence one can see that these two radiations are more appropriate than Cherenkov radiation. 873 40 H-\J Sf 1 3 E i 30 30 20 20 * 10 10 • n 0 i 0.1 11 10 0.1 10 Normalized emittance emittance [p Normalized [JT mm.mrad] mm.mrad] Figure 5. 5. Fluctuation Fluctuation as as aa function function of Figure of normalized normalized emittance. emittance. 10000 10000 Available 100 Difficult 18 MeV linac 35 MeV linac 10 1 0.001 0.01 0.001 0.01 0.1 0.1 11 10 10 100 100 Unnormalized emittance emittance [p Unnormalized [jt mm.mrad] mm.mrad] Figure 6. The number of transverse modes as a function of unnormalized emittance. Figure 6. The number of transverse modes as a function of unnormalized emittance. CHARACTERISTICS OF MEASUREMENT METHODS MEASUREMENT METHODS We CHARACTERISTICS have so far demonstrated theOF measurements by the femtosecond streak camera, have so interferometer, far demonstrated the measurements by the and femtosecond streak method. camera, theWe Michelson 10-channel polychromator the fluctuation the Michelson interferometer, 10-channel polychromator and the fluctuation method. Here let us summarize characteristics of these four schemes as shown in Table 1. Since Here let us characteristics of these four schemes as shownconsideration in Table 1. Since we have notsummarize yet achieved reliable experimental results, the theoretical and we yet achievedbyreliable experimental results, theoretical and twohave othernot achievements P. Catravas et al., [16] and V.the Sajaev [17] areconsideration referred. two other achievements by P. Catravas et al., [16] and V. Sajaev [17] are referred. The streak camera and the fluctuation method resolve the incoherent radiation, while streak camera fluctuation method resolve the incoherent theThe interferometer andand thethepolychromator observe coherent radiation.radiation, A single while shot the interferometer and the polychromator observe coherent radiation. A That singleof shot measurement of the pulse duration is available except for the interferometer. the measurement of theispulse duration forAlthough the interferometer. That ofhas the pulse distribution available only is byavailable the steakexcept camera. the streak camera pulse distribution is available only by the steak camera. Although the streak camera has the apparent time resolution of 200 fs, other three methods can measure down to less the time resolution 200 important fs, other three methods canfactors measure downscheme. to less thanapparent tens of femtosecond. It isofquite to know the error of each than tens of femtosecond. It is quite important to know the error factors of each scheme. The time resolution of the streak camera is distorted by the large light intensity and The time resolution the streak aberrations, camera is distorted the large light intensity and aberrations such as ofachromatic sphericalbyaberrations. As discussed in aberrations such as achromatic aberrations, spherical aberrations. As discussedthe in Ref.[5], the coherent radiation technique is influenced by the optical alignment, Ref.[5], the coherent radiation technique is influenced by the the optical alignment, the measurement of electron charge, transverse beam emittance, assumption of the measurement electron charge, beam emittance, the assumption the radiation fromofa single electron, andtransverse aberrations. As for the fluctuation method, weofhave radiation from a single electron, and As for the method, have not investigated experimentally yet. aberrations. One can measure the fluctuation bunch duration withwethreenot investigated experimentally yet.ofOne can measure the bunch duration with orders of magnitude from hundreds femtosecond to hundreds of picosecond by threeusing orders of magnitude from camera. hundredsThe of femtosecond to hundreds of picosecond by using the femtosecond streak coherent radiation technique requires that one the femtosecond streak camera. The coherent radiation technique requires that one 874 should know the bunch duration in advance so that the bunch form factor should be around 0.1. The dynamic ranee of the fluctuation method is determined by that of the spectrometer. The value 10 written in Table 1 is an example. The timing-jitter measurement is available only in the measurement by the streak camera. Table 1. Characteristics of measurement schemes. Radiation Single shot (Duration) (Distribution) Time resolution Error factors Dynamic range Timing jitter Streak Camera Incoherent Interferometer Coherent Polychromator Coherent Flue. Method Incoherent available available 200 fs not available not available « tens fs available difficult « tens fs available not available « tens fs Intensity Aberrations Alignment Electron charge Beam size Theory Aberrations 103 10 10 102 available not available not available not available TIMING CONTROL Timing-control systems for the linacs and the terawatt lasers have been modified. The linac is controlled by the photo-injection and the rf in cavities. The laser is controlled by the timing stabilizer for the oscillator and the pockel's cell. Currently the timing jitter of the system is 1.9 ps at rms, most of which comes from the drift. It took two hours to take all data. The instantaneous timing fluctuates 2 ps at peak-to-peak, which corresponds to 330 fs at rms. Therefore, we have worked for suppression on the drift. The main factor for the drift is the thermal expansion of the optical table, the wall and other large devices. Hence a new water cooler for the accelerating tubes and RF guns supply and a new air-conditioner were introduced. The water temperature is controlled within 0.01 °C and the air is controlled within 2 °C. As a result, the compressed electron pulse keeps its duration at least within 30 shots (12 minutes) as shown in Fig. 7 (a). In the timing-jitter measurement, however, the large drift still remains as shown in Fig. 7 (b). It took an hour to complete the measurement of 180 shots. The drift can attribute to the 3-m optical delay with 12 mirrors. The optical delay was installed before the compressor of the injection pulse in order to tune pump- and probe-pulses at the same timing in the measurement of Fig. 7 (b). Here the bunch duration fluctuates largely again. The detailed performance of the current timing system with the temperature control is now under investigation. 875 (a) (a) 15 10 1 £?! u 12 1 2 minutes minutes <————————————————— >* 5 5 0 ~J\^^^^ / P -5 / e-bunch duration x 3 -10 -15 0 0 10 20 30 (b) time difference difference (b) time e-bunch duration duration xx 33 e-bunch hour 30 30r—^——— 11 hour 25 20 15 10 5 00 50 100 180 150 150180 Shot number number Shot Shot number Figure 7. Stability of the the timing timing system, system. (a) (a) bunch bunch duration duration without without the the 3-m 3-m long long optical optical delay, delay. (b) (b) Figure Stability of bunch duration duration and and time time difference difference between between the the electron electron and and the the laser laser pulses pulses with with the the 3-m 3-m long long bunch optical delay. delay. optical SUMMARY SUMMARY We have have developed developed the the measurements measurements and and the the timing-control timing-control techniques techniques of of We subpicosecond electron pulses. The Michelson Michelson interferometer interferometer and and the the 10-channel 10-channel results with with the the femtosecond femtosecond streak streak camera camera polychromator provided the consistent results factors were were pointed pointed out. out. We We could could not not get get good good within 20-% errors. Several error factors agreement between the streak streak camera camera and and the the fluctuation fluctuation method. method. The The threethreedimensional numerical analysis showed the the effect effect of of the the transverse transverse beam beam emittance emittance upon the fluctuation, fluctuation, remaining remaining the the discussion discussion on on the the formation formation zone. zone. Suppression of hours-long drift drift in the the current current timing timing system system is is under under way way by by o running the new water C) and and air-conditioner air-conditioner(within (within water cooler cooler for for the the linacs linacs (within (within 0.01 0.01°C) 2 o°C). C). We are still facing facing with the the serious serious drift drift (> (> 10 ps ps for for 1 hours) hours) and and investigating investigating its its source. The timing system has has already already been been supplied supplied to to the the pulse pulse radiolysis radiolysis experiment experiment for radiation chemistry [19]. for [19]. ACKNOWLEDGEMENTS ACKNOWLEDGEMENTS We would like to thank Y. Shibata, K. Ishi, S. Sasaki, Sasaki, Y. Y. Sugiyama, Sugiyama, T. Yoshimatsu Yoshimatsu and Y. Kondo (Tohoku Univ.), M. S. Zolotorev, Zolotorev, W.P. W.P. Leemans, Leemans, P. P. Catravas, Catravas, E. E. Esarey Esarey and S. Chattopadhyay (LBNL) for useful useful supports supports and and numerous numerous discussions discussions in in the the measurements of electron pulses. We would also also like like to to thank thank K. K. Takasago Takasago and and K. K. Kobayashi (FESTA) for helpful helpful supports supports in in the the development development of of the the timing-control timing-control system. REFERENCES 1. M. Uesaka et al., Journal of Nuclear Materials, 248 248 (1997) (1997) 380. 380. 2. X.-J. Wang, et al., Phys. Rev. E 54 (1996) R3121. R3121. 3. S. Bulanov et al., Phys. Rev. E 58 (1998) (1998) R5257. R5257. 4. H. Nasr et al., Nucl. Instrum. Meth. A A 455 455 (2000) (2000) 149. 149. 5. T. Watanabe et al., Nucl. Instrum. Meth. A 437 437 (1999) (1999) 1. 1. 6. T. Watanabe et al., Nucl. Instrum. Meth. A 480 (2002) 315. 7. Y. Shibata, et al., Phys. Rev. A 44 (1991) 3445. 3445. 8. T. Takahashi et al., Phys. Rev. E 48 (1993) (1993) 4674. 4674. 876 9. R. Lai, A.J. Seivers, Phys. Rev. E 50 (1994) 3342. 10. R. Lai et al., Phys. Rev. E 50 (1994) 4294. 11. P. Kung, et al., Phys. Rev. Lett. 73 (1994) 967. 12. A. Murokh et al., Nucl. Instrum. Meth. A 410 (1998) 452. 13. M.S. Zolotorev and G.V. Stupakov, SLAC-PUB-7132 (1996). 14. M.S. Zolotorev and G.V. Stupakov, Proc. of the PAC'97 (1998) 2020. 15. E.L. Saldin et al., Opt. Comm. 148 (1998) 383. 16. P. Catravas, et al., Phys. Rev. Lett. 82 (1999) 5261. 17. V Sajaev, Proc. 7th EPAC (2000) 1806. 18. M.S. Zolotorev, personal communications. 19. Y. Muroya et al., Radiat. Phys. Chem. 60 (2001) 307. 877
© Copyright 2026 Paperzz