- Email: [email protected]
Sub-picosecond bunch length measurement at the TESLA test facility
Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Sub-picosecond bunch length measurement at the TESLA test facility M. Geitz *, G. Schmidt , P. SchmuK ser , G.v. Walter Deutsches Elektronen-Synchrotron FDET, Notkestr. 85, D-22603 Hamburg, Germany 3. Physikalisches Institut der RWTH Aachen, D-52072 Aachen, Germany
Abstract Sub-picosecond electron bunches are required for the operation of future VUV and X-ray Free Electron Lasers. A streak camera, a Martin}Puplett interferometer and a longitudinal phase space rotation method have been applied at the TESLA Test Facility linac to measure electron bunch lengths. 2000 Published by Elsevier Science B.V. All rights reserved.
1. Introduction Future electron-drive linacs for VUV and X-ray Free Electron Lasers (FEL) require electron bunches of at least 1 nC charge, low emittance (typically 1 mm mrad) and bunch length well in the sub-picosecond regime [1]. These high-quality electron bunches are commonly produced by an RF gun-based photo injector followed by a magnetic bunch compression chicane. The TTF photo injector is driven by an intense ultraviolet laser beam (typically 20 mJ pulse energy) to produce 6.25;10 electrons per bunch from a Cs Te photo cathode. The electron bunches are accelerated rapidly by the strong electric "elds (35}50 MV/m) of the gun cavity to avoid an emittance blow-up due to space-charge forces. The bunch length obtained from an RF gun depends on both the laser pulse
length (typically p "8 ps) and the compression caused by the RF "eld within the "rst few centimeters of the gun cavity. By a proper choice of the RF phase a velocity modulation can be impressed on the electron bunch leading to a reduction of its length within the gun cavity. Further bunch compression is achieved by combining o!-crest RF acceleration with a magnetic chicane. The o!-crest acceleration produces a correlated energy distribution along the bunch with the higher energy electrons trailing the lower energy ones. The higher energy particles travel on a shorter path through the magnetic chicane and compression is obtained. The bunches pass a second accelerating section and are then guided to a magnetic spectrometer for momentum analysis.
2. Streak camera measurement * Corresponding author. Tel.: #49-40-8998-4131. E-mail address: [email protected] (M. Geitz). Permanent address: University of Hamburg, D-20146 Hamburg, Germany
The electron bunches produced by the TTF photo injector (Q"3 nc, UV laser pulse length p "15 ps) have been transferred, without the use of the magnetic chicane compressor, to the
0168-9002/00/$ - see front matter 2000 Published by Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 9 0 0 2 ( 0 0 ) 0 0 1 4 0 - 6
SECTION VI.
344
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Fig. 1. Streak camera measurements of the longitudinal bunch charge distribution. Left: minimum bunch length. Right: bunch length versus RF gun phase.
spectrometer magnet. During the acceleration from 16 to 170 MeV, the longitudinal charge distribution can be considered as invariant. Synchrotron light pulses, produced in the spectrometer dipole, are detected with a streak camera of 2 ps resolution. Fig. 1. (left) shows the charge distribution corresponding to the minimum bunch length measured (p "1.95$0.1 mm). The right graph shows the variation of the bunch length with the RF gun phase. Superimposed is the prediction by PARMELA.
3. Fourier transform spectroscopy Coherent transition radiation can be used to determine the longitudinal bunch charge distribution [2,3]. The radiator is a thin aluminum foil arranged at an angle of 453 with respect to the beam direction so that the backward lobe of the radiation is emitted at 903 and is easily extractable from the vacuum chamber through a quartz window. The spectral intensity emitted by a bunch of N particles is I (u)"I (u) (N#N(N!1)" f (u)") (1) where I (x) is the intensity radiated by a single electron at a given frequency u and f (u) is the longitudinal bunch form factor de"ned as the Fourier transformation of the normalized longitudinal charge distribution o . ¹(u) denotes the specX tral acceptance function of the detection device. A Martin}Puplett interferometer, shown schematically in Fig. 2, is used to determine the autocorrelation function of the radiation pulse [4]. The
Fig. 2. The Martin}Puplett interferometer.
diverging transition radiation pulse is transformed into a parallel beam entering the interferometer by a parabolic mirror. The incident radiation pulse horizontally polarized by the "rst grid and then splitted by the beam divider into components of orthogonal polarization entering the two interferometer arms. The polarization is #ipped by the roof mirrors, hence the component which is "rst transmitted by the beam splitter is now re#ected and vice versa. The recombined radiation is in general elliptically polarized, depending on the path di!erence in both interferometer arms. The analyzing grid transmits one polarization component into detector 1 and re#ects the orthogonal
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Fig. 3. Spectral transfer function ¹(u) of the Martin}Puplett interferometer. The dashed curves indicate the estimated uncertainty of the transfer function.
component into detector 2. Two pyroelectric detectors equipped with horn antennas are used as detection devices for the sub-millimeter wavelength radiation. A Fourier transformation of the autocorrelation function yields the absolute magnitude of the longitudinal bunch form factor. The experimentally determined spectral intensity is strongly modi"ed by the frequency dependent transfer function ¹(u) of the interferometer. As shown in Fig. 3, ¹(u) has its maximum, ¹ , at around 400 GHz and decreases
towards lower frequencies due to the "nite size of the transition radiator and di!raction losses in the interferometer, while the reduction towards larger frequencies is caused by transmission and re#ection losses of the quartz window and the wire grids. The oscillatory behaviour is due to interference e!ects in the 100 lm thick LiTaO crystals of the pyro electric detectors [5]. The transmission of the quartz window and the wire grids has been mea-
345
Fig. 5. Longitudinal bunch charge distribution derived from the autocorrelation and coherent power spectrum of Fig. 4. The charge distribution consists of three Gaussians A "0.6, p "0.27 ps, A "0.28, p "1.0 ps and A "0.25, p "2.0 ps. The A denote the amplitudes and the p the variances. G G
sured by means of time-domain THz-spectroscopy [6]. Fig. 4 shows the autocorrelation (left) as measured by the Martin}Puplett interferometer and the evaluated coherent power spectrum. The measurement was performed at a beam energy of 170 MeV with a bunch charge of 1 nC and laser pulse length of p "8 ps. The RF gun phase was adjusted to produce the minimum electron bunch length. The data cannot be described by a single Gaussian charge distribution. Instead, the superposition of three Gaussian curves of di!erent variance and amplitude is required for a good adaption of the model to the measured data. The FWHM bunch length is (1.13$0.45) ps, (see Fig. 5) where the error is mainly caused by the uncertainty of ¹(u). The shape of the longitudinal charge distribution, a short pulse superimposed
Fig. 4. Autocorrelation (left) and coherent power spectrum (right) measured for optimum bunch compression. Data are indicated by circles. The solid curves are the autocorrelation and power spectrum of a simulated longitudinal charge distribution convoluted with the spectral transfer function.
SECTION VI.
346
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Fig. 6. A magnetic bunch compressor chicane followed by an o!-crest acceleration can be used to determine the longitudinal bunch charge distribution.
onto a wider basis, can be explained by a partially uncorrelated bunch energy distribution.
If we choose the matrix elements such that M M #M P0, the energy pro"le measured with the spectrometer is a direct image of the longitudinal bunch charge distribution in front of the compression section [7]. Fig. 7 shows an energy pro"le measurement performed at optimum compression with 1 nC bunch charge at a beam energy of 170 MeV and laser pulse length of 8 ps. The pro"le yields a FWHM energy spread of 1.00$0.07 MeV. Using Eq. (2), the energy pro"le can be transformed into a longitudinal charge pro"le yielding a FWHM bunch length of 2.3$0.3 mm in front of the chicane. The right graph shows the reduction of the electron bunch length when moving along the chicane. The FWHM length of the compressed bunch is 600$110 lm. The shaded area denotes the error of the measurement.
4. Energy spread measurement An e$cient and straightforward way to determine the bunch length is the evaluation of the bunch energy distribution. Fig. 6 shows the magnetic bunch compressor chicane followed by an RF cavity and a spectrometer dipole to determine the energy pro"le. The longitudinal dynamics is described by the transport matrix
M ¸ (2) # # M M M #M # # where M , M and M denote transfer matrix elements of the chicane and the RF cavity of Fig. 6. 1
¸
"
5. Conclusion The uncompressed longitudinal charge distribution has been measured with a streak camera and by observing the beam energy spread. The streak camera measures minimum injector bunch length of p "1.95$0.1 mm. The energy spread analysis X yields p "990$90 lm. The factor of 2 is explainX able by the reduced UV laser pulse length during the energy spread measurement. The compressed bunch length measured by the interferometer (p "200$70 lm) and by the energy spread analX ysis (p "260$50 lm) coincide within their error X bars.
Fig. 7. Left: energy pro"le measurement at optimum bunch compression. Right: the evaluated compression of the electron bunch within the chicane compressor. The initial bunch length is a free parameter of the computation.
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Acknowledgements We thank Dr. Ingrid Wilke and Mr. Maxim Khazan for carrying out the spectral transmission measurements using time-domain THz spectroscopy at the University of Hamburg.
347
[3] E.B. Blum et al., Nucl. Instr. and Meth. A 307 (1991) and references therein. [4] B. Leissner et al., Proceedings of the PAC Conference, New York, 1999 and references therein. [5] C. Settakorn et al., SLAC-PUB-7813, 1998. [6] I. Wilke et al., Proceedings of EUCAS99, in preparation. [7] K. Ricci et al., in: G.R. Neil, S.V. Benson (Eds.), Proceedings of the FEL Conference, Williams-burg, 1998, Elsevier Science B.V., Amsterdam, 1999, p. II-61.
References [1] TESLA-Collaboration, DESY-TESLA 95-01, 1995. [2] J.S. Nodvick et al., Phys. Rev. A, 96(2) (1954) and references therein.
SECTION VI.
Sub-picosecond bunch length measurement at the TESLA test facility M. Geitz *, G. Schmidt , P. SchmuK ser , G.v. Walter Deutsches Elektronen-Synchrotron FDET, Notkestr. 85, D-22603 Hamburg, Germany 3. Physikalisches Institut der RWTH Aachen, D-52072 Aachen, Germany
Abstract Sub-picosecond electron bunches are required for the operation of future VUV and X-ray Free Electron Lasers. A streak camera, a Martin}Puplett interferometer and a longitudinal phase space rotation method have been applied at the TESLA Test Facility linac to measure electron bunch lengths. 2000 Published by Elsevier Science B.V. All rights reserved.
1. Introduction Future electron-drive linacs for VUV and X-ray Free Electron Lasers (FEL) require electron bunches of at least 1 nC charge, low emittance (typically 1 mm mrad) and bunch length well in the sub-picosecond regime [1]. These high-quality electron bunches are commonly produced by an RF gun-based photo injector followed by a magnetic bunch compression chicane. The TTF photo injector is driven by an intense ultraviolet laser beam (typically 20 mJ pulse energy) to produce 6.25;10 electrons per bunch from a Cs Te photo cathode. The electron bunches are accelerated rapidly by the strong electric "elds (35}50 MV/m) of the gun cavity to avoid an emittance blow-up due to space-charge forces. The bunch length obtained from an RF gun depends on both the laser pulse
length (typically p "8 ps) and the compression caused by the RF "eld within the "rst few centimeters of the gun cavity. By a proper choice of the RF phase a velocity modulation can be impressed on the electron bunch leading to a reduction of its length within the gun cavity. Further bunch compression is achieved by combining o!-crest RF acceleration with a magnetic chicane. The o!-crest acceleration produces a correlated energy distribution along the bunch with the higher energy electrons trailing the lower energy ones. The higher energy particles travel on a shorter path through the magnetic chicane and compression is obtained. The bunches pass a second accelerating section and are then guided to a magnetic spectrometer for momentum analysis.
2. Streak camera measurement * Corresponding author. Tel.: #49-40-8998-4131. E-mail address: [email protected] (M. Geitz). Permanent address: University of Hamburg, D-20146 Hamburg, Germany
The electron bunches produced by the TTF photo injector (Q"3 nc, UV laser pulse length p "15 ps) have been transferred, without the use of the magnetic chicane compressor, to the
0168-9002/00/$ - see front matter 2000 Published by Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 9 0 0 2 ( 0 0 ) 0 0 1 4 0 - 6
SECTION VI.
344
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Fig. 1. Streak camera measurements of the longitudinal bunch charge distribution. Left: minimum bunch length. Right: bunch length versus RF gun phase.
spectrometer magnet. During the acceleration from 16 to 170 MeV, the longitudinal charge distribution can be considered as invariant. Synchrotron light pulses, produced in the spectrometer dipole, are detected with a streak camera of 2 ps resolution. Fig. 1. (left) shows the charge distribution corresponding to the minimum bunch length measured (p "1.95$0.1 mm). The right graph shows the variation of the bunch length with the RF gun phase. Superimposed is the prediction by PARMELA.
3. Fourier transform spectroscopy Coherent transition radiation can be used to determine the longitudinal bunch charge distribution [2,3]. The radiator is a thin aluminum foil arranged at an angle of 453 with respect to the beam direction so that the backward lobe of the radiation is emitted at 903 and is easily extractable from the vacuum chamber through a quartz window. The spectral intensity emitted by a bunch of N particles is I (u)"I (u) (N#N(N!1)" f (u)") (1) where I (x) is the intensity radiated by a single electron at a given frequency u and f (u) is the longitudinal bunch form factor de"ned as the Fourier transformation of the normalized longitudinal charge distribution o . ¹(u) denotes the specX tral acceptance function of the detection device. A Martin}Puplett interferometer, shown schematically in Fig. 2, is used to determine the autocorrelation function of the radiation pulse [4]. The
Fig. 2. The Martin}Puplett interferometer.
diverging transition radiation pulse is transformed into a parallel beam entering the interferometer by a parabolic mirror. The incident radiation pulse horizontally polarized by the "rst grid and then splitted by the beam divider into components of orthogonal polarization entering the two interferometer arms. The polarization is #ipped by the roof mirrors, hence the component which is "rst transmitted by the beam splitter is now re#ected and vice versa. The recombined radiation is in general elliptically polarized, depending on the path di!erence in both interferometer arms. The analyzing grid transmits one polarization component into detector 1 and re#ects the orthogonal
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Fig. 3. Spectral transfer function ¹(u) of the Martin}Puplett interferometer. The dashed curves indicate the estimated uncertainty of the transfer function.
component into detector 2. Two pyroelectric detectors equipped with horn antennas are used as detection devices for the sub-millimeter wavelength radiation. A Fourier transformation of the autocorrelation function yields the absolute magnitude of the longitudinal bunch form factor. The experimentally determined spectral intensity is strongly modi"ed by the frequency dependent transfer function ¹(u) of the interferometer. As shown in Fig. 3, ¹(u) has its maximum, ¹ , at around 400 GHz and decreases
towards lower frequencies due to the "nite size of the transition radiator and di!raction losses in the interferometer, while the reduction towards larger frequencies is caused by transmission and re#ection losses of the quartz window and the wire grids. The oscillatory behaviour is due to interference e!ects in the 100 lm thick LiTaO crystals of the pyro electric detectors [5]. The transmission of the quartz window and the wire grids has been mea-
345
Fig. 5. Longitudinal bunch charge distribution derived from the autocorrelation and coherent power spectrum of Fig. 4. The charge distribution consists of three Gaussians A "0.6, p "0.27 ps, A "0.28, p "1.0 ps and A "0.25, p "2.0 ps. The A denote the amplitudes and the p the variances. G G
sured by means of time-domain THz-spectroscopy [6]. Fig. 4 shows the autocorrelation (left) as measured by the Martin}Puplett interferometer and the evaluated coherent power spectrum. The measurement was performed at a beam energy of 170 MeV with a bunch charge of 1 nC and laser pulse length of p "8 ps. The RF gun phase was adjusted to produce the minimum electron bunch length. The data cannot be described by a single Gaussian charge distribution. Instead, the superposition of three Gaussian curves of di!erent variance and amplitude is required for a good adaption of the model to the measured data. The FWHM bunch length is (1.13$0.45) ps, (see Fig. 5) where the error is mainly caused by the uncertainty of ¹(u). The shape of the longitudinal charge distribution, a short pulse superimposed
Fig. 4. Autocorrelation (left) and coherent power spectrum (right) measured for optimum bunch compression. Data are indicated by circles. The solid curves are the autocorrelation and power spectrum of a simulated longitudinal charge distribution convoluted with the spectral transfer function.
SECTION VI.
346
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Fig. 6. A magnetic bunch compressor chicane followed by an o!-crest acceleration can be used to determine the longitudinal bunch charge distribution.
onto a wider basis, can be explained by a partially uncorrelated bunch energy distribution.
If we choose the matrix elements such that M M #M P0, the energy pro"le measured with the spectrometer is a direct image of the longitudinal bunch charge distribution in front of the compression section [7]. Fig. 7 shows an energy pro"le measurement performed at optimum compression with 1 nC bunch charge at a beam energy of 170 MeV and laser pulse length of 8 ps. The pro"le yields a FWHM energy spread of 1.00$0.07 MeV. Using Eq. (2), the energy pro"le can be transformed into a longitudinal charge pro"le yielding a FWHM bunch length of 2.3$0.3 mm in front of the chicane. The right graph shows the reduction of the electron bunch length when moving along the chicane. The FWHM length of the compressed bunch is 600$110 lm. The shaded area denotes the error of the measurement.
4. Energy spread measurement An e$cient and straightforward way to determine the bunch length is the evaluation of the bunch energy distribution. Fig. 6 shows the magnetic bunch compressor chicane followed by an RF cavity and a spectrometer dipole to determine the energy pro"le. The longitudinal dynamics is described by the transport matrix
M ¸ (2) # # M M M #M # # where M , M and M denote transfer matrix elements of the chicane and the RF cavity of Fig. 6. 1
¸
"
5. Conclusion The uncompressed longitudinal charge distribution has been measured with a streak camera and by observing the beam energy spread. The streak camera measures minimum injector bunch length of p "1.95$0.1 mm. The energy spread analysis X yields p "990$90 lm. The factor of 2 is explainX able by the reduced UV laser pulse length during the energy spread measurement. The compressed bunch length measured by the interferometer (p "200$70 lm) and by the energy spread analX ysis (p "260$50 lm) coincide within their error X bars.
Fig. 7. Left: energy pro"le measurement at optimum bunch compression. Right: the evaluated compression of the electron bunch within the chicane compressor. The initial bunch length is a free parameter of the computation.
M. Geitz et al. / Nuclear Instruments and Methods in Physics Research A 445 (2000) 343}347
Acknowledgements We thank Dr. Ingrid Wilke and Mr. Maxim Khazan for carrying out the spectral transmission measurements using time-domain THz spectroscopy at the University of Hamburg.
347
[3] E.B. Blum et al., Nucl. Instr. and Meth. A 307 (1991) and references therein. [4] B. Leissner et al., Proceedings of the PAC Conference, New York, 1999 and references therein. [5] C. Settakorn et al., SLAC-PUB-7813, 1998. [6] I. Wilke et al., Proceedings of EUCAS99, in preparation. [7] K. Ricci et al., in: G.R. Neil, S.V. Benson (Eds.), Proceedings of the FEL Conference, Williams-burg, 1998, Elsevier Science B.V., Amsterdam, 1999, p. II-61.
References [1] TESLA-Collaboration, DESY-TESLA 95-01, 1995. [2] J.S. Nodvick et al., Phys. Rev. A, 96(2) (1954) and references therein.
SECTION VI.