- Email: [email protected]
Laser pulse shaping for high gradient accelerators
Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Laser pulse shaping for high gradient accelerators F. Villa a,n, M.P. Anania a, M. Bellaveglia a, F. Bisesto a,c, E. Chiadroni a, A. Cianchi b, A. Curcio a, M. Galletti a, D. Di Giovenale a, G. Di Pirro a, M. Ferrario a, G. Gatti a, M. Moreno c, M. Petrarca c, R. Pompili a, C. Vaccarezza a a
INFN-Laboratori Nazionali di Frascati, via E. Fermi 40, 00044 Frascati, Italy INFN-Roma Tor Vergata and Università di Roma Tor Vergata, Via della Ricerca Scientifica 1, 00133 Rome, Italy c Università La Sapienza di Roma, Via A. Scarpa 14, Rome, Italy b
art ic l e i nf o
Keywords: Laser Shaping Accelerator Multibunch
a b s t r a c t In many high gradient accelerator schemes, i.e. with plasma or dielectric wakefield induced by particles, many electron pulses are required to drive the acceleration of one of them. Those electron bunches, that generally should have very short duration and low emittance, can be generated in photoinjectors driven by a train of laser pulses coming inside the same RF bucket. We present the system used to shape and characterize the laser pulses used in multibunch operations at Sparc_lab. Our system gives us control over the main parameter useful to produce a train of up to five high brightness bunches with tailored intensity and time distribution. & 2016 Elsevier B.V. All rights reserved.
1. Introduction Recently high brightness linac facilities have studied operation schemes that have more than one electron bunch per RF bucket (LCLS [1], DESY [2], Sparc_lab [3], Fermilab [4] for example). While some facilities, as FLASH and the European XFEL, use a long RF pulse with many bunches in a superconducting linac to increase the total number of bunches [5] available to experiments, we focus on other facilities, as LCLS and Sparc_lab, that have experiments that require a multibunch configuration to operate, having more than one electron bunch in the same RF period. Resonant Plasma WakeField Acceleration (PWFA) driven by electron bunches uses one or more electron bunches to drive an high gradient acceleration of a smaller witness bunch inside a plasma medium. Other high gradient acceleration experiments, such as wakefield dielectric acceleration [6], also require a specific train configuration of many electron bunches. Multibunch operation is also required in many experiments of two color FEL for pump-probe or stroboscopic experiments [7–9] or in generation of monochromatic THz radiation used in material studies [10,11]. Those configurations require the production of two or more electron bunches with very specific characteristics, as energy, transverse dimensions, time duration and separation, that can be different for each bunch of the train. The generation of electron trains with a n
Corresponding author. E-mail address: [email protected] (F. Villa).
pulse separation of few ps or less relies on different schemes, for instance a single long electron bunch is sliced by a mechanical slits system placed in a dispersive area [4,12] or a train of bunches is longitudinally manipulated using velocity bunching [13]. In the latter configuration a train of laser pulses generates a train of electron bunches, it is further manipulated with velocity bunching compression [14] and accelerated inside the linac. A laser system that can shape and control the laser pulses temporal and transversal dimensions and energy for each pulse is required. We present different laser system configurations to achieve the required pulses characteristics and experimental results from Sparc_lab about the generation, characterization and use of those pulses in an high brightness linac.
2. Pulse train generation and characterization We focus our attention on the techniques to generate a train of pulses that can be used also with UV pulses, as metallic cathodes require wavelengths shorter than visible light to photoemit. As an example, the Sparc_lab RF gun (an S-band, 1.6 cell copper gun) requires a wavelength of about 266 nm. The quantum efficiency of our cathode is in the order of 10 5, thus it requires many tens of mJ of UV energy to generate hundreds of pC. Many techniques can be used in order to generate a train of longitudinal pulses [15]. A simple technique to obtain a ps-train of laser pulses uses the birefringence properties of crystals [16]. Those crystals have two
http://dx.doi.org/10.1016/j.nima.2016.01.010 0168-9002/& 2016 Elsevier B.V. All rights reserved.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
2
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Fig. 1. Laser pulse train generation with birefringent crystals.
different diffraction indexes depending on the orientation of the laser polarization with respect to the crystal optical axis. Choosing the laser polarization it is possible to set how much energy will propagate along the fast and slow axis of the crystal (Fig. 1). The crystal induces a longitudinal separation (Δt) between the pulses of Δt ¼(ne-no)L/c, where no (ne) is the ordinary (extraordinary) diffraction index and L is the crystal thickness. The two pulses have orthogonal polarization. It is possible to have more crystals of different length in series in order to obtain more pulses. Usually αBBO crystals are used, because they have a good transmission and strong birefringence in the UV spectral region [6]. This scheme is very simple and efficient: it does not require a precise alignment of the crystal other than rotation; the transverse position of the pulses is not shifted; it requires only n optical elements (the crystals) to obtain 2^n pulses; the losses of the system are minimal (mainly due to surface reflection and crystal absorption). Major drawbacks of this configuration are: the distances between the pulses are fixed; the number of pulses are exactly doubled for each crystal; the distance between each pulse is not set independently for more than two pulses. Another technique is based on splitting the pulses and make them propagate on different paths before recombining in a interferometric like configuration (Fig. 2) [17]. In this configuration a linearly polarized laser pulse is divided by a polarizing beam splitter. One pulse is retarded with a geometrical delay line and recombined with the other by a polarizing beam splitter. The first half wave plate can rotate the polarization in order to define how much energy send in each arm. They can also accommodate other optics in order to have also a different transversal or longitudinal shape for each pulse. This scheme is simple enough in case of two pulse generation and have a great degree of customization for transverse, longitudinal and energy characteristics of each pulse, but it requires a precise alignment in the transverse direction because the following transport optics could change the relative transverse position of the two pulses. This scheme is not easily scaled to more than two pulses. One option is to split the initial pulse many times with half wave plates and polarizing beam splitters to obtain the desired number of pulses, having half of them with a polarization and the other half (minus one for odd total pulses) with the orthogonal one. When the pulses are recombined there is a loss of half the energy for each recombination after the first one: when more than two pulses are sent to a polarizing beam splitter only two orthogonal polarization direction are available, resulting in the interleaving of only half the pulses or, with a 45° rotation of the polarization, of only half of the energy of each pulse. Other options, for instance reported in [18,19], have less losses and are more compact than this scheme, but are limited to configuration of power of 2 pulses with less-than-optimal number of separation degree of freedom, like the birefringent crystal scheme described before.
Fig. 2. Scheme of train of 2 pulses generated by a interferometric-like configuration. HWP: half wave plate, PBS: polarizing beam splitter.
Other techniques involving optical dispersion can also be implemented, such as the interference of two stretched pulses in order to have a train of a large number of pulses [11] or phase [15] and amplitude [20,21] manipulation in the Fourier plane of a 4f system. Longitudinal characteristics should be precisely measured to fully characterize the laser train. Crosscorrelation with a IR laser can solve the problem of having an high resolution in long time window. These measurements are done measuring the energy of the pulse generated by the frequency difference process inside a nonlinear crystal between the UV train pulse to be characterized and the IR single peaked probe (Fig. 3) [22]. The crosscorrelation profile is obtained by a multishot measurement changing the delay of the IR pulse. The profile is strongly affected by amplitude jitter, that can be reduced by averaging on many shots per IR delay position. The resolution of the system is limited by the shorter step that the delay line can set (5 mm) and by the IR pulse length (a fwhm between 0.7 and 1.3 ps. The IR
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
pulse was slightly chirped to compensate the dispersion on the UV beam due to the optics following the harmonic generation) [23]. The time jitter can be greatly reduced by using the same IR pulse that is remaining from UV harmonic generation. In this case, the IR pulse shape can be deformed by the UV generation, that is done near the pump depletion regime (an example of an autocorrelation of an IR pulse used in crosscorrelation measurements is reported in Fig. 4), and the IR actual shape should be considered during data analysis of the crosscorrelation. Another possible limit of this technique is a deformation of the crosscorrelation profile due to spatial chirp of the pulses, as described in.
3. Pulse train at Sparc_lab Some of the previous schemes were tested at Sparc_lab [3]. The Sparc_lab photoinjector is driven by a laser system based on a chirped pulse amplified Ti:sapphire laser at 800 nm tripled to 266 nm at 10 Hz of repetition rate. Compressed pulses lengths are about 100 fs fwhm for the UV pulses. We have a quasi-flat-top transverse profile obtained by an imaging of an iris on the cathode that allows having up to 1 mJ of UV energy after the cut.
Fig. 3. Crosscorrelation scheme of a 2 pulse train. A Delay Line (DL) is used to obtain a crosscorrelation measurement by a photodiode using a Difference Frequency Generating (DFG) crystal, βBBO for example.
3
We used αBBO crystals of different thickness, ranging from 5.32 to 0.33 mm, to obtain laser pulses with a separation between 4.20 ps and 0.26 ps. Fig. 5 shows the crosscorrelation of the pulses resulting from using one to five crystals, with a series of crystals whose thicknesses are halving for each following crystal. In the last configuration the distances between the pulses are the same of the single pulse fwhm, leading to a quasi-flat-top longitudinal shape of 8.4 ps. The ripples present in the top of this pulse (Fig. 5D) are partially due to small misalignments of the rotation of the αBBO crystals, thus in balancing the intensity, and due to interference of the tail of the pulses with same polarization (that are interleaved). The rotation of the crystals can reduce the ripples depth up to about 15% of the maximum value. If more flatness is required, a temperature control of the crystals can be used to adjust the phase delay. As the velocity bunching compression and the following acceleration nonuniformly change the separation between the pulses we choose to change the configuration of the pulses in order to have more than two electron bunches with the desired time separation. We used the interferometric like configuration, to have a variable delay on one of the pulses, and use αBBO crystals in the other one to have a train of pulses (Fig. 6). The first case we studied was the generation of three equidistant pulses with different charge at the end of the linac. We used two αBBO crystals added together to have a separation between the laser pulses of 2.7 ps and a third pulse at about 4 ps later. Online fine adjustments looking directly at the longitudinal phase space of the accelerated electrons set the distance at 4.4 ps in order to have a near equal separations of the electron bunch centroids of 1.0 and 1.1 ps. The intensities are also online set to have a ratio of 5–3–1 on the electron bunches charge using the half wave plate and the rotation of the αBBO crystals (Fig. 7). We changed this configuration in order to generate 5 pulses by adding an half wave plate before the crystals to rotate the polarization of the drivers by 45°, otherwise only half of the 4 pulses generated by the crystals would had reflected along the delayed pulse to the photoinjector. This configuration can be very useful for particle PWFA using the 4 pulses to drive the acceleration and the last one as the accelerated witness. We realized two configurations: one with a train of 4 laser pulses with the same energy and a last one with half the energy; another having a ramped train with 90% of the energy in the 4 pulses with ratio 7–5–3–1 and a delayed one with the remaining 10% of the energy. We could easily change one configuration in the other by turning the crystals angles and finely tune the charge ratio and the witness delay directly measuring the longitudinal phase space of the electrons at the end of the accelerator (Fig. 8). This capability will allow us to optimize the driver charge ratio during a PWFA experiment. We obtained a separation between the electron bunches of 1.2, 1.1, 1.4 and 1.6 ps for the constant driver configuration and 1.0, 0.9, 1.4 and 1.6 ps for the ramped one. While we aimed at obtaining four equidistant beam and the last one at one and half that distance, the last driver bunch had a significant different delay compared to the others due to the electron compression in velocity bunching.
4. Discussion
Fig. 4. Autocorrelation of the IR pulse after the harmonic generation used in Figs. 7 and 8. The scan was done with 0.17 ps per step.
We studied and tested the laser system for train generation, using birefringent crystals to generate many laser pulses and using also a delay line in a interferometer like configuration.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
4
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Fig. 5. Crosscorrelation of the pulse train obtained by one (A), two (B), three (C) and five (D) αBBO crystals. The delay of each step of the delay line was between 33 fs (A) and 100 fs (D) and IR fwhm duration was 0.7 ps. Each point is an average of 30 shots.
Fig. 6. Laser system configuration for the generation of 3–5 laser pulses. HWP: half wave plate, PBS: polarizing beam splitter, DL: delay line.
Birefringent crystals are a very good system to generate a pair of pulses having just one element, being noncritically aligned and with negligible losses. When more than two pulses are required, the separation ratio between the pulses matters: if the application requires equidistant
laser pulses, birefringent crystals are the most effective solution in term of number of elements, ease of use, alignment procedure and small losses. If different separations between pulses are required, an interferometer like configuration, with a number of arms equals to the number of the pulses, each with its delay line, is required. This configuration is also required if the transverse dimensions must be changed between the pulses, i.e. in order to generate the same electron charge density with electron bunches that have different charges. The use of a mixed configuration, with at most one birefringent crystal for each arm of the interferometer can be the best tradeoff between the flexibility of the delay lines and the ease of use of the birefringent crystal, fulfilling the requirement of having each pulse with a different time separation. This solution can be more efficient than using 4f systems and amplitude masks [20] or using multiple laser sources locked together [24]. Future development of Sparc_lab laser system will implement a configuration of 4 delay lines in an interferometer like configuration in order to control the delay, transverse and longitudinal dimensions of each pulse. In conclusion we presented two different method of laser train generation based on birefringent crystals and delay lines in an interferometer like configuration, we have shown samples of laser train pulses used to make electron bunches useful in many applications (PWFA, but also for FEL and monochromatic THz generation) and pointed out the strong points and the limits of those methods.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
5
Fig. 7. Crosscorrelation of laser train (on the left) and corresponding measured electron longitudinal phase space (on the right, energy is on the horizontal axis and time in the vertical axis).
Fig. 8. Top: crosscorrelation of 5 laser pulses configurations.The red configuration (A) has one small pulse followed by 4 pulses with the same amplitude, where the misalignment in the orientation of one of the crystal was corrected during linac operation. The black configuration (B) has one small pulse followed by 4 pulses ramped in amplitude in a 7–5–3–1 ratio. Bottom: electron longitudinal phase spaces of the two configurations.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
6
Acknowledgments This work has been partially supported by the EU Commission in the Seventh Framework Program, Grant Agreement 312453 EuCARD-2 and by the Italian Ministry of Research in the framework of FIRB-Fondo per gli Investimenti della Ricerca di Base, Project no. RBFR12NK5K.
References [1] [2] [3] [4]
A. Marinelli, et al., Nat. Comm. 6 (2015). W. Ackermann, et al., Nat. Photonics 1.6 (2007) 336. M. Ferrario, et al., Nucl. Instr. Methods Phys. Res. B 309 (2013) 183. Y. E. Sun, Experiment and simulations of sub-ps electron bunch train generation at fermilab photoinjectors, In: FEL conference, 2011. [5] P. Muggli, et al., Phys. Rev. Lett. 101 (5) (2008) 054801. [6] C. Jing, et al., Phys. Rev. ST Acc. Beams 14 (2011) 021302.
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
Z. Zhang, et al., Phys. Rev. ST Acc. Beams 18 (2015) 030702. A. Petralia, et al., Phys. Rev. Lett. 115 (1) (2015) 014801. V. Petrillo, et al., Phys. Rev. Lett. 111 (11) (2013) 114802. E. Chiadroni et al., The Sparc_lab high peak power THz source: different methods of generation and characterization, In: 38th IRMMW-THz, 2013. Y. Shen, et al., Phys. Rev. Lett. 107 (2011) 204801. Y.E. Sun, Phys. Rev. Lett. 105 (2010) 234801. M. Ferrario, et al., Nucl. Instr. Methods Phys. Res. A 637 (2011) s43. M. Ferrario, et al., Phys. Rev. Lett. 2013 (2013) 054801. F. Villa, et al., Nucl. Instr. Methods Phys. Res. A 740 (2014) 188. A.K. Sharma, et al., Phys. Rev. ST Acc. Beams 12 (2009) 033501. Y. Ding, et al., Phys. Rev. ST Acc. Beams 13 (2010) 060703. C.W. Siders, et al., Appl. Opt. 37 (22) (1998) 5302. A. Aryshev et al., Femtosecond response time measurements of a Cs2Te photocathode, arXiv:1507.03302v1 [physics.acc-ph], 2015. C. Vicario et al., Drive laser system for Sparc photoinjector, In: PAC07, 2007. M. Boscolo, et al., Nucl. Instr. Methods Phys. Res. A 577 (2007) 409. A. Rakhman, et al., Appl. Opt. 53 (31) (2014) 7603. M. Petrarca et al., SPARC/LS-06/002, 2006. M. Vogt, Status of the Free Electron Laser user facility FLASH, In: IPAC, 2014.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Laser pulse shaping for high gradient accelerators F. Villa a,n, M.P. Anania a, M. Bellaveglia a, F. Bisesto a,c, E. Chiadroni a, A. Cianchi b, A. Curcio a, M. Galletti a, D. Di Giovenale a, G. Di Pirro a, M. Ferrario a, G. Gatti a, M. Moreno c, M. Petrarca c, R. Pompili a, C. Vaccarezza a a
INFN-Laboratori Nazionali di Frascati, via E. Fermi 40, 00044 Frascati, Italy INFN-Roma Tor Vergata and Università di Roma Tor Vergata, Via della Ricerca Scientifica 1, 00133 Rome, Italy c Università La Sapienza di Roma, Via A. Scarpa 14, Rome, Italy b
art ic l e i nf o
Keywords: Laser Shaping Accelerator Multibunch
a b s t r a c t In many high gradient accelerator schemes, i.e. with plasma or dielectric wakefield induced by particles, many electron pulses are required to drive the acceleration of one of them. Those electron bunches, that generally should have very short duration and low emittance, can be generated in photoinjectors driven by a train of laser pulses coming inside the same RF bucket. We present the system used to shape and characterize the laser pulses used in multibunch operations at Sparc_lab. Our system gives us control over the main parameter useful to produce a train of up to five high brightness bunches with tailored intensity and time distribution. & 2016 Elsevier B.V. All rights reserved.
1. Introduction Recently high brightness linac facilities have studied operation schemes that have more than one electron bunch per RF bucket (LCLS [1], DESY [2], Sparc_lab [3], Fermilab [4] for example). While some facilities, as FLASH and the European XFEL, use a long RF pulse with many bunches in a superconducting linac to increase the total number of bunches [5] available to experiments, we focus on other facilities, as LCLS and Sparc_lab, that have experiments that require a multibunch configuration to operate, having more than one electron bunch in the same RF period. Resonant Plasma WakeField Acceleration (PWFA) driven by electron bunches uses one or more electron bunches to drive an high gradient acceleration of a smaller witness bunch inside a plasma medium. Other high gradient acceleration experiments, such as wakefield dielectric acceleration [6], also require a specific train configuration of many electron bunches. Multibunch operation is also required in many experiments of two color FEL for pump-probe or stroboscopic experiments [7–9] or in generation of monochromatic THz radiation used in material studies [10,11]. Those configurations require the production of two or more electron bunches with very specific characteristics, as energy, transverse dimensions, time duration and separation, that can be different for each bunch of the train. The generation of electron trains with a n
Corresponding author. E-mail address: [email protected] (F. Villa).
pulse separation of few ps or less relies on different schemes, for instance a single long electron bunch is sliced by a mechanical slits system placed in a dispersive area [4,12] or a train of bunches is longitudinally manipulated using velocity bunching [13]. In the latter configuration a train of laser pulses generates a train of electron bunches, it is further manipulated with velocity bunching compression [14] and accelerated inside the linac. A laser system that can shape and control the laser pulses temporal and transversal dimensions and energy for each pulse is required. We present different laser system configurations to achieve the required pulses characteristics and experimental results from Sparc_lab about the generation, characterization and use of those pulses in an high brightness linac.
2. Pulse train generation and characterization We focus our attention on the techniques to generate a train of pulses that can be used also with UV pulses, as metallic cathodes require wavelengths shorter than visible light to photoemit. As an example, the Sparc_lab RF gun (an S-band, 1.6 cell copper gun) requires a wavelength of about 266 nm. The quantum efficiency of our cathode is in the order of 10 5, thus it requires many tens of mJ of UV energy to generate hundreds of pC. Many techniques can be used in order to generate a train of longitudinal pulses [15]. A simple technique to obtain a ps-train of laser pulses uses the birefringence properties of crystals [16]. Those crystals have two
http://dx.doi.org/10.1016/j.nima.2016.01.010 0168-9002/& 2016 Elsevier B.V. All rights reserved.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
2
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Fig. 1. Laser pulse train generation with birefringent crystals.
different diffraction indexes depending on the orientation of the laser polarization with respect to the crystal optical axis. Choosing the laser polarization it is possible to set how much energy will propagate along the fast and slow axis of the crystal (Fig. 1). The crystal induces a longitudinal separation (Δt) between the pulses of Δt ¼(ne-no)L/c, where no (ne) is the ordinary (extraordinary) diffraction index and L is the crystal thickness. The two pulses have orthogonal polarization. It is possible to have more crystals of different length in series in order to obtain more pulses. Usually αBBO crystals are used, because they have a good transmission and strong birefringence in the UV spectral region [6]. This scheme is very simple and efficient: it does not require a precise alignment of the crystal other than rotation; the transverse position of the pulses is not shifted; it requires only n optical elements (the crystals) to obtain 2^n pulses; the losses of the system are minimal (mainly due to surface reflection and crystal absorption). Major drawbacks of this configuration are: the distances between the pulses are fixed; the number of pulses are exactly doubled for each crystal; the distance between each pulse is not set independently for more than two pulses. Another technique is based on splitting the pulses and make them propagate on different paths before recombining in a interferometric like configuration (Fig. 2) [17]. In this configuration a linearly polarized laser pulse is divided by a polarizing beam splitter. One pulse is retarded with a geometrical delay line and recombined with the other by a polarizing beam splitter. The first half wave plate can rotate the polarization in order to define how much energy send in each arm. They can also accommodate other optics in order to have also a different transversal or longitudinal shape for each pulse. This scheme is simple enough in case of two pulse generation and have a great degree of customization for transverse, longitudinal and energy characteristics of each pulse, but it requires a precise alignment in the transverse direction because the following transport optics could change the relative transverse position of the two pulses. This scheme is not easily scaled to more than two pulses. One option is to split the initial pulse many times with half wave plates and polarizing beam splitters to obtain the desired number of pulses, having half of them with a polarization and the other half (minus one for odd total pulses) with the orthogonal one. When the pulses are recombined there is a loss of half the energy for each recombination after the first one: when more than two pulses are sent to a polarizing beam splitter only two orthogonal polarization direction are available, resulting in the interleaving of only half the pulses or, with a 45° rotation of the polarization, of only half of the energy of each pulse. Other options, for instance reported in [18,19], have less losses and are more compact than this scheme, but are limited to configuration of power of 2 pulses with less-than-optimal number of separation degree of freedom, like the birefringent crystal scheme described before.
Fig. 2. Scheme of train of 2 pulses generated by a interferometric-like configuration. HWP: half wave plate, PBS: polarizing beam splitter.
Other techniques involving optical dispersion can also be implemented, such as the interference of two stretched pulses in order to have a train of a large number of pulses [11] or phase [15] and amplitude [20,21] manipulation in the Fourier plane of a 4f system. Longitudinal characteristics should be precisely measured to fully characterize the laser train. Crosscorrelation with a IR laser can solve the problem of having an high resolution in long time window. These measurements are done measuring the energy of the pulse generated by the frequency difference process inside a nonlinear crystal between the UV train pulse to be characterized and the IR single peaked probe (Fig. 3) [22]. The crosscorrelation profile is obtained by a multishot measurement changing the delay of the IR pulse. The profile is strongly affected by amplitude jitter, that can be reduced by averaging on many shots per IR delay position. The resolution of the system is limited by the shorter step that the delay line can set (5 mm) and by the IR pulse length (a fwhm between 0.7 and 1.3 ps. The IR
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
pulse was slightly chirped to compensate the dispersion on the UV beam due to the optics following the harmonic generation) [23]. The time jitter can be greatly reduced by using the same IR pulse that is remaining from UV harmonic generation. In this case, the IR pulse shape can be deformed by the UV generation, that is done near the pump depletion regime (an example of an autocorrelation of an IR pulse used in crosscorrelation measurements is reported in Fig. 4), and the IR actual shape should be considered during data analysis of the crosscorrelation. Another possible limit of this technique is a deformation of the crosscorrelation profile due to spatial chirp of the pulses, as described in.
3. Pulse train at Sparc_lab Some of the previous schemes were tested at Sparc_lab [3]. The Sparc_lab photoinjector is driven by a laser system based on a chirped pulse amplified Ti:sapphire laser at 800 nm tripled to 266 nm at 10 Hz of repetition rate. Compressed pulses lengths are about 100 fs fwhm for the UV pulses. We have a quasi-flat-top transverse profile obtained by an imaging of an iris on the cathode that allows having up to 1 mJ of UV energy after the cut.
Fig. 3. Crosscorrelation scheme of a 2 pulse train. A Delay Line (DL) is used to obtain a crosscorrelation measurement by a photodiode using a Difference Frequency Generating (DFG) crystal, βBBO for example.
3
We used αBBO crystals of different thickness, ranging from 5.32 to 0.33 mm, to obtain laser pulses with a separation between 4.20 ps and 0.26 ps. Fig. 5 shows the crosscorrelation of the pulses resulting from using one to five crystals, with a series of crystals whose thicknesses are halving for each following crystal. In the last configuration the distances between the pulses are the same of the single pulse fwhm, leading to a quasi-flat-top longitudinal shape of 8.4 ps. The ripples present in the top of this pulse (Fig. 5D) are partially due to small misalignments of the rotation of the αBBO crystals, thus in balancing the intensity, and due to interference of the tail of the pulses with same polarization (that are interleaved). The rotation of the crystals can reduce the ripples depth up to about 15% of the maximum value. If more flatness is required, a temperature control of the crystals can be used to adjust the phase delay. As the velocity bunching compression and the following acceleration nonuniformly change the separation between the pulses we choose to change the configuration of the pulses in order to have more than two electron bunches with the desired time separation. We used the interferometric like configuration, to have a variable delay on one of the pulses, and use αBBO crystals in the other one to have a train of pulses (Fig. 6). The first case we studied was the generation of three equidistant pulses with different charge at the end of the linac. We used two αBBO crystals added together to have a separation between the laser pulses of 2.7 ps and a third pulse at about 4 ps later. Online fine adjustments looking directly at the longitudinal phase space of the accelerated electrons set the distance at 4.4 ps in order to have a near equal separations of the electron bunch centroids of 1.0 and 1.1 ps. The intensities are also online set to have a ratio of 5–3–1 on the electron bunches charge using the half wave plate and the rotation of the αBBO crystals (Fig. 7). We changed this configuration in order to generate 5 pulses by adding an half wave plate before the crystals to rotate the polarization of the drivers by 45°, otherwise only half of the 4 pulses generated by the crystals would had reflected along the delayed pulse to the photoinjector. This configuration can be very useful for particle PWFA using the 4 pulses to drive the acceleration and the last one as the accelerated witness. We realized two configurations: one with a train of 4 laser pulses with the same energy and a last one with half the energy; another having a ramped train with 90% of the energy in the 4 pulses with ratio 7–5–3–1 and a delayed one with the remaining 10% of the energy. We could easily change one configuration in the other by turning the crystals angles and finely tune the charge ratio and the witness delay directly measuring the longitudinal phase space of the electrons at the end of the accelerator (Fig. 8). This capability will allow us to optimize the driver charge ratio during a PWFA experiment. We obtained a separation between the electron bunches of 1.2, 1.1, 1.4 and 1.6 ps for the constant driver configuration and 1.0, 0.9, 1.4 and 1.6 ps for the ramped one. While we aimed at obtaining four equidistant beam and the last one at one and half that distance, the last driver bunch had a significant different delay compared to the others due to the electron compression in velocity bunching.
4. Discussion
Fig. 4. Autocorrelation of the IR pulse after the harmonic generation used in Figs. 7 and 8. The scan was done with 0.17 ps per step.
We studied and tested the laser system for train generation, using birefringent crystals to generate many laser pulses and using also a delay line in a interferometer like configuration.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
4
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
Fig. 5. Crosscorrelation of the pulse train obtained by one (A), two (B), three (C) and five (D) αBBO crystals. The delay of each step of the delay line was between 33 fs (A) and 100 fs (D) and IR fwhm duration was 0.7 ps. Each point is an average of 30 shots.
Fig. 6. Laser system configuration for the generation of 3–5 laser pulses. HWP: half wave plate, PBS: polarizing beam splitter, DL: delay line.
Birefringent crystals are a very good system to generate a pair of pulses having just one element, being noncritically aligned and with negligible losses. When more than two pulses are required, the separation ratio between the pulses matters: if the application requires equidistant
laser pulses, birefringent crystals are the most effective solution in term of number of elements, ease of use, alignment procedure and small losses. If different separations between pulses are required, an interferometer like configuration, with a number of arms equals to the number of the pulses, each with its delay line, is required. This configuration is also required if the transverse dimensions must be changed between the pulses, i.e. in order to generate the same electron charge density with electron bunches that have different charges. The use of a mixed configuration, with at most one birefringent crystal for each arm of the interferometer can be the best tradeoff between the flexibility of the delay lines and the ease of use of the birefringent crystal, fulfilling the requirement of having each pulse with a different time separation. This solution can be more efficient than using 4f systems and amplitude masks [20] or using multiple laser sources locked together [24]. Future development of Sparc_lab laser system will implement a configuration of 4 delay lines in an interferometer like configuration in order to control the delay, transverse and longitudinal dimensions of each pulse. In conclusion we presented two different method of laser train generation based on birefringent crystals and delay lines in an interferometer like configuration, we have shown samples of laser train pulses used to make electron bunches useful in many applications (PWFA, but also for FEL and monochromatic THz generation) and pointed out the strong points and the limits of those methods.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
5
Fig. 7. Crosscorrelation of laser train (on the left) and corresponding measured electron longitudinal phase space (on the right, energy is on the horizontal axis and time in the vertical axis).
Fig. 8. Top: crosscorrelation of 5 laser pulses configurations.The red configuration (A) has one small pulse followed by 4 pulses with the same amplitude, where the misalignment in the orientation of one of the crystal was corrected during linac operation. The black configuration (B) has one small pulse followed by 4 pulses ramped in amplitude in a 7–5–3–1 ratio. Bottom: electron longitudinal phase spaces of the two configurations.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i
F. Villa et al. / Nuclear Instruments and Methods in Physics Research A ∎ (∎∎∎∎) ∎∎∎–∎∎∎
6
Acknowledgments This work has been partially supported by the EU Commission in the Seventh Framework Program, Grant Agreement 312453 EuCARD-2 and by the Italian Ministry of Research in the framework of FIRB-Fondo per gli Investimenti della Ricerca di Base, Project no. RBFR12NK5K.
References [1] [2] [3] [4]
A. Marinelli, et al., Nat. Comm. 6 (2015). W. Ackermann, et al., Nat. Photonics 1.6 (2007) 336. M. Ferrario, et al., Nucl. Instr. Methods Phys. Res. B 309 (2013) 183. Y. E. Sun, Experiment and simulations of sub-ps electron bunch train generation at fermilab photoinjectors, In: FEL conference, 2011. [5] P. Muggli, et al., Phys. Rev. Lett. 101 (5) (2008) 054801. [6] C. Jing, et al., Phys. Rev. ST Acc. Beams 14 (2011) 021302.
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
Z. Zhang, et al., Phys. Rev. ST Acc. Beams 18 (2015) 030702. A. Petralia, et al., Phys. Rev. Lett. 115 (1) (2015) 014801. V. Petrillo, et al., Phys. Rev. Lett. 111 (11) (2013) 114802. E. Chiadroni et al., The Sparc_lab high peak power THz source: different methods of generation and characterization, In: 38th IRMMW-THz, 2013. Y. Shen, et al., Phys. Rev. Lett. 107 (2011) 204801. Y.E. Sun, Phys. Rev. Lett. 105 (2010) 234801. M. Ferrario, et al., Nucl. Instr. Methods Phys. Res. A 637 (2011) s43. M. Ferrario, et al., Phys. Rev. Lett. 2013 (2013) 054801. F. Villa, et al., Nucl. Instr. Methods Phys. Res. A 740 (2014) 188. A.K. Sharma, et al., Phys. Rev. ST Acc. Beams 12 (2009) 033501. Y. Ding, et al., Phys. Rev. ST Acc. Beams 13 (2010) 060703. C.W. Siders, et al., Appl. Opt. 37 (22) (1998) 5302. A. Aryshev et al., Femtosecond response time measurements of a Cs2Te photocathode, arXiv:1507.03302v1 [physics.acc-ph], 2015. C. Vicario et al., Drive laser system for Sparc photoinjector, In: PAC07, 2007. M. Boscolo, et al., Nucl. Instr. Methods Phys. Res. A 577 (2007) 409. A. Rakhman, et al., Appl. Opt. 53 (31) (2014) 7603. M. Petrarca et al., SPARC/LS-06/002, 2006. M. Vogt, Status of the Free Electron Laser user facility FLASH, In: IPAC, 2014.
Please cite this article as: F. Villa, et al., Nuclear Instruments & Methods in Physics Research A (2016), http://dx.doi.org/10.1016/j. nima.2016.01.010i