THz SASE FEL at PITZ: lasing at a wavelength of 100 μm

Development of an accelerator-based tunable THz source prototype for pump-probe experiments at the European XFEL is ongoing at the Photo Injector Test facility at DESY in Zeuthen (PITZ). The proof-of-principle experiments on the THz SASE FEL are performed utilizing the LCLS-I undulator (on loan from SLAC) installed in the PITZ beamline. The first lasing at a center wavelength of 100 μm was observed in the summer of 2022. The lasing of the narrowband THz source was achieved using an electron beam with an energy of ∼17 MeV and a bunch charge up to several nC. Optimization of beam transport and matching resulted in the measurement of THz radiation with a pulse energy of tens of μJ, measured with pyroelectric detectors. The THz FEL gain curves were measured by means of specially designed short coils along the undulator. The results of the first characterization of the THz source at PITZ will be presented.


Introduction
Pump-probe experiments at modern X-ray free-electron lasers (XFELs) require tunable THz sources providing high-power THz pulses.Two classes of such sources are now being considered for this purpose [1]: conventional laser driven and accelerator-based THz sources.The latter are based on different principles of radiation generation, but for coherent THz radiation the emitting part of the electron beam must be shorter than the wavelength.This implies that the electron 4 now at Oak Ridge National Laboratory, USA 5 now at SLAC National Accelerator laboratory, USA 6 on leave from CANDLE Synchrotron Research Institute, Armenia 7 on leave from Chiang Mai University, Thailand 8 now at Diamond Light Source Ltd, UK 9 now at Helmholtz-Zentrum Dresden Rossendorf, Germany 10 now at Zhangjiang Lab, China 11 now at Paul Scherrer Institute, Switzerland beams should be either shorter than ∼1 ps or constitute a microbunching structure on a sub-ps time scale.Depending on the mechanism of radiation, the accelerator-based source can have a wide or narrow bandwidth.Short bunches are applied for coherent transition and diffraction radiation (CTR/CDR) [2,3] and radiation from bending magnets [4,5,6].For such sources, afterburner option is often considered when short beams are used after they are radiated at a shorter wavelength.Narrowband THz radiation can be generated by short or modulated beams in suitable undulators or by the CTR/CDR effect with modulated beams [2,6,7,8,9].This work employs a different approach and idea of applying the same mechanism of radiation that is used to generate X-ray pulses in XFELs, namely, the Self Amplified Spontaneous Emission (SASE), where a microbunching is achieved by a single pass of high-current electron beam with a length much longer than the radiation wavelength through an undulator.Developments of a prototype for a high-power tunable accelerator-based THz source for pump-probe experiments at the European XFEL are ongoing at the Photo Injector Test Facility at DESY in Zeuthen (PITZ) [10].As PITZ is developing high brightness electron source for the European XFEL accelerator, it makes it possible to generate THz pulses with a pulse repetition rate and a pulse train structure identical to that of the X-ray pulses.High THz SASE FEL radiation power can be achieved by utilizing high charge (up to several nC) electron bunches from the PITZ RF photogun.The THz beamline has been installed as an extension of the existing PITZ linac in the tunnel annex [11].A planar LCLS-I undulator (module L143-112000-26 on-loan from SLAC) [12] is used to generate the THz radiation.The undulator parameters (period of 3 cm and undulator parameter of ∼3.5) require an electron beam energy of ∼17 MeV for the centre radiation wavelength of ∼100 µm.The undulator strong magnetic field with a horizontal gradient requires a thorough beam matching.Another challenge is the narrow vacuum chamber (5 mm height, 11 mm width, and ∼3.5 m length), which makes matching and transport of the space charge dominated electron beams a complex task.A special procedure was developed and experimentally implemented at the newly installed THz beamline.The first THz SASE FEL lasing was first obtained with 1 nC bunch charge, then the bunch charge was stepwise increased to 3 nC, the first gain curves have been measured [13].Further optimization of the THz radiation resulted in a significant increase of the detected THz pulse energy and observation of saturation onset.For this optimization, a band-pass filter was used to maximize the energy of the radiation pulse around the central frequency of ∼3 THz.The paper reports the performance of the THz SASE FEL at PITZ.

PITZ accelerator
A high-charge electron beam is generated in the RF gun with Cs 2 T e photocathode (QE∼7%), the gun gradient is tuned to yield a beam mean momentum of ∼6.3 MeV/c at the phase of maximum acceleration [13].Electron beams with charge of up to 3 nC were generated using 7 ps (FWHM) UV photocathode laser pulses by adjustment of the laser spot size at the cathode (Ø 2-3.5 mm) and the pulse energy.The electron bunches are longer (∼15-20 ps) due to the space charge effect.The final beam momentum of 16.5-17 MeV/c was achieved by tuning the gradient and the phase of the booster cavity.The booster phase was chosen to be ∼20 deg off-crest, which roughly corresponds to the minimum projected energy spread at the undulator entrance.Smooth transport of the space charge dominated electron beam over a distance of more than 25 m was realized using three quadrupole triplets.Another triplet was applied for beam matching into the LCLS-I undulator (at ∼28 m from the photocathode).More details of the matching procedure could be found in [13,15].

THz beamline
The current THz beamline in the tunnel annex is shown in figure 1.The last quadrupole triplet (three blue yokes on the left) in front of the LCLS-I undulator (in the middle) is aimed for matching the Twiss parameters of the electron beam.The design X-and Y-Twiss parameters are rather asymmetric and have relatively small acceptance [15].The matching of the high charge electron beam into the LCLS-I undulator requires the following beam transverse parameters: the rms beam sizes σ x ≈ 1.3 mm and σ y ≈ 0.2 mm, the corresponding beta-functions β x ≈ 5.4 m and β y ≈ 0.2 m, which corresponds to a flat beam configuration at the undulator entrance.The THz radiation is measured by pyroelectric detectors (THz10) located on the top of dedicated screen stations HIGH3.Scr2 (∼45 cm after the undulator exit) and HIGH3.Scr3 (∼1 m downstream of the HIGH3.Scr2) [14].Cylindrical adapters with a conic internal surface for the radiation collection are mounted on the top of a flange with a diamond window.Station HIGH3.Scr2 is equipped with a movable THz toroidal mirror with a 5 mm diameter hole for further electron beam transport, and station HIGH3.Scr3 is a solid mirror without a hole for transport of the complete THz radiation to the detector.A band-pass filter with maximum transmission centred at 102 µm and a bandwidth of ∼12 µm (FWHM) was mounted on top of the cylindrical adapter in front of the pyrodetector at HIGH3.Scr3 [14].There is a set of steering coils distributed along the undulator (figure 1).They allow to kick an electron bunch horizontally away from the nominal trajectory in the undulator to measure the THz pulse energy radiated until the kick location (active undulator length), which provides a gain curve.The last four coils are more concentrated in the second part of the undulator to provide more points of a gain curve near where saturation is expected to occur.

Experimental gain curves
As previously mentioned, the gain curve measurement procedure is based on the use of short steering coils.Starting with the last coil, all coils one after another are set to a current of +3 A which is supposed to kick the beam from the lasing trajectory (in fact, the beam is dumped on the wall of the vacuum chamber).The response time of pyroelectric detectors lies in the range of tens of ms, so it was preferable to use a single pulse in a pulse train with a repetition rate of 10 Hz to measure the THz pulse energy statistics.The measured THz pulse energy using 500-shots statistics along the undulator is shown in figure 2 for three values of the bunch charge.The relative fluctuation rates are plotted at the right axis.It should be noted that the first points for the gain curves were measured at the same gain of the pyro detector amplifier, and the accuracy suffered because of the poor signal-to-noise ratio.
The measured curves show a strong dependence of the THz pulse energy on the bunch charge (21 µJ for 2 nC versus 6 µJ for 1 nC), which is a clear sign of the coherent SASE lasing.The backward propagation of the exponential range of the gain curves to the undulator entrance (z=0) leads to initial signal of a pJ level, which is in basic agreement with expected shot noise level at this wavelength.The estimated FEL gain of ∼10 6 indicates a high gain THz SASE FEL, which is a quite remarkable result for this frequency range (the center frequency is ∼3 THz).The 3 nC case provides the highest pulse energy SASE FEL, but still does not seem fully optimized.The space charge effect leads to a dilution of the transverse phase space, which significantly complicates the matching of the electron beam into the undulator and, therefore, further optimization of the THz radiation.In addition, the waveguide and wakefield effects due to the narrow vacuum chamber further limit the current THz radiation performance.Further improvements are expected when photocathode laser pulses with a flattop temporal profile will replace the currently used Gaussian pulses.

Discission
The linear model of the FEL amplifier [16] can be applied to analyse the experimental data obtained.Using the linear 3D theory, considering the diffraction and space charge effects, one can obtain a radiation field gain Λ.Here, the field amplitude along the undulator is assumed to be E x ∝ exp[Λ • z].Following this formalism and considering the case of 2 nC (figure 2) as a reference, one can obtain the gain parameter of the FEL amplifier Γ = (0.24 m) −1 , given the measured peak current of ∼125 A and beam energy of ∼17 MeV.The FEL efficiency parameter is ρ ≈ 0.01.Applying the measured electron beam parameters to calculate the dimensionless parameters from [16] yields a beam diffraction parameter B ≈ 0.1 which corresponds to a relatively thin beam.The space charge parameter Λ 2 p ≈ 0.9, as expected, is quite large, so the space charge effect should impact the FEL field gain.An estimate of the ∼10 keV for the slice energy spread (the measured projected spread is ∼70 keV) gives the energy spread parameter Λ 2 T ≈ 0.003, which is small and can be neglected in the first approximation.On the contrary, the waveguide diffraction parameter Ω ≈ 5.3 is fairly large, which corresponds to the strong influence of the narrow vacuum chamber of the undulator onto the FEL field gain properties.Applying these dimensionless parameters to the eigenvalue problem for the field gain Λ [16] the solution (in units of Γ) can be calculated as a function of the frequency detuning from the resonance; the results are shown in figure 3 (left) for four different cases: without and with space charge effect, and without and with the waveguide effect.As it is expected, the space charge effect leads to a decrease in the growth rate.The calculated redshift (max(Λ) at ∆ω/ω < 0) agrees well with the experimental observations.The electron beam energy when optimizing the THz pulse energy with a band-pass filter slightly increased compared to the original theoretical value for a wavelength of 100 µm.
The maximum value of the field growth rate from the linear theory applied to the 2 nC case shows good agreement with the experimental gain curve (figure 3, right plot), despite the fact that the beam wiggling amplitude in the middle of the undulator is estimated to be 14th International Particle Accelerator Conference Figure 3. Left: The THz radiation field growth rate calculated using the waveguide version of the code FAST [17] for different cases without (no SC) and with space charge (SC), without and with waveguide (WVG) effect as a function of the frequency detuning ∆ω/ω.Right: linear fit to the measured THz pulse energy using the maximum growth rates Re(Λ/Γ) from the linear theory with waveguide effect included.comparable to the rms transverse beam size, which make the space charge model used limitedly applicable.The transverse waveguide modes calculated for negative frequency detuning values (local maxima of the solid curves in figure 3, left) are mostly out of the used band-pass filter bandwidth, but are considered for future studies using the Michelson interferometer.The measured fluctuations of the THz pulse energy were analyzed.The probability distribution of the radiation pulse energy from SASE FEL operating in the high gain linear regime follows gamma distribution [18]: where is number of modes in the radiation pulse and Γ(M ) denotes here the gamma function.Relevant probability distributions for points of the 2 nC gain curve (enlarged markers at the right plot in figure 3) are presented in figure 4. It can be seen that the probability distribution of the radiation pulse energy indeed follows that of the high gain SASE FEL in linear regime (left and middle histograms in figure 4).The onset of the saturation regime is illustrated by the right histogram in figure 4.

Conclusion
The high-gain THz SASE FEL radiation at the Photo Injector Test facility at DESY in Zeuthen has been optimized for the center wavelength of ∼100 µm.Gain curves for the THz SASE FEL for a bunch charge of 1-3 nC were measured.The radiation field growth rate calculated using the 3D linear FEL amplifier theory, including diffraction, space charge and waveguide effects, is in good agreement with the experimental data.It has been shown that the waveguide effect due to the narrow vacuum chamber of the LCLS-I undulator can have a significant impact on the FEL lasing process.The statistical properties of the pulse energy fluctuations demonstrate features of the high-gain SASE FEL linear regime.The onset of THz pulse energy saturation has been observed.Further detailed studies of the properties of the generated THz pulses, including spectral characterization are in progress.

Figure 1 .
Figure 1.The THz beam line in the PITZ tunnel annex with the LCLS-I undulator in the middle.The electron beam direction is from left to right.Seven steering coils are distributed along the undulator to enable gain curve measurements.

14thFigure 2 .
Figure 2. Gain curves measured for 1 nC, 2 nC and 3 nC at HIGH3.Scr3 with 3 THz band-pass filter: mean THz pulse energy W (left axis) and relative pulse energy fluctuation rate σ W / W (right axis).

Figure 4 .
Figure 4. Probability distribution of the radiation pulse energy for the 2 nC case Three cases correspond to the enlarged markers on the right plot of figure 3, respectively.