Photoinjector lasers, which stimulate the photocathode surfaces for electron-bunch generation via the photoelectric effect, are crucial components of free-electron laser (FEL) facilities. The Dalian advanced light source (DALS), which is a newly proposed light source operating in the extreme ultraviolet spectrum, is a continuous-wave FEL with a maximum repetition rate of 1 MHz and is designed for chemical physics research. To achieve the optimal brightness of the DALS, one must ensure that the electron bunches exhibit a sufficiently low emittance. An effective method for reducing emittance is by shaping the photoinjector laser three dimensionally, both transversely and longitudinally. This study specifically addresses longitudinal shaping. Current longitudinal shaping techniques include manipulating the light spectrum using devices such as spatial light modulators or acousto-optic modulators, or employing pulse stacking in the time domain. However, the direct adaptation of these techniques to our system poses significant challenges. In this study, based on the fundamental principles of pulse stacking, we utilize a grating pair, two interferometers, and a sequence of birefringent crystals to generate flat-top pulses with durations exceeding 40 ps. Results of electron-beam dynamics simulations show that the generated flat-top pulse significantly improved the emittance of electron beams, thus demonstrating its potential for optimizing the performance of FELs.

We directed infrared laser pulses generated by a Yb-doped fiber amplifier into a fourth-harmonic generator to generate ultraviolet laser pulses with a repetition rate of 1 MHz, a wavelength of 257.5 nm, and a pulse duration of approximately 200 fs. Subsequently, the ultraviolet light was stretched to 1.5 ps by passing it through a grating pair, which introduced the appropriate second-order dispersion. The stretched single pulse was segmented into four equally spaced sub-pulses using two interferometers. Finally, three birefringent crystals were used to further segregate the light, which resulted in overlapping pulse sequences that formed a pulse with a relatively flat top. The shaping results were measured using optical cross-correlation by mixing the shaped ultraviolet laser pulses with fundamental infrared pulses emitted from the amplifier. The intensity distribution was obtained by varying the temporal delay between two light paths. To evaluate the pulse-shaping performance, electron-beam dynamics simulations were conducted using a DALS injector. The simulations yielded four initial electron-beam distributions: two based on actual laser measurements and two theoretical distributions (ideal flat top and Gaussian), with almost identical pulse base widths. In the simulations, a multi-objective genetic algorithm based on the electromagnetic simulation code ASTRA was used, which optimized for a Pareto front comprising the bunch length and normalized emittance at the injector exit. The results were compared and evaluated.

The temporal distributions of the ultraviolet laser pulses generated by the fourth-harmonic generator after undergoing stretching, interferometer splitting, and birefringent crystal manipulation are presented in Fig. 3. The post-stretching pulse duration is approximately 1.5 ps, which is consistent with the calculated value. After passing through the two interferometers, the interval between the four sub-pulses is approximately 11 ps, with almost equal light intensities. The final flat-top distribution achieves through three α-BBO crystals exhibits rising and falling edges of 2.0 ps and 1.9 ps, respectively, with the middle flat-top region spanning approximately 42 ps. The intensity fluctuation in the flat-top region is 5.7% (Root mean square, RMS). The spectrum remained unchanged after shaping. Because of the aperture limitations of the optical components, the intensity at the edges of the shaped light spot is truncated. However, considering the subsequent transverse shaping through aperture truncation, this does not significantly affect the final results. The overall transmission efficiency of the shaping process is approximately 45%, and the obtained energy satisfies the requirements of an electron gun. The beam-dynamics simulation results show that regardless of whether the laser distributions are measured or ideal, the Pareto front for the flat distribution consistently performs better than that for the Gaussian distribution, with the minimum emittance improved by approximately 20%. For each distribution type (flat or Gaussian), the minimum emittances of the measured and ideal lasers are almost identical, with a variance of less than 5%. These results indicate that lasers with flat temporal distributions, instead of Gaussian distributions, are significantly better for DALS injectors.

This study demonstrates the longitudinal flat-top pulse shaping of the photoinjector laser for the DALS and validates the effect of the longitudinal distribution of the photoinjector laser on electron-beam emittance via electron-beam dynamics simulations. Specifically, we employ the pulse-stacking method, where we initially calculate the parameters and design the layout of the optical components based on the requirements for a flat-top pulse. Subsequently, we conduct shaping experiments. The experimental results show that the appropriate combination of gratings, interferometers, and birefringent crystals can effectively transform femtosecond Gaussian pulses into flat-top pulses with durations exceeding 40 ps. While the incident power is increased gradually, the transmission efficiency of the pulse-shaping optical path remained consistently above 45%, thus exhibiting stable operational performance without significant nonlinear effects such as two-photon absorption. The electron-beam dynamics simulations indicate that, under almost identical base durations, the experimentally obtained flat-top pulses can reduce the emittance by 20% compared with Gaussian pulses, thus significantly improving the beam quality. This provides substantial support for the construction of a DALS injector with improved performance. Additionally, the results suggest that the shaping method employed in this study can be broadly applied to the development and construction of photoinjector-driven laser systems.