Introduction

Hefei Advanced Light Facility (HALF) is a fourth-generation synchrotron radiation light source based on a diffraction-limited storage ring (DLSR) [1-3]. The storage ring has a designed circumference of 480 m (revolution time of 1.6 μs), bunch-to-bunch frequency of 500 MHz, and beam current of 350 mA[3]. Owing to the limited dynamic aperture (DA), DLSR requires a low-emittance injector. The HALF plans to adopt an off-axis injection scheme based on a full energy linac, and the main parameters of the injection beam are listed in Table1 [4].

Benefiting from the rapid development of linac-driven high-brightness electron sources, this requirement would be met by the systemic design and construction of the photocathode RF gun-based accelerator. These high-brightness beams with low energy spread, low transverse emittance, and excellent beam energy stability have been widely used in free-electron lasers (FELs) [5-10], Compton scattering sources [11, 12], and ultrafast electron diffraction [13-17]. Ordinarily, a high-voltage thermionic gun is used in the injector of storage rings owing to its good long-term stability and easy maintenance. However, the electron beam produced by a thermionic gun needs to be compressed from a nanosecond level to a picosecond level using a complex bunching system. The compression process causes a considerable increase in the beam transverse emittance and the emergence of large-energy-spread electrons in the beam cluster. Although the large-energy-spread electrons can be discarded through the energy slit in the dispersion section, the poor transverse emittance of the beam is an insurmountable disadvantage. For HALF, we plan to use a full energy linac as the injector, rather than a combination of injector and booster, and the photocathode RF gun should be a better candidate for the electron source. As one of the assignments of the HALF R&D project, a BNL/SLAC/UCLA/KEK type [20, 21] S-band 1.6-cell photocathode RF gun was constructed in 2018 to serve as the pilot plant. Initial commissioning with the first electron beam was completed by the end of 2020. Its measured beam emittance is approximately 1.2 mm⋅mrad with a bunch charge of 500 pC, which demonstrated the goal.

In this paper, we report the first beam results and experience obtained during the commission, including the photocathode RF gun, drive laser system, RF power commissioning, and beam diagnostics. The remainder of the paper is organized as follows: in Sect. 2, we present the description of the main components, including the main subsystems, such as the RF gun, drive laser system, and beam diagnostics system. In Sect. 3, we provide some commissioning results for the photocathode RF gun. In Sect. 4, the results of the emittance optimization are presented. Finally, the conclusion is presented.

Description of the main components

2.1

Overview

Figure 1 shows the major components of the pilot plant, which is approximately 3 meters long. The photocathode (copper) RF gun, which was developed by Tsinghua University [22], is powered by a set containing a modulator and S-band klystron (provided by Canon Electron tubes & devices Co., LTD). It produces a single electron bunch typically with an energy of ∼ 4.5 MeV and 0.5 nC of bunch charge at a 1 Hz repetition rate (the designed repetition rate is 10 Hz). The emittance compensation solenoid is placed 0.2 m downstream of the cathode; then, this beam is transported into a beam diagnostic section, consisting of an integrating current transformer (ICT) provided by Bergoz Corporation, an energy analysis dipole magnet, and several beam-profile monitors with Yttrium Aluminum Garnet (YAG) screens. The horizontal and vertical multi-slits used for emittance measurement are located 1.17 m downstream of the cathode, and the beamlet images are observable on screen 2 0.4 m downstream of the position of the slits.

Figure 1Full-screen View

(Color online) Layout of the electron gun system

2.2

Drive laser system

The drive laser system was located in a cleanroom beside the experiment hall. The drive laser is a frequency-tripled, chirped-pulse amplification ps laser system based on Ti:sapphire (provided by Coherent Corporation). It can deliver a 266 nm UV laser with a ∼ 1.5 ps FWHM duration and a maximum energy per pulse of $\tilde{\text{2}}$.5 mJ maximum energy per pulse. The transverse and longitudinal distributions of this laser pulse are both Gaussian. The laser pulse was reshaped to be uniform in time and Gaussian truncated in space to reduce the nonlinear space charge effect on the beam emittance [23-25].

We used three pieces of the α-BBO crystal for temporal shaping. The input UV laser pulse was separated into eight sub-pulses with appropriate time intervals, which were then stacked to form a uniform laser pulse for cathode driving. The temporal profile was measured based on the cross-correlation method, and the result is shown as a blue curve in Fig. 2(a). The red curve in Fig. 2(a) is the fitting curve with a plateau distribution and R²=0.9438. With a rise time of 1 ps and a fall time of 1.27 ps, the laser pulse has an FWHM duration of 11.2 ps. To ensure the uniformity of the spatial distribution of the laser beam, an image transport system was built to deliver the shaped laser beam onto the photocathode 6 m away from the beam shaping aperture. Figure 2 (b) shows the transverse profile of the UV laser at the virtual cathode plane with a Gaussian-truncated profile diameter of 2 mm. Because the free transport distance of the entire optical path set is designed to be very short, the position jitter of the UV laser is less than 4 μm. The rms fluctuation of the laser energy was less than 0.7 % (the pulse energy on the photocathode was set at approximately 50 μJ).

Figure 2Full-screen View

(Color online) Temporal and spatial profile of the UV laser.(a) Temporal distribution of UV laser. The blue curve is the measured cross-correlation signal, and the red curve is the plateau fitting of the signal. (b) Spatial distribution of UV laser. The pink solid lines represent the horizontal and vertical projections.

Commissioning of the photocathode RF gun

3.1

Rf power

The high-power conditioning process of the RF gun lasted more than 400 h after installation, and the gun power was increased gradually to approximately 6.25 MW. A circulator was installed between the klystron and the RF coupler to isolate the output terminal from the load terminal. The probability of breakdown in the RF gun is below 0.01% under normal working conditions. The gun was mostly operated with an RF pulse length of 2.8 μs. Two measuring antennas were placed on both sides of the circulator to measure the incident power of the klystron (P_FWD)and the reflected power of the electron gun (P_REV). The measuring antenna on the side of the whole cell of the electron gun was used to measure the feed power in the cavity (P_GUN). Under working conditions, as shown in figure 3, the RF power in the gun cavity reached 6.25 MW. When the fields reach a steady state, the cathode field can be related to the gun power using the shunt impedance r_shunt = 3.6 MΩ, the gun effective length l_gun = 8.68 cm, and the surface field ratio R_field = 1.8, which gives the gradient to be[26] $E_{\text{field}}=R_{\text{field}}\left(\frac{\sqrt{P_{\text{gun}}{r}_{\text{shunt}}}}{l_{\text{gun}}}\right)=98\text{MV/m}\mathrm{.}$ (1) The following sections describe the calibration of the acceleration gradient was according to the measurement of the electron beam energy spectrum and bunch charge. In addition, the measured reflection power shown in Fig. 3 indicates that the waveguide-cavity coupling parameter ${\beta }_{\text{coupler}}\approx 1$. Therefore, we performed a Gaussian fitting on the reflection waveform, and the gun's response time (1/e) to changes in RF was given as τ_cav = 0.63 μs. The maximum power that the circulator can bear is 8 MW. Therefore, we set the pulse length to 2.8 μs to ensure that the RF power was sufficient to establish the desired gradient in the electron gun.

Figure 3Full-screen View

(Color online) RF waveforms measured during high power operation of the RF gun. The Gaussian fitting of the reflected waveform gives the cavity response time τ_cav= 0.63 μs.

3.2

Dark current

The dark current was not measured precisely because there was a strong interference from the modulator of the klystron on the ICT signal. To evaluate the level of dark current, we took advantage of the downstream YAG screen 1 to observe the intensity of the dark current beam. When the exposure time of the CCD was set to 0.1 s, the dark current beam was observed as shown in Fig. 4 with the solenoid was adjusted to focus it. The photo-electron beam was observed on YAG screen 1 with a bunch charge of 0.5 nC, and the exposure time of CCD was 0.1 ms. Therefore, we can roughly evaluate that the charge level of the dark current beam was lower than that of the photo-electron beam (0.5 nC).

Figure 4Full-screen View

(Color online) Dark current image on screen 1 during high power operation.

3.3

Beam alignment

Because the electron gun and solenoid may have off-axis and tilting errors, the precise collimation of the electron beam needs to be completed by adjusting the laser incidence position and the dipole correction magnetic field of the solenoid (in both the X & Y direction). The normal and skew correction quadrupoles installed inside the bore of the solenoid were tuned to compensate for the geometric distortion of the beam transverse profile. Figure 5 shows the correction result of the beam transverse profile with appropriate compensation on screen 1. We recorded the center positions of the beam at different main solenoid fields and the variation of center position was within 0.1 mm and 0.05 mm in the horizontal and vertical directions, respectively, where the scan range of the solenoid field covered the working condition of emittance optimization. The beam center variations on screen 2 were as small as those on screen 1 in both the horizontal and vertical directions. Electron beam transport displayed good position and pointing stability, which indicates that the beam alignment had been performed well.

Figure 5Full-screen View

(Color online) Beam image before (left) and after (right) multipole field correction at screen 1.

3.4

Beam energy measurement

The measurement of the electron beam energy helps to establish the peak acceleration phase of the photocathode RF gun. The energy diagnosis system was mainly composed of a dipole magnet, YAG target, and CCD camera. As shown in Fig. 1, the energy measurement system began at the position of screen 2. From the point of incidence to the observation on screen 3, the electron beam sequentially passed through a drift section of length a=0.47 m, a dipole magnet with an equivalent deflection angle of θ =38° and a deflection radius of ρ =0.374 m, and a drift section of length b=0.415 m. Because electrons of different momentums bent at different angles in a dipole magnet, the energy spectrum diagnosis could be accomplished by measuring the beam profile on screen 3.

When measuring the energy spectrum, we placed a slit at screen 2’s position and measured the beam size on screen 3. Based on the theory of the beam transformation matrix, the single-particle motion formula is given as $x_{\text{T}}=m_{x11}{x}_i+m_{x12}{x}_i^{\text{'}}+m_{x13}\frac{\Delta E}{E_0}\mathrm{,}$ (2) where xi, $x_i^{\text{'}}$, and $\frac{\Delta E}{E_0}$ are the transverse envelope, divergence angle, and energy spread, respectively. The resolution of the energy spectrum measurement is determined by the beam emittance according to the second-order moment of the first two terms of the particle motion formula: ${\sigma }_{\text{T}}=\sqrt{{\left(m_{x11}{\sigma }_x\right)}^2+{\left(m_{x12}{\sigma }_{x^{\text{'}}}\right)}^2}\mathrm{.}$ (3) With a slit width of 50 μm (${\sigma }_x=50/\sqrt{12}\text{μm}$) and a measured beam emittance of approximately 1 mm⋅ mrad (σx' $\tilde{0}$.3 mrad), the contribution of beam emittance on energy spectrum measurement was about 0.18 mm, considering that m_x11=0.1 and m_x12=0.6 m were given by beam transformation matrices. Based on the beam energy of 4.6 MeV and the dispersion function of m_x13= 0.343 m, the energy resolution can be given as $\Delta E=\frac{\Delta {\sigma }_{\text{T}}}{m_{x13}}\cdot E_0=\frac{0.18}{0.343}\times 4.6=2.4\text{keV}\mathrm{.}$ (4) Figure 6 shows the energy spectrum measured at the on-crest acceleration phase and the red curve is the Gaussian fitting result. When the rms width of the measured spectrum was 2.978 mm, the measured energy spread was 39.93 keV, which was in good agreement with the simulation value of 33 keV.

Figure 6Full-screen View

(Color online) The measured beam energy spectrum.

When the gun power reached 6.25 MW, the measured maximum energy was 4.608 MeV after adjusting the RF phase. We scanned the beam energy compared to the RF phase, as shown in Fig. 7. The simulation for the energy scan was performed using ASTRA code [27], and it indicated that the peak acceleration phase was 27.6° from the zero-crossing. As can be seen from Fig. 7, the measured curve matched well with the simulation result when the gradient on the cathode surface was set to 97 MV/m, which is consistent with the aforementioned gradient calculated by the RF power.

Figure 7Full-screen View

(Color online) The measured beam energy versus the RF phase at P_gun = 6.25 MW and the simulations with gradient at cathode of 96 MV/m, 97 MV/m, and 98 MV/m.

3.5

Bunch charge measurement

The bunch charge was measured by an integrating current transformer located at the exit of the electron gun. The typical curves with 6.25 MW and 4.23 MW gun power and 50 μJ laser pulse energy are shown in the Fig. 8. The red curve in Fig. 8 represents the simulated Schottky scanning results using the ASTRA program when the peak acceleration gradient was set to 97 MV/m and 80 MV/m. The Schottky parameters are given as Q₀=0.36 nC, SRT_QSchottky $\tilde{0}$ $\text{nC}/\sqrt{\text{MV/m}}$, and Q_Schottky=0.005 nC/(MV/m). When the launch phase deviated from the peak acceleration phase by approximately 67°, the bunch charge reached a maximum value of approximately 0.8 nC, which is consistent with the launch phase of approximately 90°, leading to the maximum gradient on the cathode surface. Based on the energy-phase scan curve and the bunch charge scan curves, the peak energy gain phase was defined as 27.6°. We take this peak energy gain phase as the on-crest phase, and the subsequent launch phases are all denoted as off-crest phases.

Figure 8Full-screen View

(Color online) Detected bunch charge as a function of RF phase at P_gun = 6.25 MW and 4.23 MW (blue lines). The red lines represent the simulations with a gradient at the cathodes of 97 MV/m and 80 MV/m.

In the measurement of quantum efficiency (QE), we measured the change in bunch charge as a function of laser energy at different RF launch phases. The gun power was set to 6.25 MW. The off-crest phases were chosen to be 0°, 38.2°, and 73.03°, and the corresponding acceleration gradients on cathode surface were 44.49 MV/m, 88.82 MV/m, and 96.51 MV/m respectively. The measured curves are shown in figure 9. We performed a linear fit of the scanned curves, and the QEs for different launch phases are given by the slopes of the straight line. Table 2 lists the measured results of QE, which decrease because of the Schottky effect as the cathode surface field decreases.

Figure 9Full-screen View

(Color online) The measured bunch charge versus the laser pulse energy at different RF phases.

Table 2Full-screen View

QE results

Cathode surface field	QE
44.49 MV/m	4.3× 10-5
88.82 MV/m	6.4× 10-5
96.51 MV/m	6.8× 10-5

Show more

Emittance optimization

The transverse projected emittance of the electron beam was measured using a multi-slit technique [28]. Using this technique, a space-charge-dominated electron beam is converted into a set of emittance-dominated beamlets, which are used to measure the local beam divergence. The emittance measurement system was installed in the setup, as shown in Fig. 10. A set of stainless-steel slits with a width of 50 μm and separation of 0.35 mm were mounted on an actuator located 1.17 m downstream of the photocathode. Stepping motors were employed to move and adjust the spatial position of the actuator and the observation screen with respect to the beam. The drift length from the slits to the measurement screen was 40 cm. In Fig. 10, we present a multi-slit sampling pattern to demonstrate the performance of our beam emittance measurement system. The sampling pattern was observed when the main solenoid field was set to 1800 G, and the beamlets could be distinguished clearly. However, the beam size was quite large, and the beam emittance was not optimized. The optimization of beam emittance discussed in the following paragraphs shows that the main solenoid field for the optimized beam emittance was larger than 1800 G, and the number of beamlets was reduced to three due to the smaller beam size.

Figure 10Full-screen View

(Color online) Schematic of the multi-slit sampling method. The beamlet images are observed on screen 3 at a distance L_d downstream of the slit position.

Figure 11 shows typical beamlet images on the observation screen during the emittance optimization. When the RF gun launch phase was set 3° off-crest the peak energy gain phase, the solenoid field was set to 1920 G for optimized beam emittance. The rms beam size at this solenoid field was measured to be 0.26 mm and therefore the three-beamlet sampling pattern was formed. The red curves represent the horizontal and vertical projections. The transverse rms beam size ($\sqrt{\langle x^2\rangle }$) and the beam divergence ($\sqrt{\langle x^{\text{'}2}\rangle }$) were obtained by analyzing the projection of the (x,x') phase-space distribution from multi-slit sampling and $\langle x{x}^{\text{'}}\rangle$ is the co-variance term where the relative position of the beamlet on the observation screen with respect to the corresponding slit position was taken into account[28]. The width of each beamlet gave a measurement of the width of the transverse momentum distribution at each slit, and the centroid of the beamlets gave the correlated offset of the momentum distribution at each slit. The rms normalized emittance was then given as ${\epsilon }_{\text{n}\mathrm{,}x}=\beta \gamma \sqrt{\langle x^2\rangle \langle x^{\text{'}2}\rangle -{\langle x{x}^{\text{'}}\rangle }^2}\mathrm{.}$ (5) Multi-slit sampling may result in underestimation of the beam transverse size, especially in the case of fewer sampled beamlets. To better estimate the real value, we scaled up the measured rms emittance according to the ratio of the observed beam size to the reconstruction value using a multi-slit sampling pattern. With the observed three-beamlet pattern shown in Fig. 11, we obtained a normalized beam transverse emittance of 1.20 mm⋅mrad in the horizontal direction and 1.17 mm⋅mrad in the vertical direction with a bunch charge of 500 pC and beam energy of 4.6 MeV.

Figure 11Full-screen View

(Color online) Typical experimental results for emittance measurement. The red lines represent the horizontal and vertical projections.

The solenoid located 0.2 m downstream of the cathode played a crucial role in emittance compensation by aligning the slice transversely along the bunch to minimize the projected emittance. However, there were aberration effects of the solenoid field, such as the geometric aberrations caused by the anomalous quadrupole field. Moreover, the double emittance minimum in field-scan measurements allowed by a new dedicated movable emittance measurement indicated that the minimum emittance is detected in the overfocus case [29]. Therefore, as mentioned in the beam alignment, we used the normal and skew quadrupole correctors to optimize the shape and then performed a beam size scan with the solenoid field.

As shown in Fig. 1, the transverse beam profile was measured using a YAG screen, and the light from the YAG screen was detected using a CCD camera. The simulated dependence of the beam size on the main solenoid field is shown by the red curve in Fig. 12(a). In this simulation, the optimum rms laser spot size of 0.64 mm was used, and the gun RF phase was set to the on-crest acceleration phase. The bunch charge was set to 0.5 nC with the appropriate laser pulse energy and cathode surface field. The optimization results showed that the smallest beam size was obtained with a main solenoid field of 0.1915 T. The minimum beam size was obtained at a main solenoid field of 0.1920 T (the horizontal and vertical beam sizes are shown as the black and blue curves in Fig. 12(a)), which is in good agreement with the simulation. As shown in this figure, the actual measurement value of the beam size was larger than the simulation value, while the variation in the beam size with the main solenoid field was not affected. The reason for this discrepancy may be that the actual beam emittance was higher than in the numerical simulation. In fact, the thermal emittance adopted in the simulation was smaller than that in the experimental situation and is discussed in the subsequent phase scanning process. In addition, the initial distribution of the laser pulses used for the beam transformation simulation was uniform in the longitudinal direction and a truncated Gaussian distribution in the transverse direction. The nonuniformity of the longitudinal and transverse distributions of the shaped laser was the main factor affecting the nonlinear space charge effect, which contributes to the beam emittance growth.

Figure 12Full-screen View

(Color online) The results of beam emittance optimization: (a) The measured and simulated beam size versus solenoid main field. (b) The measured and simulated emittance versus RF phase at a bunch charge of 500 pC.

Figure 12 (b) shows the result of the emittance optimization at the bunch charge of 0.5 nC. For different RF launch phases, the main solenoid field was scanned to achieve minimum emittance. The optimized solenoid field are given in Fig. 12(b). The phase scan results show that the optimization of the transverse emittance occurred when the launch phase was set 3° off-crest. The black curve in Fig. 12(b) is the simulation result for phase scan by the code ASTRA. When the launch phase is 27.6° and the peak cathode field is 97 MV/m, the effective work function of the copper cathode can be given as ${\varphi }_{\text{eff}}={\varphi }_{\omega }-0.0379\sqrt{E(\text{MV/m})\cdot \sin \theta }=4.06\text{eV}\mathrm{,}$ (6) where ϕω =4.31 eV is the intrinsic work function for a clean copper cathode [12, 30]. The photon energy of 266 nm is 4.67 eV. Taking the Schottky effect into consideration, the remaining kinetic energy E_k of the electron is approximately 0.6 eV. Using the thermal emittance calculation model given by [30] ${\epsilon }_{\text{n}}={\sigma }_x\sqrt{\frac{E_{\text{k}}}{3m_0{c}^2}}\mathrm{,}$ (7) we found that the thermal emittance was approximately 0.6 mm·mrad/mm, which was used for the ASTRA simulation. As shown in Fig. 12(b), the measured emittance was larger than that of the simulation. The experimental results of the thermal emittance in the photocathode RF gun reported earlier [30, 31] indicate that the experimental thermal emittance is approximately a factor of 2 larger than the theoretical thermal emittance, and many phenomena can be attributed to this discrepancy. For example, the roughness of the cathode surface may cause the thermal emittance to be greater than the calculated value.

Conclusion

In this paper, we describe the latest results of the experimental optimization of the high-brightness electron source at the photocathode RF gun for the Hefei Advanced Light Facility (HALF). In the initial system commissioning, all subsystems operated smoothly, including the drive laser system, photocathode RF gun, beam diagnostics system. The drive laser system can deliver a spatiotemporally reshaped 266 nm UV laser with an energy jitter of 0.7 % and a position jitter of less than 4 μm. The gun is typically operated at a peak gradient of 97 MV/m with few RF breakdowns. The two parameters, launch phase, and gun solenoid magnet, were optimized to obtain the lowest emittance, and the resultant minimum transverse emittance of 500 pC bunches was lower than 1.2 mm⋅mrad. The typical bunch energy reached 4.6 MeV at 1 Hz repetition. These results provide an excellent foundation for the injection of a high-brightness electron beam required for HALF.