The accelerator line of PAL-XFEL has a capability to generate maximum 11 GeV electron bunches with a high peak current and a low emittance. It is operated in ultra-stable conditions of the beam energy, arrival time, and availability.

The 780 m long accelerator line of PAL-XFEL includes a photocathode RF electron gun, a laser heater(LH), S-band normal-conducting LINACs, an X-band linearizer(XLIN), three bunch compressors(BC) in a HX line, one more BC in a SX line, a SX branch line. The accelerator line is divided to L0~L4 regions based on LH, SX branch line, and BCs.

Electron bunches emitted from the e-gun are 6 MeV, 250 pC, and 33 A. They are accelerated, collimated, and compressed to 4~11 GeV, 150 pC, and 3.3 kA in the HX line and 3 GeV, 150 pC, 2.2 kA in the SX line for a standard operation. The transverse emittances of bunches are preserved about 0.5 μm for both directions of horizontal and vertical during the acceleration. A line switching is conducted by a DC dipole magnet nowadays, but it will be changed to a pulsed kicker by 2020.

The injector laser system of PAL-XFEL provides mainly picosecond UV pulses to generates initial high quality photoelectron bunches, which spatial and temporal characteristics can be easily controlled by shaping the laser pulses. Additionally IR pulses for the LH and cleaning the photocathode surface are provided.

A commercial picosecond Ti:sapphire system with precision environmental control is a fundamental laser source, and the splitted two beams are sent to the photocathode and the LH respectively. The pulse characteristics are optimized by an external compressor, a spatial filter and Fourier imaging.

For the synchronization between laser pulse and other accelerator components, the Ti:sapphire oscillator(79.33 MHz) itself is locked to S-band RF reference oscillator in a femtosecond scale using a sensitive Sagnac interferometer based phase detector. A pulse picker chooses the oscillator laser pulses by 120 Hz trigger signal from the event timing.

The laser pulse energy is maintained in constant by rotating a wave-plates using a fast voice coil motor to ensure constant bunch charge. The laser pointing is stabilized by quadrant detectors, Piezo-actuated mirrors and relay imaging.

In diagnostic tools, the energy and spatial profile of UV laser are monitored in real time at a virtual photocathode position. The pulse shape is estimated from the cross- correlation between the UV and femtosecond oscillator pulses.

An electron gun(e-gun) is a device to generate a collimated electron beam with a precise kinetic energy. Electrons emitted from a cathode are accelerated by electromagnetic fields in the e-gun. There are two classes of the e-gun based on the method of the electron emission: a thermionic gun which emits electrons by heating the cathode and a photocathode gun which emits electrons by the photoelectric effect from a laser. Also, there are other two classes based on the e-beam acceleration: an electrostatic gun and an RF gun. A photocathode RF gun is generally used for XFEL machines because they require high quality injection e-beams with a low emittance to improve a FEL generation.

PAL-XFEL developed S-band photocathode RF e-guns with the new design since 2005. The e-gun based on the BNL type was fabricated on 2005 and it is the first photocathode RF gun with the high brightness in the nation. But, there were many problems to achieve the specifications for the XFEL machine. We changed the design to reduce dark currents and electric discharges. Also, we improve the method for vacuum sealing, frequency tuning, cooling, and laser injecting. There are dipole and quadrupole fields to dilute emittance from asymmetry by an RF port. We made effort to reduce these fields, and several designs were suggested for the symmetric structure: a single RF port with a vacuum port on the opposite side, RF feeding by a coaxial tube, a single RF port with 3 vacuum ports on the four sides, and double RF ports of the racetrack shape with 2 vacuum ports. The last one is applied for the PAL-XFEL e-gun and we finally developed the e-gun generated the e-beams of 6 MeV, 250 pC, 3 ps with 60 Hz. Also, the under 0.5 μm transverse emittance for the injector e-beam was achieved by this e-gun.

Particle accelerators are devices which accelerate elementary particles using an electromagnetic field. There are two classes of accelerators: electro-static accelerators which use static electric fields and electro-magnetic accelerators which use changing electro-magnetic fields to accelerate particles. Electro-magnetic accelerators with oscillating radio frequency(RF) fields are generally used in high energy machines. A linear accelerator(LINAC) is a type of the particle accelerator which accelerates charged particles along a linear beamline. It is often used to provide an initial low- energy kick to particles for circular accelerators or to make high energy particles for FEL machines.

PAL-XFEL use RF LINACs for accelerating electron beams. The electro-magnetic field in the accelerator are induced by the pulsed RF power from the klystron. If the e-beams pass through the accelerator with in-phase condition, the RF energy is delivered to the e-beams. The accelerating column is the SLAC-type constant gradient traveling wave structure with the normal conducting and made of the oxygen-free high conductivity(OFHC) copper. It uses the S-band RF with 2856 MHz and the 2/3π mode for accelerating.

The accelerator line is divided to L0~L4 regions based on a LH, a SX branch line, and BC. An RF module includes one accelerating column in L0, two columns in L1, and four columns in L2~L4. Pulse compressors(SLED cavity) are used at the RF module in L2~L4 to increase an energy gain. The energy gains are 0.12 GeV in L0 with 2 RF modules, 0.23 GeV in L1 with 2 RF modules, 2.2 GeV in L2 with 10 modules, 1.0 GeV in L3 with 4 modules. They are usually fixed in the operation because the RF parameters of L0~L3 have a large influence on the bunch compression and the longitudinal distribution of the e-beams. Whereas the energy gain in L4 with 27 modules is adjusted from 0~7.5 GeV by varying the pulse timing of each RF module for changing the photon energy of a HX FEL line. There is one more RF module(L3S) with the 0.25 GeV energy gain in a SX line. It is used for the XFEL operations required relatively higher photon energy(over than ~1 keV).

Magnetic BC are used to decrease an electron bunch length for high peak current e-beams in XFEL machines. There is an energy modulation of bunch slices by an RF time-structure before the bunch goes into the BC. The bunch slices of the higher energy travel a longer trajectory than slices of the lower energy in the BC. Therefore, the energy of tail slices must be higher than one of head slices for bunch compressing. The XFEL machines usually use the multi-stage of BCs for a low emittance and a high peak current e-beams. Since coherent synchrotron radiation(CSR) by bending in dipole magnets can lead to a microbunching instability(MBI) to spoil the FEL performance, it is important to reduce the CSR-kicks in the BC.

PAL-XFEL uses three BCs for a HX line and one more BC for a SX line, because a three BC scheme has advantage to reduce overall CSR-kicks in the BCs comparing to a two BC scheme. Each BC consists of four dipole magnets. A collimator in the BC1 is used to improve the emittance by collimating the spoiled slices in the head and tail. The 3 ps long bunches at an injector are finally compressed to 18 fs in the HX line and 27 fs in the SX line for a standard operation.

Electron bunches of XFEL machines are accelerated in off-crest because an energy slope of bunch slices are required for bunch compressing in BC. Since the bunch should be extremely compressed for an intense FEL, the non-linear effects such as the sinusoidal RF time-curvature and the second order optics in the BC are dominated to the bunch. The more bunch compression, the more sharp temporal spikes are generated to drive unwanted collective effects which interrupt the FEL generation. This problem can be compensated by a higher harmonic RF to linearize the energy curvature which comes from the fundamental RF time-structure. Hence, we call the higher harmonic RF system the ‘linearizer’.

PAL-XFEL use the X-band RF, the 4th harmonic of S-band, for the linearizer. The XLIN system was developed by a collaboration with SLAC(US). We installed a 50 MW X-band klystron and an X-band accelerating structure. Also, we developed a 450 keV modulator, an 1 kW X-band solid state amplifier, and an X-band low level RF controller. The XLIN is located right in front of the first bunch compressor. The accelerating structure is mounted on a movable stage for an alignment. It is operated in the deceleration mode with ~20 MeV.

Since XFEL requires very bright electron beams with a low emittance and a high peak current, the bunch is compressed by one or more number of magnetic BC after accelerated to above several hundred MeV which the bunch becomes less sensitive to a space charge effect. When the bunch becomes shorter by a factor of 100 through the compression, the e-beam quality can be impaired by a longitudinal microbunching instability(MBI). Coherent synchrotron radiation(CSR) with a wavelength shorter than the bunch length amplifies the density modulation leading to the MBI. To avoid growing the density modulation during bunch compressions due to noisy uncorrelated energy spread, a LH system makes the bunch smoother uncorrelated energy spread as increasing total energy spread(heating). Then the heated bunch with the smoother energy spread can suppress the MBI entering the first bunch compressor. As a result, after the full compression, uncorrelated energy spread is decreased due to the reduced MBI. It can significantly improve the XFEL performance.

A LH system is located at the end of the PAL-XFEL injector where the beam energy is 138 MeV. It is composed of a 0.5 m long undulator and a magnetic chicane with four dipoles. The drive laser is a Ti:Sapphire laser with a 760 nm wavelength. An infrared(IR) laser pulse is converted to an ultraviolet(UV) laser with a 253 nm wavelength using a third harmonic crystal for the e-beam generation in the photocathode RF gun. A small fraction, about 10 %, of the IR laser pulse is used for the LH. The undulator has a 50 mm period and a 25 mm minimum gap. The gap is controlled remotely with a 5 μm accuracy and changed by moving the upper and lower magnets simultaneously. The undulator is a planar and pure permanent magnet type. Two sets of correctors for horizontal and vertical steering are installed both upstream and downstream of the undulator.

An undulator in a HX line is a planar undulator of an out-vacuum type with the 5 m long and variable gap. The undulator design benchmarked the European XFEL design. The undulator wavelength λu is 26 mm and the minimum gap is about 8.3 mm as the 1.97 undulator parameter (K). Since the photon energy is tuned by the electron energy in the HX line, we use the fixed K=1.87 as ~9 mm undulator gap. There is the 1.05 m long intersection between neighboring undulators for a quadrupole magnet (QUAD), a cavity beam position monitor (BPM), a beam loss monitor, and a phase shifter. The QUAD is mounted on a mover for a beam based alignment (BBA). The BBA with QUAD movers and cavity BPMS are regularly conducted to compensate the ground motion.

We conduct the K-tuning by the undulator gap scanning regularly to maintain the K-value of each undulator. The vertical position of the undulator is movable for ±2 mm range. The scanning of vertical position to find the minimum field region of each undulator is also conducted with the BBA. We conduct the undulator tapering and the phase shifter scanning to increase the FEL intensity. Thus, we achieve over 1 mJ pulse energy for the 2~15 keV SASE FEL operations with the electron beam of 4~11 GeV, 150 pC, ~3 kA, and ~0.5 μm transverse emittance. Also, we achieve the ultra-stable operation: ~5% intensity stability, the position jitter 10% of beam size, ~3.5 eV central spectrum jitter, 10 eV spectral bandwidth, ~20 fs arrival time jitter, and over 95% availability. There is a self-seeding region with a chicane and crystals after the 8th undulator. We also provide the self-seeded FEL with ~0.5 eV spectral bandwidth for users.

An undulator in a SX line is a planar undulator of an out-vacuum type with the 5 m long and variable gap. The undulator design benchmarked the European XFEL design. The undulator wavelength λu is 35 mm and the minimum gap is about 9.0 mm as the 3.32 undulator parameter (K). Since the injection electron energy is fixed at 3 GeV, the photon energy is tuned by the undulator gap in the SX line. There is the 1.05 m long intersection between neighboring undulators for a quadrupole magnet (QUAD), a cavity beam position monitor (BPM), a beam loss monitor, and a phase shifter. The QUAD is mounted on a mover for a beam based alignment (BBA). The BBA with QUAD movers and cavity BPMS are regularly conducted to compensate the ground motion.

We conduct the K-tuning by the undulator gap scanning regularly to maintain the K-value of each undulator. The vertical position of the undulator is movable for ±2 mm range. The scanning of vertical position to find the minimum field region of each undulator is also conducted with the BBA. The phase shifter gap is tuned automatically to the optimum condition as the undulator gap variation. Thus, we provide over 1012 photons for the 0.25~1.25 keV SASE FEL operations with the 3 GeV electron beam of 150 pC, ~2 kA, and ~0.5 μm transverse emittance.

The PAL-XFEL project was started in April 2011 at Pohang Accelerator Laboratory (PAL) on Pohang, Korea. The Korean government launched the project with a budget of ~400 billion Won (~400 million USD). The project involved 75 members - 35 newly hired members and 39 experienced members from PLS-II. The project was completed in November 2016.

The PAL-XFEL facility has the capacity for five undulator lines: three HX undulator lines and two SX undulator lines. However, the budget was limited to two undulator lines: one for the HX and the other for the SX line. Since the PAL is the host institution carrying out this project, the project budget was able to avoid significant costs - the payment for the land of the building site and the infrastructures such as power transmission lines. The PAL-XFEL was constructed on the ground. The ground breaking was started in June 2012 and the building was constructed from September 2012 to January 2015. The installation of devices was completed in December 2015. The commissioning was started in April 2016 after the permission for the radiation safety. We achieved the first SASE FEL lasing at 0.5 nm in June 2016. The project was completed with the first saturation of the 0.15 nm FEL in November 2016. We achieve the design goal of 0.1 nm FEL saturation. Finally, the user service was started in June 2017.