- Email: [email protected]
Design of a high repetition rate S-band photocathode gun
Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Design of a high repetition rate S-band photocathode gun Jang-Hui Han , Matthew Cox, Houcheng Huang, Shivaji Pande Diamond Light Source, Oxfordshire OX11 0DE, UK
a r t i c l e i n f o
abstract
Article history: Received 24 February 2011 Received in revised form 11 May 2011 Accepted 12 May 2011 Available online 25 May 2011
Photocathode RF guns have been developed in many laboratories for generating high quality electron beams for free-electron lasers based on linear accelerators. Such guns can generate electron beams with an exceptionally high peak current as well as a small transverse emittance. Their applications have been recently expanded for ultrafast electron diffraction, coherent terahertz radiation, and X-ray or g-ray radiation by Compton scattering. In this paper, we design an S-band normal-conducting gun with capabilities of high quality beam generation and high repetition rate operation. The RF design and thermal analysis of the gun cavity and coupler are introduced. Optimal position of the gun focusing solenoid for low emittance beam generation is found by performing particle tracking simulations. Then, the gun system is designed to be able to afford the optimal solenoid position. The cooling-water channel surrounding the gun cavity and coupler is designed and analyzed numerically. The pressure in the gun is simulated with a vacuum model containing the detailed inner structure of the gun. An injector for a free-electron laser application is designed by using this gun and the beam dynamics simulation is shown. A cold test with a prototype gun for confirmation of the RF design is reported. & 2011 Elsevier B.V. All rights reserved.
Keywords: Photoinjector RF gun High repetition rate Low emittance beam
1. Introduction Beam brightness from electron injectors has been dramatically increased with the development of photocathode RF guns for the last two decades. In many laboratories, such guns have been adopted as the injectors for free-electron lasers (FELs) based on linear accelerators (linac) [1–6], for ultrafast electron diffraction [7–11], for coherent terahertz radiation [12,13], and for X-ray or g-ray Compton scattering [14]. In such guns, an electron beam is generated at the cathode by using a drive-laser pulse and the beam is accelerated from rest to relativistic energy by the strong RF accelerating field. By optimizing the drive-laser pulse and accelerating the emitted beam quickly, high peak current and low transverse emittance can be achieved. When RF power is transmitted into a gun cavity, RF power dissipation takes place at the cavity surface. Then, the surface temperature rises at the location with strong surface current. Due to the local temperature rise, the temperature distribution of the cavity becomes non-uniform, cavity deformation takes place, and therefore the cavity may suffer from mechanical stress. Such a cavity deformation also affects the RF field distribution, which may result in changes of the resonant frequency and field balance. Therefore, the cavity temperature must be controlled with cooling-water in order to keep the cavity at the nominal temperature
Corresponding author. Tel.: þ 44 1235 778935.
E-mail address: [email protected] (J.-H. Han). 0168-9002/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2011.05.032
and to keep the temperature distribution as uniform as possible. As the RF duty factor increases, the problem of the non-uniformity of temperature distribution becomes more severe. This limits the maximum achievable repetition rate and also impairs stable RF operation. An optimally designed cooling-water channel is therefore required. Another obstacle to increasing the repetition rate may come from the excessive heating at the RF coupling region. If the RF coupler is connected to the side of the cavity cell where the surface current induced by the RF field is strong, the coupling region can have high temperature and high surface stress due to the RF heating. However, if the coaxial RF coupler is instead connected to the gun front iris where RF power dissipation density is low, the coupling region can be relatively cool and does not limit the repetition rate. Furthermore, when a coaxial coupler is utilized, the entire outer tube of the cavity can be enclosed with axisymmetric cooling-water channels around the cavity cylinder. Therefore, the cooling capacity can be maximized and the cavity deformation caused by the RF heating also becomes axisymmetric. High order transverse RF modes will not exist because of the perfectly symmetric cavity inner surface as well. Such a type of RF coupler has been employed in the DESY L-band guns [15], which are operating for FLASH [4] and PITZ [16]. Even with the fast acceleration of a beam in an RF gun, the beam radially expands through the gun and the transverse emittance blows out due to the space charge force. This transverse increase can be compensated by using a focusing solenoid [17]. This process is realized by reconfiguring the beam distribution in the transverse
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
phase space by using the focusing solenoid. The solenoid should be placed around the gun for an efficient transverse emittance control as will be discussed in Section 2. If the RF coupler is connected to the side of the cavity cell, the optimal location for solenoid is occupied by the coupler and therefore the solenoid must be placed downstream of the gun. In this gun design, the solenoid can be placed at the optimum location thanks to the coaxial coupler. The resonant frequency was chosen to be 2.998 GHz, which is the European standard S-band frequency. Compared with the Eindhoven gun [18], which is another S-band gun with a coaxial RF coupler, the possibility of further cooling channel installation was considered and the coupler tube length was increased to allow solenoid installation around the gun. Compared with lower frequency band guns such as L-band guns, an S-band gun made from oxygen-free electronic (OFE) copper generates very low emittance and short pulse electron beams because of the high accelerating field at the cathode. Compared with higher frequency bands, such as C- or X-bands, an S-band gun allows enough space for the coolingwater circuit installation inside the iris and therefore temperature control of the gun against RF thermal heating can be properly performed. Their repetition rates are, however, generally limited to about 100 Hz [5]. The gun introduced here has a potential of a repetition rate as high as 1 kHz as will be discussed in Section 3. This gun will provide a possibility to operate an FEL based on normalconducting technology at such a high repetition rate [19]. The RF design, thermal analysis, and vacuum simulation of the new gun are reported in this paper. A prototype of the gun was produced for low power RF tests with a network analyzer. The test results with the prototype are also presented.
2. Beam dynamics consideration in the gun design The transverse beam emittance can be minimized by optimizing the gun cavity geometry and the location of the gun solenoid. The lengths of the gun cells are optimized based on the previous study in Ref. [20] and further beam dynamics simulations for this particular gun design have been carried out. The optimum location of the main solenoid is also found with beam dynamics simulation. The length of the first cell strongly affects the beam formation and acceleration when the beam is in a non-relativistic regime. When a beam is emitted from the cathode, the velocity of the beam is almost zero. After the emission, the beam is accelerated by the RF field to become relativistic through the first cell. When the beam energy is very small, the space charge effect changes the beam properties considerably. As the beam is accelerated, the space charge effect becomes weaker. The beam dynamics also depends on the length of the second cell. The relation between the cell lengths and beam dynamics has been studied in detail elsewhere [20]. After simulations with various sets of cell lengths as carried out in Ref. [20], the lengths of the first (half) and second (full) cells were chosen to be 28 and 50 mm, respectively. This optimization was carried out at 200 pC bunch charge, 8 ps laser pulse length and 120 MV/m peak field at the cathode. Even though the effect on the beam dynamics is relatively small compared to the initial beam shape defined by the drive-laser parameters or the focusing solenoid field configuration, the gun should be carefully designed for best beam performance. Otherwise, a new manufacturing and installation of a gun will be needed to improve the beam quality later with an improved gun design. This is practically difficult because it may interrupt the operation of the entire accelerator system for quite a long time. When an electron beam is accelerated in a photocathode RF gun, the strength of space charge effect is different for each
0.4
0.35 emittance (mm mrad)
18
0.3
0.25
0.2 0.05
0.1
0.15 0.2 solenoid position (m)
0.25
Fig. 1. Normalized transverse emittance (full rms) as a function of the distance between the center of the solenoid and the cathode. Simulations were carried out at 200 pC bunch charge and 100 MV/m peak RF field at the cathode. The drivelaser beam size and the beam launch phase at the cathode were optimized at each solenoid position.
temporal slice of a bunch. The head and tail slices experience weaker repulsive force to the radial direction, while the central slices experience stronger defocusing force. As a result, the slices show different shape and angle in transverse phase space. The area in the phase space, emittance, can be minimized under a certain condition combined with the solenoid field, the RF field, and the initial beam shape. As pointed out in Refs. [15,21], if the gun consists of two (1/2þ1) cells, the optimum location of the solenoid is around the cavity. Fig. 1 shows the beam emittance at the exit of the gun for a 200 pC beam as a function of main focusing solenoid position. At the cathode, the solenoid field is compensated to zero by a bucking solenoid sitting behind the gun. For the simulations using the ASTRA code [22], 100 MV/m gun RF peak field at the cathode and 8 ps drive-laser pulse length were used. A flat-top temporal laser profile with 1 ps rise/fall time was used. The radial size of the drive-laser pulse was optimized to obtain the lowest transverse emittance for each condition. Since these simulations were carried out without further linac sections, the projected beam emittance blows up after the minimum point. The minimum projected emittance takes place when the temporal slices of the beam are aligned in the transverse phase space, and the distance from the cathode to the minimum emittance point varies depending on the solenoid position. Here, we took the minimum projected emittance value for each simulation condition. The laser full radius was in a range between 0.28 and 0.46 mm. The optimum radius for a lowest emittance increased with the solenoid position from the cathode. The launch phase of the beam at the cathode was also optimized at each solenoid position. As shown in Fig. 1, lower emittance can be achieved as the solenoid moves toward the cathode. When the physical thickness of the solenoid, 0.135 m in the design here, is considered, the surrounding of the gun cavity should be cleared for solenoid installation in order to achieve the lowest emittance.
3. Gun cavity design In this section, we discuss the detailed gun design. An RF design is carried out, which satisfies the requirements discussed in Section 2. Given the RF field distribution, the thermal behavior is analyzed numerically.
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
quality factor power
14400
3
6.1
14200
2 1 0 1
2
3
4
5
6 7 z (cm)
8
9
10
11
Fig. 2. RF electric field distribution of the gun cavity. The cathode plug and the slot for the plug at the rear wall are shown. The RF contact between the cavity and the cathode is made with a Cu–Be spring. A part of the inner antenna of the RF coupler is shown at the right side. The emitting surface of the cathode is at z ¼ 5 mm.
quality factor
r (cm)
6.2
14600
6
14000
5.9
13800 5.8 13600 5.7
13400
5.6
13200
5.5
13000
5.4
12800 20 A slot for exchangeable cathode plugs was implemented at the rear wall of the gun (see Fig. 2). With the cathode slot, damaged cathodes can be easily replaced with a fresh one and various photo-emitters such as Cu, Mo, Cs–Te, or any other new material can be used for test and operation. When the front surface of the cathode is not perfectly aligned with the rear wall of the gun cavity, which may take place during cathode exchange, the RF characteristics may be influenced. By reducing the diameter of the cathode plug, this effect can be minimized. If the radial size of the cathode extends to the region where the RF magnetic field is strong, a small misalignment in the longitudinal direction induces a large RF resonant frequency shift and field balance change. Another possible side-effect of a large diameter cathode is thermal heating in the outer part of the cathode front surface caused by the magnetic field. The diameter of the cathode plug was chosen to be 8 mm. The plug size was also chosen considering a minimum impact on the beam dynamics due to field distortion at the cathode plug edge. If the edge of the cathode plug is too close to the center of the beam axis, the RF field distortion caused by the edge may affect the beam dynamics. The RF field distribution and power dissipation density at the gun cavity were calculated by using the SUPERFISH code [23]. Fig. 2 shows that the p mode is activated. The maximum field strength in the first cell was set to be higher than in the second cell by 4% for uniform RF power dissipation density and equal radial deformation in both cells. The details such as the cathode plug and the slot for the plug at the rear wall of the gun as well as the coupler antenna were included in the RF simulation. The quality factor of the gun cavity is 13,570 at 45 1C (see Fig. 3). The p mode is utilized for beam acceleration. Even if the 0 mode is separated from the p mode, the tail of the 0 mode peak may be still activated at the resonant frequency of the p mode, 2.998 GHz. When the 0 mode is activated, an electron beam may be affected by the unwanted 0-mode field in a wrong phase. If the 0 mode is separated far enough from the p mode, the unwanted influence can be reduced. In this design, the mode separation is 21 MHz, which is larger than those of the other S-band guns as reported in [5,18,24]. The temperature on the RF surface of the gun rises when an RF pulse is applied and falls when the RF pulse ends. The heat at the surface, caused by the RF pulse, is transferred to the surrounding Cu volume and is then transferred to the cooling-water by heat convection. The local surface temperature of the gun varies from location to location because the local RF heat dissipation density (caused by the magnetic field) is different and heat transfer to the body and cooling channels also varies. After a number of RF cycles, the gun temperature distribution reaches the steady state so that the local temperature when an RF pulse starts is the same as the temperature when a next pulse starts because the RF heat
power (MW)
4
19
30
40 50 60 temperature (°C)
70
80
Fig. 3. Quality factor of the gun cavity and RF power required for 100 MV/m peak field at the cathode at several operating temperatures. The temperature was assumed to be uniform over the cavity.
Fig. 4. Gun cooling-water channels. The channels enclose the cavity tube (left). In the irises and cavity rear wall, the cooling channels are shaped on half-circle discs (right) and brazed into the main body.
dissipation to Cu surface and the heat transfer to cooling-water are balanced. The cooling-water channels were designed for high cooling capacity, uniform temperature distribution over the gun body, and mechanical sustainability against high water pressure (see Fig. 4). Through the cooling channels of about 36 mm2 crosssectional area, cooling-water flows with a speed of 3 m/s. The minimum wall thickness between cooling channel and cavity inner surface is 3 mm. In order to achieve 100 MV/m peak field at the cathode when the gun operates at 45 1C (Fig. 3), 5.7 MW peak RF power should be provided from an RF power source to the gun. When this gun operates with 3 ms RF pulse length and 1 kHz repetition rate, the average dissipated power is 17 kW. Now, we study if this gun can operate with the average power under the given conditions. The RF power dissipation density at the RF surface was used as the input for the steady state temperature, deformation, and stress analyzes with ANSYS [25]. The real three-dimensional cooling-water channel shape was modeled. For these analyzes, 16 1C cooling-water temperature through the cooling channels was assumed. The result of the temperature analysis is shown in Fig. 5. The maximum temperature of the gun was found to be about 50 1C, which occurs at the iris between the cells. The average temperature
20
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
Fig. 5. Steady-state temperature distribution of the gun cavity simulated by using ANSYS.
Fig. 7. Steady-state von Mises stress on the RF surface of the gun cavity.
pulse heating can be found with the equation [27]: pffiffiffi 2PRF t DTs ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi prkC e where PRF is the dissipated RF power density and t is the pulse length. r ¼ 8:93 103 kg=m3 , k¼391 W/m K and Ce ¼ 385 J=kg K are the material density, the heat capacity and the specific heat for the OFE copper. In this gun design, a maximum peak dissipation of 225 MW/m2 occurs at the iris. For 100 MV/m and 3 ms, the temperature rise caused by an RF pulse is 12 1C, which is far below the damage threshold for cyclic stress, 40 1C [26].
4. RF coupler design
Fig. 6. Steady-state gun cavity deformation by using ANSYS.
over the inner surface of the cavity is about 45 1C. Even though there is no direct water-cooling applied to the cathode plug, the plug temperature is below 40 1C. The right hand side of the front iris, where the RF coupler is connected, is also below 40 1C. The cavity deformation (see Fig. 6) was analyzed with the temperature distribution in Fig. 5. The end of the left hand part was fixed for the analysis. The maximum (total) deformation was about 40 mm. However, the deformation is axially symmetric and therefore no dipole or higher order mode is configured from the RF heating. The radial deformations of both cells are almost the same and below 15 mm (not specifically shown in Fig. 6). Therefore, the RF field balance will be affected only by 0.001%. The yield stress of the annealed oxygen-free copper is known to be 62 MPa [26]. The steady state stress may not induce material damage when the gun operates continuously. However, the cavity temperature and deformation should still be controlled for stable RF operation and long cavity lifetime. The local stress caused by the RF heating was also studied by using ANSYS. The maximum stress occurring at the inner corner of the second cell is 50 MPa (Fig. 7). This is in the safe regime by about 20% margin. The stress at the cathode plug, which will be made from molybdenum, is below 17 MPa (not shown in Fig. 7). The yield stress of molybdenum is far higher than copper. During pulsed RF operation, a transient temperature rise takes place during the RF pulse and the temperature falls again when the pulse ends. As this process repeats, severe surface fatigue may be produced. The transient temperature rise, DTs , at surface by RF
In this section we describe the RF coupler. The coaxial coupler tube is designed to be long enough to place the main focusing solenoid around the second cell of the gun. The coupling coefficient is optimized to be 1. Thermal analysis is carried out with ANSYS too. RF power will be transmitted to the coupler through the WR284 rectangular waveguide. The door-knob transformer converts the TE10 mode in the waveguide to the transmission mode in the coaxial line between the door-knob and the cavity. The transmission mode is a hybrid of the TEM and TE11 modes. The gap between the inner antenna tip and the cavity front iris is crucial for deciding the coupling. There are reflections from the shorted end of the waveguide and the open end of the coupling antenna. Therefore, the location of the shorting plane is also important for having a proper match between the gun cavity and the input waveguide. The part of the reflected wave from the shorting plane, the part of the reflected wave from the antenna tip and the radiated wave from the cavity will add up to a negligible amplitude to have almost no reflected wave in the input waveguide for a properly matched cavity. For the calculation of the S-parameters, the CST Studio timedomain (TD) solver [28] was used. S11 was computed for different shorting plane locations for a fixed value of gap. Such sets were obtained for different gap values to arrive at the optimum location of the shorting plane and the coupling gap. The computed values of S11 are 10 and 33 dB for the 0 and p modes, respectively. Fig. 8 shows S11 computed by the CST Studio TD solver with RF power losses included. Fig. 9 shows the electric field (absolute) of the p mode computed by the CST Studio TD solver for 0.5 W power input through the coupling waveguide. The computed electric field at the cathode center is 30.4 kV/m. For 100 MV/m peak field at the
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
21
0 -5
S11 (dB)
-10 -15 -20 -25 -30 -35 2.975
2.98
2.985 2.99 frequency (GHz)
2.995
3
Fig. 8. S11 parameter as computed with the CST Studio time-domain solver. The 0 mode (left peak) is under-coupled and the p mode (right) is critically coupled.
RF power in
shorting plane Fig. 9. Electric field (absolute) of the p mode as computed with CST Studio for 0.5 W at the input port of the waveguide.
cathode center, the required power is 5.4 MW which is in good agreement with the SUPERFISH results mentioned in Section 3. In order to control the temperature of the coupler antenna, one turn of the cooling-water channel was accommodated downstream of the inner antenna (see Fig. 11). Both 20 1C coolingwater temperature and 2 m/s water flow rate were used for cooling the coupler. The RF heat dissipation at the RF surface of the coupler inner antenna was calculated with CST Studio and the result was used as input for the ANSYS analysis in the steady state condition. The analysis results are shown in Fig. 10. The maximum temperature was 61.5 1C at the end tip of the antenna. Thermal convection at the end tip relies on the thin, long antenna tube. Since the heat transfer through the antenna tube is not efficient, the temperature at the region is relatively hot. The deformation is about 60 mm at the end tip. However, the impact on the RF coupling to the gun cavity is negligible according to CST Studio simulation. The maximum von Mises stress is 1.6 MPa at the corner of the RF mode converting region. Even though the temperature at the tip is relatively high at the steady state (61.5 1C), transient temperature rise during an RF pulse at the coupler is very small because the heat dissipation density (caused by the magnetic field) is far less than 1% of that at the gun cavity. A maximum dissipation of 1.1 kW/m2 occurs at a point on the antenna surface. For 100 MV/m peak field at the cathode and 3 ms pulse length, the temperature rise caused by an RF pulse is only 6 10 5 1C at that point.
Fig. 10. Thermal analysis of the RF coupler at the steady state; temperature distribution at the steady state (top), deformation caused by the RF heating (middle), and von Mises stress on the RF surface (bottom) at the steady state. Thermal analysis was carried out by using ANSYS. On the outer surface of the inner antenna, RF power is dissipated during RF transmission to the gun cavity.
Since the heat dissipation density on the inner surface of the coupler outer tube is lower than that on the inner antenna outer surface and the outer tube is exposed to air and also connected to other copper parts, the thermal effect on the outer tube is expected to be even smaller than on the inner antenna.
5. Gun section design A gun section was designed as shown in Fig. 11. The main gun solenoid is placed around the gun cavity and RF coupler tube. With this gun cavity and coupler, the solenoid can be placed within a range of 0.07–0.12 m from the cathode. A smaller solenoid is placed just behind the gun in order to compensate the magnetic field at the cathode. Through the longer, open rectangular arm of the RF coupler, high power RF is forwarded to the gun cavity. An RF window will be connected to this port. The shorting plane at the shorter arm makes the coupling coefficient critically matched and is also used as a vacuum pumping port through the grille at the plane. The rear part of the gun cavity is to be connected to a cathode exchange chamber. Cathodes can be exchanged under ultrahigh vacuum by using a manipulator. Through the long inner hole of the inner antenna tube of the coupler, a drive-laser pulse is transmitted from a laser mirror
22
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
0.8
×10-9
0.7
total pressure (mbar)
0.6 0.5 0.4 0.3 0.2 0.1 0 0
Fig. 11. Gun section layout. It consists of an RF gun, an RF power coupler, and focusing solenoids.
0.05
0.1
0.15
0.2 z (m)
0.25
0.3
0.35
Fig. 13. Simulated total pressure on the symmetry axis of the gun. The cathode is at z¼ 0 m.
ultrahigh vacuum by a 100 l/s ion pump (see Figs. 11 and 12). Additional pumping will be provided with a second 100 l/s ion pump at a downstream chamber, where a drive-laser mirror will be installed. At the other end of the rectangular waveguide connected to the coupler, an RF window will be installed. Near the window, an additional pump will be installed; these are not shown in Fig. 12 and were omitted from the vacuum simulation. This pump will mainly absorb gases emitted from the RF window. The pressure distribution was simulated by using the MonteCarlo Test Particle software, MCTPVac [30], developed in-house. An out-gassing rate of 10 11 mbar l/s cm2 at the vacuum surface, which should be achieved after full RF conditioning, and a gas molecular mass of 28 (nitrogen equivalent) were assumed. The pump entrance ports were modeled as pumping surfaces with an appropriate sticking probability 0 o s o1. Diffuse scattering of molecules is assumed at all internal surfaces. Adequate statistics were obtained with 100,000 molecules. As expected, the highest calculated pressure is found in the first cell of the gun but stays below 8 1010 mbar (see Fig. 13).
7. FEL injector beam dynamics Fig. 12. Vacuum surface model. The detailed inner structure of the gun cavity and coupler, including the grille at the coupler shorting plane, was modeled.
placed downstream of the gun to the cathode. A beam generated at the cathode also propagates through the antenna hole.
6. Vacuum simulation Ultrahigh vacuum is required in the gun cavity in order to suppress RF breakdown and also to prevent deterioration of the quantum efficiency of the photocathode due to pollution. For keeping the cathode lifetime reasonably long, 4 1000 h, the vacuum must be better than 10 9 mbar for either a metal cathode or a Cs–Te cathode [29]. In the gun system, there is no pumping port connected directly to the gun body. Pumping should be achieved through the RF coupler and the downstream beam pipe. At one end of the coupler waveguide there is RF shielding grille to reflect the RF wave with parallel slots allowing pumping to
By using this gun, an injector was designed for high beam quality and 1 kHz repetition rate [31] with an assumption of being used for an FEL based on a normal-conducting linac [19]. This injector consists of the gun system and four 3 m traveling wave linac tubes. For 1 kHz operation, 100 MV/m gun peak field and 12.5 MV/m linac gradient were used. Beam dynamics simulations were carried out by using ASTRA with 100 000 macroparticles. The center of the main gun solenoid was placed 0.09 m from the cathode. A 200 pC beam is generated at the cathode by a drive-laser pulse with 8 ps full width at half maximum (FWHM) and 0.32 mm full radius. The temporal shape of a laser pulse was assumed to be flat top with 1 ps rise/fall time. The transverse distribution was assumed to be uniform. For a realistic estimation of thermal emittance, the kinetic energy of emitted electron was assumed to be 0.6 eV, which is higher than the experimentally known value [32]. As discussed in Section 2 transverse beam emittance may be improved if the solenoid is shifted toward the cathode. However,
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
180
1.8 emittance energy
160
1.4
140
1.2
120
1
100
0.8
80
0.6
60
0.4
40
0.2
20
beam energy (MeV)
1.6 emittance (mm mrad)
23
0
0 0
5
10 z (m)
15
20
Fig. 14. Normalized transverse emittance and beam energy evolution from the cathode (z¼ 0 m) to the end of the injector.
Table 1 Parameters of the injector simulation using ASTRA. Beam parameters are shown at the end of the injector. Bunch charge Laser pulse length (FWHM) Laser rise/fall time (10–90%) Laser full radius Kinetic E of emitted electrons Intrinsic (thermal) emittance Peak field at the cathode Beam launch phase at the gun Linac gradient Phase of the 1st linac tube Phase of the 2nd–4th linac tube Projected rms emittance (100%) Emittance of central slices Beam size Bunch length (FWHM) Peak current DE=E at centre Mean E
200 pC 8 ps 1 ps 0.32 mm 0.6 eV 0.142 mm mrad 100 MV/m 371 from 0-crossing 12.5 MV/m 601 from on-crest 01 from on-crest 0.172 mm mrad 0.16 mm mrad 0.12 mm 7.2 ps 30 A
Fig. 15. Simplified gun prototype for cold test with a network analyzer. The prototype includes the critical RF features.
1:14 105 140 MeV
the 0.09 m solenoid position was chosen to minimize possible multipacting processes at the cathode which may take place when the main solenoid is placed too close to the cathode [33]. The normalized transverse emittance and beam energy evolutions with distance from the cathode are shown in Fig. 14. The emittance (full rms) is about 0.17 mm mrad at the end of the injector, where the beam energy is 140 MeV. This low emittance was achieved by optimizing the gun cell length, the gun solenoid position, and the first accelerating section position [31]. The peak current is 30 A. The input parameters for the beam simulation and the results are summarized in Table. 1.
8. RF cold test with a prototype A simplified prototype was fabricated to test the RF characteristics with a network analyzer (Fig. 15). The prototype does not have cooling-water channels or vacuum connection. Low quality oxygen-free copper was used and the machining requirements for tolerance and surface finishing were relaxed to save the cost. However, the critical features for testing the RF characteristics, the cathode slot, the connecting part between the gun cavity and coupler, and the vacuum grille at the coupler were included. The machined parts were vacuum brazed for good RF contacts
Fig. 16. S11 measurement with a network analyzer.
between the parts. Since the low grade oxygen-free copper was not heat-treated prior to machining, the machined body could be slightly deformed during the brazing process. With the prototype, the resonance frequency, field balance, and quality factor of the cavity and the RF coupling were measured. The resonance frequencies were shifted to the direction of lower frequency by about 4 MHz (see Fig. 16). It might result from the low quality machining tolerance and also from the copper body deformation during the vacuum brazing. The measured quality factor was smaller than the design value by 11.5%. It might result from the low quality surface finishing. The RF field at the first cell was higher than at the second cell by 2%, which is fairly close to the design value, 4%. As shown in Fig. 10, the p mode was critically coupled. The voltage standing wave ratio (VSWR) was 1.02. The 0 mode was strongly under-coupled as
24
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
Acknowledgments
Table 2 Design parameters and measured ones with the prototype. Parameters
Design
Prototype
bp
1 2998 13,875 1.04 20.8 0.520 11,870
0.99 2994 12,280 1.02 21.2 0.453 10,504
mode
fp mode Q p mode at 25 1C Field balance Df ðp20 modeÞ
b0 Q0
mode mode
at 25 1C
We thank D. Brice, N. Hammond, M. Jensen, J. Kay, A. Peach, and L. Zaja for their engineering support on the design and measurement. J.H.H. thanks K. Floettmann at DESY for valuable discussions and R.P. Walker for encouraging this study and comments on the manuscript. I.P.S. Martin kindly proof read the manuscript.
References expected from the calculation. The mode separation between the p and 0 modes was 21.2 MHz. The result of the cold test is summarized in Table 2. RF tuning was exercised by using the plastic deformation of the rear and front walls as carried out for the DESY gun [34]. Tuning up to 400 kHz with the rear wall and 100 kHz with the front wall were tested. In practice, the resonance frequency and the field balance will be checked with a network analyzer prior to the final brazing of the cavity parts for an operational gun. Therefore, only a small amount of tuning will be needed by plastic deformation.
9. Discussion We here designed an improved S-band gun for high repetition rate operation and a low transverse beam emittance. By adopting the coaxial RF coupler connected to the front iris of the gun, the water-cooling capacity could be increased and the focusing solenoid could be placed at the optimum location for emittance compensation. When the gun operates at 100 MV/m peak field at the cathode, 1 kHz repetition rate will be possible. The factor limiting the repetition rate is the thermal stress at the gun cavity. The coaxial coupler does not limit the repetition rate. This gun may be used for high brightness electron injector for FELs based on normal-conducting linacs. An S-band injector capable of 1 kHz operation was designed using the gun. A 200 pC beam will have a transverse emittance about 0.17 mm mrad and a peak current of 30 A. Other possible applications using this gun are ultra-short electron beam generation for diffraction experiments and terahertz or X/g-ray radiation. This high brightness and high repetition gun will be beneficial to such applications. The RF design of the gun and coupler was successfully tested with a prototype. An operational gun is under production for a high power (but low repetition rate) RF test with the existing RF system at Diamond Light Source. A high repetition rate RF station will, however, be needed to test the full capacity of the gun.
[1] P.G. O’Shea, et al., in: Proceedings of the Particle Accelerator Conference, 1991, p. 2754. [2] D.H. Dowell, et al., Appl. Phys. Lett. 63 (1993) 2035. [3] R. Dei-Cas, et al., Nucl. Instr. and Meth. A 296 (1990) 209. [4] S. Schreiber, in: Proceedings of the European Particle Accelerator Conference, 2000, p. 309. [5] R. Akre, et al., Phys. Rev. ST Accel. Beams 11 (2008) 030703. [6] H.S. Kang, S.H. Nam, in: Proceedings of the 32nd International Free Electron Laser Conference, 2010, p. 155. [7] J.B. Hasting, et al., Appl. Phys. Lett. 89 (2006) 184109. [8] R. Li, et al., Rev. Sci. Instrum. 80 (2009) 083303. [9] J. Yang, et al., Rad. Phys. Chem. 78 (2009) 1106. [10] P. Musumeci, et al., Rev. Sci. Instrum. 82 (2010) 013306. [11] J.-H. Han, Phys. Rev. ST Accel. Beams 14 (2011) 050101. [12] R. Kuroda, Nucl. Instr. and Meth. A 593 (2008) 91. [13] C. Yim, et al., in: Proceedings of the International Particle Accelerator Conference, 2010, p. 1059. [14] R. Kuroda, et al., Nucl. Instr. and Meth. A 637 (2011) S183. [15] B. Dwersteg, et al., Nucl. Instr. and Meth. A 393 (1997) 93. [16] F. Stephan, et al., Phys. Rev. ST Accel. Beams 13 (2010) 020704. [17] B.E. Carlsten, Nucl. Instr. and Meth. A 285 (1989) 313. [18] B. van der Geer, et al., Presented at the Future Light Source Workshop, 2006. [19] R.P. Walker, Presented at the X-Band Structures, Beam Dynamics and Sources Workshop, 2010. [20] J.-H. Han, in: Proceedings of the Particle Accelerator Conference, 2009, p. 497. [21] R. Alley, et al., Nucl. Instr. and Meth. A 429 (1999) 324. [22] K. Floettmann, A Space Charge Tracking Algorithm ASTRA, /http://www. desy.de/ mpyflo/S. [23] J.H. Billen, L.M. Young, ‘‘Poisson Superfish’’, LAUR-96-1834. [24] L. Faillace, et al., in: Proceedings of the International Particle Accelerator Conference, 2010, p. 3688. [25] ANSYS, /http://www.ansys.comS. [26] D. Pritzkau, R.H. Siemann, Phys. Rev. ST Accel. Beams 5 (2002) 112002. [27] D. Pritzkau, SLAC-R-577, 2001. [28] CST Studio, /http://www.cst.comS. [29] D.H. Dowell, et al., Nucl. Instr. and Meth. A 622 (2010) 685. [30] B.F. Macdonald, et al., Vacuum 84 (2009) 283. [31] J.-H. Han, in: Proceedings of the International Linear Accelerator Conference 2010, p. 977. [32] C.P. Hauri, et al., Phys. Rev. Lett. 104 (2010) 234802. [33] J.-H. Han, K. Floettmann, W. Hartung, Phys. Rev. ST Accel. Beams 11 (2008) 013501. [34] S. Rimjaem, et al., in: Proceedings of the European Particle Accelerator Conference, 2008, p. 244.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research A journal homepage: www.elsevier.com/locate/nima
Design of a high repetition rate S-band photocathode gun Jang-Hui Han , Matthew Cox, Houcheng Huang, Shivaji Pande Diamond Light Source, Oxfordshire OX11 0DE, UK
a r t i c l e i n f o
abstract
Article history: Received 24 February 2011 Received in revised form 11 May 2011 Accepted 12 May 2011 Available online 25 May 2011
Photocathode RF guns have been developed in many laboratories for generating high quality electron beams for free-electron lasers based on linear accelerators. Such guns can generate electron beams with an exceptionally high peak current as well as a small transverse emittance. Their applications have been recently expanded for ultrafast electron diffraction, coherent terahertz radiation, and X-ray or g-ray radiation by Compton scattering. In this paper, we design an S-band normal-conducting gun with capabilities of high quality beam generation and high repetition rate operation. The RF design and thermal analysis of the gun cavity and coupler are introduced. Optimal position of the gun focusing solenoid for low emittance beam generation is found by performing particle tracking simulations. Then, the gun system is designed to be able to afford the optimal solenoid position. The cooling-water channel surrounding the gun cavity and coupler is designed and analyzed numerically. The pressure in the gun is simulated with a vacuum model containing the detailed inner structure of the gun. An injector for a free-electron laser application is designed by using this gun and the beam dynamics simulation is shown. A cold test with a prototype gun for confirmation of the RF design is reported. & 2011 Elsevier B.V. All rights reserved.
Keywords: Photoinjector RF gun High repetition rate Low emittance beam
1. Introduction Beam brightness from electron injectors has been dramatically increased with the development of photocathode RF guns for the last two decades. In many laboratories, such guns have been adopted as the injectors for free-electron lasers (FELs) based on linear accelerators (linac) [1–6], for ultrafast electron diffraction [7–11], for coherent terahertz radiation [12,13], and for X-ray or g-ray Compton scattering [14]. In such guns, an electron beam is generated at the cathode by using a drive-laser pulse and the beam is accelerated from rest to relativistic energy by the strong RF accelerating field. By optimizing the drive-laser pulse and accelerating the emitted beam quickly, high peak current and low transverse emittance can be achieved. When RF power is transmitted into a gun cavity, RF power dissipation takes place at the cavity surface. Then, the surface temperature rises at the location with strong surface current. Due to the local temperature rise, the temperature distribution of the cavity becomes non-uniform, cavity deformation takes place, and therefore the cavity may suffer from mechanical stress. Such a cavity deformation also affects the RF field distribution, which may result in changes of the resonant frequency and field balance. Therefore, the cavity temperature must be controlled with cooling-water in order to keep the cavity at the nominal temperature
Corresponding author. Tel.: þ 44 1235 778935.
E-mail address: [email protected] (J.-H. Han). 0168-9002/$ - see front matter & 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.nima.2011.05.032
and to keep the temperature distribution as uniform as possible. As the RF duty factor increases, the problem of the non-uniformity of temperature distribution becomes more severe. This limits the maximum achievable repetition rate and also impairs stable RF operation. An optimally designed cooling-water channel is therefore required. Another obstacle to increasing the repetition rate may come from the excessive heating at the RF coupling region. If the RF coupler is connected to the side of the cavity cell where the surface current induced by the RF field is strong, the coupling region can have high temperature and high surface stress due to the RF heating. However, if the coaxial RF coupler is instead connected to the gun front iris where RF power dissipation density is low, the coupling region can be relatively cool and does not limit the repetition rate. Furthermore, when a coaxial coupler is utilized, the entire outer tube of the cavity can be enclosed with axisymmetric cooling-water channels around the cavity cylinder. Therefore, the cooling capacity can be maximized and the cavity deformation caused by the RF heating also becomes axisymmetric. High order transverse RF modes will not exist because of the perfectly symmetric cavity inner surface as well. Such a type of RF coupler has been employed in the DESY L-band guns [15], which are operating for FLASH [4] and PITZ [16]. Even with the fast acceleration of a beam in an RF gun, the beam radially expands through the gun and the transverse emittance blows out due to the space charge force. This transverse increase can be compensated by using a focusing solenoid [17]. This process is realized by reconfiguring the beam distribution in the transverse
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
phase space by using the focusing solenoid. The solenoid should be placed around the gun for an efficient transverse emittance control as will be discussed in Section 2. If the RF coupler is connected to the side of the cavity cell, the optimal location for solenoid is occupied by the coupler and therefore the solenoid must be placed downstream of the gun. In this gun design, the solenoid can be placed at the optimum location thanks to the coaxial coupler. The resonant frequency was chosen to be 2.998 GHz, which is the European standard S-band frequency. Compared with the Eindhoven gun [18], which is another S-band gun with a coaxial RF coupler, the possibility of further cooling channel installation was considered and the coupler tube length was increased to allow solenoid installation around the gun. Compared with lower frequency band guns such as L-band guns, an S-band gun made from oxygen-free electronic (OFE) copper generates very low emittance and short pulse electron beams because of the high accelerating field at the cathode. Compared with higher frequency bands, such as C- or X-bands, an S-band gun allows enough space for the coolingwater circuit installation inside the iris and therefore temperature control of the gun against RF thermal heating can be properly performed. Their repetition rates are, however, generally limited to about 100 Hz [5]. The gun introduced here has a potential of a repetition rate as high as 1 kHz as will be discussed in Section 3. This gun will provide a possibility to operate an FEL based on normalconducting technology at such a high repetition rate [19]. The RF design, thermal analysis, and vacuum simulation of the new gun are reported in this paper. A prototype of the gun was produced for low power RF tests with a network analyzer. The test results with the prototype are also presented.
2. Beam dynamics consideration in the gun design The transverse beam emittance can be minimized by optimizing the gun cavity geometry and the location of the gun solenoid. The lengths of the gun cells are optimized based on the previous study in Ref. [20] and further beam dynamics simulations for this particular gun design have been carried out. The optimum location of the main solenoid is also found with beam dynamics simulation. The length of the first cell strongly affects the beam formation and acceleration when the beam is in a non-relativistic regime. When a beam is emitted from the cathode, the velocity of the beam is almost zero. After the emission, the beam is accelerated by the RF field to become relativistic through the first cell. When the beam energy is very small, the space charge effect changes the beam properties considerably. As the beam is accelerated, the space charge effect becomes weaker. The beam dynamics also depends on the length of the second cell. The relation between the cell lengths and beam dynamics has been studied in detail elsewhere [20]. After simulations with various sets of cell lengths as carried out in Ref. [20], the lengths of the first (half) and second (full) cells were chosen to be 28 and 50 mm, respectively. This optimization was carried out at 200 pC bunch charge, 8 ps laser pulse length and 120 MV/m peak field at the cathode. Even though the effect on the beam dynamics is relatively small compared to the initial beam shape defined by the drive-laser parameters or the focusing solenoid field configuration, the gun should be carefully designed for best beam performance. Otherwise, a new manufacturing and installation of a gun will be needed to improve the beam quality later with an improved gun design. This is practically difficult because it may interrupt the operation of the entire accelerator system for quite a long time. When an electron beam is accelerated in a photocathode RF gun, the strength of space charge effect is different for each
0.4
0.35 emittance (mm mrad)
18
0.3
0.25
0.2 0.05
0.1
0.15 0.2 solenoid position (m)
0.25
Fig. 1. Normalized transverse emittance (full rms) as a function of the distance between the center of the solenoid and the cathode. Simulations were carried out at 200 pC bunch charge and 100 MV/m peak RF field at the cathode. The drivelaser beam size and the beam launch phase at the cathode were optimized at each solenoid position.
temporal slice of a bunch. The head and tail slices experience weaker repulsive force to the radial direction, while the central slices experience stronger defocusing force. As a result, the slices show different shape and angle in transverse phase space. The area in the phase space, emittance, can be minimized under a certain condition combined with the solenoid field, the RF field, and the initial beam shape. As pointed out in Refs. [15,21], if the gun consists of two (1/2þ1) cells, the optimum location of the solenoid is around the cavity. Fig. 1 shows the beam emittance at the exit of the gun for a 200 pC beam as a function of main focusing solenoid position. At the cathode, the solenoid field is compensated to zero by a bucking solenoid sitting behind the gun. For the simulations using the ASTRA code [22], 100 MV/m gun RF peak field at the cathode and 8 ps drive-laser pulse length were used. A flat-top temporal laser profile with 1 ps rise/fall time was used. The radial size of the drive-laser pulse was optimized to obtain the lowest transverse emittance for each condition. Since these simulations were carried out without further linac sections, the projected beam emittance blows up after the minimum point. The minimum projected emittance takes place when the temporal slices of the beam are aligned in the transverse phase space, and the distance from the cathode to the minimum emittance point varies depending on the solenoid position. Here, we took the minimum projected emittance value for each simulation condition. The laser full radius was in a range between 0.28 and 0.46 mm. The optimum radius for a lowest emittance increased with the solenoid position from the cathode. The launch phase of the beam at the cathode was also optimized at each solenoid position. As shown in Fig. 1, lower emittance can be achieved as the solenoid moves toward the cathode. When the physical thickness of the solenoid, 0.135 m in the design here, is considered, the surrounding of the gun cavity should be cleared for solenoid installation in order to achieve the lowest emittance.
3. Gun cavity design In this section, we discuss the detailed gun design. An RF design is carried out, which satisfies the requirements discussed in Section 2. Given the RF field distribution, the thermal behavior is analyzed numerically.
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
quality factor power
14400
3
6.1
14200
2 1 0 1
2
3
4
5
6 7 z (cm)
8
9
10
11
Fig. 2. RF electric field distribution of the gun cavity. The cathode plug and the slot for the plug at the rear wall are shown. The RF contact between the cavity and the cathode is made with a Cu–Be spring. A part of the inner antenna of the RF coupler is shown at the right side. The emitting surface of the cathode is at z ¼ 5 mm.
quality factor
r (cm)
6.2
14600
6
14000
5.9
13800 5.8 13600 5.7
13400
5.6
13200
5.5
13000
5.4
12800 20 A slot for exchangeable cathode plugs was implemented at the rear wall of the gun (see Fig. 2). With the cathode slot, damaged cathodes can be easily replaced with a fresh one and various photo-emitters such as Cu, Mo, Cs–Te, or any other new material can be used for test and operation. When the front surface of the cathode is not perfectly aligned with the rear wall of the gun cavity, which may take place during cathode exchange, the RF characteristics may be influenced. By reducing the diameter of the cathode plug, this effect can be minimized. If the radial size of the cathode extends to the region where the RF magnetic field is strong, a small misalignment in the longitudinal direction induces a large RF resonant frequency shift and field balance change. Another possible side-effect of a large diameter cathode is thermal heating in the outer part of the cathode front surface caused by the magnetic field. The diameter of the cathode plug was chosen to be 8 mm. The plug size was also chosen considering a minimum impact on the beam dynamics due to field distortion at the cathode plug edge. If the edge of the cathode plug is too close to the center of the beam axis, the RF field distortion caused by the edge may affect the beam dynamics. The RF field distribution and power dissipation density at the gun cavity were calculated by using the SUPERFISH code [23]. Fig. 2 shows that the p mode is activated. The maximum field strength in the first cell was set to be higher than in the second cell by 4% for uniform RF power dissipation density and equal radial deformation in both cells. The details such as the cathode plug and the slot for the plug at the rear wall of the gun as well as the coupler antenna were included in the RF simulation. The quality factor of the gun cavity is 13,570 at 45 1C (see Fig. 3). The p mode is utilized for beam acceleration. Even if the 0 mode is separated from the p mode, the tail of the 0 mode peak may be still activated at the resonant frequency of the p mode, 2.998 GHz. When the 0 mode is activated, an electron beam may be affected by the unwanted 0-mode field in a wrong phase. If the 0 mode is separated far enough from the p mode, the unwanted influence can be reduced. In this design, the mode separation is 21 MHz, which is larger than those of the other S-band guns as reported in [5,18,24]. The temperature on the RF surface of the gun rises when an RF pulse is applied and falls when the RF pulse ends. The heat at the surface, caused by the RF pulse, is transferred to the surrounding Cu volume and is then transferred to the cooling-water by heat convection. The local surface temperature of the gun varies from location to location because the local RF heat dissipation density (caused by the magnetic field) is different and heat transfer to the body and cooling channels also varies. After a number of RF cycles, the gun temperature distribution reaches the steady state so that the local temperature when an RF pulse starts is the same as the temperature when a next pulse starts because the RF heat
power (MW)
4
19
30
40 50 60 temperature (°C)
70
80
Fig. 3. Quality factor of the gun cavity and RF power required for 100 MV/m peak field at the cathode at several operating temperatures. The temperature was assumed to be uniform over the cavity.
Fig. 4. Gun cooling-water channels. The channels enclose the cavity tube (left). In the irises and cavity rear wall, the cooling channels are shaped on half-circle discs (right) and brazed into the main body.
dissipation to Cu surface and the heat transfer to cooling-water are balanced. The cooling-water channels were designed for high cooling capacity, uniform temperature distribution over the gun body, and mechanical sustainability against high water pressure (see Fig. 4). Through the cooling channels of about 36 mm2 crosssectional area, cooling-water flows with a speed of 3 m/s. The minimum wall thickness between cooling channel and cavity inner surface is 3 mm. In order to achieve 100 MV/m peak field at the cathode when the gun operates at 45 1C (Fig. 3), 5.7 MW peak RF power should be provided from an RF power source to the gun. When this gun operates with 3 ms RF pulse length and 1 kHz repetition rate, the average dissipated power is 17 kW. Now, we study if this gun can operate with the average power under the given conditions. The RF power dissipation density at the RF surface was used as the input for the steady state temperature, deformation, and stress analyzes with ANSYS [25]. The real three-dimensional cooling-water channel shape was modeled. For these analyzes, 16 1C cooling-water temperature through the cooling channels was assumed. The result of the temperature analysis is shown in Fig. 5. The maximum temperature of the gun was found to be about 50 1C, which occurs at the iris between the cells. The average temperature
20
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
Fig. 5. Steady-state temperature distribution of the gun cavity simulated by using ANSYS.
Fig. 7. Steady-state von Mises stress on the RF surface of the gun cavity.
pulse heating can be found with the equation [27]: pffiffiffi 2PRF t DTs ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi prkC e where PRF is the dissipated RF power density and t is the pulse length. r ¼ 8:93 103 kg=m3 , k¼391 W/m K and Ce ¼ 385 J=kg K are the material density, the heat capacity and the specific heat for the OFE copper. In this gun design, a maximum peak dissipation of 225 MW/m2 occurs at the iris. For 100 MV/m and 3 ms, the temperature rise caused by an RF pulse is 12 1C, which is far below the damage threshold for cyclic stress, 40 1C [26].
4. RF coupler design
Fig. 6. Steady-state gun cavity deformation by using ANSYS.
over the inner surface of the cavity is about 45 1C. Even though there is no direct water-cooling applied to the cathode plug, the plug temperature is below 40 1C. The right hand side of the front iris, where the RF coupler is connected, is also below 40 1C. The cavity deformation (see Fig. 6) was analyzed with the temperature distribution in Fig. 5. The end of the left hand part was fixed for the analysis. The maximum (total) deformation was about 40 mm. However, the deformation is axially symmetric and therefore no dipole or higher order mode is configured from the RF heating. The radial deformations of both cells are almost the same and below 15 mm (not specifically shown in Fig. 6). Therefore, the RF field balance will be affected only by 0.001%. The yield stress of the annealed oxygen-free copper is known to be 62 MPa [26]. The steady state stress may not induce material damage when the gun operates continuously. However, the cavity temperature and deformation should still be controlled for stable RF operation and long cavity lifetime. The local stress caused by the RF heating was also studied by using ANSYS. The maximum stress occurring at the inner corner of the second cell is 50 MPa (Fig. 7). This is in the safe regime by about 20% margin. The stress at the cathode plug, which will be made from molybdenum, is below 17 MPa (not shown in Fig. 7). The yield stress of molybdenum is far higher than copper. During pulsed RF operation, a transient temperature rise takes place during the RF pulse and the temperature falls again when the pulse ends. As this process repeats, severe surface fatigue may be produced. The transient temperature rise, DTs , at surface by RF
In this section we describe the RF coupler. The coaxial coupler tube is designed to be long enough to place the main focusing solenoid around the second cell of the gun. The coupling coefficient is optimized to be 1. Thermal analysis is carried out with ANSYS too. RF power will be transmitted to the coupler through the WR284 rectangular waveguide. The door-knob transformer converts the TE10 mode in the waveguide to the transmission mode in the coaxial line between the door-knob and the cavity. The transmission mode is a hybrid of the TEM and TE11 modes. The gap between the inner antenna tip and the cavity front iris is crucial for deciding the coupling. There are reflections from the shorted end of the waveguide and the open end of the coupling antenna. Therefore, the location of the shorting plane is also important for having a proper match between the gun cavity and the input waveguide. The part of the reflected wave from the shorting plane, the part of the reflected wave from the antenna tip and the radiated wave from the cavity will add up to a negligible amplitude to have almost no reflected wave in the input waveguide for a properly matched cavity. For the calculation of the S-parameters, the CST Studio timedomain (TD) solver [28] was used. S11 was computed for different shorting plane locations for a fixed value of gap. Such sets were obtained for different gap values to arrive at the optimum location of the shorting plane and the coupling gap. The computed values of S11 are 10 and 33 dB for the 0 and p modes, respectively. Fig. 8 shows S11 computed by the CST Studio TD solver with RF power losses included. Fig. 9 shows the electric field (absolute) of the p mode computed by the CST Studio TD solver for 0.5 W power input through the coupling waveguide. The computed electric field at the cathode center is 30.4 kV/m. For 100 MV/m peak field at the
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
21
0 -5
S11 (dB)
-10 -15 -20 -25 -30 -35 2.975
2.98
2.985 2.99 frequency (GHz)
2.995
3
Fig. 8. S11 parameter as computed with the CST Studio time-domain solver. The 0 mode (left peak) is under-coupled and the p mode (right) is critically coupled.
RF power in
shorting plane Fig. 9. Electric field (absolute) of the p mode as computed with CST Studio for 0.5 W at the input port of the waveguide.
cathode center, the required power is 5.4 MW which is in good agreement with the SUPERFISH results mentioned in Section 3. In order to control the temperature of the coupler antenna, one turn of the cooling-water channel was accommodated downstream of the inner antenna (see Fig. 11). Both 20 1C coolingwater temperature and 2 m/s water flow rate were used for cooling the coupler. The RF heat dissipation at the RF surface of the coupler inner antenna was calculated with CST Studio and the result was used as input for the ANSYS analysis in the steady state condition. The analysis results are shown in Fig. 10. The maximum temperature was 61.5 1C at the end tip of the antenna. Thermal convection at the end tip relies on the thin, long antenna tube. Since the heat transfer through the antenna tube is not efficient, the temperature at the region is relatively hot. The deformation is about 60 mm at the end tip. However, the impact on the RF coupling to the gun cavity is negligible according to CST Studio simulation. The maximum von Mises stress is 1.6 MPa at the corner of the RF mode converting region. Even though the temperature at the tip is relatively high at the steady state (61.5 1C), transient temperature rise during an RF pulse at the coupler is very small because the heat dissipation density (caused by the magnetic field) is far less than 1% of that at the gun cavity. A maximum dissipation of 1.1 kW/m2 occurs at a point on the antenna surface. For 100 MV/m peak field at the cathode and 3 ms pulse length, the temperature rise caused by an RF pulse is only 6 10 5 1C at that point.
Fig. 10. Thermal analysis of the RF coupler at the steady state; temperature distribution at the steady state (top), deformation caused by the RF heating (middle), and von Mises stress on the RF surface (bottom) at the steady state. Thermal analysis was carried out by using ANSYS. On the outer surface of the inner antenna, RF power is dissipated during RF transmission to the gun cavity.
Since the heat dissipation density on the inner surface of the coupler outer tube is lower than that on the inner antenna outer surface and the outer tube is exposed to air and also connected to other copper parts, the thermal effect on the outer tube is expected to be even smaller than on the inner antenna.
5. Gun section design A gun section was designed as shown in Fig. 11. The main gun solenoid is placed around the gun cavity and RF coupler tube. With this gun cavity and coupler, the solenoid can be placed within a range of 0.07–0.12 m from the cathode. A smaller solenoid is placed just behind the gun in order to compensate the magnetic field at the cathode. Through the longer, open rectangular arm of the RF coupler, high power RF is forwarded to the gun cavity. An RF window will be connected to this port. The shorting plane at the shorter arm makes the coupling coefficient critically matched and is also used as a vacuum pumping port through the grille at the plane. The rear part of the gun cavity is to be connected to a cathode exchange chamber. Cathodes can be exchanged under ultrahigh vacuum by using a manipulator. Through the long inner hole of the inner antenna tube of the coupler, a drive-laser pulse is transmitted from a laser mirror
22
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
0.8
×10-9
0.7
total pressure (mbar)
0.6 0.5 0.4 0.3 0.2 0.1 0 0
Fig. 11. Gun section layout. It consists of an RF gun, an RF power coupler, and focusing solenoids.
0.05
0.1
0.15
0.2 z (m)
0.25
0.3
0.35
Fig. 13. Simulated total pressure on the symmetry axis of the gun. The cathode is at z¼ 0 m.
ultrahigh vacuum by a 100 l/s ion pump (see Figs. 11 and 12). Additional pumping will be provided with a second 100 l/s ion pump at a downstream chamber, where a drive-laser mirror will be installed. At the other end of the rectangular waveguide connected to the coupler, an RF window will be installed. Near the window, an additional pump will be installed; these are not shown in Fig. 12 and were omitted from the vacuum simulation. This pump will mainly absorb gases emitted from the RF window. The pressure distribution was simulated by using the MonteCarlo Test Particle software, MCTPVac [30], developed in-house. An out-gassing rate of 10 11 mbar l/s cm2 at the vacuum surface, which should be achieved after full RF conditioning, and a gas molecular mass of 28 (nitrogen equivalent) were assumed. The pump entrance ports were modeled as pumping surfaces with an appropriate sticking probability 0 o s o1. Diffuse scattering of molecules is assumed at all internal surfaces. Adequate statistics were obtained with 100,000 molecules. As expected, the highest calculated pressure is found in the first cell of the gun but stays below 8 1010 mbar (see Fig. 13).
7. FEL injector beam dynamics Fig. 12. Vacuum surface model. The detailed inner structure of the gun cavity and coupler, including the grille at the coupler shorting plane, was modeled.
placed downstream of the gun to the cathode. A beam generated at the cathode also propagates through the antenna hole.
6. Vacuum simulation Ultrahigh vacuum is required in the gun cavity in order to suppress RF breakdown and also to prevent deterioration of the quantum efficiency of the photocathode due to pollution. For keeping the cathode lifetime reasonably long, 4 1000 h, the vacuum must be better than 10 9 mbar for either a metal cathode or a Cs–Te cathode [29]. In the gun system, there is no pumping port connected directly to the gun body. Pumping should be achieved through the RF coupler and the downstream beam pipe. At one end of the coupler waveguide there is RF shielding grille to reflect the RF wave with parallel slots allowing pumping to
By using this gun, an injector was designed for high beam quality and 1 kHz repetition rate [31] with an assumption of being used for an FEL based on a normal-conducting linac [19]. This injector consists of the gun system and four 3 m traveling wave linac tubes. For 1 kHz operation, 100 MV/m gun peak field and 12.5 MV/m linac gradient were used. Beam dynamics simulations were carried out by using ASTRA with 100 000 macroparticles. The center of the main gun solenoid was placed 0.09 m from the cathode. A 200 pC beam is generated at the cathode by a drive-laser pulse with 8 ps full width at half maximum (FWHM) and 0.32 mm full radius. The temporal shape of a laser pulse was assumed to be flat top with 1 ps rise/fall time. The transverse distribution was assumed to be uniform. For a realistic estimation of thermal emittance, the kinetic energy of emitted electron was assumed to be 0.6 eV, which is higher than the experimentally known value [32]. As discussed in Section 2 transverse beam emittance may be improved if the solenoid is shifted toward the cathode. However,
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
180
1.8 emittance energy
160
1.4
140
1.2
120
1
100
0.8
80
0.6
60
0.4
40
0.2
20
beam energy (MeV)
1.6 emittance (mm mrad)
23
0
0 0
5
10 z (m)
15
20
Fig. 14. Normalized transverse emittance and beam energy evolution from the cathode (z¼ 0 m) to the end of the injector.
Table 1 Parameters of the injector simulation using ASTRA. Beam parameters are shown at the end of the injector. Bunch charge Laser pulse length (FWHM) Laser rise/fall time (10–90%) Laser full radius Kinetic E of emitted electrons Intrinsic (thermal) emittance Peak field at the cathode Beam launch phase at the gun Linac gradient Phase of the 1st linac tube Phase of the 2nd–4th linac tube Projected rms emittance (100%) Emittance of central slices Beam size Bunch length (FWHM) Peak current DE=E at centre Mean E
200 pC 8 ps 1 ps 0.32 mm 0.6 eV 0.142 mm mrad 100 MV/m 371 from 0-crossing 12.5 MV/m 601 from on-crest 01 from on-crest 0.172 mm mrad 0.16 mm mrad 0.12 mm 7.2 ps 30 A
Fig. 15. Simplified gun prototype for cold test with a network analyzer. The prototype includes the critical RF features.
1:14 105 140 MeV
the 0.09 m solenoid position was chosen to minimize possible multipacting processes at the cathode which may take place when the main solenoid is placed too close to the cathode [33]. The normalized transverse emittance and beam energy evolutions with distance from the cathode are shown in Fig. 14. The emittance (full rms) is about 0.17 mm mrad at the end of the injector, where the beam energy is 140 MeV. This low emittance was achieved by optimizing the gun cell length, the gun solenoid position, and the first accelerating section position [31]. The peak current is 30 A. The input parameters for the beam simulation and the results are summarized in Table. 1.
8. RF cold test with a prototype A simplified prototype was fabricated to test the RF characteristics with a network analyzer (Fig. 15). The prototype does not have cooling-water channels or vacuum connection. Low quality oxygen-free copper was used and the machining requirements for tolerance and surface finishing were relaxed to save the cost. However, the critical features for testing the RF characteristics, the cathode slot, the connecting part between the gun cavity and coupler, and the vacuum grille at the coupler were included. The machined parts were vacuum brazed for good RF contacts
Fig. 16. S11 measurement with a network analyzer.
between the parts. Since the low grade oxygen-free copper was not heat-treated prior to machining, the machined body could be slightly deformed during the brazing process. With the prototype, the resonance frequency, field balance, and quality factor of the cavity and the RF coupling were measured. The resonance frequencies were shifted to the direction of lower frequency by about 4 MHz (see Fig. 16). It might result from the low quality machining tolerance and also from the copper body deformation during the vacuum brazing. The measured quality factor was smaller than the design value by 11.5%. It might result from the low quality surface finishing. The RF field at the first cell was higher than at the second cell by 2%, which is fairly close to the design value, 4%. As shown in Fig. 10, the p mode was critically coupled. The voltage standing wave ratio (VSWR) was 1.02. The 0 mode was strongly under-coupled as
24
J.-H. Han et al. / Nuclear Instruments and Methods in Physics Research A 647 (2011) 17–24
Acknowledgments
Table 2 Design parameters and measured ones with the prototype. Parameters
Design
Prototype
bp
1 2998 13,875 1.04 20.8 0.520 11,870
0.99 2994 12,280 1.02 21.2 0.453 10,504
mode
fp mode Q p mode at 25 1C Field balance Df ðp20 modeÞ
b0 Q0
mode mode
at 25 1C
We thank D. Brice, N. Hammond, M. Jensen, J. Kay, A. Peach, and L. Zaja for their engineering support on the design and measurement. J.H.H. thanks K. Floettmann at DESY for valuable discussions and R.P. Walker for encouraging this study and comments on the manuscript. I.P.S. Martin kindly proof read the manuscript.
References expected from the calculation. The mode separation between the p and 0 modes was 21.2 MHz. The result of the cold test is summarized in Table 2. RF tuning was exercised by using the plastic deformation of the rear and front walls as carried out for the DESY gun [34]. Tuning up to 400 kHz with the rear wall and 100 kHz with the front wall were tested. In practice, the resonance frequency and the field balance will be checked with a network analyzer prior to the final brazing of the cavity parts for an operational gun. Therefore, only a small amount of tuning will be needed by plastic deformation.
9. Discussion We here designed an improved S-band gun for high repetition rate operation and a low transverse beam emittance. By adopting the coaxial RF coupler connected to the front iris of the gun, the water-cooling capacity could be increased and the focusing solenoid could be placed at the optimum location for emittance compensation. When the gun operates at 100 MV/m peak field at the cathode, 1 kHz repetition rate will be possible. The factor limiting the repetition rate is the thermal stress at the gun cavity. The coaxial coupler does not limit the repetition rate. This gun may be used for high brightness electron injector for FELs based on normal-conducting linacs. An S-band injector capable of 1 kHz operation was designed using the gun. A 200 pC beam will have a transverse emittance about 0.17 mm mrad and a peak current of 30 A. Other possible applications using this gun are ultra-short electron beam generation for diffraction experiments and terahertz or X/g-ray radiation. This high brightness and high repetition gun will be beneficial to such applications. The RF design of the gun and coupler was successfully tested with a prototype. An operational gun is under production for a high power (but low repetition rate) RF test with the existing RF system at Diamond Light Source. A high repetition rate RF station will, however, be needed to test the full capacity of the gun.
[1] P.G. O’Shea, et al., in: Proceedings of the Particle Accelerator Conference, 1991, p. 2754. [2] D.H. Dowell, et al., Appl. Phys. Lett. 63 (1993) 2035. [3] R. Dei-Cas, et al., Nucl. Instr. and Meth. A 296 (1990) 209. [4] S. Schreiber, in: Proceedings of the European Particle Accelerator Conference, 2000, p. 309. [5] R. Akre, et al., Phys. Rev. ST Accel. Beams 11 (2008) 030703. [6] H.S. Kang, S.H. Nam, in: Proceedings of the 32nd International Free Electron Laser Conference, 2010, p. 155. [7] J.B. Hasting, et al., Appl. Phys. Lett. 89 (2006) 184109. [8] R. Li, et al., Rev. Sci. Instrum. 80 (2009) 083303. [9] J. Yang, et al., Rad. Phys. Chem. 78 (2009) 1106. [10] P. Musumeci, et al., Rev. Sci. Instrum. 82 (2010) 013306. [11] J.-H. Han, Phys. Rev. ST Accel. Beams 14 (2011) 050101. [12] R. Kuroda, Nucl. Instr. and Meth. A 593 (2008) 91. [13] C. Yim, et al., in: Proceedings of the International Particle Accelerator Conference, 2010, p. 1059. [14] R. Kuroda, et al., Nucl. Instr. and Meth. A 637 (2011) S183. [15] B. Dwersteg, et al., Nucl. Instr. and Meth. A 393 (1997) 93. [16] F. Stephan, et al., Phys. Rev. ST Accel. Beams 13 (2010) 020704. [17] B.E. Carlsten, Nucl. Instr. and Meth. A 285 (1989) 313. [18] B. van der Geer, et al., Presented at the Future Light Source Workshop, 2006. [19] R.P. Walker, Presented at the X-Band Structures, Beam Dynamics and Sources Workshop, 2010. [20] J.-H. Han, in: Proceedings of the Particle Accelerator Conference, 2009, p. 497. [21] R. Alley, et al., Nucl. Instr. and Meth. A 429 (1999) 324. [22] K. Floettmann, A Space Charge Tracking Algorithm ASTRA, /http://www. desy.de/ mpyflo/S. [23] J.H. Billen, L.M. Young, ‘‘Poisson Superfish’’, LAUR-96-1834. [24] L. Faillace, et al., in: Proceedings of the International Particle Accelerator Conference, 2010, p. 3688. [25] ANSYS, /http://www.ansys.comS. [26] D. Pritzkau, R.H. Siemann, Phys. Rev. ST Accel. Beams 5 (2002) 112002. [27] D. Pritzkau, SLAC-R-577, 2001. [28] CST Studio, /http://www.cst.comS. [29] D.H. Dowell, et al., Nucl. Instr. and Meth. A 622 (2010) 685. [30] B.F. Macdonald, et al., Vacuum 84 (2009) 283. [31] J.-H. Han, in: Proceedings of the International Linear Accelerator Conference 2010, p. 977. [32] C.P. Hauri, et al., Phys. Rev. Lett. 104 (2010) 234802. [33] J.-H. Han, K. Floettmann, W. Hartung, Phys. Rev. ST Accel. Beams 11 (2008) 013501. [34] S. Rimjaem, et al., in: Proceedings of the European Particle Accelerator Conference, 2008, p. 244.