- Email: [email protected]
Design and test of a superconducting RFQ for heavy ions
Nuclear Instruments and Methods in Physics Research B79 (1993) 711-713 North-Holland
Design and test of a superconducting
NOMB
Beam Interactions with Materials 8 Atoms *
RFQ for heavy ions *
I. Ben-Zvi ‘, A. Jain *, J.W. No&, P. Paul and H. Wang * Physics Department, SUNY at Stony Brook, NY 11794-3800,
USA
A. Lombardi INFN-LNL,
via Romea 4, Legnaro (PO) I-35020, Italy
A prototype SRFQ resonator optimized for a velocity of p = 0.03 and a charge to mass ratio (q/M) of l/6 has been designed, fabricated and tested. This is a short RFQ structure (only 2PA long) operating at 57.4 MHz. The resonator is made of copper, and a 1.5 pm thick layer of electroplated Pb-2 at.% Sn is used as the superconductor. The electrical characteristics of the resonator derived from detailed 3D numerical simulations using the MAFIA codes, and also from measurements on the actual resonator are described. The multipactoring levels could be processed away with ease. Quality factor (Q) > 1 X lo8 is obtained for E, up to 1 MV/m. Other parameters relevant to superconducting operation are also presented.
1. Introduction
The radio frequency quadrupole (RFQ) combines acceleration and focussing in one element, and is therefore very suitable for acceleration of low velocity ions. A normal conducting RFQ resonator, however, requires a high rf power. This makes a cw operation quite difficult. A superconducting RFQ (SRFQ) resonator could combine the advantages of the RFQ for slow ions with the cw operation and efficiency of superconducting linac structures. The design principles and possible application of the SRFQ as a linac injector were explored recently [l]. This paper briefly describes the design principles, and results of numerical and experimental studies of the characteristics of a prototype SRFQ resonator.
2. Resonator design The basic guidelines for our design were the use of the simple, low cost technology of lead (or lead-tin) plated on copper, the acceleration of very heavy ions (such as lead) with a charge to mass ratio (q/M) of l/6 or better and the concept of the RFQlet, a short
* Work Supported in part by NSF Grant no. PHY-8902923. * Also NSLS Department, Brookhaven National Laboratory, Upton, NY 11973, USA. * Now at Brookhaven National Laboratory, Accelerator Development Department, Building 902B, Upton, NY 11973, USA. 0168-583X/93/$06.00
RFQ resonator. Unlike a conventional RFQ, a RFQlet is designed to accept a prebunched beam. It combines the advantages of RFQ focussing with a wide transit time factor curve, low stored energy, small size and flexibility of the independently phased resonator approach. The beam dynamics aspects of the RFQlet design have been discussed in detail in ref. [l]. In a SRFQ design, one has to also consider the cryogenic requirements and electromechanical stability in conjunction with the transverse and longitudinal stability of the beam. In particular, the peak surface electric fields that can be supported by mass-produced leadplated resonators is rather low, say - 16 MV/m. This implies a weak transverse focussing and a need to optimize the accelerating field for a given peak surface electric field. Such an optimization leads to a linear dependence [l] of the minimum aperture (a) on /3( = u/c) for a given modulation factor (m) and the frequency (f). The nominal values of these parameters for the prototype resonator are a = 1.57 cm, m = 4, /I = 0.03 and f = 50 MHz. For a low frequency of 50 MHz, we have chosen the four rods RFQ structure [2]. In order to accommodate a large modulation (m = 4), the electrodes in the present design have a “vane-like” transverse section. This design also reduces the high order multipole components. The length of each electrode is 39.7 cm, giving a 2@h long (four accelerating gaps) structure. Fig. 1 shows a schematic view of the resonator. The electrodes are hollow in order to reduce weight and to provide a channel for the liquid helium. Each pair of opposing electrodes is connected to a hollow sphere using two curved tubes about 4 cm in diameter. The 12
0 1993 - Elsevier Science Publishers B.V. All rights reserved
XII. ACCELERATOR
TECHNOLOGY
I. Ben-.&i et al. / Superconducting RFQ for heavy iom
712
cm diameter spheres are then connected to the top plate via 8 cm diameter straight tubes. These tubes act as supports for the electrodes as well as liquid helium channels. The external tank, in the shape of a can open on its bottom, is cooled by conduction. The resonator is connected to the liquid helium tank of the cryostat via a shallow helium manifold on top of the resonator. The rf ports for the input coupler and pickup are located on the top, next to the helium manifold. Further details on resonator fabrication can be found in ref. [3].
3. Electrical ch~c~~stics DISTANCE
The electrical characteristics of the prototype resonator have been analyzed in detail using both an approximate analytical (lumped circuit) model [4], and elaborate numerical simulations [5] using the 3D MAFIA codes. In the 3D calculations, the complete resonator geometry (e.g. the vane modulations) has been included using nearly 350000 mesh points with an average mesh density of 1 mesh point/cm3. Parameters such as the capacitance, stored energy, geometric factor, and the peak electric and magnetic fields were calculated from the field distributions. The calculated parameters have been found to be in reasonable agreement with measured values in the actual prototype resonator [S1. In particular, the resonant frequency was calculated to be 56.5 MHz as against a measured value of 57.4 MHz.
,
Helium Port
RF Port
ALGNG AXIS
The electrical field profiles on and off the beam axis have been measured at Grumman Corporation using the bead pull technique. In order to obtain sufficient signal, it was necessary to use a metallic bead instead of a dielectric one. The measurements are therefore sensitive to both the electric and the magnetic fields. Fig. 2 shows the bead perturbation signal measured along the beam axis (solid line). The signal estimated from the numerically calculated fields is shown by the crosses. The good agreement with the measurements is evident. The peaks on both ends are due to the fringe fields, which make a non-negligible contribution to the total acceleration. The progressive decrease in the peak fields of the cells is a result of increase in cell length as the beam accelerates along the structure. The bead perturbation shows a noticeable magnetic field at about 40.1 m from the center (see fig. 2). This is perhaps due to the proximity to the inductive support tubes. The transit time curve calculated from the measured field profile peaks at @ = 0.033 and acceleration in excess of 80% of the peak can be obtained for 0.029 s p I 0.040. The stored energy was calculated as 0.965 J at 1 MV/m acceleration field (E,), defined as energy gain per unit charge divided by the total external length (58.5 cm) of the resonator.
4. Pb plating and superconducting
-
6eam Port
Fig. 1. A schematic view of the prototype superconducting RFQ resonator.
(m)
Fig. 2. The measured and the calculated bead perturbations along the axis of the SRFQ resonator.
tests
A 1.5 pm thick layer of electroplated Pb-2 at.% Sn is used as the superconductor. The resonator can, with an extension collar, was used as the vessel containing the electrolyte. Two L-shaped Pb sheets, and two 1.5 in. diameter hollow Pb pipes were used as anodes. A low current density of - 0.25 mA/cm’ was used to avoid any whiskers. Due to the large surface area involved ( - 193 00 cm*), the entire operation was carried out in an inert atmosphere of dry nitrogen.
I. Ben-Zvi et al. / Superconducting RFQ for heavy ions
In the room temperature rf tests after the plating, several very low field multipactoring levels were seen. The lowest, and the most stubborn level was at E, = 0.94 kV/m, and corresponded to an inter-electrode gap of 1.5 cm, consistent with the minimum aperture. All the levels could be cleaned with rf processing (maximum of 1.5 kW peak/l50 W average rf power) in about 48 h. Some higher levels (E, = 60 and 80 kV/m) showed up at LHe temperature, but could be quickly processed away. Multipactoring therefore is not a serious problem even with the complex geometry of this resonator. This is a very significant result from the point of view of practical operation of a SRFQ. Mechanical stability of the resonator is an important requirement for superconducting operation. The resonant frequency excursions due to vibrations was measured to be _ 60 Hz (peak-to-peak) with considerable disturbance coming from a rather noisy mechanical pump. The frequency excursions reduced to - 20 Hz with the pump isolated. These stability figures are reasonable, though not as good as the best linac structures. In principle, the structure could be improved by dampening the mechanical Q. The resonant frequency variation due to the LHe pressure change is 160 Hz/psi. The Q as a function of E, was measured in a standard fashion using a phase locked loop and critical coupling. The resulting curve is shown in fig. 3. The low field Q is 1.5 X 10’ and remains above 1 X lo8 up to E, = 1 MV/m, achieved with only 3 W of rf power. It begins to drop sharply after this (circles in fig. 3). Considerable improvement was made by helium conditioning (squares in fig. 3). Processing with a high power rf pulse led to a further improvement at E, - 1.3 MV/m (triangles in fig. 31.. However, the Q still dropped sharply after 1.2 MV/m, making it impossible to exceed 1.3 MV/m, even in a pulsed mode. The limitation was a slow (ms) breakdown to lower field
713
levels which did not respond to standard helium gas conditioning. Temperature sensors located around the resonator did not show a temperature rise of more than 5 0.2 K. This barrier is most likely due to a local imperfection in the copper material or a poorly brazed joint.
5. Conclusions It has been established that it is feasible to build and operate a practical SRFQ device. The present limitations on achievable field level do not appear to be fundamental. Several multipactoring levels were exhibited, but were easily processed away. It should be noted that this device already provides significant acceleration (600 keV/charge at 1 MV/m) with a dissipation of only 3 W. This SRFQ can be used as a linac structure for heavy ions produced by an ECR source, replacing electrostatic accelerators, or as a compact, variable energy ion implanter. In both applications cw operation offers significant advantages. For a research accelerator the high duty factor is often essential for coincidence experiments; for high current requirements cw operation allows larger average currents for a given space charge limit. However, any practical application would have to consider the additional questions of phase locking, effect of fringe fields, etc. A test of the resonator with a beam is being planned to answer some of these questions.
Acknowledgements
We wish to thank the staff of the Physics Department machine shop, J. Rico and H. Uto for their many contributions to the resonator fabrication, plating and testing effort. We also thank Jim Rose for bead-pull measurements and H. Kirk and Rosemary Wallace for helping with the MAFIA calculations.
References
Ben-Zvi, A. Lombardi and P. Paul, Part. Accel. 35 (1991) 177. [2] A. Schempp, M. Ferch and H. Klein, Proc. 1987 Particle Accelerator Conf. IEEE Conf. Record 87CH2387 (1987) 267. [3] A. Jain, I. Ben-Zvi, P. Paul, H. Wang and A. Lombardi, Proc. 1991 Particle Accelerator Conf. IEEE Chf. Record
[l] I.
~lo7~,,,y~~FT~,,,~ lo"0
0.5
1
1.5
2
ACCELERATING FIELD E, @‘J/m)
Fig. 3. Q vs E. curve of the prototype SRFQ resonator.
91CH3038-7 (1991) 2444. [4] I. Ben-Zvi, A. Jain, H. Wang and A. Lombardi, Proc. 1990 Linac Conf. 1990, Albuquerque, New Mexico, p. 73. [5] H. Wang, I. Ben-Zvi, A. Jain, P. Paul and A. Lombardi, in ref. [3], p. 3038.
XII. ACCELERATOR
TECHNOLOGY
Design and test of a superconducting
NOMB
Beam Interactions with Materials 8 Atoms *
RFQ for heavy ions *
I. Ben-Zvi ‘, A. Jain *, J.W. No&, P. Paul and H. Wang * Physics Department, SUNY at Stony Brook, NY 11794-3800,
USA
A. Lombardi INFN-LNL,
via Romea 4, Legnaro (PO) I-35020, Italy
A prototype SRFQ resonator optimized for a velocity of p = 0.03 and a charge to mass ratio (q/M) of l/6 has been designed, fabricated and tested. This is a short RFQ structure (only 2PA long) operating at 57.4 MHz. The resonator is made of copper, and a 1.5 pm thick layer of electroplated Pb-2 at.% Sn is used as the superconductor. The electrical characteristics of the resonator derived from detailed 3D numerical simulations using the MAFIA codes, and also from measurements on the actual resonator are described. The multipactoring levels could be processed away with ease. Quality factor (Q) > 1 X lo8 is obtained for E, up to 1 MV/m. Other parameters relevant to superconducting operation are also presented.
1. Introduction
The radio frequency quadrupole (RFQ) combines acceleration and focussing in one element, and is therefore very suitable for acceleration of low velocity ions. A normal conducting RFQ resonator, however, requires a high rf power. This makes a cw operation quite difficult. A superconducting RFQ (SRFQ) resonator could combine the advantages of the RFQ for slow ions with the cw operation and efficiency of superconducting linac structures. The design principles and possible application of the SRFQ as a linac injector were explored recently [l]. This paper briefly describes the design principles, and results of numerical and experimental studies of the characteristics of a prototype SRFQ resonator.
2. Resonator design The basic guidelines for our design were the use of the simple, low cost technology of lead (or lead-tin) plated on copper, the acceleration of very heavy ions (such as lead) with a charge to mass ratio (q/M) of l/6 or better and the concept of the RFQlet, a short
* Work Supported in part by NSF Grant no. PHY-8902923. * Also NSLS Department, Brookhaven National Laboratory, Upton, NY 11973, USA. * Now at Brookhaven National Laboratory, Accelerator Development Department, Building 902B, Upton, NY 11973, USA. 0168-583X/93/$06.00
RFQ resonator. Unlike a conventional RFQ, a RFQlet is designed to accept a prebunched beam. It combines the advantages of RFQ focussing with a wide transit time factor curve, low stored energy, small size and flexibility of the independently phased resonator approach. The beam dynamics aspects of the RFQlet design have been discussed in detail in ref. [l]. In a SRFQ design, one has to also consider the cryogenic requirements and electromechanical stability in conjunction with the transverse and longitudinal stability of the beam. In particular, the peak surface electric fields that can be supported by mass-produced leadplated resonators is rather low, say - 16 MV/m. This implies a weak transverse focussing and a need to optimize the accelerating field for a given peak surface electric field. Such an optimization leads to a linear dependence [l] of the minimum aperture (a) on /3( = u/c) for a given modulation factor (m) and the frequency (f). The nominal values of these parameters for the prototype resonator are a = 1.57 cm, m = 4, /I = 0.03 and f = 50 MHz. For a low frequency of 50 MHz, we have chosen the four rods RFQ structure [2]. In order to accommodate a large modulation (m = 4), the electrodes in the present design have a “vane-like” transverse section. This design also reduces the high order multipole components. The length of each electrode is 39.7 cm, giving a 2@h long (four accelerating gaps) structure. Fig. 1 shows a schematic view of the resonator. The electrodes are hollow in order to reduce weight and to provide a channel for the liquid helium. Each pair of opposing electrodes is connected to a hollow sphere using two curved tubes about 4 cm in diameter. The 12
0 1993 - Elsevier Science Publishers B.V. All rights reserved
XII. ACCELERATOR
TECHNOLOGY
I. Ben-.&i et al. / Superconducting RFQ for heavy iom
712
cm diameter spheres are then connected to the top plate via 8 cm diameter straight tubes. These tubes act as supports for the electrodes as well as liquid helium channels. The external tank, in the shape of a can open on its bottom, is cooled by conduction. The resonator is connected to the liquid helium tank of the cryostat via a shallow helium manifold on top of the resonator. The rf ports for the input coupler and pickup are located on the top, next to the helium manifold. Further details on resonator fabrication can be found in ref. [3].
3. Electrical ch~c~~stics DISTANCE
The electrical characteristics of the prototype resonator have been analyzed in detail using both an approximate analytical (lumped circuit) model [4], and elaborate numerical simulations [5] using the 3D MAFIA codes. In the 3D calculations, the complete resonator geometry (e.g. the vane modulations) has been included using nearly 350000 mesh points with an average mesh density of 1 mesh point/cm3. Parameters such as the capacitance, stored energy, geometric factor, and the peak electric and magnetic fields were calculated from the field distributions. The calculated parameters have been found to be in reasonable agreement with measured values in the actual prototype resonator [S1. In particular, the resonant frequency was calculated to be 56.5 MHz as against a measured value of 57.4 MHz.
,
Helium Port
RF Port
ALGNG AXIS
The electrical field profiles on and off the beam axis have been measured at Grumman Corporation using the bead pull technique. In order to obtain sufficient signal, it was necessary to use a metallic bead instead of a dielectric one. The measurements are therefore sensitive to both the electric and the magnetic fields. Fig. 2 shows the bead perturbation signal measured along the beam axis (solid line). The signal estimated from the numerically calculated fields is shown by the crosses. The good agreement with the measurements is evident. The peaks on both ends are due to the fringe fields, which make a non-negligible contribution to the total acceleration. The progressive decrease in the peak fields of the cells is a result of increase in cell length as the beam accelerates along the structure. The bead perturbation shows a noticeable magnetic field at about 40.1 m from the center (see fig. 2). This is perhaps due to the proximity to the inductive support tubes. The transit time curve calculated from the measured field profile peaks at @ = 0.033 and acceleration in excess of 80% of the peak can be obtained for 0.029 s p I 0.040. The stored energy was calculated as 0.965 J at 1 MV/m acceleration field (E,), defined as energy gain per unit charge divided by the total external length (58.5 cm) of the resonator.
4. Pb plating and superconducting
-
6eam Port
Fig. 1. A schematic view of the prototype superconducting RFQ resonator.
(m)
Fig. 2. The measured and the calculated bead perturbations along the axis of the SRFQ resonator.
tests
A 1.5 pm thick layer of electroplated Pb-2 at.% Sn is used as the superconductor. The resonator can, with an extension collar, was used as the vessel containing the electrolyte. Two L-shaped Pb sheets, and two 1.5 in. diameter hollow Pb pipes were used as anodes. A low current density of - 0.25 mA/cm’ was used to avoid any whiskers. Due to the large surface area involved ( - 193 00 cm*), the entire operation was carried out in an inert atmosphere of dry nitrogen.
I. Ben-Zvi et al. / Superconducting RFQ for heavy ions
In the room temperature rf tests after the plating, several very low field multipactoring levels were seen. The lowest, and the most stubborn level was at E, = 0.94 kV/m, and corresponded to an inter-electrode gap of 1.5 cm, consistent with the minimum aperture. All the levels could be cleaned with rf processing (maximum of 1.5 kW peak/l50 W average rf power) in about 48 h. Some higher levels (E, = 60 and 80 kV/m) showed up at LHe temperature, but could be quickly processed away. Multipactoring therefore is not a serious problem even with the complex geometry of this resonator. This is a very significant result from the point of view of practical operation of a SRFQ. Mechanical stability of the resonator is an important requirement for superconducting operation. The resonant frequency excursions due to vibrations was measured to be _ 60 Hz (peak-to-peak) with considerable disturbance coming from a rather noisy mechanical pump. The frequency excursions reduced to - 20 Hz with the pump isolated. These stability figures are reasonable, though not as good as the best linac structures. In principle, the structure could be improved by dampening the mechanical Q. The resonant frequency variation due to the LHe pressure change is 160 Hz/psi. The Q as a function of E, was measured in a standard fashion using a phase locked loop and critical coupling. The resulting curve is shown in fig. 3. The low field Q is 1.5 X 10’ and remains above 1 X lo8 up to E, = 1 MV/m, achieved with only 3 W of rf power. It begins to drop sharply after this (circles in fig. 3). Considerable improvement was made by helium conditioning (squares in fig. 3). Processing with a high power rf pulse led to a further improvement at E, - 1.3 MV/m (triangles in fig. 31.. However, the Q still dropped sharply after 1.2 MV/m, making it impossible to exceed 1.3 MV/m, even in a pulsed mode. The limitation was a slow (ms) breakdown to lower field
713
levels which did not respond to standard helium gas conditioning. Temperature sensors located around the resonator did not show a temperature rise of more than 5 0.2 K. This barrier is most likely due to a local imperfection in the copper material or a poorly brazed joint.
5. Conclusions It has been established that it is feasible to build and operate a practical SRFQ device. The present limitations on achievable field level do not appear to be fundamental. Several multipactoring levels were exhibited, but were easily processed away. It should be noted that this device already provides significant acceleration (600 keV/charge at 1 MV/m) with a dissipation of only 3 W. This SRFQ can be used as a linac structure for heavy ions produced by an ECR source, replacing electrostatic accelerators, or as a compact, variable energy ion implanter. In both applications cw operation offers significant advantages. For a research accelerator the high duty factor is often essential for coincidence experiments; for high current requirements cw operation allows larger average currents for a given space charge limit. However, any practical application would have to consider the additional questions of phase locking, effect of fringe fields, etc. A test of the resonator with a beam is being planned to answer some of these questions.
Acknowledgements
We wish to thank the staff of the Physics Department machine shop, J. Rico and H. Uto for their many contributions to the resonator fabrication, plating and testing effort. We also thank Jim Rose for bead-pull measurements and H. Kirk and Rosemary Wallace for helping with the MAFIA calculations.
References
Ben-Zvi, A. Lombardi and P. Paul, Part. Accel. 35 (1991) 177. [2] A. Schempp, M. Ferch and H. Klein, Proc. 1987 Particle Accelerator Conf. IEEE Conf. Record 87CH2387 (1987) 267. [3] A. Jain, I. Ben-Zvi, P. Paul, H. Wang and A. Lombardi, Proc. 1991 Particle Accelerator Conf. IEEE Chf. Record
[l] I.
~lo7~,,,y~~FT~,,,~ lo"0
0.5
1
1.5
2
ACCELERATING FIELD E, @‘J/m)
Fig. 3. Q vs E. curve of the prototype SRFQ resonator.
91CH3038-7 (1991) 2444. [4] I. Ben-Zvi, A. Jain, H. Wang and A. Lombardi, Proc. 1990 Linac Conf. 1990, Albuquerque, New Mexico, p. 73. [5] H. Wang, I. Ben-Zvi, A. Jain, P. Paul and A. Lombardi, in ref. [3], p. 3038.
XII. ACCELERATOR
TECHNOLOGY