- Email: [email protected]
Proposal of an electrostatic multi-stage accelerator for high-current pulsed ion beam
NUCLEAR INSTRUMENTS 8 METHODS IN PHVSICS RESEARCH ELSEWIER
Nuclear Instruments and Methods in Physics Research
A 399 (1997)
1~4
Section A
Proposal of an electrostatic multi-stage accelerator for high-current pulsed ion beam Katsumi Department
Masugata”
ofElectrical Engineering, Nagaoka Uniwrsity of’ Technolo~. Nagaoka. Niigatu Y40 2 I Jtrpm Received
14 April 1997: received in revised form IO June 1997
Abstract
A new type of multi-stage accelerator is proposed to produce high-current-pulsed ion beam. In the accelerator, a pulsed high-current ion beam is accelerated without using induction cells. The accelerator is simple and a higher acceleration gradient is realized compared with an induction linac. The performance is demonstrated through the example of a ten-stage accelerator designed to accelerate a 50 MeV 100 kA, 50 ns beam of C+ ions to 60 MeV. The operation of the accelerator circuit was simulated by a computer code, which shows that acceleration pulses were successfully applied to each gap. P,‘lCS: 19.17. + w: 41.75.Lx Inertial confinement multi-stage accelerator
Keywrds:
fusion; Intense pulsed ion beam: Induction
In recent years, an intense pulsed ion beam (PIB) has been actively studied as an energy driver of inertia1 confinement fusion. Linear induction linacs have been considered to be a hopeful source of PIB since multi-stage acceleration of high-current PIB is possible [l-7]. In the induction linacs, however, induction cores of ferromagnetic materials are utilized to sustain the acceleration voltages, which makes the accelerator expensive and the structure complicated [2]. In addition, the average gradient
*Tel.: + 8125847 9894; [email protected].
fax:
+ 8125847
9500;
e-mail:
0168.9002/97/$17.00 1~:’ 1997 Elsevier Science B.V. All rights reserved PII SOl68-9002(97)00912-I
linac; Pulse power technology:
Electrostatic
of the acceleration in the induction linacs is limited by the size of the ferromagnetic cores. Even for a short pulse of 50 ns duration, the obtainable gradient is only 1 MV/m, whereas the maximum stress at the vacuum interface in the acceleration tube is more than 10 MV/m [S]. In this article, a new type of high-current, electrostatic, multistage accelerator is proposed. The accelerator is simple and is expected to have a higher acceleration gradient. Fig. l(a) shows the concept of the proposed accelerator (as an example, a ten-stage system is shown). In the accelerator, the ion beam is injected from the left side. Here we assume that the length of the beam (I,,) is much shorter than the accelerator length (I,). By us&g the velocity (u,,) and the
2
lnstr. and Meth. in Phys. Res. A 399 (1997) l-4
K. Masuguta/Nucl.
Acceleratrion gap
II
II
II II
IIJ II II
II II
II
Ii II
Cl II
II II
v,
v,, v,, v,
v,
V,IO’
II
Ii
II
J----f!-yyFf~~~[“l:
’ v,, @) I-
v,
v,, v,
r
Fig. 1. (a) The concept of the accelerator proposed (a ten-stage system is shown as an example). (b) Example of the gap voltage waveforms.
To satisfy Eq. (2), a voltage of - (+)I’, (the negative voltage is called “compensating voltage”) is applied to the gaps where beam is absent. The waveforms of the gap voltages are shown in Fig. l(b). As seen in the figure, the acceleration voltage is compensated by the “compensating voltage” and multi-stage acceleration is realized. This is in complete contrast to the induction linacs where acceleration voltage is compensated by an inductively produced field. In other words, in the proposed accelerator, the application of the acceleration voltages becomes possible due to a timeof-flight effect of the ion beam. Fig. 2 shows an example of the accelerator design to demonstrate the operation. The accelerator is designed to accelerate a 50 MeV 100 kA, 50 ns beam of C+ ions to 60 MeV in ten acceleration gaps. The accelerator consists of an acceleration tube and pulse transmission lines which are put in a deionized water pool. The acceleration tube consists of acceleration gaps, drift tubes and water-vacuum interfaces. The gap intervals (Ii, 12, . . . , 19) are adjusted to be 25 ns, being equal to the time-of-flight delay of the beam (a half of the beam duration). Since the velocity of the beam, and hence, the length of the beam increases as the beam
duration (r) of the beam, Ib is described as 4 50 MeV C’
Ib = Ubx t.
(1)
For example, Ib for a Cf beam of 50 MeV with duration of 50 ns is only 142 cm and the condition I, >>Ib is realized in a 10 m accelerator. In the figure, Ii, is supposed to be two times the interval of the acceleration gap (li). In the accelerator, points A and B in Fig. l(a) are kept at ground potential. Consequently, the sum of the voltages applied to the gaps (I’,,, I/gZ, . . . , V,,,) is always kept zero and the following relation is satisfied: v,, + v,, + v,, + .‘. + l/,10 = 0.
(2)
An acceleration voltage (V,) is applied to the gaps where the beam is just passing (in the figure, the beam is passing g, and g, at time tl) and V,, = V,, = V, (at ti).
(3)
/ Pulse Transmission Line
Fig. 2. The example operation.
of the accelerator
‘460MeVC’
design to demonstrate
the
3
K. MasugaialNucl. Insir. and Meth. in Phj,s. Res. A 399 (1997) i-4
is accelerated, the intervals are elongated downstream (e.g., l1 and l9 are designed to be 71 and 77 cm, respectively). The structure of the acceleration gaps is not designed in detail. However, the gap structure is flexible and many types of the structure including a magnetically insulated gap [4-61 or the multibeam accelerator concept [2,3] will be acceptable. One of the subject for developing the high-currentpulsed ion beam linac is to obtain a stable propagation of the beam, which is not considered in this article. However, the enhanced acceleration gradient will reduce the problem of propagation. For the acceleration gaps, pulse transmission lines are connected to apply the acceleration voltages. The lines have a characteristic impedance of 10 R and an electrical length of 150 ns (which corresponds to a physical length of 5 m for a waterfilled line). For each line, the acceleration pulse is applied with an interval of 25 ns. The electrical length of 1.50 ns is important to isolate the gap voltage as shown below. Fig. 3 shows the simulation model of the accelerator. Here, L,,(n = 1 - 10) are the parasitic lines produced between the outer conductors of the adjacent transmission lines. The characteristic impedance of each line is assumed to be 30 Q. Since the ion beam stays in the accelerator for 300 ns, the go and back delay time of the pulse in the lines (Zq,, where TV is the electrical length of the lines) is designed to be larger than 300 ns to avoid the effect of reflection. Parasitic lines are also produced between the outer conductor and grounded plate, however, the effect was ignored for simplicity (the effect is expected to be reduced by using a deepwater pool). The impedance of the acceleration gaps is considered to be high when the beam is absent and low when the beam is present. In the simulation, the gap impedance is assumed to be 100 and 10 fz when the beam is absent and present, respectively, and the transient times of the change in the impedance are assumed to be 1 ns. The input pulses of rise and fall time of 5 ns and the duration of 50 ns are applied to each pulse transmission line. Each pulse is applied with an interval of 25 ns to be synchronized with the arrival of the beam in each gap. The height of the pulse is
Fig. 3. The simulation
model of the accelerator.
determined to be 1.3 MV to obtain gap voltages of w 1 MV. Compensating voltages are not applied since they are automatically produced when the acceleration pulses are applied. Fig 4 shows the typical waveforms of the gap voltages (V,) obtained by the simulation with an input pulse for L,. As seen in Fig. 4, V,, rises at t = 150 ns and has a peak of z 1 MV. With the rise voltages are produced in the of F,,, compensating other gaps ( Vg,-l,). Since the parasitic lines and C, divide the applied voltages, the waveforms of the compensating voltages are observed to be almost the same. At r = 360 ns, a negative pulse is observed in I’,,, which is due to the reflection of the acceleration voltage in L,i. The potential at point X in Fig. 3 against the ground potential has a peak value of 1.6 MV and is not high compared to the total acceleration voltage of 10 MV. Therefore, the insulation between the
K. MasugafalNucl. Instr. and Meth. in Phys. Res. A 399 (1997) 1-4
4 1.4 1.2 1 0.8 s 5 & 3 s
0.6 04 0.2 0 -0.2 -0.4 -0.6 0
50
100
150
200
250
300
350
400
In conclusion, a new type of multi-stage accelerator is proposed to produce high-current pulsed ion beam. Compared to the conventional induction linacs, the proposed accelerator is of great advantage since it is simpler and has a higher acceleration gradient. As an example, a ten-stage accelerator, which is designed to accelerate a 50 MeV, 100 kA, 50 ns beam of C+ ions to 60 MeV, is shown to demonstrate its performance. The operation of the accelerator circuit is simulated numerically and shows that the acceleration pulses are successfully applied to each gap.
Time fns)
Fig. 4. Typical waveforms of gap voltages simulation with input pulse for L,.
(V,) obtained
by the
acceleration tube and the grounded plate is not important in the accelerator. The efficiency of the accelerator is evaluated to be around 40% from the input pulse energy and the input power to the gap. The low efficiency is considered to be due to the energy transfer to L, and the mismatch between the input pulse and the gap impedance. The former can be reduced by increasing the characteristic impedance of L, by reducing the capacitance between the lines, while the latter can be improved by optimizing the circuit parameters.
References Cl1 D. Keefe, Part. Accel. 11 (1981) 187.
PI
A. Faltens. E. Hoyer, D. Keefe, L. J. Laslett, IEEE trans. Nucl. Sci. NS-26 (1979) 3106. c31V.O. Brady, A. Faltens, D. Keefe, E.P. Lee, Fusion Technol. 13 (1988) 255. II41S. Humphries Jr., T.R. Lockner, in: A. Septier (Ed.), Appl. Charged Particle Optics (Suppl. 13C). Academic Press, New York, 1980. CSI S. Humphries Jr., T.R. Lockner, J.R. Freeman, IEEE Trans. Nucl. Sci. NS-28 (1981) 3410. A. Tokuchi, K. C61T. Tanabe, A. Kanai, K. Takahashi, Masugata, M. Ito, K. Yatsui, Phys. Rev. Lett. 56(1986) 831. Y. Kubota, A. Miyahara, K. Yamamoto. c71S. Kawasaki, IEEE Trans. Nucl. Sci. NS-30 (1983) 3016. PI J.A. Nation, Part. Accel. 10 (1978) 1.
Nuclear Instruments and Methods in Physics Research
A 399 (1997)
1~4
Section A
Proposal of an electrostatic multi-stage accelerator for high-current pulsed ion beam Katsumi Department
Masugata”
ofElectrical Engineering, Nagaoka Uniwrsity of’ Technolo~. Nagaoka. Niigatu Y40 2 I Jtrpm Received
14 April 1997: received in revised form IO June 1997
Abstract
A new type of multi-stage accelerator is proposed to produce high-current-pulsed ion beam. In the accelerator, a pulsed high-current ion beam is accelerated without using induction cells. The accelerator is simple and a higher acceleration gradient is realized compared with an induction linac. The performance is demonstrated through the example of a ten-stage accelerator designed to accelerate a 50 MeV 100 kA, 50 ns beam of C+ ions to 60 MeV. The operation of the accelerator circuit was simulated by a computer code, which shows that acceleration pulses were successfully applied to each gap. P,‘lCS: 19.17. + w: 41.75.Lx Inertial confinement multi-stage accelerator
Keywrds:
fusion; Intense pulsed ion beam: Induction
In recent years, an intense pulsed ion beam (PIB) has been actively studied as an energy driver of inertia1 confinement fusion. Linear induction linacs have been considered to be a hopeful source of PIB since multi-stage acceleration of high-current PIB is possible [l-7]. In the induction linacs, however, induction cores of ferromagnetic materials are utilized to sustain the acceleration voltages, which makes the accelerator expensive and the structure complicated [2]. In addition, the average gradient
*Tel.: + 8125847 9894; [email protected].
fax:
+ 8125847
9500;
e-mail:
0168.9002/97/$17.00 1~:’ 1997 Elsevier Science B.V. All rights reserved PII SOl68-9002(97)00912-I
linac; Pulse power technology:
Electrostatic
of the acceleration in the induction linacs is limited by the size of the ferromagnetic cores. Even for a short pulse of 50 ns duration, the obtainable gradient is only 1 MV/m, whereas the maximum stress at the vacuum interface in the acceleration tube is more than 10 MV/m [S]. In this article, a new type of high-current, electrostatic, multistage accelerator is proposed. The accelerator is simple and is expected to have a higher acceleration gradient. Fig. l(a) shows the concept of the proposed accelerator (as an example, a ten-stage system is shown). In the accelerator, the ion beam is injected from the left side. Here we assume that the length of the beam (I,,) is much shorter than the accelerator length (I,). By us&g the velocity (u,,) and the
2
lnstr. and Meth. in Phys. Res. A 399 (1997) l-4
K. Masuguta/Nucl.
Acceleratrion gap
II
II
II II
IIJ II II
II II
II
Ii II
Cl II
II II
v,
v,, v,, v,
v,
V,IO’
II
Ii
II
J----f!-yyFf~~~[“l:
’ v,, @) I-
v,
v,, v,
r
Fig. 1. (a) The concept of the accelerator proposed (a ten-stage system is shown as an example). (b) Example of the gap voltage waveforms.
To satisfy Eq. (2), a voltage of - (+)I’, (the negative voltage is called “compensating voltage”) is applied to the gaps where beam is absent. The waveforms of the gap voltages are shown in Fig. l(b). As seen in the figure, the acceleration voltage is compensated by the “compensating voltage” and multi-stage acceleration is realized. This is in complete contrast to the induction linacs where acceleration voltage is compensated by an inductively produced field. In other words, in the proposed accelerator, the application of the acceleration voltages becomes possible due to a timeof-flight effect of the ion beam. Fig. 2 shows an example of the accelerator design to demonstrate the operation. The accelerator is designed to accelerate a 50 MeV 100 kA, 50 ns beam of C+ ions to 60 MeV in ten acceleration gaps. The accelerator consists of an acceleration tube and pulse transmission lines which are put in a deionized water pool. The acceleration tube consists of acceleration gaps, drift tubes and water-vacuum interfaces. The gap intervals (Ii, 12, . . . , 19) are adjusted to be 25 ns, being equal to the time-of-flight delay of the beam (a half of the beam duration). Since the velocity of the beam, and hence, the length of the beam increases as the beam
duration (r) of the beam, Ib is described as 4 50 MeV C’
Ib = Ubx t.
(1)
For example, Ib for a Cf beam of 50 MeV with duration of 50 ns is only 142 cm and the condition I, >>Ib is realized in a 10 m accelerator. In the figure, Ii, is supposed to be two times the interval of the acceleration gap (li). In the accelerator, points A and B in Fig. l(a) are kept at ground potential. Consequently, the sum of the voltages applied to the gaps (I’,,, I/gZ, . . . , V,,,) is always kept zero and the following relation is satisfied: v,, + v,, + v,, + .‘. + l/,10 = 0.
(2)
An acceleration voltage (V,) is applied to the gaps where the beam is just passing (in the figure, the beam is passing g, and g, at time tl) and V,, = V,, = V, (at ti).
(3)
/ Pulse Transmission Line
Fig. 2. The example operation.
of the accelerator
‘460MeVC’
design to demonstrate
the
3
K. MasugaialNucl. Insir. and Meth. in Phj,s. Res. A 399 (1997) i-4
is accelerated, the intervals are elongated downstream (e.g., l1 and l9 are designed to be 71 and 77 cm, respectively). The structure of the acceleration gaps is not designed in detail. However, the gap structure is flexible and many types of the structure including a magnetically insulated gap [4-61 or the multibeam accelerator concept [2,3] will be acceptable. One of the subject for developing the high-currentpulsed ion beam linac is to obtain a stable propagation of the beam, which is not considered in this article. However, the enhanced acceleration gradient will reduce the problem of propagation. For the acceleration gaps, pulse transmission lines are connected to apply the acceleration voltages. The lines have a characteristic impedance of 10 R and an electrical length of 150 ns (which corresponds to a physical length of 5 m for a waterfilled line). For each line, the acceleration pulse is applied with an interval of 25 ns. The electrical length of 1.50 ns is important to isolate the gap voltage as shown below. Fig. 3 shows the simulation model of the accelerator. Here, L,,(n = 1 - 10) are the parasitic lines produced between the outer conductors of the adjacent transmission lines. The characteristic impedance of each line is assumed to be 30 Q. Since the ion beam stays in the accelerator for 300 ns, the go and back delay time of the pulse in the lines (Zq,, where TV is the electrical length of the lines) is designed to be larger than 300 ns to avoid the effect of reflection. Parasitic lines are also produced between the outer conductor and grounded plate, however, the effect was ignored for simplicity (the effect is expected to be reduced by using a deepwater pool). The impedance of the acceleration gaps is considered to be high when the beam is absent and low when the beam is present. In the simulation, the gap impedance is assumed to be 100 and 10 fz when the beam is absent and present, respectively, and the transient times of the change in the impedance are assumed to be 1 ns. The input pulses of rise and fall time of 5 ns and the duration of 50 ns are applied to each pulse transmission line. Each pulse is applied with an interval of 25 ns to be synchronized with the arrival of the beam in each gap. The height of the pulse is
Fig. 3. The simulation
model of the accelerator.
determined to be 1.3 MV to obtain gap voltages of w 1 MV. Compensating voltages are not applied since they are automatically produced when the acceleration pulses are applied. Fig 4 shows the typical waveforms of the gap voltages (V,) obtained by the simulation with an input pulse for L,. As seen in Fig. 4, V,, rises at t = 150 ns and has a peak of z 1 MV. With the rise voltages are produced in the of F,,, compensating other gaps ( Vg,-l,). Since the parasitic lines and C, divide the applied voltages, the waveforms of the compensating voltages are observed to be almost the same. At r = 360 ns, a negative pulse is observed in I’,,, which is due to the reflection of the acceleration voltage in L,i. The potential at point X in Fig. 3 against the ground potential has a peak value of 1.6 MV and is not high compared to the total acceleration voltage of 10 MV. Therefore, the insulation between the
K. MasugafalNucl. Instr. and Meth. in Phys. Res. A 399 (1997) 1-4
4 1.4 1.2 1 0.8 s 5 & 3 s
0.6 04 0.2 0 -0.2 -0.4 -0.6 0
50
100
150
200
250
300
350
400
In conclusion, a new type of multi-stage accelerator is proposed to produce high-current pulsed ion beam. Compared to the conventional induction linacs, the proposed accelerator is of great advantage since it is simpler and has a higher acceleration gradient. As an example, a ten-stage accelerator, which is designed to accelerate a 50 MeV, 100 kA, 50 ns beam of C+ ions to 60 MeV, is shown to demonstrate its performance. The operation of the accelerator circuit is simulated numerically and shows that the acceleration pulses are successfully applied to each gap.
Time fns)
Fig. 4. Typical waveforms of gap voltages simulation with input pulse for L,.
(V,) obtained
by the
acceleration tube and the grounded plate is not important in the accelerator. The efficiency of the accelerator is evaluated to be around 40% from the input pulse energy and the input power to the gap. The low efficiency is considered to be due to the energy transfer to L, and the mismatch between the input pulse and the gap impedance. The former can be reduced by increasing the characteristic impedance of L, by reducing the capacitance between the lines, while the latter can be improved by optimizing the circuit parameters.
References Cl1 D. Keefe, Part. Accel. 11 (1981) 187.
PI
A. Faltens. E. Hoyer, D. Keefe, L. J. Laslett, IEEE trans. Nucl. Sci. NS-26 (1979) 3106. c31V.O. Brady, A. Faltens, D. Keefe, E.P. Lee, Fusion Technol. 13 (1988) 255. II41S. Humphries Jr., T.R. Lockner, in: A. Septier (Ed.), Appl. Charged Particle Optics (Suppl. 13C). Academic Press, New York, 1980. CSI S. Humphries Jr., T.R. Lockner, J.R. Freeman, IEEE Trans. Nucl. Sci. NS-28 (1981) 3410. A. Tokuchi, K. C61T. Tanabe, A. Kanai, K. Takahashi, Masugata, M. Ito, K. Yatsui, Phys. Rev. Lett. 56(1986) 831. Y. Kubota, A. Miyahara, K. Yamamoto. c71S. Kawasaki, IEEE Trans. Nucl. Sci. NS-30 (1983) 3016. PI J.A. Nation, Part. Accel. 10 (1978) 1.