though it. The efficiency of this conversion was greater than 40% from beam energy to microwave energy. Problem areas that remain to be studied, and there are many, include phase control of the rf, transverse stability of the driving beam, and the transport of intense microwave power from the FEL to the acceleration gaps.

All of these problems appear to be solvable, but at the expense of simplicity. The phase control of the rf, for example, is particularly difficult. Because of the powerful amplifier action of the FEL, the rf phase is sensitive to beam current, wiggler field and beam energy. So far, the only method devised to control phase is to discard the rf when its phase is unacceptable and reinsert proper-phase rf at a small cost of emittance growth in the driving beam wiggling its way through the FEL.

The transfer-cavity alternative, often called the relativistic klystron version of the two-beam accelerator, shown in figure 7, is now the object of considerable effort by a SLAC-Berkeley-Livermore collaboration, and also by the CERN group. It does seem to remove some of the problems associated with the FEL, but one should note that the FEL has been demonstrated to produce power, while the relativistic klystron thus far exists only on paper. People have investigated transfer-cavity problems such as longitudinal and transverse beam dynamics and the proper design of the cavities. Transverse beam dynamics is of particular concern because the very small transfer cavities enhance transverse instability, causing "beam breakup."

Klystrons, lasertrons and gyroklystrons

Finally, I will describe three other possible sources of short-wavelength power for a high-gradient rf linear accelerator.

32 PHYSICS TODAY JANUARY 1988

The klystron, until now the power amplifier of choice for electron-positron linear colliders, is an electron-beam tube usually containing two cavities. In the first cavity a velocity modulation is imposed on the tube's electron beam and the bunched beam then radiates coherent microwaves into the second, tuned output cavity. The SLAC linac operates in the "S band," at a wavelength of 10 cm, so that S-band klystrons have been the subject of extensive development. Having been upgraded to serve as the accelerator for the Stanford Linear Collider, the linac now accelerates electrons and positrons to 50 GeV. More than two hundred 65-MW klystrons line the great gallery atop the linac, feeding it rf power in 3.5-microsecond pulses.

A klystron operating in the S band has produced 150MW peak power in a pulse of 1 microsecond. The high-energy linear colliders we envision for the next generation will have to operate at significantly shorter wavelengths. Experiments at SLAC have shown that an accelerating gradient of more than 200 MeV/m can be achieved at the 3 cm wavelength regime, the so-called X band. As klystrons are designed for higher frequency, the output power decreases as the cavities must be made smaller, reducing the klystron's beam current.

A development program is under way at SLAC to build a 100-MW, 2-sec klystron operating in the X band with 65% efficiency." The rf pulse will have to be further compressed because at these wavelengths the rf must fill the accelerating structure in 100 nanoseconds. A compression scheme can be devised, but it requires very long sections of waveguide (2000 feet for a 2-sec pulse). For still shorter rf wavelengths, conventional klystrons do not seem to be suitable for linear colliders.

A second development program at SLAC, in collabora

tion with Livermore, employs pulse-forming devices, using saturable inductors pioneered at Livermore to make a short-pulse (50-100 nsec) klystron." They are planning to put a megavolt pulse on an X-band klystron in the expectation that a peak power on the order of a gigawatt may be obtained from klystrons at very short pulse widths. The development confronts problems associated with large cathode-anode stresses and with the cathode current-density responses to the fast rise times (15 nsec) of very short pulses. The group aspires to efficiencies on the order of 70%.

The lasertron is a microwave power tube that employs a photoemission cathode illuminated by an optical pulse train of the proper microwave structure to produce a bunched electron beam directly at the cathode. This lasertron beam is accelerated to high voltage and immediately passed through an rf output structure to produce microwave power for the linac. In a high-peakpower device there is considerable charge in each bunch, requiring that the lasertron beam be confined transversely by an axial magnetic field, and that it be rapidly accelerated to limit longitudinal spreading. As in all microwave power tubes, the spent lasertron beam is dumped in a collector. Lasertrons are under active development at SLAC, KEK and Orsay. The idea is also being studied at the Texas Accelerator Center and at Los Alamos.

The attractiveness of a photoemission cathode stems from the ease with which one can directly modulate the emission at high frequency, and from the very high current densities one can readily obtain. The lasertron development program at SLAC involves learning to prepare robust high-current-density photocathodes capable of operating for long periods in the environment of a high-power microwave tube. Because photoemission cathodes are typically far more sensitive than thermionic cathodes to degradation or destruction by residual gases, one needs a very good vacuum system. Furthermore, photocathodes generally require alkali metals for their activity, which makes the operation of high-voltage guns more difficult.

The laser itself is a complex system, far more costly than a filament transformer. The necessity of getting the illuminating light to the cathode can add significantly to the complexity of the basic device. In general, photoemission cathodes operating in actively pumped ultrahighvacuum systems have not shown very long lifetimes, and the ability of these cathodes to deliver the large total charge required over the operating life of a microwave power source has not yet been demonstrated.

On the other hand, the lasertron is in many ways

simpler than thermionic devices, so it may be possible to produce a relatively inexpensive unit. Because the photocathode emits only while illuminated, only a dc power supply or a very simple modulator is needed, as opposed to the complex and expensive modulators necessary for devices with thermionic emission cathodes. This is true, however, only if the very large voltages required to accelerate bunches before they disassemble due to space charge can be held in dc.

If they are to be useful for future linear colliders, lasertrons must be extended to shorter wavelengths. The lasertron is, however, subject to the same wavelength scaling law as the klystron; both give decreased output power as the frequency is increased. Recently, the Texas group has proposed a non-axial-symmetric "sheet beam" lasertron for delivering high power at high frequency.

Gyrotron oscillators have been developed for heating thermonuclear fusion plasmas at short wavelengths (5 mm), with very high average power (300 kW), high peak power (300 MW) and high efficiency (50-65%). Although some effort has been made to develop gyrotron amplifiers (as distinguished from oscillators), a group at the University of Maryland has only recently begun to develop a highpeak-power gyrotron tube. 19

The basic idea of the gyrotron (also called a cyclotronmaser or gyroklystron) is to produce a near resonance between a harmonic of the electron cyclotron frequency and an output cavity mode. A longitudinal magnetic field causes the gyrotron's spiraling electron beam to bunch, thus generating coherent short-wavelength radiation.

The suppression of unwanted modes is the paramount issue here. Also of concern are wall losses and spacecharge depression, both of which must be kept within acceptable bounds. These considerations work in opposite directions; the suppression of unwanted modes is most easily achieved by using a small cavity in which few modes are excited. On the other hand, wall losses are minimized by working in a large cavity, and space-charge depression is minimized by using a large-radius, low-density beam.

Phase stability in the gyroklystron is a crucial consideration for use in large linear accelerators, where hundreds or even thousands of microwave amplifiers must be synchronized. While this is a demanding requirement, especially for the modulator, it is within the state of the art. Because prior studies of gyroklystrons have been plagued by spurious oscillations, stability is of special

concern.

The Maryland design is for a 10-GHz tube producing 2-μsec pulses with a peak power in excess of 30 MW and an efficiency of 45%. Studies indicate that the design can be scaled up to power levels exceeding 150 MW. This would

PHYSICS TODAY JANUARY 1988 33

be accomplished by increasing the voltage, radius and current of the tube's beam while maintaining its intensity. Just as in the conventional klystron, the gyroklystron's output pulse would have to be further compressed so that it matches the requirements of the accelerating structure.

Continuing the growth

There is clearly no lack of ideas for new acceleration schemes. They range from evolutionary developments of conventional power sources to completely new concepts such as the laser-plasma accelerator. Furthermore, there are already modest experimental programs associated with each of the approaches discussed here. Success in some of these developments is crucial to the building of future high-energy electron-positron colliders.

But conventional accelerators are very effective and very large. Something really new is not going to compete until it has undergone extensive development and modeling. That process will be lengthy and expensive, but necessary if we are to continue to see, as we have in the past, the exponential rise in the accelerator energy available to high-energy physicists shown in figure 1, and hence the continuation of high-energy physics as an experimental science.

References

1. B. Richter, in Laser Acceleration of Particles, AIP Conf. Proc. 130, C. Joshi, T. Katsouleas, eds., AIP, New York (1985), p. 8. P. B. Wilson, in 1986 Linear Accelerator Conf. Proc., SLAC303, Stanford Linear Accelerator Center, Stanford, Calif. (1986), p. 585. B. Richter, in Proc. 1984 ICFA Seminar on Future Perspectives in High Energy Physics, KEK-841, KEK, Tsukuba, Japan (1984), p. 226.

2. P. J. Channell, ed., Laser Acceleration of Particles, AIP Conf. Proc. 91, AIP, New York (1982).

3. J. Mulvey, ed., The Challenge of Ultra-High Energies, ECFA 83/68, Rutherford Laboratory, Didcot, UK (1982).

4. The Generation of High Fields for Particle Acceleration to Very High Energies, ECFA 85/91, CERN 85-07, CERN, Geneva (1984).

5. C. Joshi, T. Katsouleas, eds., Laser Acceleration of Particles,

AIP Conf. Proc. 130, AIP, New York (1985).

6. F. Mills, ed., Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, AIP, New York (1987).

7. A. M. Sessler, Am. J. Phys. 54, 505 (1986).

8. T. Tajima, J. M. Dawson, Phys. Rev. Lett. 43, 267 (1979). 9. A. E. Dangor et al., in Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, F. Mills, ed., AIP, New York (1987), p. 112.

10. C. E. Clayton, C. Joshi, C. Darrow, D. Umstadter, Phys. Rev. Lett. 54, 2343 (1985). F. Martin et al., in Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, F. Mills, ed., AIP, New York (1987), p. 121.

11. G. A. Voss, T. Weiland, in The Challenge of Ultra-High Energies, ECFA 83/68, J. Mulvey, ed., Rutherford Laboratory, Didcot, UK (1982), p. 287. H. Dehne et al., Proc. 12th Int. Conf. on High Energy Accelerators, Fermilab, Batavia, Ill. (1983), p. 454.

12. P. Chen, J. M. Dawson, R. W. Huff, T. Katsouleas, Phys. Rev. Lett. 54, 693 (1985).

13. J. Rosensweig et al., in Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, F. Mills, ed., AIP, New York (1987), p. 231.

14. W. Willis et al., in Laser Acceleration of Particles, AIP Conf. Proc. 130, C. Joshi, T. Katsouleas, eds., AIP, New York (1985), p. 421. S. Aronson, in F. Mills, ed., Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, AIP, New York (1987), p. 283.

15. A. M. Sessler, in Laser Acceleration of Particles, AIP Conf. Proc. 91, P. J. Channell, ed., AIP, New York (1982), p. 154 (1982). A. M. Sessler, S. S. Yu, Phys. Rev. Lett. 58, 2439 (1987). W. Schnell, in Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, F. Mills, ed., AIP, New York (1987), p. 17. 16. T. Orzechowski, B Anderson, J. Clark, W. Fawley, A. Paul, D. Prosnitz, E. Scharlemann, S. Yarema, D. Hopkins, A. Sessler, J. Wartele, Phys. Rev. Lett. 58, 2172 (1986).

17. M. Allen, private communication

18. C. K. Sinclair, in Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, F. Mills, ed., AIP, New York (1987), p. 298. M. Yoshioka, in Int. Symp. on Advanced Accelerator Concepts, AIP Conf. Proc. 156, F. Mills, ed., AIP, New York (1987), p. 313.

19. K. R. Chu et al., IEEE Trans. Plasma Sci. P5-13, 424 (1985).