The NBS Proposal for a One GeV CW Racetrack Microtron Facility

As part of a joint accelerator research project with the Los Alamos National Laboratory, NBS is now building a 200 MeV, high current, CW racetrack microtron (RTM). Upon its completion, scheduled for 1984, we propose to use this machine to provide CW electron beams to nuclear physics experimenters, and also as the injector for a second stage RTM to boost the final energy to one GeV. A building addition of 35,000 square feet will house the second stage RTM and new experimental facilities. Subharmonic RF beam splitting is planned to allow up to three simultaneous beams for experiments, with currents up to l00¿A each. In addition, low and high energy beams at low currents for tagged-bremsstrahlung experiments can be delivered at the same time. This proposed multi-user facility is intended to be a national center for electromagnetic nuclear physics research.

MeV and accelerated to any desired final energy up to es, and thus 1.0 GeV maximum. The design current capability of RTM-2 Eventually is 300 PA. RTM-2 would make extensive use of the tech-(which reyqu nology developed in our present accelerator research steericng ar project in order to minimize risk and development costs. acceler9ating As many as three simultaneous high current beams can be celeratu obtained from the cascaded microtrons by use of subhar-cessive retu monic rf beam splitting. The currents of these three beams are made independently controllable by a modification of the present rf chopper system in the injector of will eventu RTM-1.
Two different beam energies can be provided steering and simultaneously: 200 MeV from RTM-1, and the output of the toler energy of RTM-2. (We will refer to this energy as "One the problems GeV," although it should be understood that it is conwill be easi tinuously variable in the range 200-1000 MeV  (1) the energy gain (in MeV) of the accelerating s the rf vacuum wavelength (in centimeters), field (in Tesla) of the end magnets. The is the integer number of wavelengths by bit circumference increases between succes-It is well known that v should be as small to maximize the phase stability and minimize and operating tolerance requirements. All planned microtrons of which we are aware v = 1 or 2. The staple range of resonant ven by 0-< r < tan'1(-). rtant cost consideration for a large microsize of the end magnets. The radius of the in the end magnets is proportional to the divided by the magnetic field. The cost of will be approximately proportional to their therefore they should be designed for the :tical magnetic field. For the high-quality Ids needed here, this practical limit is in 5-1.6 Tesla.
that v = 1 and large B are desired, equals us that the energy gain per pass required )nal to the wavelength, X. The energy gain 1lerator section, neglecting beam loading AV = (P RL)"2 (2) the input rf power (MW), R the shunt imped-, and L the length (m) of the accelerating Since rf power and accelerator length are Dst factors, equations (1) and (2) tend to small rf wavelength.
Furthermore, a high ance is highly desirable and other factors R scales as (A)-1"2. On the other hand, as the allowable energy gain per pass decreass the number of passes through the accelerato obtain a given final energy increases. the cost of the additional return paths ire additional diagnostic instrumentation, nd focusing) will overcome the cost of the section. In addition, the spacing of sucirn paths, given by ially become too small to allow separate i focusing on each orbit. Furthermore, many rances of the system will scale with A, and s of producing and transmitting the rf power ier at longer wavelengths. ese considerations in mind, the final choice ency is determined by the commercial availiigh-power CW klysrons. Within the frequency )le for an RTM, the only available high-power Is are in the 2400 MHz region (S band, ). Then with v = 1 and B = 1.6 Tesla, the n per pass (from equation II B.1) is In the conventional RTM the beam is returned to the accelerating section by means of uniform field end magnets, as shown in Fig. 1. On successive passes, the beam must pass through the accelerating section at the same synchronous phase, fr' of the rf field. This resonance condition can be expressed by the relation2 The current obtainable from an RTM is known to be limited by beam blowup due to excitation by the beam of rf modes in the accelerating structure which have strong transverse magnetic fields on the structure axis. This effect has been observed in the Stanford and Illinois superconducting accelerators but never, to our knowledge, in microtrons having room-temperature accelera-U.S. Government work not protected by U.S. copyright. Compensation for the effect of D6 on recirculated orbits is accomplished by the chicane system consisting of D9 and D10. The array of dipole magnets D13 and D14 compensate the recirculating orbits for the displacements produced in D7 and D8. The array of quadrupole magnets Q10 and Qll provide adjustable focussing on the return lines. The beam can be extracted after any number of passes by moving the extraction magnet D15 to the appropriate orbit. ting structures. However, the observation of beam blowup in superconducting accelerators together with theoretical studies of the blowup phenomenon3 predict that the starting current for beam blowup, I, in microtrons with room-temperature accelerating struc%ures will be in the range 100-lOOO1 A. Given that the accelerator current and energy are determined by nuclear physics requirements, and AV is determined by equation (1) and the -1.6 Tesla practical limit on end-magnet field strength, the beam blowup threshold can be increased by: (1) raising the injection energy, (2) increasing the accelerating gradient, AV/L, and (3) using stronger focussing. One of the goals of the NBS-LANL accelerator research project has been to increase the available gradient, AV/L, by attaining the highest possible shunt impedance, R, and the dissipation limit (P/L) of the rf structure.
The important cost trade-off in the design of a large microtron is between rf power and accelerating gradient, since the energy gain per pass is determined by equation (1) together with the dependence of magnet cost on energy and field strength. The dependence of cost on rf power and rf structure length can be written where C is the cost per megawatt of rf power, and CL is the coEt of the machine per unit structure length.
Using equation (2) to establish the relationship between P and L, a cost minimum is found for an accelerating gradient of AV CLR L Cp (5) We have developed reasonably rel i able values for the differential cost factors from our experience with the The 200 MeV RTM (1, 4) is currently under construction at NBS as part of a joint NBS-LANL accelerator research project funded by DOE. Design parameters are given in Table I. Our schedule calls for completion of the accelerator in mid-1984, to be followed by a year of beam studies which are an important part of the accelerator research project. A major effort will be made to determine the starting current, Is, for beam blowup (BBU) over a range of beam tunes and accelerating gradients. As an injector, RTM-1 will be operated at a tune which has been found to inhibit BBU. Theoretical pred i ct i ons 3 of I for RTM-1 range from 250 to 1300jjA. It is very unlikely that BBU will limit the achievable current to less than 300A.
The only major component of RTM-1 which is still under development is the rf accelerating structure. It is a water-cooled version of the side-coupled structure.5 Fu 1 power rf testing of the 2.7 m long preaccelerator section was started on February 28, 1983. After less than one day conditioning, this structure was operated at 83 kW CW power input, exceeding the design power level by 14 percent. The shunt impedance was measured to be 82.5 MQ/m. No parasitic modes are excited near the operating frequency and no power-induced tilt in the electric field has been observed.
In the present proposal, conversion of RTM-1 to an injector will begin in mid-1985. The only modification necessary is to replace the present chopper system with a subharmonic chopper. As shown in Fig. (2), this system provides independent control of the current in three successive beam pulses which are to be split into three simultaneous beams following acceleration. Every third beam pul se passes through the same aperture, (indicated by the rectangles).
Each aperture can be adjusted independently to control the current in the pulses which it transmits.
One GeV Microtron RTM-2 The proposed 1 GeV RTM is shown schematically in Fig. (3). Design parameters of RTM-2 are given in Table  1. Much of the technology and many of the components developed for the 200 MeV RTM will be utilized. This provides a firm basis for cost estimating and reduces the design effort and the risk of the project.
The accelerating section consists of two 4 m long side-coupled structures identical to those used in RTM-1. They will operate at a lower gradient than in RTM-1; 1.25 MeV/m rather than 1.5 MeV/m. The rf power source employs a 500 kW CW klystron, identical to the one used  in RTM-1. The entire rf system-power supply, control, and distribution will be copied from incorporating any modifications found necessary testing of the first stage machine. drive, RTM-1, during A preliminary design for the end magnets is shown in Fig. (4). The design is based on the same principles as the 200 MeV end magnets 7 but i s mod i f i ed to a semicircular shape to save weight. The design is suffi-ciently symmetric that its field distribution can be calculated by using the accurate, two-dimensional codes which solve Poisson 's equation (POISSON or TRIM). Results of field mapping the 200 MeV end magnets will be considered before arriving at a final design. The end magnets for RTM-2 require a field uniformity of ±1 part in 104 to eliminate significant beam emittance growth.
The TRIM calculations for the end magnets of RTM-1 indicate that this uniformity would be achieved in construc-1 GeV RTV END MAGNET WT 300 t POWE-R 300 k'W @ 1,6 TESLA  tion, except for the effects of manufacturing tolerances and steel non-uniformity. To compensate for these imperfections, the magnetic field will be mapped using NMR probes and a computer-controlled two-dimensional scanning device.8 The field deviation measurements will be used to construct pole-face current sheets to reduce the non-uniformity to less than 1 part in 104. We expect to use the same technique to obtain the required field uniformity in RTM-2.
The experimental studies of BBU starting current, Is on RTM-1 will be completed in time to aid in selecting a beam tune for operation of RTM-2. The theory of Rand et a13 is expected to give reliable ratios of Is in RTM's which have the same type of accelerating structure. The theory predicts a starting current in RTM-2 which is 40% above that in RTM-1 for the same tune. Thus, once a tune for RTM-1 has been determined to avoid BBU, an equal or weaker tune can be specified for RTM-2. Focussing on the return lines, if required, will be provided by quadrupole doublets of the current-sheet type9 which fit in the 4 cm return-line spacing.
The effects of synchrotron radiation from the beam in the end magnets has been calculated for a range of tunes. In the limiting case of no transverse focussing, the radiation causes a horizontal displacement of the central orbit from the unperturbed return lines. The displacement increases quadratically with the turn number up to a maximum displacement of 3.3 mm on the last turn. The final beam energy is simultaneously increased by 21 keV compared to the no-radiation case. With focussing on the return lines for a transverse tune of 3283 1/8 in both x and y, the displacement is reduced to virtually zero.
The main effect of the synchrotron radiation is to increase the longitudinal emittance of the beam.
In the presence of return-leg quadrupoles which introduce transverse-longitudinal coupling, the horizontal-plane transverse emittance is also increased. Typically, both the horizontal and longitudinal emittance envelopes show an increase in areas of about 40 percent or less. Quantum fluctuations in the syncrotron radiation contribute an additional 20% increase in the longitudinal emittance, and -40% in the horizontalplane transverse emittance. The emittances given in Table (1) for RTM-2 include the effect of the synchrotron radi ation as calcul ated for the case vx = v = 0.25, after arbitrarily increasing the injection emXttances by 30% (transverse) and 40% (longitudinal) from the previously-calculatedl values for RTM-1.
The control for RTM-2 will utilize components developed for the RTM-1 control system10 and will be integrated with that system.

Beam Distribution
The beam transport system shown in Fig. 5 allows as many as five beams with different currents to be sent simultaneously to the five experimental areas. At four points in the system, two-way beam splitting is achieved by a combination of an RF beam splitter, dc deflection magnets, and a septum magnet. Operating at one-third the accelerating frequency, the rf splitter may be phased to send two of three beam pulses into either of two beam lines and the third pulse into the other beam line.  At one GeV, a 2 meter structure powered by a second 50 kW transmitter provides a 1.0 mrad deflection amplitude. Following each rf splitter, the split beams are allowed to drift until a separation of about one centimeter is obtained and then a DC septum magnet is used to further separate the beams. Because of the very small beam emittance, and especially the phase spread of < 20 (at fo), degradation of the beam emittance in the rf splitters is negligible.
Each of the five 90' bends in the beam transport system is accomplished using a nondispersive DQD sys- In cases when one split-beam current is lOOpA, the minimum stable current that can be provided to either of the photon-tagging areas using the adjustable chopper slits is about liA. Further reduction in beam current to meet the photon-tagging requirements is accomplished by closing adjustable slits in the 200 MeV and 1 GeV beam transport tunnels.

Buildings
For the 200 MeV program with RTM-l, only minor building modifications are needed. Structural changes to the present building required for the 1 GeV program are mainly improved utility access between existing rooms, a new fire suppression system for the existing areas being used, improvement of equipment handling capability, and relocation of the present RTM klystron power supply.
The scale of a building to house the RTM-2, two experimental halls and associated data rooms is set by the size of the accelerator and spectrometer magnets and shielding walls to contain the radiation generated The dimensions of the two new experimental halls (MR4 and MR5 if Fig. 5) were chosen to accommodate the equipment designed for a tagged photon program (measurement room 4, a low current room) and a coincident electron scattering program (measurement room 5, a high current room).
A set-up area, served by a freight elevator, is the common access point for the high current measurement laboratory (70'x7O'x56'H), low current measurement laboratory (50'x70'x25'H) and the RTM machine room. Located on the other side of these areas is the beam distribution room connecting to the RTM machine room. Above the beam distribution room, a service area is planned for the mechanical and electrical distribution of the RTM machine and the associated laboratories.
Above th i s service area, a counting floor is planned to house two data collecting rooms, two user rooms, a conference room and a library. Above the counting floor, a plant floor is reserved for the building mechanical and electrical services. Roof hatches are planned for equipment access to measurement rooms 4 and 5 and the RTM Machine Room.
Effective utilization of a multiple-beam research facility is greatly enhanced if all areas of the building can be occupied by the staff on a full time basis whenever there is no beam in that particular area. This was a design criterion for the existing NBS Linac facility. The shielding walls between rooms were calculated to be adequate to reduce radiation levels in all rooms not traversed by an electron (or photon, or neutron) beam to < 0.1 mrem/hr for any possible mode of linac operation up to a total beam power of 100 kW at 100 MeV. Extensive radiation surveys performed under a great variety of operating conditions have shown this goal was met. We are thus assured of the adequacy of existing shielding for RTM-1 and associated experimental areas, control rooms, and other occupiable spaces.
New shielding calculations have been performed for RTM-2, including a consideration of the radiation levels in the existing building due to sources in the new addition. The 3.6m thick concrete shielding walls which have proven to be adequate under all operating conditions for a 100 MeV, 100 kW accelerator are not sufficient by themselves for complete personnel protection at 1 GeV and 300 kW. Nevertheless, we have not used wall thicknesses greater than 3.6m (12 feet) (except for one small region around the beam dump in the high current room, MR5) because of prohibitive cost. We pl an to augment the main shielding walls with auxiliary local shielding near the predicatable main radiation sources such as beam dumps and collimating apertures. Since the beam quality and stability of the RTM will be so much better than older accelerators the amount of beam spill which is distributed over long beam paths (and thus not amenable to local shielding) is quite small.
A system of radiation monitoring, personnel exclusion, and area inspection modelled after the system in use at the NBS Linac will be implemented to insure radiation safety for all staff and visitors. The existing system has proven to be fully adequate and not unnecessarily restrictive in seventeen years of operating experi ence.