New projects and important upgrades are under way in many laboratories, while technological progress and continual ingenuity augur well for the future.
High-energy accelerators in the spotlight
An initial report from the recent Particle Accelerator Conference (PAC) in New York was published last month. The PAC parallel sessions spanned a range of accelerator activities, which are summarized here.
Magnets
Most of the PAC magnet session covered the status and progress on the research and development for superconducting magnets. For CERN’s LHC collider, a large effort is devoted to the optimization of the dipole for series production. C Wyss presented the updated version of the dipole parameters, featuring stainless steel collars and six-block coils. He also discussed the aim and status of the short model and full-size prototype dipole programmes and the schedule for series manufacture.
D Tommasini described in detail the scope of the 20 short models, some of which have been rebuilt in different variants, with more than 35 versions already having been tested. The aims are to compare the five-block and six block coil geometry, training behaviour, temperature margin, mechanical stability and magnetic field quality.
In the US the post-SSC period has seen a renewed interest in high-field dipole development. R M Scanlan presented an extensive review of the activities in various laboratories, including Brookhaven; Fermilab; KEK, Japan; LASA/INFN, Milan; Berkeley; Texas A&M; and Twente, the Netherlands. He stressed the common requirement to exceed 10 Tesla the practical limit for niobium-titanium superconductors.
Potential applications include a Very Large Hadron Collider (VLHC), a Muon Collider and upgrades to the LHC. The cosine theta coil-winding approach is replaced in recent work by block coil designs, which may be more compatible with the brittle superconductors and high Lorentz field stresses that are inherent in high-field magnets.
The new block designs include the “common coil” designs that are being explored at Brookhaven and Berkeley, as well as a segmented block design with reduced winding stresses at Texas A&M. In addition to magnet design work, several new superconductors are being developed for use in high-field accelerator magnets. These include niobium-aluminium as well as the high-temperature superconductors in both tape and cable configurations.
S Gourlay and A Zlobin gave more details on the dipoles proposed at Berkeley and Fermilab respectively. At Berkeley, a prototype niobium-tin superconducting magnet,utilizing a racetrack coil design, has been built and tested. This was constructed with coils wound from conductor developed for the ITER fusion project, limiting the magnet to a field of approximately 6 Tesla. Subsequent magnets will utilize improved conductor, culminating in a design that is capable of approaching 15 Tesla. The simple geometry is more suitable for the brittle superconductors that are needed to reach high fields.
At Fermilab, high-field magnets of between 10 and 12 T are proposed in view of the VLHC. The main aims are to exploit the relatively small machine circumference and emittance damping owing to synchrotron radiation and still be able to accommodate the radiation power absorbed in the beam tube. Recent progress in the development of niobium-tin superconducting strands makes it possible to design cost-effective accelerator magnets based on a cosine-theta coil geometry above 10 T.
A 1 m high-field dipole model with 1011 T nominal field in a 50 mm bore is being developed at Fermilab in collaboration with Berkeley and KEK as part of the effort for a VLHC.
P Lee presented the prospects for the use of high-temperature superconductors (HTS) in high-field accelerator magnets. In the short term the most promising high-field magnet application is 2212. However, HTSs are still at an early stage of development and continued improvement over the next 10 years should reveal other HTSs for accelerator application.
The Bi-2212 recipe appears to have the greatest potential today, because it can be made in round wire form with a reasonably high critical current, thus permitting access to the cabling technology of low-temperature materials. A dipole is being made at Berkeley using round wire Bi-2212. Among other recipes being explored, Bi-2223 and YBCO are both largely committed to wide-tape designs, for which cabling is a significant challenge.
G Foster described the Transmission Line Magnet a dual-aperture warm-iron superferric magnet, built around an 80 kA superconducting transmission line. He pointed out that the large inventory of surplus cable manufactured for the defunct Superconducting Supercollider could be used for the construction of a VLHC injector with an energy up to 3 TeV.
R Goupta described the tuning shims technique that is used in the interaction region quadrupoles of Brookhaven’s Relativistic Heavy Ion Collider (RHIC) to obtain much lower field errors. Measurements have shown that both systematic and random error harmonics have been reduced to several parts in 100 000 instead of a few parts in 10 000 at two-thirds of the coil radius. The ultimate field errors are now limited by the “changes” in harmonics after quenching and thermal cycling rather than measurements, design or magnet construction errors. These changes appear to depend on the details of the magnet.
J Plueger was one of the few to describe high-technology warm magnets. He presented an overview of an insertion designed for the next generation of free flectron faser (FEL) synchrotron light sources using the principle of self -amplified spontaneous emission. Very long undulators are needed to reach saturation, easily exceeding 100 m for the X-ray FELs. To minimize thetotal length and maximize output, an optimum overlap must be foreseen between electron and laser beam, as well strong external focusing fields to keep the electron beam size small over the whole undulator length.
Walter Scandale.
Current and future machines
Many PAC papers covered machines that are under construction or undergoing major upgrades. News of the RHIC heavy ion collider at Brookhaven and the PEP-II and KEKB electronpositron collider B-factories was included in our PAC preview last month.
Design studies for the next generation of colliders linear electronpositron colliders, muon colliders and very large hadron colliders were well covered.
R Brinkmann of DESY talked about technology and challenges of linear colliders. The session chairman J Peoples paid tribute to B Wiik (April), who had been scheduled to give this talk. T Raubenheimer of SLAC covered the accelerator physics challenges of linear colliders; M Pekeler of DESY the experience of superconducting cavity cperation in the TESLA test facility; and J P Delahaye of CERN the CLIC study of a multi-TeV linear collider.
K T McDonald of Princeton reported on the status of research and development and future plans relating to muon colliders, mentioning that 40 PAC papers were related to muon colliders. G Dugan of Cornell described reseach and development work for Very Large Hadron Colliders.
Quite a few posters also treated various aspects of these machines.
Radiofrequency technology
Striking in the PAC radiofrequency sessions was the remarkable and encouraging progress made in the maximum hold-off voltage in superconducting cavities. A reliable technique to attain the necessary accelerating gradient of 34 MV/m is of paramount importance for the TESLA approach. Electropolishing (EP), jointly studied at KEK in Japan and CEA-Saclay in France, is a clear candidate.
Tests on three sample cavities showed an increase of the maximum accelerating field from 25 to 33 MV/m (in one case a record 37 MV/m was attained), at full TESLA pulse length.
It is interesting that, in contrast with the established methods of high-pressure rinsing and chemical polishing (CP), EP does not seem to act on the residual resistance ratio. Once this is high enough, EP pushes the maximum gradient right up. Even a cavity with bad initial performance attained 33 MV/m. The fact that these cavities turned “bad” again when doing CP after EP can be considered as a validation of the EP approach.
Even though ideas on photonic bandgap (PBG) structures were already presented in PAC three years ago, they now appear interesting for future multi-TeV electronpositron colliders. Of primary concern in these accelerators are the transverse wakefields fields that are left behind in the accelerating structure by the passing bunches, which, in turn, can kick subsequent bunches so violently that they eventually get lost before the collision point.
The damping and detuning of these detrimental transverse modes are the techniques deployed and investigated until now to alleviate this serious problem. (See MOBC2, THAL6 on the PAC Web site.
The PBG structure now sheds light on accelerating structures from a different viewpoint The cell of a PBG structure consists of a transverse periodic lattice of metallic rods between a pair of metal plates. Thus it can be described like a two-dimensional crystal, the central beam hole being a “defect” in this lattice.
If the fundamental, accelerating mode frequency lies in the bandgap of the PBG structure, it cannot propagate transversely and thus remains well confined around the defect. The structure can, however, be made so that it is (transversely) transparent to all higher-order modes, which in turn can quite easily be damped by absorbing material at the periphery a few lattice constants from the central beam hole. (See MOP72, MOP73.)
For anyone who thought that vacuum tubes were history, the development of high-power RF tubes, especially for modern accelerator applications, would prove them wrong. High-power klystrons will be needed in large numbers for future accelerators. For gridded tubes, the technology is still advancing.
Efficiency and high average power at moderate anode voltages and good operational stability are of increasing importance. For klystrons, this leads to multibeam devices, like the one being developed at Thomson Tubes (TTE) for TESLA (seven beams, 10 MW, 1.5 ms, 1.3 GHz). An interesting novel device being developed at Communications and Power Industries (formerly Varian Electron Device Group) is higher-order mode inductive output tube (HOM IOT), which has been around since 1982 under the name klystrode (i.e. a gridded input, klystron type output device). Its promising design data aim for 1 MW continuous wave at 700 MHz with (only) 45 kV anode voltage, and with an efficiency of 73% (THBL3).
An impressive gridded tube was on display at TTE’s stand in the industrial exhibit. The company’s Diacrode is a tetrode with optimized current distribution, which allows for larger surfaces compared with the wavelength, and thus with higher power. This tube has now been built and tested, producing 1 MW in continuous wave, 4.1 MW for 1 ms pulses, both at 200 MHz.
Leaving high-power RF, an impressive realization of a modern low-level RF system is that of the asymmetric B factory PEP-II now being commissioned at SLAC, Stanford.
Since the advent of feedback in accelerators in the 1970s, many systems have been realized that are conceptionally similar, but this is a beautiful example of feedback at work in a modern environment the VXI(VME)-based, completely digital control system. Both storage rings are longitudinally unstable, but, as if this were not enough trouble, the control system has to handle heavy transient beam loading.
The beam is a more powerful source of induced voltages in the cavities than the power amplifier and “transient” refers in this case to the ion-clearing gap the ring is not homogeneously filled, but the beam current changes drastically during a single turn. Many interwoven control loops are necessary to handle this nightmare situation, the fastest having a group delay of less than 500 ns, 150 of which are contributed by the klystron alone. Another loop deploys digital comb filters. Its signal is fed back exactly one turn later to control the longitudinal vibration of particles in exactly the same bunch where they were detected.
Since the control systems have to have such a hard grip on the beam, some amplifiers risk saturation. This is prevented by learning algorithms for the creation of reference signals. In addition, a fibre-optic system distributes the signal of yet another control loop longitudinal multibunch feedback to the different RF stations around the rings. Network analysers are integrated in the part of the control (see THBL1).
Erk Jensen.
