1.1 Electron-Cloud Effects in the LHC

Transcription

1 Electron-Cloud Effects in the LHC F. Zimmermann, E. Benedetto 1 mail to: frank.zimmermann@cern.ch CERN, AB Department, ABP Group 1211 Geneva 23, Switzerland Introduction The LHC is the first proton accelerator for which synchrotron radiation becomes noticeable. At a beam energy of 7 TeV, the relativistic γ factor is comparable to that of electron or positron beams in the B factories or at many light sources. This means that the same number of synchrotron-radiation photons are emitted per proton and turn. The critical photon energy of about 44 ev in the LHC is near the energy of maximum photoemission yield for many materials. Therefore, a significant electron cloud can be expected from synchrotron radiation and photo-emission alone. The possibility that beam-induced multipacting may also occur in the LHC was suggested in 1996 [Grobner96]. Since 1997, electron-cloud build up in the LHC arcs due to both photoemission and beam-induced multipacting were predicted and studied in simulations [Zimmermann97,Grobner97,Bruning98]. Electron cloud effects in the LHC were reviewed previously [Rumolo01,Arduini03] Experiments in the LHC Injectors Experiments with LHC type beam in the CERN SPS and PS, which serve as LHC injector and pre-injector, respectively, have indeed revealed the rapid build up of an electron cloud by beam-induced multipacting, even without any contribution from synchrotron radiation at the low beam energy of 26 GeV. At the nominal LHC bunch spacing of 25 ns, the multipacting is observed for bunch populations above 3x10 10 protons per bunch at the start of a run. The threshold increases to protons per bunch after 10 days of scrubbing (continuous operation with LHC beam at the maximum possible intensity and duty cycle permitted by electron-induced pressure rise). In the SPS the two main effects of the electron cloud are a pressure increase by several orders of magnitude [Jimenez03] and beam instabilities that can lead to emittance growth and even beam loss (coupled bunch instability in the horizontal plane and single-bunch instability in the vertical plane) [Arduini01b,Cornelis02]. Degradation of BPM signals or feedback pick-ups due to electron bombardment were also observed; these could be partially cured by solenoid windings or by data processing at higher frequencies [Hofle01]. Since about 2000, a large number of detectors were installed in the SPS to benchmark the electron-cloud simulations and to explore possible countermeasures. Promising results were achieved. In particular, vacuum chambers coated with the newly developed TiZrV getter material [Benvenuti01] showed no sign of multipacting, which 1 Also INFM and Dip. di Energetica, Politecnico di Torino, Italy

2 12 suggests that the solution adopted for the warm parts of the LHC, about 10% of the circumference, will work fine. Also a fast surface conditioning by scrubbing was demonstrated in the SPS arcs. After 1 or 2 weeks of scrubbing the electron cloud did no longer limit the SPS operation with LHC beam. In situ measurements confirmed a considerable reduction of the maximum secondary emission yield, decreasing from about 2.0 to 1.5 over the same time period. However, measurements with two cold chambers in the SPS have shown a much slower scrubbing; see, e.g., [Baglin03]. This could be due to the fact that the cold sections are too short and influenced by gas influx from the adjacent warm vacuum chambers. A large number of gas molecules cryosorbed on the cold surface could lead to an enhanced secondary emission yield. In the laboratory, cold surfaces did show a conditioning similar to that of warm samples [Cimino03] Predictions for the LHC A primary concern for the LHC is the additional heat load deposited by the electron cloud on the beam screen (a Cu-coated stainless steel shield inserted into the arc vacuum chamber, which absorbs the photons from synchrotron radiation). Only a limited cooling capacity is available for the additional heat load due to the electron cloud. If it is surpassed, a quench of the superconducting magnets would result. Figure 1 shows the heat load per unit length simulated for an LHC arc cell under various conditions. Each heat-load value was computed as a weighted average of three independent simulations for dipoles, field-free regions, and quadrupoles, according to the cell fraction covered by each type of field (for sextupoles we assumed the same heat load as for quadrupoles). The different curves refer to different values of the maximum secondary emission yield, ranging from 1.1 to 1.7, and to different numbers of successive bunch trains. The available cooling capacity for the electron cloud is also indicated. It decreases towards higher intensity, since the cooling required for synchrotron radiation, image currents, and gas scattering increases. The latter process seems to be dominant: It presently appears that, due to the gas scattering, at the ultimate bunch intensity of 1.67x10 11 almost no cooling capacity might be left for the electron cloud [Tavian04]. However, no final conclusion has yet been reached on this point. Figure 1: Simulated average arc heat load per unit length as a function of bunch intensity at injection (left) and at top energy (right) for various values of the maximum secondary emission yield and for a variable number of bunch trains, and the available cooling capacity.

3 13 The LHC beam consists of batches of 72 bunches with 25-ns bunch spacing, which are separated by gaps of 225 ns. At injection energy, the multipacting process is launched by residual-gas ionization, and the electron build up saturates only at the end of the first or during the second batch. As a result, the simulated heat load depends on the number of batches. At top energy, photoelectrons are abundant and the electron density saturates already after a few bunches of the 1 st batch, so that in this case the heat load is rather insensitive to the number of batches. The left picture of Fig. 1 suggests a resonance with enhanced heat load for bunch populations around 6x10 10 protons, visible for the lower values of secondary emission yield. This picture also shows that with a maximum secondary emission yield of 1.3 it is possible to reach or exceed the nominal bunch intensity of 1.15x10 11 at injection. On the other hand, a maximum emission yield below 1.1 is needed at top energy (right picture). Before the required low values of the secondary emission yield are achieved by surface scrubbing, the LHC could be operated with a reduced charge per bunch (equal or below 5x10 10 protons) or with an increased spacing between bunches. Due to the asymmetric arrangement of the collision points, a strict 50-ns bunch spacing (twice the nominal) would yield zero luminosity at one of the collision points (LHCb). Therefore, 75-ns is a more agreeable value. Simulated heat loads for 75-ns bunch spacing are compared with those for 25-ns spacing in Fig. 2, where we consider a single batch and the nominal bunch population of 1.15x The two pictures again refer to injection and to top energy. Figure 2 shows that with 75-ns spacing, any realistic value of δ max can be accommodated, up to δ max =2.0 or beyond. Higher luminosity would be achieved with 50-ns spacing. For example, if nominal bunches at 50-ns spacing were interleaved with low-charge satellites at 25 ns separation, the desired lower luminosity could be delivered to the LHCb experiment, while the heat load would still be under control. Figure 2: Simulated average arc heat load as a function of the maximum secondary emission yield for bunch spacings of 25 ns and 75 ns at injection (left) and at top energy (right). In addition to the heat load, the electron cloud could introduce other complications for the LHC operation: instabilities, pressure increase, and emittance growth: Instabilities could be of coupled-bunch or single-bunch type, as in the SPS. The pressure rise might be several orders of magnitude, again as experienced in the SPS. Simulations using the HEADTAIL code indicate the possibility of a long-term emittance growth that could be detrimental for a proton storage ring where the beam is to be stored over tens of hours. As an example, the left picture of Fig. 3 shows the simulated electron volume density as a function of the simulated arc heat load, for various scenarios. There is no 1-to-1 relation between electron density and heat load,

4 14 but in general the heat load appears acceptable, if the electron density drops below 5x10 11 m -3. The right picture displays the emittance growth time simulated by HEADTAIL as a function of the electron-cloud density on a double-logarithmic scale. Pessimistically extrapolating the left five points on this plot linearly to larger rise times and lower densities, we estimate that an emittance rise time larger than 30 minutes is reached for a density of about 3x10 10 m -3, with zero chromaticity and without feedback. According to these preliminary results, the acceptable limit on the electron cloud density that is imposed by long-term emittance preservation may be lower than the limit arising from the heat load. Figure 3: Electron volume density simulated by ECLOUD as a function of average-arc heat load at top energy (left) and emittance growth rise time simulated by HEADTAIL as a function of electron volume density at injection (right) Countermeasures The LHC design adopted a number of countermeasures against the electron cloud. Most vacuum chambers in the warm sections of the LHC are coated by a newly developed getter material, TiZrV [Benvenuti01], which has a low maximum secondary emission yield of about 1.1. In the cold arcs, a sawtooth pattern (steps of 35 micron separated by 500 micron) is impressed on the horizontally outward side of the beam screen that forms the inner layer of the vacuum chamber [Collins98]. The sawtooth pattern results in a locally perpendicular impact of synchrotron-radiation photons yielding both a strongly reduced reflectivity and a lower photoemission yield. The reduced reflectivity is important as, in dipole magnets, photoelectrons emitted at the outer side of the chamber are confined and do little harm to the beam, while photoelectrons emitted at the top and bottom of the chamber, via scattered photons, may approach the beam and contribute to multipacting and heat load. The LHC beam screen contains pumping slots at its top and bottom. Multipacting electrons which pass through these slots along the magnetic field lines would hit the cold bore of the magnets at 2 K, where the available cooling capacity is much smaller than at the beam-screen temperature of 4-20 K. To prevent this fatal heat load, pumping-slot shields ( baffles ) were added on the outer side of the beam screen, so as to intercept such electrons, at the expense of a slightly reduced pumping speed [Kos03,Krasnov03]. Heat load on the beam screen and vacuum pressure can be confined to tolerable values by reducing either the number of bunches or the bunch charge. As shown above,

5 15 for a three times increased bunch spacing of 75 ns, no significant heat load from the electron cloud is expected. Alternatively, bunch populations below 5x10 10 at the nominal bunch spacing of 25 ns may also yield an acceptable heat load. In addition, low-charge satellite bunches, following 5 or 10 ns behind the main bunches, could be employed as a fall back option to suppress the electron-cloud build up and to reduce the heat load during commissioning [Ruggiero99]. The surface of the vacuum chamber will be conditioned by operating near the heatload limit for extended periods of time (the scrubbing effect is described in [Hilleret01]). At the LHC this scrubbing will be more difficult than in the SPS, since the electron cloud activity will increase during acceleration, due to additional contributions from synchrotron radiation and the reduced beam sizes. If the beam needs to be dumped, when the heat load approaches the magnet quench limits, the time needed to re-iterate is of the order one hour rather than 20 s as in the SPS. It is expected, that after several weeks or months of operation, the surface conditioning during commissioning and early operation will reduce the secondary emission yield to a level where operation with nominal LHC beam parameters becomes possible Open Questions The simulated heat load strongly depends on the reflection probability of lowenergetic electrons when they hit the chamber wall. Recent measurements and a simple quantum-mechanical calculation suggest that the reflectivity may approach 1 in the limit of zero energy [Cimino04]. This conclusion has not yet been generally accepted in the electron-cloud community. The reflectivity has a great influence on the survival of secondary electrons between bunches and, in particular, during the gaps between bunch trains. The LHC strategy heavily relies on surface conditioning by scrubbing (electron bombardment due to the electron cloud itself). In the SPS experiments, some of the cold and also one warm detector showed little scrubbing, while most of the regular warm stainless steel chambers in the arcs did. The apparent lack of scrubbing for the cold detectors could be explained by the peculiarities of the SPS set up, which consists of short (1 or 2 m long) cold sections bordered on either side by warm parts with significant gas influx. This possibility will be explored in the 2004 SPS run, where heat loads will be measured in a cold detector that is isolated by cryogenic barriers from the rest of the ring. A strong increase in the gas pressure during scrubbing would reduce the beam lifetime and increase the heat load on the cold bore of the magnets due to scattered proton losses. Since already at nominal pressure levels the absorption of scatteredproton energies by the cold bore constitutes a significant load on the LHC cryogenic system, only much lower pressure rises than in the SPS can be tolerated at the LHC. This source of heat load may further complicate the scrubbing process with respect to the SPS, in addition to the reduced duty cycle and to the new effects of synchrotron radiation and photoemission encountered towards top energy. The LHC requires a low value of secondary emission, in order to reach the design parameters for bunch charge and bunch spacing, according to the simulations, one which has not yet been reached in the SPS experiments. A related concern is that low

6 16 energy electrons hitting the wall, if there are many, could amount to a significant heat load, without contributing to surface conditioning [Baglin04]. For the latter a minimum electron energy of about 30 ev is required [Hilleret03]. If instabilities occur in the LHC, one could attempt to suppress them by a combination of bunch-to-bunch feedback and increased chromaticity, as is the case in the SPS. The LHC ring being larger than the SPS, a still higher value of Q might be required to suppress instabilities (at the SPS Q values up to 30 were needed at the start of a scrubbing run), that could adversely affect the dynamic aperture. The deleterious effect of a large Q on the dynamic aperture might also be enhanced by the more complex optics at the LHC, in particular the low-beta insertions. The long-term emittance growth due to the electron cloud is another open issue [Benedetto03]. Recent simulation results, already mentioned above, suggest that the emittance growth in the LHC will be acceptable for small, but achievable average electron densities. Further studies of this topic are ongoing. In collaboration with T. Katsouleas group at USC the continuous interaction of the proton beam and the electron cloud is being modeled by the code QUICKPIC [Rumolo03]. This provides a valuable benchmark for the HEADTAIL code. The latter code concentrates the beamelectron interaction at a few, typically ten, points around the ring, which speeds up the calculation, and allows for a larger number of turns, but it is less accurate than QUICKPIC. Neither simulation has so far considered the effect of varying beta functions around the ring, which might introduce additional emittance dilution [Ohmi03,Rumolo04]. References [Arduni01] G. Arduini et al., Present Understanding of Electron Cloud Effects in the Large Hadron Collider, PAC2003 Portland (2003). [Arduini01b] G. Arduini et al., Transverse Behavior of the LHC Proton Beam in the SPS: An Update, PAC2001 Chicago (2001). [Baglin03] V. Baglin, LHC Project Report 667 (2003). [Baglin04] V. Baglin, private communication (2004). [Benedetto03] E. Benedetto et al., PAC2003 Portland (2003). [Benvenuti01] C. Benvenuti et al., Vacuum 60, 57 (2001). [Bruning98] O. Bruning, EPAC98 Stockholm (1998). [Cimino03] R. Cimino et al., LHC Project Report 669 (2003). [Cimino04] R. Cimino, I.R. Collins, M.A. Furman, M. Pivi, F. Ruggiero, G. Rumolo, F. Zimmermann, Can Low-Energy Electrons Affect High- Energy Physics Accelerators?, Submitted to Phys. Rev. Let. (2004). [Collins98] V. Baglin, I.R. Collins, O. Grobner, EPAC98 Stockholm, p (1998). [Cornelis02] K. Cornelis, The Electron Cloud Instability in the SPS, ECLOUD02 Geneva (2002). [Grobner96] O. Grobner, Technological Problems Related to the Cold Vacuum [Grobner97] System of the LHC, Vacuum 47, p. 591 (1996). O. Grobner, Beam Induced Multipacting, PAC 1997 Vancouver (1997). [Hilleret01] N. Hilleret et al., Proc. EPAC 2000 Vienna, p. 217 (2000). [Hilleret03] N. Hilleret, private communication (2003). [Hofle01] W. Hofle, Progress with the SPS Damper, Proc. Chamonix01 (2001). [Jimenez03] M. Jimenez et al., CERN-LHC Project-Report-632 (2003).

7 17 [Kos03] N. Kos, Pumping Slot Shields for LHC Beam Screens, CERN Vacuum Technical Note (2003). [Krasnov03] A. Krasnov, LHC Project Report 671 (2003). [Ohmi03] K. Ohmi, private communication (2003). [Ruggiero99] F. Ruggiero, X. Zhang, AIP Conf. Proceedings 496, p. 40 (1999). [Rumolo01] G. Rumolo, F. Ruggiero, F. Zimmermann, PRST-AB 4, (2001). [Rumolo03] G. Rumolo, A.Z. Ghalam, T. Katsouleas et al. PRST-AB 6, [Rumolo04] (2003). G. Rumolo et al., Single-Bunch Instabilities Induced by Electron Cloud in Short Positron/Proton/Ion Bunches, this ICFA newsletter. [Tavian04] L. Tavian, private communication (2004). [Zimmermann97] F. Zimmermann, LHC Project Report 95 (1997).