ELECTRON CLOUD STUDIES AND BEAM SCRUBBING EFFECT IN THE SPS

Transcription

1 EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH European Laboratory for Particle Physics Large Hadron Collider Project LHC Project Report 4 ELECTRON CLOUD STUDIES AND BEAM SCRUBBING EFFECT IN THE SPS J.M. Jimenez, G. Arduini, P. Collier, G. Ferioli, B. Henrist, N. Hilleret, L. Jensen, J-M. Laurent, K. Weiss, F. Zimmermann Abstract The performance of the SPS as LHC injector has been limited, with LHC-type beams, by the electron cloud multipacting for high bunch intensities (>.x protons/bunch). To study in detail the dependence of the electron cloud on the beam parameters (bunch intensity, filling pattern, filling factor ), three strip-detectors have been installed inside dipole corrector magnets. The measuring set-up was completed by shielded pick-ups and by an in-situ secondary electron yield detector installed in field-free regions. This paper presents the variation of the electron cloud activity as a function of the bunch intensity and bunch length, filling pattern and filling factor. The spatial and energy distributions of the electrons in the cloud were studied and the latter compared with results obtained with a conventional retarding field detector. Evidence of the scrubbing effect as a solution to reduce the electron cloud activity is shown in the SPS. Heat load estimations based on the electron cloud intensity and on their energy distributions are given as a function of the bunch intensity and filling factor. The results can be compared with independent direct calorimetric measurements. The implications for the LHC are discussed, in particular heat load extrapolations and a possible start up scenario. To be published in Phys. Rev. Special Topics CERN CH - 2 Geneva 2 Switzerland Geneva, 24 April 2

2 ELECTRON CLOUD STUDIES AND BEAM SCRUBBING EFFECT IN THE SPS J.M. Jimenez, G. Arduini, P. Collier, G. Ferioli, B. Henrist, N. Hilleret, L. Jensen, J-M. Laurent, K. Weiss, F. Zimmermann CERN, Geneva, Switzerland Abstract The performance of the SPS as LHC injector has been limited, with LHC-Type beams, by the electron cloud multipacting for high bunch intensities (>.x protons/bunch). To study in detail the dependence of the electron cloud on the beam parameters (bunch intensity, filling pattern, filling factor ), three strip-detectors have been installed inside dipole corrector magnets. The measuring set-up was completed by shielded pick-ups and by an in-situ secondary electron yield detector installed in field-free regions. This paper presents the variation of the electron cloud activity as a function of the bunch intensity and bunch length, filling pattern and filling factor. The spatial and energy distributions of the electrons in the cloud were studied and the latter compared with results obtained with a conventional retarding field detector. Evidence of the scrubbing effect as a solution to reduce the electron cloud activity is shown in the SPS. Heat load estimations based on the electron cloud intensity and on their energy distributions are given as a function of the bunch intensity and filling factor. The results can be compared with independent direct calorimetric measurements. The implications for the LHC are discussed, in particular heat load extrapolations and a possible start up scenario. STRIP DETECTORS DESCRIPTION Since 998, the electron cloud activity is being studied in the SPS using the pressure gauges and shielded pickups [][2][]. During the shutdown of 2-2, three different versions of strip-detectors (Fig.) have been installed to study independently or simultaneously the spatial distribution and the energy distributions of the electrons in the cloud. In presence of LHC-type beams with high bunch intensities, the electron multipacting takes place in the stainless steel vacuum chamber creating an electron cloud. The collecting copper strips are placed under vacuum and separated from the cloud by a grid which transparency was chosen (7%) to reduce the perturbations to the electron cloud itself by an excessive collection of electrons. The distribution of the holes (Ø 2 mm) has been calculated in order to have a similar transparency over the different channels. The three versions of the strip-detectors installed in the SPS machine were designed for specific measurements: -4 batches of 72 bunches injected from the PS machine, 2 ns bunch spacing, 22 ns batch spacing,.x protons/bunch (.7x ultimate) and 4ns bunch length (4σ) at injection energy (2 GeV). the spatial distribution of the electrons with a resolution of.2 mm is measured with a -channels stripdetector 2, a strip pick-up detector for the energy distributions and a retarding field strip 4 detector to study simultaneously the energy and the spatial distributions of the electrons. Beam B field MBA chamber Holes (transparency 7%) Copper strips Fig.: Schematic view of a strip-detector showing the position of the collecting strips with respect to the beam. In both the strip pick-up and the retarding field stripdetector, three filtering grids placed between the beam pipe and the collecting strips allow the energy distribution measurements. The distribution of the holes of the filtering grids is identical to the holes distribution in the beam pipe wall. The two first grids allow to apply a varying filtering voltage potential, the rd grid close the collecting strips avoids that the secondary electrons generated by the primary electrons coming from the cloud, escape from the detector. The readout electronics (Fig.2) of the strip detectors, based on VME system, include a CPU used to control the whole, a timing unit to synchronise the electronics with the beam cycle and five acquisition cards with 8 channels each. The analogue signals issued from the collecting strips of the detectors are integrated with an interval time which can be adjusted from 2 to 2 ms. The revolution time in the SPS being of 2 µs, the fastest acquisition rate (2 ms) will integrate the signals over 87± turns. At the end of each interval time or integrated time, the outputs of the integrators on each analogue card are multiplexed, amplified if necessary (gain 2), digitised by an 8-bits ADC and stored in a 2 Kbytes Ram memory. Several consecutive intervals are stored in the memories during an acquisition cycle. At the end of the cycle, a CPU reads and transfers the data for treatment and display via an Ethernet connection. 2 width of 4 mm, 2 mm in length rectangular strips (x2mm), square (xmm) 4 width of 4 mm, 2 mm in length, -channels, 2. mm resolution

3 T int.= ms C= nf Strips detector Control Strip Ch Integ Ch 2 Ch Ch 4 AMP ADC RAM Ch MPX C= nf Ch Ch 7 Strip 8 Ch 8 Integ 8 Analog card Strip 2 Integ 42-4 MPX AMP ADC RAM Strip 4 Analog card Fig.2: Strip detector readout electronics CPU Timing Unit Similarly, a decrease of the bunch length by % doubles the electron cloud intensity (Fig.) and a rd central strip appears already at.x p/bunch instead of.x p/bunch for the nominal bunch length (4 ns). The simulations already for 8.x p/bunch predicted the appearance of the rd central strip. The rd central strip disappears rapidly and an attempt of explanation will be given in 2.. Conversely, if the bunch length is increased by %, the electron cloud disappeared (Fig.)..2E-.E- N e- (A/m 2 ) 2.E-2 2.E-2.E-2.E-2.E-.E s/div Fig.: Electron cloud build up signal measured using the -channels strip-detector in field-free conditions. 8.E-4.E-4 4.E-4 2.E-4.E Fig.4: Electron cloud build up signal measured using the -channels strip-detector in a dipole field at the start of the run with a 4 batch-injections. 2 s/div 2 ELECTRON CLOUD OBSERVATIONS AND EFFECT OF THE BEAM PARAMETERS The electron cloud multipacting is driven mainly by the beam parameters and by the secondary electron yield (δ max, Energy of the maximum of the secondary electron yield) of the wall surface. Immediately after the air venting, the threshold of the electron cloud, measured with a single batch-injection, was.x p/bunch in the dipole field regions and.x p/bunch for the field-free. In a dipole field, the appearance threshold of the two lateral strips was.x p/bunch and at.x p/bunch, a rd central strip appeared as predicted by the simulations [4][]. Fig. and 4 show respectively the electron cloud build up in a field-free and in a dipole field region with up to 4 batch-injections and.x p/bunch. The non-flat distribution observed on the field-free signal is due to the angular acceptance of the detectors and to the rectangular shape of the vacuum chamber. 2. Electron Cloud Activity 2.. Bunch intensity and bunch length The electron cloud build up is strongly dependent on the beam potential (bunch intensity and length) which determines the kick to the electrons. As expected, the measurements showed an increase of the electron cloud intensity by a factor of 7 to when the bunch intensity is doubled from.x to.x p/bunch. Fig.: The electron cloud is enhanced by a bunch length decrease (left: -%) and disappears with an increase of the bunch length (right: +%) Filling pattern and filling factor To optimise the injection in the LHC, or 4 batches will be accumulated in the SPS before being injected into the LHC. The measurements made using the pick-ups confirmed that the electron cloud build up during the passage of the 2 nd, the rd and 4 th batch is enhanced by the passage of the preceding batches (Fig.). If the lifetime of the electrons is higher than the 22 ns batch spacing, the surviving electrons created during the previous batch passage enhance the build up during the passage of the following batches. But after the 4 th batch passage and due to the revolution time in the SPS, i.e. 2 µs, the st batch will pass once again after 4. µs. This delay appears to be long enough to loose all the surviving electrons. Fig.7 shows the electron cloud signal as a function of the number of batches injected in the SPS. With batches injected, the total number of electrons is times bigger in the dipole field.x - A/m (~2x e - /m) than in the field-free.x -4 A/m (.x e - /m). The pressure rises in the SPS ring confirmed the larger electron cloud activity in the dipole field regions (arcs). Fig.8 shows a linear increase of the electron cloud activity with the 2

4 number of batches injected with a rate of increase larger in the absence of a magnetic field: 2. instead of in a field-free region. Beam signal e - cloud signal e - cloud signal Beam signal Fig.: Pick-up signals showing the build up during the passage of the 2 nd, the rd and 4 th batch enhanced by the passage of the preceding batches (left). ns batch spacing is not long enough to decouple the build up during the passage of two successive batches (right). N e - (A/m) Dipole field 2.E- 2.E-.E-.E-.E-4 e - cloud intensity in a Dipole field batch 2 batches batches e - cloud intensity in a field-free 4 batches.e-4 4.E-4.E-4 2.E-4.E-4.E+.E Time (ms) Fig.7: Electron cloud build up as a function of the number of batches injected in the SPS. N e - (A/m) Field-free through the slots was accepted and is now in the LHC baseline [7]. During the scrubbing runs (addressed in ), the electron cloud threshold in a dipole field (single batchinjection), increased from.x p/bunch to.x p/bunch. It was expected that the threshold of appearance of the two lateral strips would not change since the position of the two lateral strips depends weakly on the energy of the δ max which decrease from 24 down to 22 ev. The measurements confirmed that the threshold of appearance of the two lateral strips stayed constant, the two lateral strips were visible with batch-injections (Fig.) and their position is similar to the one before the scrubbing (Fig.4) even if the amount of electrons decreased by a factor. d (mm) between lateral strips - -. mm 9 mm Beam axis 2 22 Pumping holes in the beam screen -.E+ 2.E+ 4.E+.E+ 8.E+.E+.2E+.4E+ Bunch intensity (p/bunch) Fig.9: Distance between the two lateral strips as a function of the bunch intensity. 2 2 Ratio to batch slope = x +/-.2 slope = 2. x +/-.2 Build up in field-free Build up in dipole fields electrons Fig.: Picture of the beam screen showing the position of the pumping slots and a schematic view of the slot shielding design to intercept the electrons passing through the slots. (Courtesy of N. Kos) 2 4 Number of batches Fig.8: Enhancement factor due to the number of batches injected in the SPS in a dipole and in a field-free condition. 2.2 Spatial distribution of the electrons The spatial distribution of the electrons in the cloud has been confirmed using a -channels strip-detector which allowed twice the resolution obtained in 2 (Fig.9) and recent simulations fit fairly well the measurements []. At.x and.x p/bunch, the two vertical lateral strips of the cloud will stand on top of the beam screen pumping slots (9 and. mm respectively from the centre) as shown in Fig.9, inducing additional heat load to the cold bore. The decision to insert the pumping slot shielding (Fig.) to intercept the electrons passing.2e-.e- 8.E-4.E-4 4.E-4 2.E-4.E Fig.: Distance between the 2 lateral strips after days of LHC-type beams in a dipole field and.x p/bunch. The effect of a dipole field applied on the detector on the electron cloud appears to be more important below 2 s/div

5 Gauss. In these field conditions, both the electron cloud intensity and the width of the lateral strips increase by 2 and 2 respectively. This dependence at low field is not yet understood. Energy acceptance considerations could be one explanation since at low field, the Lamor radius of the electrons is close to the diameter of the holes of the grid. Below 2 Gauss, the electron cloud disappears if the bunch intensity is below the electron cloud threshold in field-free regions. The influence of the magnetic field is even more spectacular when a transition through Gauss field is applied to the magnet and measured with the stripdetector (Fig.2). The behaviour of the electron cloud at these small dipole field and the possible interferences with the acceptance of the strip-detectors is being investigated. and on the pressure rises induced by the electron stimulated desorption (ESD) mechanism. The energy distributions were measured using two different protocols: in a multicycle mode where a different filtering voltage is applied for each SPS supercycle (~2 s duration) or in a single-cycle mode, where the measurements are made within the same cycle by applying an exponential voltage decay to the filtering grids shortly after the last injection. dn/de (a.u.) Strip pick-up (multicycle measurement) d 2 N/dxdE Strip-detector (single-cycle d2n/dxde meas.-fit) Strip detector Strip pick-up (single-cycle measurement).e-2 4.E-2.E-2 2.E-2.E-2.4 s/div Energy (ev) Fig.4: Electrons energy distribution (dn/de) measured in a dipole field using the strip pick-ups detector in a single and multiple cycle mode and using the retarding field strip-detector..e Fig.2: Spatial distribution of the electrons in the cloud during a magnetic field transition through Gauss..E+.E- Ffree batch Ffree 2 batches Ffree batches Dfree batch Dfree 2 batches Dfree batches Strip pick-up Distribution (a.u.) Field free - Strip pick-ups Field free - Retarding Field Detector (fit) RFD Distribution (a.u.).e-2.e-.e Fig.: Evolution of the spatial distribution of the electrons with the number of batches injected in a dipole field (Dfield) or field-free (Ffree) regions. The correspond to the beam position which was slightly offcentred. Fig. gives the variation of the spatial distribution of the electrons with the number of batches injected and with the presence or not of a dipole field. 2. Energy distribution of the electrons The energy distribution of the electrons in the cloud was measured using the strip-detectors in both the dipole field and field-free conditions and using a conventional retarding field detector in a field-free region. Due to hardware limitations, the electrons with energies below 2 ev could not be measured. However, the impact of this limitation is small (<2%) both on the heat load budget Electron Energy (ev) Fig.: Electrons energy distribution (dn/de) measured in a field-free region using the strip pick-ups in a singlecycle mode and using a conventional retarding field detector (RFD) detector. In the dipole field where most of the electrons are trapped in two vertical lateral strips by the magnetic field, the energy distribution showed a peak of energy between 8 and 2 ev (Fig.4) to be compared with the 8 ev measured in the field-free regions (Fig.). The energy distributions measured by the strip pick-ups are in good agreement with the results given by a conventional retarding field detector and by the recent simulations []. The simultaneous energy and spatial distribution studies showed that most of the high-energy electrons i.e. above 2 ev, are located in the central strip (Fig.). The two lateral strips have electrons with energies below 8 ev. This last observation could explain why the central strip tends to disappear after several hours of LHC-type beam. In fact, the decrease of the secondary electron yield is enhanced by the amount of primary electrons. After a given dose, the amount of electrons produced, i.e. δ SEY above the multipacting threshold is reached earlier for the 4

6 electrons with energies above 8 ev thus causing the extinction of the cloud in the central area. The decrease of the SEY will be addressed in. N e - (a.u.) V - V (exp. Decrease) -8 V - V -7 V - V Filtering potential Fig.: Spatial energy distribution (d 2 N/dxdE) of the electrons measured in a single-cycle mode using the retarding field strip-detector. EVIDENCE OF A SCRUBBING EFFECT IN PRESENCE OF LHC TYPE BEAMS The two scrubbing periods, days of beam in total, gave evidence of a scrubbing effect. All around the SPS ring, both in the field-free and dipole field regions, the pressures decreased by in 4 days, 4 in days (Fig.7). Meanwhile, the thresholds measured with the strip-detector increased in the dipole field, from.x to.x p/bunch and from.x to more than.x p/bunch in the field-free regions. No signal could be detected in the field-free regions after days of LHCtype beams with a 4 batch-injection. The Fig.8 shows the increase of the electron cloud threshold measured with pressure gauges in a field-free region, the behaviour was similar in a dipole field. This threshold increase is also a clear signature of the scrubbing. P / I batch.e-7.e-8.e-9.e-2.e-2 Dipole field (HP) Field free (HP8) Bunch Intensity (x E).E Cumul. time (hours) Fig.7: Pressure decreases both in dipole and field-free regions with the LHC-type beam exposure. The pressures are normalised to the batch intensity (P/I bunch x72xn bacthes ). Similarly, the evolution of the electron cloud activity measured by the strip-detectors throughout the cleaning process showed a decrease by a factor 2 in days in a field-free region (Fig.9) and between to in the dipole field regions (Fig.2). Several parameters measured (pressures, electron cloud intensity, heat load measured by the calorimeters [8][9]) showed that the beam-induced multipacting and thus the 9 - V 2 I Bunch ( p) scrubbing stopped after 4 days in the field-free regions. This observation is consistent with the in situ measurement of the secondary electron yield (SEY), which remained constant after 4 days; the δ max was., also consistent with analytical calculations []. After short periods without LHC-type beams, the SEY drifted up i.e. from. to.7 in two weeks time. However, the initial value was recovered after 4 hours with nominal intensity LHC-type beams with at least batches injected. P/P 2 8.E+7 2 days.e+8 4 days.2e+9 days E+ 2.E+ 4.E+.E+ 8.E+.E+.2E+.4E+ Proton bunch intensity Fig.8: Increase of the electron cloud threshold in fieldfree regions with beam exposure. Single batch-injection and.x p/bunch. N e- (A/m).E-2.E-.E-4.E Cumul. time (hours) Fig.9: Evolution of the electron cloud activity measured by the strip-detectors throughout the scrubbing process in field-free conditions showing a decrease by a factor 2 in days. After a venting to air to reset the detectors and beam pipe surfaces, the scrubbing run was repeated to confirm the results of the st scrubbing run (May 22). Fig.2 confirmed the scrubbing efficiency in both the dipole and field-free regions. The step between the first and second part of the chart appeared after the introduction of a ramp in energy ( GeV). Simultanously, the strip detector showed a lateral displacement of the cloud resulting from a beam orbit displacement. The observed pressure rises are easily explained by the bombardement of nonscrubbed surfaces. The introduction on purpose of a small orbit displacement, to 4 mm with respect to the nominal orbit, produced as predicted, an increase of the pressures ( P/P) by to depending on the chamber shape and position in the SPS ring.

7 N e - (A/m).E-.E-4.E-.E- batch 2 Batches Batches 4 Batches Hours of beam Fig.2: Evolution of the electron cloud activity measured by the strip-detectors throughout the cleaning process in a dipole field conditions..e- Dipole field (HP4) Field free (HP4) nominal bunch intensity; the beam dump is adjusted to stay just below the pressures interlock levels. Subsequently, the delay of the beam dump should be increased to stay continuously just below the pressure interlock levels. As soon as possible, increase the number of batches injected ( or 4) or alternatively, increase the bunch intensity and/or shorten the bunches. Finally, introduce a ramp in energy. If not allowed (cost of operation), alternatively introduce local orbit bumps to create orbit displacements to make a wider scrubbing of the inner walls of the bending magnet vacuum chambers. Appearance of the rd central strip Ramp.E-2 4.E-2.E-2 2.E-2.E-2.E+ N e - (A/m 2 ) Pressures (mbar).e-7.e-8.e-9.e+2 2.E+22 4.E+22.E+22 8.E+22.E+2.2E+2.4E+2 Cumulated dose (protons) Fig.2: Pressure decreases both in dipole and field-free regions with the LHC-type beam exposure. Pressure bump after the introduction of the ramp to GeV. However, the decreases both in dipole and field-free regions confirmed the scrubbing effect SEY coefficient 2.4 /9/2 2: 2.2 8/9/2 4:2 2. 8/9/2 9: 9/9/2 :.8 2/9/2 9:4. //2 8: Electron energy (ev) Fig.22: Evolution of the Secondary Electron Yield (δ) measured in situ with the LHC-type beam exposure. During the 2 nd scrubbing period (Fig.22), the δ max decreased faster indicating a memory effect. The ultimate value of. was achieved by injecting a 4 th batch. This result is consistent with the absence of multipacting with batch-injections or less, in the field-free regions where the SEY detector is installed. During the scrubbing, the multipacting threshold (.) is reached earlier for the electrons with energies above 8 ev and this could explain why the rd central strip disappears prematurely as compared to the lateral strips. The two scrubbing runs allowed establishing an optimised scrubbing run scenario for the SPS. The machine should start with or 2 batches injected at Lateral position (mm) s /Division Fig.2: Small orbit displacement seen by the stripdetectors due to the ramp to GeV after the 4 th batchinjection Σ N e - (A/m) 2 GeV lateral strips (A/m 2 ) 4 GeV central strip (A/m 2 ) 4.E- 4.E-.E-.E- 2.E- 2.E-.E-.E-.E-4..E Time (ms) Fig.24: Variation of the electron cloud signal with the number of batches injected and with the ramp in energy (4 GeV). The dark line gives the intensity in A/m 4. EFFECT OF THE RAMP IN ENERGY At the end of the st scrubbing period, a small ramp in energy was introduced after the 4 th injection; the proton energy was increased up to GeV to check whether or not the scrubbing run should be made at injection energy (costs of operation) or if unexpected enhancement effects will imply going to higher energies. The effect of this small ramp was larger than expected since it induced a small displacement of the orbit (Fig.2) producing pressure rises all around the SPS machine (Fig.2) and a doubling of the electron cloud intensity measured by the strip-detectors (Fig.24). A squeezing of the bunch during the ramp, which implies an increase of the beam potential, and therefore an increase of the electron energies could easily explain the increase of the electron cloud activity. As expected from 2.., the two lateral strips became wider and more N e - (A/m)

8 intense and rd central strip, which disappeared after the scrubbing run, appeared again. Similarly, the pressure variations during the ramp indicate also an increase of the electron cloud activity during the ramp, the dipole field regions being more sensitive than the field-free regions (Fig.2). Despite the negative effect of the ramp and a scrubbing run made mainly with 2 batch-injections, the electron cloud intensity was reduced by a factor of 2 after the 2 nd scrubbing run (Fig.2). P/P HP HP7 HP2 Beam Energy (GeV) Time (s) Fig.2: Pressure variations in dipole (HP, HP2) and field-free (HP7) regions during the ramp in energy up to 4 GeV. N e- (A/m) in a dipole field 2.E- 2.E-.E-.E-.E-4 Before scrubbing run (7) After days of scrubbing, ramp GeV (77) After scrubbing run, ramp 4 GeV (78) Energy (GeV).E Time (ms) Fig.2: Evolution of the electron cloud activity before and after the 2 nd scrubbing run and with a ramp in energy at 4 GeV. HEAT LOADS - SCALING TO LHC The heat load due to the electron cloud was measured directly using two different types of calorimeters: pick-up calorimeters [8] and standard calorimeters with a close geometry [9]. In this paper, only the heat loads calculated from the electron cloud intensity and from the energy distribution of the electrons are presented. This approach is very useful since the strip-detectors provide a fast response (a few seconds) as compared to the close geometry calorimeters, which have an intrinsic time constant of a few hours. The pick-up calorimeters are an intermediate solution with a time constant of 2 minutes. But due to their overall size, these pick ups cannot be installed in dipole field regions. As stated earlier ( 2.), the energy distribution of the electrons below 2 ev could not be measured. The Beam Energy (GeV) Beam energy (GeV) contribution of the low energy electrons (<2 ev) on the heat load was measured by applying a bias voltage (-V) on the collecting plates of the pick up calorimeters to repel the low energy electrons and their measured contribution represents, as expected, about % of the total heat load measured. In the field-free regions, the heat load measured by the pick-up calorimeters decreased by a factor of 2 after 4 days and then remained constant due to the extinction of the electron cloud in the field-free regions. The calculated heat load (strip-detectors) before the scrubbing run and with a single-batch injection was around.4 W/m, and it decreased down to mw/m after days. The estimated heat load with batches injected is mw/m. N electrons (A/m) Field free Ratio to batch Heat load (mw/m) N electrons (A/m) Dipole field Ratio to batch Heat load (mw/m) st scrubbing run (May2), days of LHC-type beams before scrubbing.8e- 9 not measured to "protect" the batch after scrubbing 2.E- 8 strip detector from an expected After the st scrubbing run (May 22) and an air exposure during 4 hours batch.4e- 2.4E batches.8e E N e- batches.e E batches.9e nd scrubbing run (September 22), 4 days of LHC-type beams before 2 nd scrubbing 2.4E-4 42 Batch after 2 nd scrubbing 2.E before 2 nd scrubbing 4.E Batches after 2 nd scrubbing 4.E-. 24 before 2 nd scrubbing 8.E-4 47 Batches after 2 nd scrubbing.e and with the ramp at 4 GeV N e- batch 4.4E- 2 2 batches 9.E- 2.2 batches 4.E batches.2e batches + ramp.e Table : Heat loads calculated from both the electron could intensity (N electrons ) and the electron energy distribution in dipole and field-free regions. In the dipole field regions, the heat load could not be measured at the start up after the shutdown due to the saturation of the acquisition cards. Before the 2 nd scrubbing run and after resetting the strip-detectors by an air exposure, the heat load was measured with 4 batches injected and was around. W/m. This value decreased down to 7 mw/m after 4 days. The ramp to or 4 GeV doubled the electron cloud activity in the dipole field regions and the heat load estimated with the ramp was mw/m. No effect could be measured in the field-free regions with up to 4 batchinjections. Results are summarised in Table. Field-free N electrons Heat load (mw/m) Dipole field N electrons Heat load (mw/m) After scrubbing runs Filling factor or 4 / (SPS).4E- 4.4E E-4 2.2E-4 2 Number of.e E-4 27 batches E ramp - -.E- 92 Filling factor (LHC) No ramp.e- 4 2.E- 84 Ramp at 4 GeV - -.2E- 894 Table 2: Electron cloud activities and calculated heat loads scaled to the LHC filling factor. HP = Half Period (in the SPS ring) 7

9 .2 Scaling to the LHC accelerator Experiments have been made to measure the dependence of the electron cloud activity on the filling factor. At nominal intensity, the calculated heat loads assuming the correction factors in the dipole and fieldfree regions introduced by the different filling factors between the SPS and the LHC, are given in Table 2. The scaling made in Table 2 assumes that the electron cloud build up will increase linearly with the number of batches injected which is thought to be a pessimistic hypothesis mainly due to the space charge limitations. CONCLUSIONS As the LHC injector, the scrubbing runs confirmed that after days of scrubbing, the SPS should be able to inject or 4 batches at nominal intensity into the LHC. However, the electron cloud is still visible with 4 batches injected at nominal intensity in the arcs (dipole field) of the SPS after about hours of LHC-type beams, both as pressure rises and by the electron signals collected by the strip-detectors. In the field-free regions, i.e. in the long straight sections, the electron cloud activity decreased below the detection level of the stripdetectors (< -9 A/m). This observation is not in contradiction with the scrubbing efficiency since most of the LHC beam time used for the scrubbing was with or 2 batches injected with bunch intensities close to the electron cloud threshold after 4 days of scrubbing. As the scrubbing efficiency depends on the amount and energy of the impinging electrons (bunch intensity, bunch length, number of batches) and similarly to RF and HV devices, the scrubbing efficiency is limited by the parameters used during the scrubbing. The scrubbing will not be effective if running conditions are more favourable for the electron cloud build up. The ultimate SEY value reached after 4 days in the field-free regions in the SPS, δ=., shows the multipacting threshold under these conditions, consistent with recent analytical calculations []. The increase of the SEY (δ) observed when the SPS is not operated with LHC-type beams is not an issue since several measurements showed that the SEY recovered its initial value (before drifting) after 4 hours of operation with LHC-type beams above the electron cloud thresholds. The measurements made with a ramp in energy ( or 4 GeV) showed that additional measurements are required to quantify the effect of the bunch length shortening and of the beam orbit displacements expected during the ramp. Preliminary measurements gave an increase by a factor of 2 when shortening the bunch (%) and also a factor of 2 due to the ramp. Therefore, the bunch length variations could explain part of the increase observed during the ramp. For the LHC, a major concern is the excessive heat load due to the electron cloud. Data have been collected and are being compared with the predictions from simulations. A tentative of scaling to the LHC has been made in Table 2, estimations vary between. and.9 W/m in presence or not of a ramp in energy. However, these values must be balanced against the limited amount of beam time with 4 batches injected, an additional decrease ( ) is expected if doing the scrubbing run with 4 batches injected. The LHC situation is also expected to be more complicated since most of the machine is at a cryogenic temperature and therefore the gasses cryosorbed on the inner surfaces of the vacuum chambers could modify the present picture obtained at room temperature. The scrubbing is not yet confirmed on cold surfaces where the efficiency of the impinging electrons could be reduced due to the energy reduction introduced by the gasses cryosorbed on the surfaces. The different filling schemes in the two machines also complicate the extrapolation from the SPS situation to the LHC. In the SPS and with 4 batches injected, the surviving electrons are though to be lost in the 4. µs between the 4 th batch passage and a new passage of the st batch. In the LHC [], the situation could be less favourable since the ring will be full of batches and the maximum spacing between batches will not exceed µs, required by the rise time of the LHC dump kickers. If the electrons from the cloud survive these gaps, the build up may be significantly enhanced. Finally, even if the measurements made in the SPS at room temperature and in the future at cryogenic temperature are difficult to extrapolate to the LHC, these results are important both qualitatively and quantitatively to benchmark the simulations. The latest comparison showed that the simulations are qualitatively in good agreement with the SPS measurements. The appearance of two lateral strips in a dipole field above.x p/bunch and of a rd central strip at higher bunch intensities were predicted and are consistent with the strip-detector measurements. 7 REFERENCES [] J.M. Jimenez and al., Workshop Chamonix XI (2) [2] J.M. Jimenez and al., PAC, CHICAGO (2) [] J.M. Jimenez et al., LHC Project Report 2(2) [4] F. Zimermann and al., Workshop Chamonix X, (2) [] F. Zimmermann and al., Mini-Workshop CERN 22 [] F. Zimmermann, Workshop Chamonix XII (22) [7] N. Kos, EDMS 288 (CERN) [8] N. Hilleret and al, Scrubbing days CERN 22 [9] V. Baglin, Scrubbing days CERN 22 [] L. Vos, CERN (to be published) [] O. Brunning, Minutes of the 9 th PLC9 (CERN) ACKNOWLEDGMENTS K. Cornelis, D. Manglunki, SPS and PS Operators, T. Bohl, P. Baudrenghien, W. Hofle, J. Arnold 2, J-C. Billy 2, G. Mathis 2, J. Ramillon, N. Munda, G. Favre, D. Valero, M. Blanc and the Main Workshop. AB, 2 AT and EST Divisions, CERN. 8