Beam Loss Handling at Tevatron: Simulations and Implementations A.I

BEAM LOSS HANDLING AT TEVATRON: SIMULATIONS AND IMPLEMENTATIONS A.I. Drozhdin and N.V. Mokhov, Fermilab, P.O. Box 500, Batavia, IL 60510 USA

Abstract This causes irradiation of conventional and superconducting (SC) components of the machine, an increase of back- Summary of studies is presented towards minimization of ground rates in the detectors, possible radiation damage, beam loss in the critical locations at the Fermilab Tevatron quench, overheating of equipment and even a total destruc- to reduce background rates in the collider detectors and to tion of some units. A very reliable multi-component beam protect machine components. Based on detailed Monte- collimation system is the main way to handle beam loss and Carlo simulations, measures have been proposed and in- is mandatory at any SC accelerator, providing [4, 5, 6]: corporated in the machine to reduce accelerator-related instantaneous and residual background levels in the DØ and reduction of beam loss in the vicinity of IPs to sustain CDF detectors. Recent measurements are in good agree- favorable experimental conditions; ment with the predictions. A re-alignment of the electro- minimization of radiation impact on personnel and static deflector and the Lambertson magnet and the addi- environment by localizing beam loss in the predeter- tion of shielding in the AØ straight section has resulted mined regions and using appropriate shielding in these in reduction of beam induced energy deposition in the su- regions; perconducting magnets, which allowed an increase in the protection of accelerator components against irradi- extracted beam intensity. Using the same simulation tech- ation caused by operational beam loss and enhance- nique, it has been calculated that the total beam of 10 ment of reliability of the machine; protons can be removed from the Tevatron at the end of prevention of quenching of SC magnets and protec- the store, leaving the antiproton beam in the machine for tion of other machine components from unpredictable recycling. Using the EØ collimator with attached scatter- abort and injection kicker prefires/misfires and unsyn- ing targets, this process will require about 100 seconds to chronized abort. keep the power deposition in the superconducting magnets below the quench level. At the early Tevatron days the first collimation system was designed [2] on the basis of the MARS/STRUCT [7, 8] 1 INTRODUCTION full-scale simulations of beam loss formation in the machine. The optimized system, consisted of primary and sec- Tevatron is the world’s first superconducting and most pow- ondary collimators about 1 m long each, was installed in the erful hadron collider. Enormous efforts at Fermilab, reli- Tevatron which immediately made it possible to raise by a ably provided 900 900 GeV pp collisions with the peak factor of 5 the efficiency of fast resonant extraction system − − luminosity up to 2.510 cm s , recently resulted in and intensity of the extracted 800 GeV proton beam. The the discovery of the top quark, among many other impor- data on beam loss rates and on their dependence on the col- tant achievements. The current fixed target run, begun in limator jaw positions were in an excellent agreement with May 1996, exhibits the impressive performance of both the the calculational predictions. machine and experiments. At the same time, work pro- Later, we have refined the idea of a primary-secondary gresses to upgrade the accelerator and detectors into even collimator set and shown that this is the only way to use more powerful research tools [1]. The high performance such a system in the TeV region with a length of a pri- of Tevatron both in the fixed target and collider modes is mary collimator going down to a fraction of the radiation achievable only with a dedicated beam cleaning system em- length. The whole system should consist then of a primary bedded in the lattice [2, 3, 4]. ‘thin scattering target’, followed immediately by ‘a scraper’ with a few ‘secondary collimators’ in the appropriate lo- 2 BEAM LOSS HANDLING cations in the lattice [5, 6]. The purpose of a thin target is to increase amplitude of the betatron oscillations of the In the Tevatron, as in any other accelerator, the creation halo particles and thus to increase their impact parameter of a beam halo is unavoidable: proton (antiproton) scatter- on the scraper face on the next turns. This results in a sig- ing in the IPs, in beam-gas interactions and on the limit- nificant decrease of the outscattered proton yield and total ing apertures, the diffusion of particles due to various non- beam loss in the accelerator, scraper jaws overheating and linear phenomena out of the beam-core, as well as vari- mitigating requirements to scraper alignment. Besides that, ous hardware and software errors – all result in emittance the scraper efficiency becomes independent of accelerator growth and eventually in beam loss in the lattice [2, 5, 6]. tuning, there is only one drastic restriction of accelerator ∗ Work supported by the U. S. Department of Energy under contract aperture and only the scraper region needs heavy shielding No. DE-AC02-76CH03000 and probably a dogleg structure. The method would give

0-7803-4376-X/98/$10.00  1998 IEEE 133 an order of magnitude in beam loss reduction at multi-TeV narrow region of resonance phases used for extraction, the machines, but even at the Tevatron we have got a noticeable angle of the Lambertson magnet alignment depends mostly effect. Recently the existing scraper at AØ was replaced on the septa position, not on the resonance phase. This an- with a new one with two 2.5 mm thick L-shaped tungsten gle is equal to x’=-0.330 mrad for the septa at 20 mm from targets with 0.3 mm offset relative to the beam surface on the beam orbit. Any misalignment can drastically increase the either end of the scraper (to eliminate the misalignment the effective septum thickness and thus beam loss. problem), resulting in reduction of beam loss rate upstream 0.5 of both collider detectors [3]. A few other recent studies "parpho.ONEW2" 0.4 "PARPHO.ONN2" are described in the following sections. 0.3

0.2

0.1

3 FORWARD PROTON DETECTOR 0

-0.1 The detector [9] consists of four Roman Pot units placed -0.2 -0.3 in the DØ straight section upstream and downstream of the X'(Lambertson entr.), mrad -0.4 separators, and of three units at the C48 location. Each unit -0.5 -0.6 consists of two square 22cm detectors placed from both -40 -30 -20 -10 0 10 20 30 40 X(Lambertson entr.), mm sides of the closed orbit. 6000

Particle background in the Roman pot detectors is orig- 5000 inated in proton and antiproton interactions with the pri- 4000 mary and secondary collimators. The primary collimators 3000 are positioned at 5σ while secondary ones at 8σ.TheRo- dN/dX man pot detectors are at 8σx for proton beam and 9.4σx 2000 for antiprotons. Moreover, antiproton intensity intercepted 1000 by the collimation system is 3.6 times lower compared to 0 -40 -30 -20 -10 0 10 20 30 40 the proton intensity. Therefore, antiproton background in X(Lambertson entr.), mm the Roman pots amounts only 2% of the total background, Figure 2: Fast extraction phase space at the AØ Lambertson and backgrounds in the DØ detector due to Roman pots on magnet entrance. Top – extracted beam (black crosses) and the proton side about two orders of magnitude higher com- protons outscattered from the DØ deﬂector (grey squares). pared to the antiproton side. Bottom – transverse distribution of outscattered protons. Total background hit rates in Roman pots are (2.3- 3.3)10 p/s for detectors at 8σx and (0.87-1.09)10 p/s Proton distributions at AØ are shown in Fig. 2 with pro- for detectors at 9σx, i. e. the rates are three times lower for ton beam kicked by the DØ electrostatic septum. Protons the detectors at larger distance from the closed orbit. Beam outscattered from the septum wires are intersepted by the loss distributions in Tevatron with the Roman pot detectors Tevatron collimators. The EØ straight section is a very con- at 8σx are presented in Fig. 1. venient place for the absorption of scattered protons, but unfortunately, there is no collimator in this location. It was 100 found that the antiproton injection Lambertson magnet can 10

1 be used as a collimator with particles catched by the normal 0.1 conducting magnet yoke. Fast resonant extraction related 0.01 beam losses (in SC magnets only) with and without colli- 0.001

Particle loss, W/m mation are presented in Fig. 3. Collimation system reduces 0.0001

1e-05 beam losses in the superconducting magnets downstream

1e-06 0 1000 2000 3000 4000 5000 6000 of D17 by one order of magnitude. The DØ collimator right Path length, m after the septum, catching the low-energy debries from the B0 C0 D0 E0 F0 A0 B0 wires, unfortunately doesn’t protect the DØ - D17 region. . Calculations show that a 1 m long collimator

0 1000 2000 3000 4000 5000 6000 (rin=15 mm) upstream of the extraction line superconduct- Path length, m ing skew dipoles will protect them from the secondaries Figure 1: Beam loss distributions in Tevatron for D17 col- produced in the Lambertson magnet. Similar collimator limation with Roman pots at 8σx. with a round aperture of rin=20 mm upstream of the ﬁrst and second quadrupoles will protect them and other ring 4 FAST RESONANT EXTRACTION superconducting magnets.

Our recent simulations have shown that with the appropri- 5 PROTON BEAM REMOVAL ate collimation, additional shielding in AØ and electrostatic deﬂector and Lambertson magnet realignment, one could The upgrade plan requires to remove proton beam from reduce beam loss rates in Tevatron and increase the ex- Tevatron before the deceleration leaving antiproton beam tracted intensity without quenches. It was found that in a for recycling. There are two main concerns with the intense

134 1e+09

1e+08 warm magnet warm magnet primary collimator

1e+07 quad quad secondary collimators warm magnet warm magnet superconducting RF

Particle loss, proton/m 1e+06

100000 0 1000 2000 3000 4000 5000 6000 Path length, m 15.0m 6.0706 6.0706 6.07062.5 m 6.0706 1e+09 0.5 0.5 1.0 1.0 1.0 0.5 3.0 2.5248 1.9134 1.9982 2.5248 1e+08 Figure 4: Dog-leg scheme for proton beam removal at EØ 1e+07 straight section.

Particle loss, proton/m 1e+06 beam was removed from Tevatron without problem until a 100000 0 1000 2000 3000 4000 5000 6000 SC magnet quench happened at the rate of 0.36 10 p/s. Path length, m This is three times below of that was expected, what is eas- C0 D0 E0 F0 A0 B0 C0 ily explained by absence of a scattering target in the col- . limator and possible closed orbit displacement. Moreover, 0 1000 2000 3000 4000 5000 6000 analysis of the spill at beam removal has shown peaks of Path length, m losses at frequency of 1, 4.6, 13.9, 21, 37, 73 and 90 Hz, Figure 3: Fast resonant extraction losses without collima- which are understood from the Main Ring and Central He- tion (top), and with DØ , D17, F17, F49, AØ collimators lium Liqueﬁer performance and the beam position oscil- and Lambertson magnet at EØ (bottom). lations with synchrotron frequency. The experiment has shown that a feed-back system from the Beam Loss Moni- beam fast removal: superconducting magnet quenches tors to the dipole correctors used for the beam displacement caused by secondaries from a collimator and a target- is necessary. This system would provide rectangular spill collimator assembly overheating. A quench level of the shape and eliminate low-frequency peaks (1-37 Hz). Tevatron magnets at 1 TeV is about 310 p/m/s. This cor- responds to about 50 W/m. A good practice is to keep a 6 REFERENCES heat load to cryogenics below 1.5 W/m, or 110 p/m/s. [1] D.A. Finley, J. Marriner and N.V. Mokhov, ‘Tevatron Status With the Main Injector, the EØ straight section becomes and Future Plans’, Fermilab–Conf–96/408, 1996. free of the magnets used for the beam injection into the [2] A.I. Drozhdin, M. Harrison and N.V. Mokhov, ‘Study of Tevatron. With the ﬁrst 15 m of EØ straight section re- Beam Losses During Fast Extraction of 800 GeV Protons served for RF, the rest 35 m can be successfully arranged from the Tevatron’, Fermilab FN-418, 1985. for the proton beam removal (Fig. 4). Four DØ conven- [3] J. Butler, D. Denisov, T. Diehl, A. Drozhdin, N. Mokhov and tional bump-magnets are supposed to be used for the EØ D. Wood, ‘Reduction of Tevatron and Main Ring Induced dog-leg to protect the Tevatron magnets against neutral and Backgrounds in the DØ Detector’, Fermilab–FN–629, 1995. low-energy charged particles out of a primary collimator. Two 1.5 m long L-shaped secondary collimators placed at [4] N.V. Mokhov, ‘Optimization of the Radiation Environment at the Tevatron’, in Proc. of the Second Workshop on Simulating 10σ downstream of the dog-leg at the entrance to the cold Accelerator Radiation Environments, CERN, October 1995, region intercept most secondaries. With such a system, the CERN/TIS-RP/97-05, 1997. maximum beam loss in the Tevatron SC magnets is esti- mated to be 1.4 W/m. Moreover, the calculations show that [5] M. Maslov, N. Mokhov and I. Yazynin, ‘The SSC Beam Scraper System’, SSCL-484, 1991. it allows to get rid of other secondary collimators in the machine and, what is remarkable, reduce the beam loss level [6] A. Drozhdin, N. Mokhov, R. Soundranayagam and J. Tomp- in the DØ - D17 region by about a factor of four. kins, ‘Toward Design of the Collider Beam Collimation Sys- tem’, SSCL-Preprint-555, 1994. The EØ target heating is a serious problem for short spills. An instantaneous target temperature rise is 10000C. [7] N. V. Mokhov, ‘The MARS Code System User’s Guide, Ver- The target-collimator assembly cooling tremendously de- sion 13(95)’, Fermilab–FN–628, 1995. creases this temperature. With this, for a 10 msec spill a [8] I. Baishev, A. Drozhdin and N. Mokhov, ‘STRUCT Program stainless steel collimator edge is heated up to 1600C, but User’s Reference Manual’, SSCL–MAN–0034, 1994. already for a 1 sec spill, ∆Tmax =40C, only. So, the [9] A. Brandt, A. Drozhdin, N. Mokhov et al., ‘Proposal for For- target-scraper assembly overheating seams is not a restric- ward Proton Detector at DØ´, Fermilab, October 1996. tion for the proton beam removal from Tevatron. Calculations show that total beam intensity of 10 can be removed from Tevatron during 100 s using EØ collimator. During the January 1997 experimental studies, proton

135