Electron cloud measurements in heavy-ion driver for HEDP and inertial fusion energy

www.elsevier.com/locate/nimb

Nuclear Instruments and Methods in Physics Research B 261 (2007) 980–985

NIMBBeam Interactions

with Materials & Atoms

Electron cloud measurements in heavy-ion driver for HEDPand inertial fusion energy

Michel Kireeff Covo a,c,*, Arthur W. Molvik a, Alex Friedman a, Ronald Cohen a,Jean-Luc Vay b, Frank Bieniosek b, David Baca b, Peter A. Seidl b,

Grant Logan b, Jasmina L. Vujic c

a Lawrence Livermore National Laboratory, Heavy-Ion Fusion Science Virtual National Laboratory, 7000 East Avenue, Livermore, CA 94550, USAb Ernest Orlando Lawrence Berkeley National Laboratory, Heavy-Ion Fusion Science Virtual National Laboratory, 1 Cyclotron Road,

Berkeley, CA 94720, USAc Department of Nuclear Engineering, University of California at Berkeley, 4155 Etcheverry Hall, MC 1730, Berkeley, CA 94720, USA

Available online 3 May 2007

Abstract

The high-current experiment (HCX) at LBNL is a driver scale single beam injector that provides a 1 MeV K+ ion beam current of0.18 A for 5 ls. It transports high-current beams with large fill factor (ratio of the maximum beam envelope radius to the beam piperadius) and low emittance growth that are required to keep the cost of the power plant competitive and to satisfy the target requirementsof focusing ion beams to high-power density. Beam interaction with the background gas and walls desorbs electrons that can multiplyand accumulate, creating an electron cloud. This ubiquitous effect grows at higher fill factors and degrades the quality of the beam. Wereview simulations and diagnostics tools used to measure electron production, accumulation and its properties.Published by Elsevier B.V.

PACS: 29.27.Bd; 29.30.Aj; 34.50.Dy; 41.75.Ak; 79.20.Rf

Keywords: Electron cloud; Diagnostics; Retarding field analyzer; Heavy-ion beam; High-current accelerator; Inertial fusion

1. Introduction

The high-current experiment (HCX) [1] at LBNL is a1 MeV K+ linear DC accelerator that produces an ionbeam current of 0.18 A for 5 ls from an alumino-silicatesource. It is the first transport experiment with driver-scaleline charge density and pulse duration that has an injector,an electrostatic matching section, an electrostatic transportsection, and a magnetic transport section (Fig. 1) consistingof four room-temperature pulsed magnetic quadrupoles(QM1-4). The primary goal of the experiment is to study

0168-583X/$ - see front matter Published by Elsevier B.V.

doi:10.1016/j.nimb.2007.04.278

* Corresponding author. Address: Lawrence Livermore National Lab-oratory, Heavy-Ion Fusion Science Virtual National Laboratory, 7000East Avenue, Livermore, CA 94550, USA. Tel.: +1 925 422 9817; fax: +1925 424 6401.

E-mail address: [email protected] (M. Kireeff Covo).

the transport of high-current and high-energy space–chargedominated heavy-ion beams with large fill factors (beyond60%) and low emittance (�0.5 mm mrad for 1r) growth, inorder to reduce the cost of the fusion power plants (mini-mizing the transport array diameter and consequently theamount of induction core material needed for acceleration)and satisfy the requirements of focusing high-power den-sity for high-energy-density physics (HEDP) and inertial-fusion targets.

If the fill factor is increased, the beam runs closer to thewalls and starts to produce ion-induced electrons and des-orbed gas, which could move to the beam path and be ion-ized. Inside the matching section and electrostaticquadrupole section, the electrons are swept out by the elec-tric field towards the positive rods, but inside the magneticsection, the electrons from the ionization of backgroundand desorbed gas are trapped inside a potential well

Fig. 1. Magnetic quadrupole transport section of HCX has four quadrupole magnets (QM1-4). Electrons can be confined inside by the beam potential(�2100 V), if the suppressor (S) and last electrostatic quadrupole are biased negatively. Local sources of electrons can be removed, if the clearingelectrodes (A, B and C) are turned on. Retarding field analyzer (RFA) measures ions expelled by the beam potential, when the clearing electrode A is takenaway.

M. Kireeff Covo et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 980–985 981

produced by the space–charge beam potential of �2100 V.The ion-induced electrons desorbed from the walls can alsobe electrostatically trapped at the beginning of the beampulse, when the beam potential is rising at a rate of�2000 V/ls, but if they were produced during the flattop, they will reach the walls where they can be lost. Elec-tron lifetime will be given by its electron yield andreflectivity.

The transverse trapped electrons are confined axially bythe suppressor electrode at one end and by the last electro-static quadrupole electrode at the other end, which arebiased to �10 kV and �18.6 kV, respectively. Trappedelectrons decrease the beam potential and change the beamenvelope, producing a positive feedback that results inelectron cloud effects (ECE). Deleterious ECE includeelectron-stimulated gas desorption, cloud-induced noiseon instrumentation, tune shifts, instabilities and heat depo-sition on cryocooled components [2].

ECE were observed in the proton storage rings at BINP,the intersecting storage rings at CERN, the proton storagering at LANL, the relativistic heavy ion collider at BNL,the photon factory at KEK, the low energy ring at KEKBand other storage rings. They can potentially limit the per-formance of the large hadron collider at CERN and havebeen subject of and featured in various meetings (EPAC2004, ECLOUD’07, ICFA-HB2004, HHH2004, PAC05,DIPAC2005, etc.).

2. Simulations

Simulations use the WARP three-dimensional self-con-sistent particle-in-cell (PIC) code [3]. The code contains acomprehensive set of models governing the interaction ofpositively-charged beams with stray electrons and gas,including secondary electron emission from walls, chargeexchange, neutral emission and other processes. The mag-netic transport section from HCX, Fig. 1, is being usedto study ion beams containing electrons and to validatethe WARP code.

A novel mover for electrons that interpolates betweenfull electron dynamics and drift kinetics was developedand implemented in the code. The algorithm is discussedin [4] and takes advantage of Parker’s observation thatthe conventional Boris particle advance scheme, when run-ning with large time steps compared to the cyclotron per-iod, continues to exhibit correct drift velocities, butcauses particles to gyrate with a large radius compared tothe physical gyro orbit and with a frequency that is lowerthan the physical gyrofrequency. This Cohen mover per-forms an interpolation between full electron dynamics(Boris mover) and drift kinetics (motion along B plusdrifts), to preserve the physical gyroradius, but with largertime steps, reducing the computational time by a factor of10–100 times.

The PIC method for simulation of plasmas and particlebeams was also merged with the adaptive mesh refinement(AMR) technique [5]. This technique covers areas that needa higher resolution with a finer mesh, if the areas of thephysical domain that need refinement evolve in time, thenan automatic redistribution of the refinement applies, sav-ing computational effort in simulations of time-dependentspace–charge-limited flow by up to 20,000 times withproved numerical convergence.

The Cohen mover and AMR technique have led to largespeedups for affordable numerically-converged and accu-rate results. WARP simulations show that the transverseelectron density distributions inside the magnetic quadru-poles depend on the nature of the electron source. Elec-trons originating from ion impact on structures at theend of HCX will produce a virtual cathode and can moveupstream thought two opposite quadrants by E

!� B!

andrB!

drifts if the suppressor electrode is turned off. Electronsoriginating from ionization of background gas will havethe beam profile and the even and odd quadrants will driftin opposite directions. Electrons desorbed from the beampipe will populate the entire transverse section, followingthe magnetic field lines and peaking near the wall at thecenter of the quadrants.

982 M. Kireeff Covo et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 980–985

3. Experiments

The gas-electron source diagnostic (GESD) [6] is aninstrumented target to measure ion-induced gas and elec-tron production as a function of the ion angle of incidence.The GESD measured that each 1 MeV K+ ion impactingnear grazing incidence on stainless steel desorbs �10,000molecules of gas and produces �100 electrons. A theoreti-cal model for the electron desorption [7], using TRIM codeto evaluate dEe/dx at several depths in the target, demon-strates good agreement with the experimental data andalso models gas desorption [8].

The measured electron and gas desorption yield areneeded for calibration of signal intensities collected onthe wall electrodes inside the last two magnetic quadru-poles, QM3-4. These sets of passive nonintercepting diag-nostics include flush probes, capacitive probes andgridded probes [6,9]. All the electrodes have dedicated cou-pling circuits to simultaneously permit biasing and measur-ing small currents, using current amplifiers. The amplifiedsignals are digitalized by an A/D converter and analyzedand archived by a Labview program.

The third quadrupole magnet has an array of eight longflush collectors (FLLs), Fig. 2(a) and (b). These electrodesare mounted longitudinally on the beam bore pipe, thearray encircles the beam with a length exceeding the effec-tive magnetic field of the quadrupole. The FLL signal is asum of the beam loss, the ion-induced electrons and the

Fig. 2. Third magnetic quadrupole – QM3 (a) Transverse magnetic field lindiagnostics that share the same magnetic field lines are called paired. (b) PicturQM4 (c) Diagnostic sketch showing three beam position monitors (BPMs), twBPM and two gridded electron collector (GECs) are placed on the far side and ar

induced charge by the ion beam. The beam loss is negligiblecompared with the ion-induced electrons emitted and canbe neglected. The design was made in such way that themagnetic field lines through one electrode will go throughthe adjacent electrode. The paired electrodes that sharethe same transverse magnetic field lines have opposite bias(+50 V). Ion-induced electrons from the negatively biasedflush electrode inside the quadrupole will follow the mag-netic field lines and end up in the positively biased elec-trode. Ideally, measurement of ion-induced electronemission from the negatively biased probes can be madeby summing the differences between paired electrodes,which removes the capacitive signal and dividing by two,which takes into account that the electron current leavingone electrode is collected by its pair. If the bias is invertedand the same procedure is followed, the total ion-inducedelectron current can be obtained by adding up the mea-sured currents for each bias. The beam induced capacitivesignal can be obtained by adding the paired signals. In thisway, the electron current for the paired probes will cancelout, leaving only the capacitive pickup. Beam loss andthe dynamic density of desorbed gas can also be inferredfrom the total electron emission from the FLLs.

The fourth quadrupole magnet has three beam positionmonitors (BPMs), two gridded electron collectors (GECs),two gridded ion collectors (GICs) and two short flushcollectors (FLSs) that are even to the walls, Fig. 2(c) and(d).

es are superimposed to the FLL diagnostics sketch. The green and bluee of the diagnostics before the installation. Fourth magnetic quadrupole –o gridded ion collectors (GICs) and two short flush collector (FLSs). Onee not seen in the sketch. (d) Picture of the diagnostics before the installation.

M. Kireeff Covo et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 980–985 983

The BPMs are recessed from the bore beam pipe andhave small scrappers upstream to intercept ions near graz-ing incidence, before they reach a BPM. They are placed ina region where the magnetic field is parallel to the probesurfaces, magnetically suppressing electron emission. Asany capacitive electrodes, they are sensitive to changingelectric fields, which are highest at the head and tail ofthe beam. The net charge per unit length is obtained byintegrating the induced signal and scaling to the probe azi-muthal angle and length relative to the beam. As electronsdecrease the net beam charge, the BPM might determinethe density of electrons. A problem was observed becausethe net induced charge measured does not go back to zeroat the end of the beam pulse, probably indicating that someion-induced electrons escape along magnetic fields to theborder. Design modifications to fix this problem are beingstudied.

The GEC electrodes are recessed from the bore beampipe behind flush grids and are placed in a region wherethe magnetic field that passes through the beam entersthe electrode. The entrance double grid attenuates thepickup capacitive signal by a factor of �500, allowing mea-surements of untrapped electrons expelled at the end of thebeam, when the beam potential decays.

The GICs are similar to the GECs but are placed in aregion where the magnetic field is parallel to the collectorsurface, suppressing electron emission or collection. Ionsproduced from beam-background gas interaction (ioniza-tion and charge exchange) are driven across the magneticfield (BMAX � 0.48 T) by the beam space–charge potential.The entrance double grids attenuate the capacitive pickupby a factor of 500, enabling measurement of very smallexpelled ion currents by reducing the strong noise pro-duced by capacitive pickup and consequently increasingthe signal to noise ratio. These gridded electrodes were suc-cessfully used to measure the dynamic gas density withinthe ion beam, effectively working as an in-situ fast ioniza-tion gauge [9].

The FLSs are mounted flush to the beam bore pipe, butthey are not paired and have shorter length. The ion-induced electrons can leave the negative electrode and fol-low the magnetic field lines. Similarly to the long flushprobes, they can be calibrated with the GESD data to inferbeam loss and dynamic density of desorbed gas.

Several diagnostics (one suppressor ring, three clearingelectrodes, one retarding field analyzer – RFA, one CCDcamera and two Faraday cups) are placed before, afterand between the quadrupole magnets and help to studythe electron cloud features.

The suppressor ring is installed after the magnetic sec-tion. It can be biased to �10 KV, so we can choose tosuppress electron emission from K+ ion impact on endstructures.

The clearing electrodes are added to the lattice betweenmagnets. They are stainless steel rings biased positively(+9 kV) to remove electrons that reach the gaps. The clear-ing electrodes constitute an efficient mitigation technique

[9] and can be used to measure electron cloud line chargedensity at gap A [10]. The static background cloud line-charge density is obtained if the electron current measuredwith the clearing electrode is divided by the average driftvelocity of electrons inside the magnets (�0.60 m/ls) [11].

The RFA, which is described in detail in [12], can beinserted in the drift region between quadrupole magnetsQM1-2 (gap A) instead of the clearing electrode A. TheRFA and the clearing electrodes measurements can becombined to provide absolute time-dependent electroncloud density accumulation during the beam pulse [10].The beam-background gas interaction produces cold ionsfrom ionization and charge exchange that are expelled bythe beam space–charge potential, converting potentialenergy into kinetic energy. The expelled ions reach thewalls in few hundred nanoseconds. As electrons accumu-late, the beam potential decreases and so does the energyof the expelled ions [13]. The electron density as a functionof time is obtained from the beam potential decay measure-ment accounting for the ion and electron transverse distri-butions. The dynamic density can be supplemented andcorroborated by the static background density obtainedfrom clearing electrodes measurements, giving the absoluteelectron density. An experiment that compared the neutral-ization (ratio of electron to the ion charge density) mea-sured with the RFA and the clearing electrode techniquesfor three different conditions is described in [10]. For thefirst condition (B, C and S on) the clearing electrodes andthe suppressor of Fig. 1 are all on, minimizing all sourceof electrons. For the second condition the clearing elec-trodes are off and the suppressor is on (B, C off and Son), which allows electrons from local sources (ionizationand desorbed from the beam pipe) to accumulate. Forthe third condition the suppressor and the clearing elec-trodes are off (B, C and S off), which also allows electronsoriginating from the end structures to drift upstream. TheRFA, Fig. 3(a), and the clearing electrode measurements,Fig. 3(b), give the neutralization shown in Table 1, whichare in reasonable agreement. The RFA is a versatile toolthat also provides information of ion and electron energydistributions, beam potential distribution, halo losses andbeam-background gas cross-sections.

The CCD camera can be placed before, between andafter the quadrupole magnets. It is an optical diagnosticthat provides time-resolved 4-D transverse informationfrom the beam hitting a kapton scintillator. Beam focusingto small spots has been seen and attributed to ECE. Fromside measurements of beam interaction with desorbed gaswe infer an average velocity of desorbed gas of 1.5 mm/ls [14], consequently, during the beam duration of 5 ls,most of the gas cloud does not expand into the beam pathand therefore it will not be ionized.

A Faraday cup is positioned before and another afterthe magnetic section. They are intercepting diagnostics thathave two electrodes, a suppressor ring upstream and a col-lector with a honeycomb structure downstream, that areinside a grounded case for electrical shielding. The current

Fig. 3. (a) Faraday cup current and dynamic beam potential measured for three different configurations, increasing the sources of electrons. For the firstcondition (B, C and S on), the clearing electrodes B, C and suppressor are on. For the second condition (B, C, off and S on), we allow local sources ofelectrons to accumulate by turning off the clearing electrodes B and C. For the third condition (B, C and S off) we also allow electrons generated at the endstructures to drift upstream by turning off the suppressor. (b) Electron current from clearing electrode A obtained for the same configurations of Fig. 3(a),after subtracting the beam induced capacitive signal.

Table 1Comparison of the beam neutralization measured in gap A using theclearing electrode and RFA techniques

Percentage of beamneutralization

B, C and Son

B, C off and Son

B, C and Soff

Clear electrodes 7.3 25.2 89.2RFA 7.3 27.5 79.5

984 M. Kireeff Covo et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 980–985

difference measured before and after the magnetic sectionprovides beam losses. For a fill factor of 60%, the measuredcurrent loss is �2 mA. As the total measured beam-back-ground gas interaction cross-section (ionization plus chargeexchange) with the RFA is 3.1 · 10�19 m2 [15], which givesan upper limit to the beam neutralization at the end of thepulse of 3%, the major electron source will be from ionbeam losses to the walls.

4. Conclusions

Electrostatic quadrupoles provide efficient ion-beamtransport at low energy and provide clearing fields thatsweep out unwanted electrons. At higher energies the trans-port is usually by quadrupole magnets. HCX is studyinghigh-line-charge-density, high-perveance beam transportby quadrupole magnets for application to high-energy-den-sity physics and to heavy-ion fusion. At this regime space–charge forces strongly influence the beam properties. Elec-tron clouds are a ubiquitous source of negative charge thatalters the space–charge forces, which can change beamemittance, envelope size and halo and can drive instabilities.

The HCX is highly instrumented to measure sources andaccumulation of electrons in the magnetic section. The sig-nal intensities collected on the wall electrodes within thelast two magnetic quadrupoles are calibrated with datafrom GESD experiment to determine ion beam and elec-tron dynamics inside the magnets. Clearing electrodes,RFA and suppressor are placed between and at the endof the magnets. With these and the other diagnostics dis-

cussed, we study the electron cloud, comparing the resultswith state-of-art simulations and developing a non-intru-sive technique that measures the time-dependent electroncloud density during the beam [10].

Future plans include the installation of an electron gunfor studies of electron transport, accumulation and effectson the beam. During the beam pulse, the magnetic sectionforms an electron trap that can accumulate electrons, if thesuppressor and the last electrostatic quadrupole are biased.The beam space–charge potential and the adjustable poten-tial energy of the gun-extracted electrons, given by thecathode bias, will enable us to change the electron spatialdistribution during the K+ beam pulse, providing a con-trollable source of electrons that can be measured withHCX diagnostics.

Acknowledgements

We wish to thank Tak Katayanagi who built the diag-nostics, Wayne G. Greenway, Larry W. Mills and GaryRitchie who maintain HCX and Craig Rogers, Ed Romeroand William L. Waldron who provided electronic support.We also want to express our gratitude to Richard A.Rosenberg and Katherine C. Harkay for sharing detailsthat aided our RFA design, and to Miguel Furman forhis insightful comments. This work was performed underthe auspices of US Department of Energy by the Universityof California, LLNL and LBNL under Contracts No. W-7405-ENG-48 and No. DE-AC02-05CH11231.

References

[1] L.R. Prost et al., Nucl. Instr. and Meth. A 544 (2005) 151.[2] H. Fukuma, in: Proc. of Diag. Instr. Part. Acc. Conf., Lyon, France,

2005, p. 122.[3] J.-L. Vay et al., in: Proc. of Part. Acc. Conf., Knoxville, TN, 2005, p.

525.[4] R.H. Cohen, A. Friedman, S.M. Lund, A.W. Molvik, E.P. Lee, T.

Azevedo, J.-L. Vay, P. Stoltz, S. Veitzer, Phys. Rev. ST Accel. Beams7 (2004) 124201.

M. Kireeff Covo et al. / Nucl. Instr. and Meth. in Phys. Res. B 261 (2007) 980–985 985

[5] J.-L. Vay et al., Phys. Plasmas 11 (2004) 2928.[6] A.W. Molvik, M. Kireeff Covo, F.M. Bieniosek, L. Prost, P.A. Seidl,

D. Baca, A. Coorey, A. Sakumi, Phys. Rev. ST Accel. Beams 7 (2004)093202.

[7] M. Kireeff Covo et al., Phys. Rev. ST Accel. Beams 9 (2006) 063201.[8] M. Kireeff Covo, A.W. Molvik, A. Friedman, J.-L. Vay, F.M.

Bieniosek, D. Baca, P.A. Seidl, J. Vujic, in: Proc. of AVS 53rdInternational Symposium, San Francisco, CA, 2006, VT-WeM5, p.113.

[9] A.W. Molvik, R.H. Cohen, A. Friedman, M. Kireeff Covo, S.M.Lund, G. Westenskow, Nucl. Instr. and Meth. A 544 (2005) 194.

[10] M. Kireeff Covo, A.W. Molvik, A. Friedman, J.-L. Vay, P.A. Seidl,G. Logan, D. Baca, J.L. Vujic, Phys. Rev. Lett. 97 (2006) 054801.

[11] R.H. Cohen et al., LBNL Report No. 56496, 2004.[12] M. Kireeff Covo, A.W. Molvik, A. Friedman, J.J. Barnard, P.A.

Seidl, G. Logan, D. Baca, J.L. Vujic, Nucl. Instr. and Meth. A 577(2007) 139.

[13] J. Klabunde, H. Reiserb, A. Schijnlein, P. Spadtke, J. Struckmeier, in:Proc. of Part. Acc. Conf., Santa Fe, NM, 1983, p. 2543.

[14] F.M. Bieniosek, Phys. Rev. ST Accel. Beams, submitted forpublication.

[15] M. Kireeff Covo, to be published.