high-rate particle identification of high-energy heavy ions using a tilted electrode gas ionization...

ARTICLE IN PRESS

0168-9002/$ - se

doi:10.1016/j.ni

�Correspondi95-830-1126.

E-mail addre1Present addr

Japan.2Present addr

Saitama 338-853Present add

(GSI), D-64291

Nuclear Instruments and Methods in Physics Research A 538 (2005) 608–614

www.elsevier.com/locate/nima

High-rate particle identification of high-energy heavy ionsusing a tilted electrode gas ionization chamber

K. Kimuraa,�, T. Izumikawab, R. Koyamac, T. Ohnishid, T. Ohtsuboc,A. Ozawad,1, W. Shinozakic, T. Suzukic,2, M. Takahashic, I. Tanihatad,

T. Yamaguchid,3, Y. Yamaguchic

aNagasaki Institute of Applied Science, 536 Abamachi, Nagasaki 851-0193 JapanbRadioisotope Center, Niigata University, Niigata 951-8510 Japan

cDepartment of Physics, Niigata University, Niigata 950-2181 JapandThe Institute of Physical and Chemical Research (RIKEN), Wako, Saitama 351-0198 Japan

Received 23 December 2003; received in revised form 9 July 2004; accepted 4 August 2004

Available online 15 September 2004

Abstract

A high-rate particle identification device for high-energy heavy ions has been developed which utilizes a stacked

configuration of grid-less parallel plate gas ionization chambers with thin anode–cathode gaps. The high-rate capability

of this chamber was realized by adopting bipolar shaping of anode signals and by the suitable choice of a counter gas.

Z-resolutions of 0.2–0.3 were obtained for nuclear fragments of 40Ar at 95AMeV with an intensity as high as 106 cps:r 2004 Elsevier B.V. All rights reserved.

PACS: 29.40.Cs

Keywords: Gas ionization chamber; Heavy ion; Particle identification

e front matter r 2004 Elsevier B.V. All rights reserve

ma.2004.08.100

ng author. Tel.: +81-95-838-5189; fax: +81-

ss: [email protected] (K. Kimura).

ess: University of Tsukuba, Ibaraki 305-8571

ess: Department of Physics, Saitama University,

70 Japan.

ress: Gesellschaft fur Schwerionenforschung

Darmstadt, Germany.

1. Introduction

Recently, radioactive isotope (RI) beams havebecome important probes in nuclear experiments.High-energy RI beams are usually produced assecondary beams by magnetically separating pro-jectile fragments from high-energy heavy ionreactions using a fragment separator, as atRIKEN and GSI. In such cases, the RI beamscontain many different types of isotopic elements.

d.

www.elsevier.com/locate/nima

ARTICLE IN PRESS

K. Kimura et al. / Nuclear Instruments and Methods in Physics Research A 538 (2005) 608–614 609

When we use the RI beams in the secondaryreactions, it is imperative, therefore, to identify theatomic numbers and mass numbers of the indivi-dual particles of the beams. A combination of theenergy loss measurement of beam particles using aDE-detector and the velocity measurement using atime-of-flight (TOF) detector is usually employedfor this particle identification. Here we describe thedevelopment of a DE-detector that can be used forsuch beam monitoring. A suitable DE-detectorneeds to have good energy resolution, highcounting rate capability and robustness againstbeam bombardment. Semiconductor detectors canbe used because of their good energy resolution,but they have several drawbacks: they deterioraterather rapidly under beam bombardment, espe-cially under that of high-intensity heavy ions, andit is difficult to obtain large-scale detectors, e.g.,detectors larger than 100 cm2 that can cover theentire beam profile. Unlike semiconductor detec-tors, gas ionization chambers are extremely stableunder beam bombardment if equipped with gasflow systems. Also, they can provide energyresolution which is as good as that of semicon-ductor detectors, and large-scale detectors are easyto fabricate. But, normally they are not fastenough to resolve the signals of individual beamparticles at high intensities, e.g., above 100 kcps.We attempted to extend the high-rate capability ofthe gas ionization chamber to a region above100 kcps by employing a stacked configuration ofthin grid-less gas ionization chambers.

Gridded gas ionization chambers are usuallyemployed for DE-measurement instead of grid-lessgas ionization chambers, because grid-less ioniza-tion chambers suffer from extremely slow positiveion movement and normally have position-depen-dent pulse shapes, making it difficult to determineDE-values within short intervals. A high-ratechamber has already been developed at GSI [1]using a gridded gas ionization chamber. However,in the normal usage of gridded ionization cham-bers, anode pulses have various delay times whichdepend on the distance of the particle tracks fromthe grid plane because incident particles runparallel to the grid plane. The spread of thesedelay times hampers the high-rate operation ofgridded ionization chambers, especially in the case

of large volumes, and also hampers the employ-ment of output signals for event characterization.We developed a multilayer grid-less gas ionizationchamber that is free of the delay time, slowpositive ion effect and position dependence of theanode signals.

2. Construction and operation of the tilted electrode

gas ionization chamber

Fig. 1 shows a cross-sectional view of our grid-less gas ionization chamber. Its structure is verysimple. Twelve anode planes and thirteen cathodeplanes are alternately placed in 20-mm steps. Atotal of 24 parallel plate ionization chambers arestacked together back to back, which results in a48-cm-thick chamber.The anode and cathode planes are made of thin

conductive foils (4-mm-thick mylar aluminized onboth sides). Incident particles pass through all ofthe electrode foils. In order to avoid the recombi-nation of electrons and positive ions liberatedfrom the gas along the particle trajectories, theelectrode planes are tilted 30� toward the centeraxis, as shown in Fig. 1. By tilting the electrodes,the liberated electrons and positive ions drift awayfrom the original particle trajectories in oppositedirections of each other. Then, they have littlechance to collide with each other and recombine.Hereafter, we refer to this ionization chamber as atilted electrode gas ionization chamber (TEGIC).The anode and cathode foils are glued onaluminum rings, the inner diameter of which is116 mm. These foils are stretched tightly byapplying the appropriate tension, using subsidiaryrings to avoid the distortion due to the electro-static force when high voltage is applied to theanodes. The whole assembly of the electrodes ishoused in an aluminum vessel. Kapton sheets (50-mm thick) are attached to the entrance and exitwindows. In order to reduce the number of outputconnectors, pairs of anode electrodes are con-nected together electrically inside the chamber,and hence the summed anode signals of each set offour ionization chamber units are taken outthrough six separate feedthroughs. Also, all thecathode planes are connected together and

ARTICLE IN PRESS

650 mm

195

mm

cathodeanode

480 mm

Fig. 1. Cross-sectional view of the tilted electrode gas ionization chamber (TEGIC).

K. Kimura et al. / Nuclear Instruments and Methods in Physics Research A 538 (2005) 608–614610

grounded. These electrical connections are illu-strated in Fig. 1.

The operation of the TEGIC as a high-rate DE-counter is as follows. When an incident particlepasses through the chamber, anode potentialsbegin to drop in accordance with the drift ofelectrons and positive ions liberated along theparticle trajectory. These drops continue until allthe electrons and positive ions reach the anodesand cathodes, respectively. The potential drop ofone anode finally reaches a value equal to the sumof the electron and positive ion charges liberatedon both sides of the anode divided by twice theanode capacity. It is thus proportional to theenergy loss, and so we can use the TEGIC as a DE-detector. The starting time of each potential dropoccurs without delay after the time of the particletransit, regardless of where the particle has passed.This enables the adoption of anode signals in eventidentification. However, the slow positive ioneffect must be removed for high-rate operationof this chamber. Typically, the time for thepositive ion drift is of the order of 1ms, whereasthe time for the electron drift is less than several100 ns. In the TEGIC, electrons and positive ionshave an equal potential drop. Only the electronpart of the anode pulse is sufficient to obtain DE

information, and so the positive ion part can becompletely disregarded. After amplifying theanode signals using a conventional charge-sensi-

tive preamplifier, the output pulse shapes aresimply proportional to the anode potential drops.The slow positive ion part can be filtered to someextent using a spectroscopy amplifier with ordin-ary unipolar shaping. But, it is difficult to filter outthe positive ion part completely. Excessive baselinefluctuation of the amplifier output occurs due tothe pileup of residual positive ion tails even at thelow counting rate of 100 cps. We found thatcomplete filtering of positive on tails is possibleby adopting bipolar shaping of the spectroscopyamplifier. Bipolar shaping restores the baseline ofthe spectroscopy amplifier extremely well andfacilitates high counting rate operation of theTEGIC.

3. Experimental setup and procedure

A test experiment of the TEGIC was performedat the RIKEN Ring Cyclotron facility. TheTEGIC was located on the secondary beam lineE1C and was tested using secondary beamsproduced by nuclear fragmentations of 56Fe and40Ar primary beams at 90AMeV and 95AMeV,respectively. The TEGIC was first evacuated downto vacuum using a rotary pump, and then filledwith counter gas using a gas flow system. Thecounter gas was continuously flowed at a rate ofabout 1 l/h. A gas mixture of Ar–CH4 (90%, 10%;

ARTICLE IN PRESS

Fig. 2. Two-dimensional scatter plot of DE vs. TOF for the

secondary beam produced by nuclear fragmentation of 56Fe at

90AMeV.


purity 499:99%) was mainly used as the countergas. The performance of Ar–CF4 gas (90%, 10%;purity 499:99%) was also studied in the high-rateexperiment. The gas pressure was adjusted to740Torr throughout the experiment and wasregulated within a variation of 1Torr using thegas flow system. The anode voltage was set atþ500V for the Ar–CH4 gas and at þ1000V for theAr–CF4 gas to ensure that the drift velocity ofelectrons reached the maximum value for each gas[2].

The anode signals taken from the six feed-throughs of the TEGIC were separately amplifiedusing a combination of charge-sensitive preampli-fiers (Tennelec 171) and spectroscopy amplifiers(Canberra 2024). As explained in the previoussection, bipolar shaping of the spectroscopyamplifiers was adopted for pulse shaping. Aminimum selectable time constant of 0:25ms wasused for high counting rate tests. A time constantof 0:5ms was also tried for low counting rate tests,but no effective difference in energy resolution wasobserved.

The TOF of the nuclear fragments was deter-mined from the time difference between cyclotronrf-signals and plastic scintillator timing signals andwas measured using a TDC.

ADC values of the six anode signals of theTEGIC were measured using a peak-sensitiveADC. Plastic scintillator energy signals weremeasured using a charge-sensitive ADC. TheCAMAC system was used to collect all data.

4. Experimental results

4.1. Particle identification

Fig. 2 shows a typical example of DE vs. TOFplots for nuclear fragments from 56Fe at 90AMeVmeasured at a relatively low secondary beamintensity of 1.7 kcps. The DE values were deter-mined simply by summing the ADC values of thesix anode signals of the TEGIC.

Clusters of events can be seen in this figure.Each cluster represents a particular isotope. Theseries of clusters vertically aligned around the TOFchannel of 650 corresponds to nuclides having the

same value of A=Z ¼ 2: It is evident that the DE-values are dependent on the velocities. In order toreduce this velocity dependence and obtain particleidentification (PID) signals, we transformed theraw data using the relationship DE / PID=b;where b is the fragment velocity. In our semi-relativistic energy region, the velocity dependenceof DE is simulated simply by b�1 rather than b�2:Fig. 3 shows the PID spectrum gated on thenuclear fragments of A=Z ¼ 2: All particles fromZ ¼ 6 up to Z ¼ 27 are clearly resolved. The PIDsignals were found to be almost exactly propor-tional to Z2; as shown in Fig. 4, and are consistentwith the Bethe–Bloch formula for energy loss.

4.2. Z-resolution

Quantitative discussions for the resolving powerof atomic number Z can be expressed in terms ofZ-resolution (DZ) defined by the following equa-tion:

DZðZÞ ¼2DPIDðZÞ

PIDðZ þ 1Þ � PIDðZ � 1Þ(1)

where DPIDðZÞ denotes the FWHM of the PIDspectrum for atomic number Z, and PIDðZÞ

ARTICLE IN PRESS

1200

1000

800

600

400

200

0

PID

(ch

anne

l)

302520151050

Z

PID = 1.23·Z 2.03

+1.75

Fig. 4. PID vs. Z relationship for A=Z ¼ 2 nuclides. A solid

curve is the best fit to the PID and indicates that the PID is

proportional to Z2:03: The errors of the PID channels are less

than the marker size.

0.5

0.4

0.3

0.2

0.1

0.0

∆Z

30252015105

Z

85 kcps 1.7 kcps

Fig. 5. Z-resolution vs. Z relationship obtained for A=Z ¼ 2

nuclear fragments of 56Fe at 90AMeV. The data at two

different counting rates, 1.7 kcps and 85 kcps, are compared.

1

10

100

Cou

nts

(cou

nt/c

hann

el)

10008006004002000Particle ID (channel)

Fig. 3. PID spectrum for A=Z ¼ 2 nuclides produced by the

fragmentation of 56Fe at 90AMeV. Particles from Z ¼ 6 up to

Z ¼ 27 are observed.


denotes the channel number of this spectrum. Theresults of these calculations for two differentcounting rates, 1.7 and 85 kcps, of the secondarybeam are shown in Fig. 5. Although the datascatter widely, the values of DZ for small Z

(Zp14) seem to be constant around the value of0.22, but above this they increase with Z. Withinthe region of Zp26; the DZ’s remain at less than

0.35 and unambiguous identification of the atomicnumber is possible.A slight counting rate dependence of DZ was

also observed. At the counting rate of 85 kcps, theDZ’s are generally worse than those at 1.7 kcps asshown in Fig. 5.Since total energy loss in the TEGIC is divided

into six anode signals, we can obtain energy lossdata at every 8 cm of gas thickness. The depen-dence of DZ on gas thickness can readily bedetermined by summing the anode signals onsuccessively different numbers of these six anodesignals and transforming them into PID signals.Fig. 6 shows the results of this analysis for Z ¼ 14and Z ¼ 18: The horizontal axis is scaled in t�1=2;where t denotes the effective length of gas incentimeter. In this figure nine data points insteadof six are shown for each Z. The additional threepoints to the far left were measured by connectinga small TEGIC to the present TEGIC in tandem.The small TEGIC has three anode outputs, eachcarrying energy loss signals corresponding to 8 cmof gas thickness.The solid and dotted lines in Fig. 6 show the

best fit to the two data groups of Z ¼ 14 and Z ¼

18; respectively. It is clear that the DZ’s arelinearly dependent on t�1=2 but have some offsetvalues. Z ¼ 14 data has a small offset of 0.04,

ARTICLE IN PRESS

0.5

0.4

0.3

0.2

0.1

0.0

∆Z

0.40.30.20.10.0

t-1/2

(cm-1/2

)

Z=14 Z=18

∆Z=1.19·t -1/2

+0.04

∆Z=1.10·t -1/2

+0.08

Fig. 6. Dependence of Z-resolutions on effective length t of

counter gas obtained for nuclides of Z ¼ 14 and 18. DZ’s show

linear dependence on t�1=2:

0.20

0.15

0.10

0.05

0.00

Off

set v

alue

302520151050

Z

Fig. 7. Dependence of offset-values of DZ on Z. A solid curve

is fit to the data by a polynomial function of the second order.


while Z ¼ 18 data has a larger offset of 0.08. Wedetermined the offsets for various Z values. Asshown in Fig. 7, the offsets increase fairly steeplywith Z and tend to zero as Z approaches Z ¼ 0:This is directly related to the increase of DZ withincreasing Z, previously seen in Fig. 5.

Let us briefly consider the origin of these offsets.If DZ is determined simply by the statisticalfluctuations of energy loss, i.e., the energy strag-gling of heavy ions within gas, DZ should beexactly proportional to t�1=2; i.e., 1=

ffiffiffiffiffiffiffi

DEp

; and hasno offset value because the numerator in Eq. (1)becomes proportional to

ffiffi

tp

[3], i.e.,ffiffiffiffiffiffiffi

DEp

; whilethe denominator is proportional to DE: This alsoresults in DZ’s which are independent of Z. Suchan ideal situation seems to apply to the lightelements (Zp14) because of the small offset andthe constancy of DZ; seen in Fig. 5. Electronicnoise does not account for the offset because it hasno Z dependence. The charge collection ineffi-ciency does not seem to be important because DE

shows an extremely good Z2 dependence, as seenin Fig. 4, and no charge loss appears to occur.

One possible factor affecting the increase ofoffsets is the existence of the electrode foils.Electron ejection from the foils by beam bombard-ment and absorption of ionized electrons from gas

by the foils should bring about additional fluctua-tions of the number of electrons within gas. Thiscontribution becomes more important with in-creasing the Z of beam particles. The effect ofthese foils can be checked by using thinnerelectrodes.

4.3. Counting rate dependence

The counting rate dependence of the detectorresolution was measured using the secondarybeam from 40Ar at 95AMeV, which can providehigh fragment intensities. No pileup rejectioncircuit was employed but off-line pileup rejectionwas performed by taking correlations between twoplastic scintillator energy signals with differentdelays: one signal with proper timing to the energygate and the other signal with an additional delay.In a current integration ADC, if two pulses in theplastic scintillator happen to occur very close intime (closer than the gate width of the ADC), theenergy signal of the proper timing has a largerADC value than the delayed energy signal. Hence,the event can be rejected as a pileup occurring inthe TEGIC. Pileup events amounted to about 40%of the total events at a rate of 320 kcps andincreased to about 75% at 1Mcps.For this measurement, we compared the high-

rate performance of two kinds of counter gas,

ARTICLE IN PRESS

0.4

0.3

0.2

0.1

0.0

∆Z

102

103

104

105

106

Counting rate (cps)

Ar-CH4

Ar-CF4

Z=14

Fig. 8. Dependence of Z-resolutions on the intensities of the

secondary beam produced by nuclear fragmentation of 40Ar at

95AMeV. Performances of two kinds of counter gas, Ar–CH4

(90%, 10%) and Ar–CF4 (90%, 10%), are compared.


Ar–CH4 (90%, 10%) and Ar–CF4 (90%, 10%).Fig. 8 shows the observed dependence of DZ forZ ¼ 14 on the secondary beam intensities. In thecase of Ar–CH4 gas, the values of DZ change veryslightly up to an intensity of about 300 kcps butabove this they increase fairly rapidly.

On the other hand, although the resolution forAr–CF4 gas is slightly poorer than Ar–CH4 gas,Ar–CF4 gas shows a better rate dependence. Thevalues of DZ are almost constant up to themaximum beam intensity of 106 cps: The betterrate dependence of Ar–CF4 gas is probably due tothe fact that it has a higher drift velocity ofelectrons compared with Ar–CH4 gas; that is, thedrift velocity is twice as large [2] for the electricfield strength of saturation drift velocities, whichwe employed in this measurement.

5. Summary

The tilted electrode gas ionization chamber,which we called the TEGIC, was found to work

stably up to a secondary beam intensity of 106 cps;produced by nuclear fragmentations of 40Ar at95AMeV. The high counting rate capability of theTEGIC, which is essentially a grid-less parallelplate ionization chamber, was facilitated simply bymaking the anode–cathode gap suitably small andemploying bipolar shaping of the spectroscopyamplifiers. The bipolar shaping served to reject theslow positive ion effect and restore baselinefluctuations. The proper gas selection, Ar–CF4

(90%, 10%), was also essential. Z-resolutions ofbetter than 0.35 were obtained for nuclides ofZp26 and E 90AMeV; and unambiguousidentifications of the atomic number was possiblefor these nuclei. However, DZ values show atendency to increase with increasing atomicnumber probably because of the existence of theelectrode foils. The use of thinner foils than thoseused in the present study might improve theresolution of the TEGIC for heavy nuclei.

Acknowledgements

The authors gratefully acknowledge the effortsof the staff of the RIKEN Ring Cyclotron toensure the smooth operation of the accelerator.This experiment was supported in part by a Grant-in-Aid for Scientific Research from the Ministry ofEducation, Culture, Sports, Science and Technol-ogy in Japan.

References

[1] A. Stolz, et al., GSI Scientific Report 1998, 174.

[2] A. Peisert, F. Sauli, preprint CERN-EP/84-08, 1984.

[3] W.R. Leo, Techniques for Nuclear and Particle Physics

Experiments, Springer, Berlin, Heidelberg, 1987.

high-rate particle identification of high-energy heavy ions using a tilted electrode gas ionization...

Documents