counterth12. 10/ - estudo geral · gas proportional scintillation counter with xenonand mixtures...

GAS PROPORTIONAL SCINTILLATION COUNTER WITH XENON AND XENON MIXTURES

C.A.N. Conde, A.J.P.L. Policarpo and M.A.F. AlvesLaboratorio de Fisica, Universidade de Coimbra

Coimbra, Portugal

Summary

The light output of the recently deve-loped gas proportional scintillation counter- a gas scintillation counter with light mul-tiplication produced by a cylindrical geome-try electric field - is investigated for hea-vy gaseous media under alpha particle excita-tion with a 56 UVP phototube. The gases usedare Xe, and Xe-Ar and Xe-N2 mixtures for a

wide range of concentrations, at a totalpressure of about 965 Torr.

The light output of the Xe-Ar mixturesshifts towards the ultraviolet ( A < 3250 i)region as the electric field intensity increa-ses. Xe-Ar mixtures with Xe concentrationsranging from about 1 to 10/ give - with a

p-quaterphenyl wavelength shifter - a lightoutput more than two orders of magnitude lar-ger than that of a CsI(Tl) scintillator.

In Xe-N2 mixtures - with wavelengthshifter - nitrogen has a quenching effect.

For the mixtures of noble gases therise time of the secondary component of thelight pulse is slow, typically 15 As. Methaneand nitrogen while having a quenching effect,reduce this rise time.

Introduction

Gas scintillation counters, speciallythose using noble gases, are well known fortheir good time resolution and versatility ofthe detection media but, in general, the ener-

gy resolution is poorl13.Their behaviour when an uniform elec-

tric field is applied, has also been investi-gated4-6 and it was observed that, followingthe fast primary scintillation it appearsanother and much slower component which can

be used for energy resolution measurements.The uniform electric field geometry, althoughconvenient to study the mechanism of the pro-

cess, gives rise to different pulse heightsdepending on the orientation and location ofthe charged particle track. Therefore theenergy resolution is relatively poor.

The use of a cylindrical geometryelectric field has lead to the developmentof the gas proportional scintillation counter- PS counterTh12. Such a counter by confiningthe region of light multiplication to theimmediate vicinity of the central anode, cangive pulses with amplitude independent of theposition of the initial track. Their amplitu-de rises with the electric field. These pul-ses - secondary scintillation pulses - havea rise time dependent on the electron migra-tion time which is slow. When the electricfield is strong enough there might appear ter-tiary and higher order componentsl. Theseare the consequence of avalanches resultingfrom photoelectrons ejected from the walls ofthe counter by ultraviolet light produced inthe scintillation processes. We call lightmultiplication pulse the total scintillationpulse minus the contribution of the primaryscintillation, i.e. the pulse arising fromthose components due to the electric field.Its rise time being slow cannot be used forfast coincidence work but, as the primaryscintillation is not affected, there is stillthe possibility of obtaining the fast time re-solution characteristic of gas scintillatorswith no electric field.

To the increase of the light output,which can be quite large, corresponds an im-provement in the energy resolution, which islimited by the appearance of the higher ordercomponents8-10. The experimental energy reso-lution, for the argon-nitrogen mixture thatgives the largest light output seems to belimited by the amount of light reaching thephototube, provided that those higher ordercomponents are avoided. The energy resolu-tions already obtained are comparable withthose of a proportional ionization coun-ter819.

The performance of PS counters is sur-passed by semiconductor detectors, but forspecial applications, those for which gasscintillation counters and proportional ioni-zation counters are still used, the PS coun-ter might be advantageous even at the presentstage of development.

84

Authorized licensed use limited to: Universidade de Coimbra. Downloaded on March 15,2010 at 08:16:12 EDT from IEEE Xplore. Restrictions apply.

As little work has bee done with PScounters using xenon - an useful medium forX and gamma ray detection - we carried outthis investigation aimed at obtaining dataon the performance of pure xenon counters andof xenon counters with variable amounts ofargon under alpha particle excitation. Xenonmixtures are particularl interesting due totheir large light output 2. Small amouts ofnitrogen and methane were added in order toreduce the pulse rise time.

Experimental techniques

The gases used (commercially available)are: Xenon of purity 99.8% (impurities: Kr2000 v.p.n., N2 20 v.p.m., H20 10 v.p.m., 0210 v.p.m., CO+C02 10 v.p.m., CH4 5 v.p.m.)Argon of purity 99.99% (humidity < 10 mg/mi),Nitrogen 99.9% (02 ' 5 p.p.m., H20 < 5 p.p.m.- "dew point" < -72 °C, argon.1000 9.p.m.),Methane 99.98% (N2 0.02%).

The noble gases were continuously puri-fied with calcium turnings at 370 °C, in thesystem described beforel0. When working withnitrogen or carbon dioxide mixtures the tempe-rature of the purifier was reduced to 250 OCafter a previous purification at 370 °C oftenoble gases. It was verified that when wor-king with methane at these temperatures therewas a slow consumption of this gas. So mixtu-res with methane were made with continuouscirculation and the purifier temperature re-duced to 1250C. For all mixtures the totalworking pressure was always 965 Torr.

The geometry of the PS counter is theone of a proportional counter, apart from thesystem of light collection and detection. Thelight is observed with a 56 UVP photomulti-plier tube through a quartz window 1 cm thick.Detailed diagrams of the counters I and IIused in this work are found in refs. 10 and11 respectively, counter I being longer (17cm)than counter II (7cm). Counter I provides amuch better electric field geometry than coun-ter II, but has the disadvantage of a smallerlight collection efficiency. The diameter ofthe anode nickel wire used is 0.005" for com-ter I and 0.010" for counter II. Both countershave side aperture 2mm diameter through whichalpha particles are injected. Counter I aper-ture has a 0.001" thick Mylar window whilecounter II has no window. The sources usedwere ThC+ThC' and Am-241. The inside surfaceof the quartz window is covered with p-quater-phenyl ( < lOO ag/cm2) and the walls of coun-ter I (stainless steel) with a layer of thesame wavelength shifter ( . lOO pg/cm2). Thewalls of counter II (brass) have no deposit.

The photomultiplier anode resistancewas 4.7 MQ ; this allows a detailed observa-tion of the pulse structure. The pulses werestudied with a 585 Tektronix oscilloscope.Energy resolution measurements were made fee-ding the photomultiplier pulses through aWhite cathode follower into a linear ampli-fier (integration time 5 ps, differentiationtime off) and analysed with a single channelanalyser (HP 5583A).

Experimental results

Pure Xenon PS counter

Pulse shapes for pure xenon under Am-241 alpha particle excitation in counter IIwith no p-quaterphenyl wavelength shifter,are shown in fig. 1. Fig. 1 a, for no elec-tric field, shows only the primary scintilla-tion fast pulse. Fig. 1 b for 300 Volts inthe central anode shows the secondary scintil-lation pulse which follows that fast pulseafter a time lag of 12 )is, its rise timebeing about 10 ps. The time lag between theprimary and secondary scintillation is welldefined on account of the short range of al-pha particles in xenon. Otherwise only foralpha particles parallel to the central ano-de can the time lag be well definedl. Whenthe anode voltage rises there is a decreaseof the time lag and of the rise time of thesecondary component and an increase of itsamplitude. Fig. 2 shows the variation withanode voltage of the light multiplicationpulse amplitude. The plotted anplitude isthe total amplitude minus the contributionof the primary scintillation, rather thanthe total scintillation pulse9-l2. This cor-rection is made in order to enhance the in-fluence of the anode voltage on the lightmultiplication process.

By interposing filters with cut-offwavelengths of 3250 and 3800 2 10 the pulseamplitudes Vl and V2 corresponding to thelight that passed through the respectivefilter were compared to VO, the pulse ampli-tude with no filters. VO corresponds to wave-lengths above the sensitivity limit (1800 i)of the phototube. The ratios Vl/Vo and V2/Vosuitably corrected for the transmittance ofthe filters and the contribution of the pri-mary scintillation, are plotted, as rc(°,in fig. 2, as a function of the anode volta-ge. The rc for A> 3250 a and forX>3800 aare approximately the same. This neans that,for any anode voltage, the contribution ofthe light within the 3250-3800 i range issmall. For less than about 0.5 kV, rc decrea-

85


ses sharply but it remains approximatelyconstant for more than 1 kV. This implies apronounced shift towards shorter wavelengths,so that from 1 kV upwards only about 10' ofthe pulse height is due to light with A

> 3250 a. This shifting is attributed tothe excitation of higher energy levels by mo-re energetic electrons.

In view of this results it is to beexpected that, as far as the pulse height isconcerned, a wavelength shifter is advisable,even when a quartz window phototube is used.A layer of p-quaterphenyl ( S 100 ,g/cm2) wasvacuum deposited on the inside surface of thequartz window of the counter. Fig. 2 shows al-so the ratio, q, of the light multiplicationpulse height with and without p-quaterphenylas a function of the anode voltage. In the re-gion where there is a fast decrease of rc, qincreases rapidly, remaining practically cons-tant for larger anode voltages. Here the ef-fect of the wavelength shifter is to increasethe pulse height by a factor 8.

Xenon-Argon PS counter

In order to study the behaviour of xe-non-argon PS counters, several xenon-argonmixtures were prepared, continuously purifiedand circulated through counter II.

The light multiplication pulse ampli-tude(corrected for the energy loss of theAm-241 alpha particles in the gas channel cor-responding to the thickness of the counterwall) is plotted in fig. 3, as a function ofthe central anode voltage, for various xenon--argon mixtures. Two sets of curves are shownone for data with the inside surface of thequartz window covered with a p-quaterphenyllayer ( < 100 )jg/cm2) and another for datawithout such a layer. The effect of the wave-length shifter is to increase the pulse ampli-tude by about an order of magnitude. Besidesthe data shown in fig. 3, measurements werealso made for the following xenon percentages:99.2, 69.7, 23.1, 9.9, 2.5, 2.3, 0.25 and0.05, without p-quaterphenyl, and 9.7 and 2.5,with p-quaterphenyl. These curves are not plot-ted for the sake of clearness. For each curvethe maximum voltage used was limited eitherby a sudden and strong distortion of the pul-se shape or by sparking.

The general behaviour of the curves isthe same whether there i8 a wavelength shif-ter or not: a fast increase of the secondarypulse amplitude for voltages up to about 0.4kV followed by an increase at a slower rate.For xenon concentrations of a few percent therate of increase rises again for voltagesover about 1.5 kV. Two mixtures with a small

concentration of one of the components,Ar-G.05% Xe and Ar-99.2% Xe, give the samelight output than the pure gas, i.e. argonand xenon respectively. But while mixturesvery rich in argon like Ar-43.8% Xe, giveabout the same light output as does pure xe-non, mixtures like Ar-0.51% Xe, that have asmall content of xenon, give a light outputmuch larger than that from pure argon (fig.3).

The variation of the light output withthe concentration is voltage dependent as canbe' seen in fig. 4 where a family of curvesfor 1.0, 1.4 and 1.8 kV is plotted. The lightoutput is practically constant for xenon con-centrations larger than about 40/o. It risessharply from 0.1 to a few percent. For a xe-non concentration of around 3%/ the light out-put, at relatively high voltages, reaches amaximum that is a sharper for 1.8 kV than for1.4 kV, specially when the wavelength shifteris used(fig. 4). As it has already been poin-ted out12 it is for argon-xenon mixtures inthis region of concentrations that a gas pro-portional scintillation counter gives one ofthe largest light yields.

Fig 5 shows the ratio, rc (%), forA>13250 a for three anode voltages as afunction of the xenon concentration. As itshould be expected, at any concentration the-re is a decrease of rc when the voltage in-creases, the curve for 1.8 kV lying below theone for 1.0 kV and this one below the curvefor 0.5 kV; this corresponds to a shiftingtowards shorter wavelengths. The decrease ofrc for xenon concentrations from about 0.1 toa few percent implies a relative increase ofthe light output from 1800 to 3250 i. Furtherincrease in the xenon content does not changeappreciaby the proportion of light withinthis range. The comparison of figs. 4 and 5shows that the ranges of concentrations forwhich these features occur are the same asthose for which the pulse amplitude first ri-ses sharply and thereafter remains constantfor not very high anode voltages.

Xenon-Nitrogen PS counter

Counter I was used as before withits window and walls covered with p-quaterphe-nyl in order to study the behaviour of xenon--nitrogen mixtures. The light multiplicationpulse amplitude versus anode voltage is plot-ted in fig. 6. For large nitrogen concentra-tions (dashed curves for 16.5 and 33.8% N2)the plotted pulse amplitude includes the pri-mary scintillation because this was too smalland of the order of the noise. Whatever theanode voltage nitrogen has a quenching effectas it had for the argon-nitrogen counter des-

86


cribed beforell. Nevertheless the nitrogenquenching effect is less accentuaded for xe-non than it was for argon. Fig. 7 shows theeffect of nitrogen on the amplitude and therise time of the light multiplication pulsefor 2.2 and 1.4 kV. Nitrogen concentrationsof about 1% have a practical interest sincethey reduce the rise time from about 18 to 7,us, while reducing the amplitude only , 30%/.Larger concentrations do not change apprecia-bly the rise time while decreasing drastical-ly the pulse amplitude. The observed decreasein rise time is to be expected on the groundsthat, as it is well known, the drift velocityof electrons in noble gases is reduced whenimpurities like nitrogen and methane are ad-ded. This suggests the study of the effect ofthese impurities in a mixture giving a largerlight output like the Ar-5% Xe one.

Fig. 8 shows the rise time and thelight multiplication pulse amplitude versusnitrogen or methane content in this mixture.The quenching effect of nitrogen is strongerfor this mixture than it was for pure xenonand the rise times are about the same. Metha-ne while reducing the rise time more than ni-trogen has a very strong quenching effect.For example, for 2% of methane the rise timeis only 1.5 ps, but the pulse amplitude hasdecreased already by three orders of magnitu-de.

A similar behaviour was observed whenmethane was added to pure xenon.

Practical light output and energy resolutions

The arbitrary units scale used in thiswork for pulse amplitudes iS the same (within10%) as for the previous works1'12. In thisscale the primary scintillation pulse ampli-tude for pure xenon was found to be 2.8 unitsin counter I and 6.4 units in counter II(counters with wavelength shifter and the ano-de at zero Volts). The different outputs ofthe two counters result from different lightcollection efficiencies.

Pure xenon in counter I at 1.4 kV gi-ves an amplitude of 400 units, a gain ofabout 140 over the primary scintillation; thesame gain is observed in counter II. This isabout 80 times the pulse amplitude from aCsI(Tl) scintillatorl2. At 1.8 kV this lastfigure rises to 120 but it was difficult tosee whether tertiary components were present.From our observations we estimate that thepractical light output - defined as the outputto which corresponds the largest pulse obtain-ed with no tertiary components12 - is, for pu-re xenon, about two orders of magnitude largerthan the CsI(Tl) light output.

Xenon-argon mixtures can give largerpulse amplitudes, over a th6usand units, spe-cially on the region 3 to 5% of xenon con-tent. The practical light output of anAr-5°% Xe mixture (counter I) is reached for1.46 kV to which corresponds an amplitude of1800 units i.e. over 300 times the one forCsI(Tl).

To practical light outputs as large asthese should correspond a substantial impro-vement in the energy resolution. The bestresolution obtained (counter I) was 3.3% foran Ar-5% Xe mixture at 1.36 kV and E. = 6.7MeV; for pure xenon and Xe-l% N2 mixture theenergy resolutions were only 5 or 6%. Thesefigures are not corrected for the stragglingin the Mylar window.

Due to the low light collection effi-ciency of the counter, the amount of lightthat reaches the phototube is only a frac-tion of the total light. This fact and thepossible instabilities in the multiplicationprocesses may explain the relatively poorresolutions obtained.

Conclusions

Gas proportional scintillation coun-ters using xenon and xenon-argon mixturesare found to give, with a wavelength shifter,light outputs that may be more than two or-ders of magnitude larger than the ones fromstandard inorganic scintillators.

The maximum light output for xenon-ar-gon mixtures is obtained for a xenon concen-tration of 3-55%. It is interesting to pointout that this is the range of concentrationsfor which Golubnichii and Yakolev13 whileinvestigating xenon-argon mixtures for sparkchambers found a marked minimum for the so-called breakdown potential. The variation ofthe breakdown potential with xenon concentra-tion seems to be related with the variationwe found for the light multiplication pulseamplitude.

The spectral distribution of the se-condary scintillation light is found to de-pend on the anode voltage. For xenon this isin apparent disagreement with the results ofTeyssier et al.14 which did not observe sucha dependence for uniform electric fields.However, in PS countersthe electric field atthe anode surface is typically 20 kV/cmwhich is much stronger than the fields usedby those authors. Any future attempt to in-terpret the mechanism of a PS counter has totake into account the scintillation spectrafor such high fields. At present there is nodata available on this subject.

The slow pulse rise time of PS coun-

87


ters with pure noble gases and their mixturescan be reduced by the addition of smallamounts of nitrogen and methane but this isachieved at the expense of a large reductionin the pulse amplitude. Nevertheless pulserise times of 7 to 5 ps are attainable withno drastic reduction in the pulse amplitudeif 1 or 2% of nitrogen is added to pure xenon.These mixtures might have practical interestfor X and gamma ray detection.

The energy resolutions obtained arecomparable with those from standard scintilla-tors. However, we expect this characteristicto be improved when a more efficient lightcollection system is devised and when the pro-blem of the suspected instabilities of thelight multiplication process is dealt with.Work in progress at this Laboratory is beingcarried out along these lines.

Thanks are due to Prof. J.R. de AlmeidaSantos for his interest in this work.Financial support given by Comissao de Estu-dos de Energia Nuclear do Instituto de AltaCultura, Portugal, as well as funds suppliedby the Calouste Gulbenkian Foundation, Lisbon,and by the NATO Scientific Committee are gra-tefully acknowledged.

References

1. J.L. Teyssier, D. Blanc and A. Godeau,J. Phys. Radium 24 (1963) 55.

2. J.B. Birks, "The theory and practice ofscintillation counting" (Pergamon Press,Oxford, 1964).

3. A. Sayres and C.S.Wu, Rev. Sci. Instr. 28(1957) 758.

4. L. Koch, Rapport C.E.A. No. 1532 (1960),Centre d'Etudes Nucleaires de Saclay.

5. J.L. Theyssier, D. Blanc and H. Brunet,Nucl. Instr. Meth. 33 (1965) 359.

6. J.L. Teyssier, D. Blanc, J. Brunet and A.Godeau, Nucl. Instr. Meth. 30 (1964) 331.

7. G. Charpak and G.A. Renard, J. Phys. Ra-dium 17 (1956) 585.

8. C.A.N. Conde and A.J.P.L. Policarpo,Diploma Report (University of Manchester,1960), unpublished.

9. C.A.N. Conde and A.J.P.L. Policarpo, Nucl.Instr. Meth. 53 (1967) 7.

10. A.J.P.L. Policarpo, M.A.F. Alves and C. A.N. Conde, Nucl. Instr. Meth. 55 (1967)105.

11. M.A.F. Alves and A.J.P.L. Policarpo,Nucl. Instr. Meth. _57 (1967) 321.

12. A.J.P.L. Policarpo, C.A.N. Conde and M.A.F. Alves, Nucl. Instr. Meth. 58 (1968)151.

13. P.I. Golubnichii and V.I. Yakolev, Instr.Exp. Techn. (USSR) No. 2 (1966) 311.

14. J.L. Teyssier, D. Blanc and J.P. Boutot,J. Phys. (France) 28 (1967) 427.

(A)

(B)

Fig. 1. Pulse shape for Am-241 alpha par-ticles in xenon at 965 Torr (counter IIwithout wavelength shifter): (a) no elec-tric field (hor.: 10 us/div.). (b) 300 Vanode voltage (hor.: 10 us/div.).

88


Xe-A

rPS

Coun

ter

IIp:965

Torr

E:.<

5.1MeV

Ar-4.9S%Xt

Ar-1.

22%Xe

/_o

-Ar-

4I.9

%xe

//eAr-z%Xt

v~~~~

,,

Ar-S.51%X

voo+

Ar-

1.22%

Xe"U

-,/

Ar

it;;*<

,I

o

Ia,

/_

140%

AP

-_____

windcw

with

p-quaterphenyl

-without

"

ANODE

VOLTAGE(k

V)

Fig.

2.Characteristics

of

the

pure

xenon

proportional

scintillation

counter

as

a

function

ofthe

anode

voltage

(counter

II).

The

errors

shown

are

estimated.

crosses:

light

multiplication

pulse

ampli-

tude

(no

p-qu

ater

phen

yl).

full

circ

les:

rati

o,rc,

(see

text)

ofthe

amplitudes

with

and

without

filter

forx>

3250

A.open

circles:

rati

o,rc,

forA-3800-.

full

squares:

wavelength

shifter

efficien-

cy

parameter,

q,

(see

text).

lI

II

I-,

1-.

tI

iI

-IAL0

ANODE

VOLTAGE

(kV

)

Fig.

3.Light

multiplication

pulse

ampli-

tude

versus

anode

voltage

for

various

xe-

non-argon

mixtures

(counter

IIwith

and

without

p-quaterphenyl).

The

errors

shown

are

estimated.

1- z tl: or t w J

I10

-

c: D10*

Is

0

uT

17

II

II

II

.4

I1

Q.5

1.0

1*5

2.5


LD 0

.4PE

RCEN

TAGE

OFXe

pureXe

pureAr

Fig.

4.Light

multiplication

pulse

ampl

i-tude

versus

xenon

conc

entr

atio

n,for

vari-

ous

anode

voltages

(counter

IIwith

and

without

p-qu

ater

phen

yl).

The

errors

shown

are

esti

mate

d.

rc% soI

n.-

.....................................

-0.1

1000

*/.

PERCENTAGE

OfXe

Fig.

5.Ra

tios

,r0

,(see

text)

ofthe

am-

plitudes

with

and

without

filt

er(A>

3250

A)versus

xenon

concentration

inar

gon,

for

0.5,

1.0

and

1.8

kV(counter

IIwi

thno

p-quaterphenyl).

The

errors

shown

are

esti

mate

d.


r I I I T I I I I I I I

Fig. 6. Light multiplicationpulse amplitude versus anodevoltage for various xenon-ni-trogen mixtures (p-quaterphe-nyl in counter I window andwalls). Dashed curves are notcorrected for the primary scin-tillation. The errors shownare estimated.

Us

=) id'

4-C

-Jw qdaJ

-JD 10a.

100

0.1 10

PERCENTAGE OF N2 IN Xe

Xe-N2 PS Counter IPs 965 TorrE.,z6.7 M*V

Xe-2.4%N2

. Xe-3.7%N2

Xe-5. % N2

Xe- s6% ml

-. xe-16.5%N

Xe- 33.S%N2.-

r I I0.5

I3C 20

un

lz

trtra]<s

10 Jia.7

uJ

tnJ)

15

I

1-

!2 10

Jx

0Oe%

1.0 1.5

Ar-5% Xe PS Counter Iwith N2 or CH4

p-965 TorrE.,:6.7 MeV

Anode Voltage 1.4 kV

2.0 2.5ANODE VOLTAGE (kV)

Ar -5%X -rise timewith Nz | -ampl tude

Ar- 5% Xe - rise timewith CH4 A- amplitude

Ar-5%Xe -rrise time(no N, or CH4,1 amplitude

.1

0.1 10

PERCENTAGE OF N2 OR CH4 IN A Ar-5%Xe MIXTURE

Pre Xe

3.0

_ 103

I

1002/u

iz-lo tr

a:

_ cz4

_ IL

10100%

Fig. 7. Rise time and amplitude of thelight multiplication pulse, versus nitro-gen content in xenon for 1.4 and 2.2 kVanode voltage (counter I with p-quaterphe-nyl). The errors shown are estimated.

Fig. 8. Rise time and amplitude of thelight multiplication pulse, verus nitrogenor methane content in an Ar-5% Xe mixturefor 1.4 kV anode voltage (counter I withp-quaterphenyl).

91

20

15

I

O-

u 10

5

o Pure Xenon

a

Xe-N2 PS Counterp:965 TorrE.e 6.7 Me V

{o - rise time1.4 kV

* - amplitude

O- rise time2.2 kV

*- amplitude

...............11.11 ................... I 'I I I I

0 L ................. .1 ..I 0 L. ............... .-I- j 1- 'I --I -.

-4

CN2

-1

zx.,. -.1--1I

I

.1III

I

I

5


counterth12. 10/ - estudo geral · gas proportional scintillation counter with xenonand mixtures...

Documents