arXiv:hep-ex/0106104 v2 17 Jul 2001SubMeVParticlesDetectionandIdenti�cationin

theMUNU

detector

TheMUNUcollab

orationM.Avenier

1,C.Broggin

i3,J.Busto

2,C.Cern

a3,V.Chazal

2,P.Jean

neret

2,G.Jon

kmans2,

D.H.Koang1�,J.Lam

blin

1,D.Lebrun1,O.Link4,R.L�uesch

er2,F.Ouldsad

a4,G.Puglierin

3,A.Stutz1,A.Tadsen

3,J.-L

.Vuilleu

mier

2

1Institu

tdes

Scien

cesNucl�ea

ires,IN

2P3/CNRS-UJF,53Aven

uedes

Marty

rs,F-38026Gren

oble,

France

2Institu

tdePhysiqu

e,A.-L

.Bregu

et1,CH-2000Neuchatel,

Switzerla

nd

3INFN,Via

Marzo

lo8,I-3

5131Padova,Ita

ly4P

hysik-In

stitut,Winterth

urerstra

sse190,CH-8057Z�urich

,Switzerla

nd

Werep

orton

theperform

ance

ofa1m

3TPC

�lled

with

CF4at

3bar,

immersed

inliquid

scintillator

andview

edbyphotom

ultip

liers.Particle

detection

,eventidenti�

cationandlocalization

achieved

bymeasu

ringboth

theelectric

curren

tandthescin

tillationligh

tare

presen

ted.Particu

larfeatu

resof�particle

detection

arealso

discu

ssed.Finally,

the54M

nphotop

eak,recon

structed

fromtheCom

pton

scatteringandrecoil

angle

isshow

n.

PACSnumbers:

29.40.+G+M,13.15

,26.65,

96.60.J

I.INTRODUCTION

Large

timeprojectio

ncham

bers

(TPC)have

been

used

inhigh

energy

particle

experim

ents

ascen

traltrackers

andmore

recently

foridenti�

cationofheav

ynuclei

and

radioactiv

ebeams[1],[2].

Atlow

erenerg

y,gas

TPCwere

operated

both

forparticle

trackingandcalorim

etry.A

TPC

hasbeen

used

inan

experim

entsearch

ingfor

�-e

conversio

n[3].

IntheMeV

energy

range,

theTPC'sin-

volved

areofsm

aller

size.First

doublebeta

decay

spectra

were

obtain

edbyElliot

etal.

[4]usin

gaTPCof

about

100l.

IntheGoth

ard

experim

ent[5],

aTPC

with

anactive

volumeof200lprov

ided

oneofthebest

limits

onneutrin

oless

doublebeta

decay

in136X

e.TPCwith

opti-

calread

outwere

also

consid

eredfordark

matter

searches

[6],[7].

TheMUNU

collaboration

has

built

a1m

3gas

TPCfortheexperim

entalstu

dyof

�e e�scatterin

gnear

anuclear

reacto

rin

Bugey

(France).

Thedetector

isdesign

edto

identify

recoilelectron

sandrecon

struct

par-

ticletra

ckswell

below

1MeV

.Thetrack

ingcap

ability

oftheTPC

(angle,

thresh

old,andtra

ckcontain

ment)

isessen

tialforbackgro

undrejection

andallow

sonlin

ebackgrou

ndmeasurem

ents.

TheTPC

vessel

ismadeof

transparen

tacry

licandisinstalled

insid

ea10m

3ligh

tcollection

vessel,

�lled

with

liquidscin

tillatorandview

edbyphoto

tubes.

Both

theelectrec

curren

tandthescin

-tillation

lightof

particles

interactio

nin

theTPCgas

canbecollected

.In

thefollow

ingsectio

n,adescrip

tionof

thedetecto

ris

given

whereas

insectio

nIII

theenergy

andeÆ

ciency

calib

ratio

nsof

thedetector

are

describ

ed.

Waveform

sandeventidenti�

cationare

presen

tedin

sec-

tionIV.In

sectionVtheenergy

andangular

resolution

sare

discu

ssed.Incid

entgam

maenergy

reconstru

ctionis

alsopresen

ted.Finally,

conclu

sionsare

givenin

sectionVI.

II.THEMUNUDETECTOR

Thegen

eralcon

ceptof

theMUNUexperim

entandits

detector

have

been

presen

tedprev

iously

[8].Here

wegive

anupdate

anddiscu

sstheperform

ances

ofthedetector.

Thedetector

consists

ofacen

tralTPC

�lled

with

CF4

gas,insid

eaperip

heral

lightcollection

vessel�lled

with

10m

3of

liquid

scintillator.

Itis

equiped

with

photo-

multip

liersandcan

actas

veto

forCom

pton

andcosm

iceven

ts.Thedetector

show

nin

Figu

re1isinstalled

in-

sidelow

activity

passive

shield

ings

composed

ofborated

polyeth

ylen

e(8

cmthick

)andlead

(15cm

thick

).

II.1

TheTPC

The1m

3tim

eprojection

cham

ber

canbe�lled

with

CF4gas

atapressu

rebetw

een1and5bar.

Thisgas

was

chosen

becau

seof

itshigh

density

(1.06�10

21electron

sper

cm3at

STP)andits

lowZ.Thisgives

ahigh

targetdensity

combined

with

relativelylittle

multip

lescatter-

ing.

Most

data,

inparticu

larall

those

reported

inthis

paper,

havebeen

takenat

apressu

reof

3bar,

atwhich

electrontrack

scan

bereason

ably

well

resolveddow

nto

300keV

.

�Corresp

ondingauthor.

Tel.:

+33

47628

4048;fax

:+33

47628

4004e-m

ailkoan

g@isn

.in2p3.fr

1

The gas is contained in a cylindrical acrylic vessel of90 cm inner diameter, and with an inner length of 162cm. A cathode made from high purity electrolytic copperfoil is mounted on one lid. A read-out plane consisting,from inside to outside, of a grid, an anode plane and anx-y pick-up plane, is mounted on the other side. Thearrangement is shown in Figure 2. The grid wires have adiameter of 100 �m. The distance between grid wires is4.95 mm. The anode plane has 20 �m anode wires, sep-arated by 100 �m potential wires. The distance betweenanode and potential wires is also 4.95 mm. The distancefrom the anode to the x-y plane is 3 mm whereas thegrid and the anode plane are separated by a 8.5 mm gap.The spacing between x strips on one side of the support-ing PET foil, or y strips on the other side, is 3.5 mm,giving in total 256 x and 256 y strips.The drift �eld is de�ned by the cathode, the read-out

plane, primarily the grid, and the �eld shaping ringsmounted on the outside of the acrylic vessel. The x-y plane is grounded through the read-out preampli�ers.The drift time and drift velocity are optimized in orderto get the best spatial resolution. At 3 bar, we are oper-ating the TPC with a drift �eld of 206 V/cm. This givesa drift velocity of 2.14 cm.�s�1, and a total drift time of74.8 �s. This was determined from the total drift time ofmuon tracks spanning the entire chamber, from cathodeto grid. The voltages applied to the last �eld shapingring, the grid, the anode and the potential wires mustlead to suÆcient ampli�cation around the anode wires.Moreover they must provide a symmetric �eld con�gu-ration around the anode wires to minimize mechanicalstrain and sagging, and a homogeneous drift �eld all theway down to the grid over the entire area. The Gar�eldcode was used to �nd a con�guration meeting these re-quirements. It is shown in Figure 2.Gas purity is essential for good charge collection, but

also to minimize contributions to the background fromradon emanations. The CF4 gas is therefore constantlycirculated at a ow rate of 500 l/h through a commercialhigh temperature getter made by SAES [10] followed bya cold trap. The cold trap consists of a coil tubing fol-lowed by a a cell �lled with 50 g of active charcoal. Itis immersed in an ethanol bath kept at a temperature of190 0K.The signals from the x and y strips and the anode are

ampli�ed in current voltage preampli�ers and sampledat 12.5 MHz (80 ns sampling time) in 12 bit FADC'swith logarithmic response. Electronics noise on the xand y strips is serious, because of the length, over 2 m, ofthe at signal cables from the TPC to the preampli�erslocated outside the polyethylene, lead and steel shield-ing. A simple procedure allows to signi�cantly reduce it.First a Fourier transform of the signals is performed. Fre-quency peaks identi�ed as noise sources are eliminated.The loss in spatial resolution is acceptable. To give anidea of the quality of the tracks, an example is shown in

Figure 3.

II.2 The external light collection detector

The TPC vessel is immersed in a cylindrical steel tankpainted with TiO2 for di�use re ection and �lled witha mineral oil based liquid scintillator (NE235, providedby Nuclear Enterprise, Scotland). The scintillator hasan attenuation length of 8 m at 430 nm and is viewedby 48 hemispherical photomultipliers of 20 cm diameter(EMI 9351 low activity series, provided by HamamatzuPhotonics), subdivided in two groups of 24 each on theanode and cathode sides. The mean quantum eÆciencyof the photocathodes is around 25 %. The PMT's aresampled with the same FADC's as the TPC anode andx-y strips. Radioactive sources can be introduce into thelight collection vessel at an axial distance of 26cm fromthe center of the detector, on the anode side through anacrylic pipe. The radial position can be as close as 1 cmfrom the TPC lateral wall.The main role of the external detector is to identify and

to reject the Compton electrons induced by rays com-ing from outside as well as from inside the TPC. Since theTPC walls are made of transparent acrylic, the scintilla-tion light of the CF4 can also be collected. Within theTPC, the primary scintillation light of CF4 can be emit-ted anywhere while avalanche scintillation light is pro-duced only in the vicinity of the anode plane during theampli�cation process. The scintillation light spectrum ofCF4 has a narrow maximum at 160 nm and a broaderone around 300 nm [13]- [15]. The light is detected bythe phototubes through the acrylic walls and the liquidscintillator. Due to the transmittance of the acrylic andof the liquid scintillator, only a fraction of about 10 %of the spectrum with wavelengths larger than 400 nmmakes signi�cant contribution to the light signal in ourdetector.

II.3 The readout system

As already mentionned, anode, x-y strips and photo-tube signals are read out with 8 bits ash ADC's withlogarithmic response. The memory size is 1024 words.The sampling interval is 80 ns and is given by a commonclock. This sampling time is well ajusted to the electricpulses from the anode which have a relatively slow risetime of few hundred ns. Signal pulses from the photo-tubes associated with Compton electrons are much faster(< 30 ns) and are shaped to meet the sampling frequency.The digitized data of the ash ADC's provide the pulseshape of the signals delivered by the phototubes, the an-ode, and the x-y strips. This way, projection on the x-yplanes can be reconstructed. Time bins can be converted

2

to relative position coordinate along the TPC Z-axis us-ing the drift velocity in the TPC (2.14 cm.�s�1 in normalrunning conditions).The scintillation light produced in the liquid scintilla-

tor or the primary scintillation light of the CF4 whichis strong enough to be seen for � particles and not fornearly minimum ionizing electrons and muons, are col-lected within one or two sampling bins of 80 ns. Chargesarrive to the anode with a time delay proportional totheir distances from the anode plane. Absolute z positioncan thus be de�ned for events with prompt scintillationlight. For events such as con�ned single electrons, whichhave no primary scintillation light in CF4 and no interac-tion with the liquid scintillator, only relative z positionscan be obtained.

III. CALIBRATIONS OF THE DETECTOR

In the following, we �rst discuss the light collectionand energy calibration for the liquid scintillator detector.In particular, we discuss the ray detection eÆciency andwe compare the measurements to simulations. We thendescribe the energy calibration of the TPC.

III.1 Energy calibration and light collection in the

external detector

In the light collection vessel, rays undergo multipleCompton scattering and are in most cases totally ab-sorbed within the scintillator liquid. Radioactive sources(54Mn, 135Cs, 22Na) are used for light collection and en-ergy calibrations. Along the cylindrical vessel axis (z),a convenient position parameter is the assymetry (i.e.di�erence-to-sum ratio) of light signals collected by pho-totubes of the cathode and anode sides. The light col-lection has been measured as function of this parameterusing the total absorption peak of a 54Mn source.Experimental data agree with simulations based on

known emission spectrum and attenuation length of theliquid scintillator, quantum eÆciency of the phototubes,and optical properties of other components. At the cen-tral part of the vessel, the response is quite homogeneousover a large region (� 40 cm from the center along thedetector axis) where about 7.5% of photons emitted inthe liquid scintillator are collected. At 80 cm from thecenter, close to both ends of the TPC, the fraction of lightcollected is 9.5%. It increases to 15 % at 120 cm from thecenter of the detector. Position dependence of the lightcollection can be corrected for an optimum energy reso-lution or exploited for event localization. The quantityof photoelectrons collected was also evaluated from totalabsorption spectra of 54Mn measured for each individualphototube and compared to single photoelectron spectra.For an energy deposition of 1 MeV at the central part of

the detector, 144 � 8 photoelectrons were collected in the48 phototubes while 150 were predicted. Total absorp-tion spectra obtained with di�erent radioactive sourcesand corrected for position dependence are compared tosimulations in Figure 4. The energy resolution is 13%at 1 MeV. In the following text, only corrected values ofenergy will be used.

III.2 's detection eÆciency

CF4 is composed of low Z atoms, only a small amountof very low energy rays will lose the entire energy inthe gas. Most of them undergo a single Compton scat-tering in the gas and are absorbed in the scintillator.The MUNU detector has been designed for an optimumdetection of rays associated with a Compton electroninside the TPC. To minimize absorption, the lateralwalls of TPC are made of very thin (5 mm) acrylic. Atan energy, threshold set at 100 keV which corresponds toa total collection of 15 photoelectrons, the contributionsof dark noise from individual phototubes is not neglige-able. For this reason a multiplicity criteria asking forat least 5 phototubes hit is required for a trigger. Thecounting rate for the 10 m3 of liquid scintillator withinthe shielding is 710 s�1 including 270 s�1 coming fromcosmic muons. The energy spectrum measured at � 100keV (low) threshold shown in Figure 5 corresponds to acounting rate of 0.05 s�1 kg�1 of liquid scintillator. Sig-nals from the scintillator can be applied to collect or toveto TPC events associated with a scattered ray. (Sim-ilarly, a signal at an energy threshold of � 8 MeV electronequivalent can be applied to veto cosmic muons). Thedetection eÆciency of scattered has been measured byusing a 54Mn source of 15 kBq inserted at 1 cm from theTPC wall. Electron energy spectra were collected withand without the anti-Compton veto operating at 100 keVhardware threshold. The anti-Compton eÆciency is eval-uated from the ratio of the two electron spectra, correctedfrom background contributions measured in the absenceof the radioactive source. Only 3% of the associatedwith a Compton electron of more than 300 keV escapedetection. The scattered photon is lost when it does notinteract with the liquid scintillator or when it is absorbedby inert material inside the vessel(acrylic walls, cables).Figure 6 shows the comparison of the experimental re-sults to simulations at two thresholds: 100 and 125 keV[17]. For a ray coming from outside the light collectionvessel, the ineÆciency is much smaller, since it has totravel twice through about 50 cm of liquid scintillator.Taking also into account the solid angle and the absorp-tion due to the liquid scintillator, the simulation showsthat the probability to miss a ray leaving more than300 keV in the TPC is less than 10�6.

3

III.3 Energy calibration of the TPC

Particle tracks in the TPC produce charges which driftto the anode plane. There the charge is ampli�ed andavalanche light is produced. Light emission during themultiplication phase in CF4 has been previously observed[14]. In our detector, about 2.104 photoelectrons are col-lected per MeV of electron. Electric current signals aremore sensitive to electromagnetic noises and have largerpedestal uctuations. Current and light waveforms fromthe same electron event will be shown in the followingsection. Avalanche light signals have better bin to binpedestal stability and are used instead of the electric cur-rent signals for energy evaluation and for determinationof track ends. An event per event comparison shows thelinear correlation between the amplitudes the two typesof signals as given in Figure 7. Compton electron spec-tra were obtained using a 54Mn source both from theelectric current and the light and are compared to sim-ulations in Figure 8. Mean energy loss of muon eventsper unit length in the TPC is used to monitor day-to-daygain variations (Figure 9). The gain changes on the an-ode plane are mapped with alpha and muon events andare presented in Figure 10.

IV. WAVEFORMS AND EVENT

IDENTIFICATION

IV.1 Muon events

The experimental area in Bugey has an overheadshielding in heavy concrete equivalent to 20 m of wa-ter. Compared to ground level, the cosmic muon ux isreduced by a factor of �ve. The muon counting rate isexpected to be 65 s�1 in the TPC and 270 s�1 in the lightcollection vessel. A cosmic muon attaining the TPC hasto cross more than 100 cm of liquid scintillator. With amean energy lost of about 1.8 MeV/cm, it leaves a verylarge prompt signal in the phototubes. This signal canbe used as trigger to select muon events or to veto muoncorrelated events. The energy loss of cosmic muons in theCF4 covers an energy range from hundred keV to a fewMeV. The muon mean energy loss per unit length in theTPC can be used to continuously monitor the gain varia-tions in a complementary way to the measurements withradioactive sources. Muon events crossing both cathodeand anode have x-y projection con�ned within the TPCradius and are used to evaluate the drift velocity. Thetime di�erence between the beginning and the end of thetrack correponding to a maximum drift path of 1.62 m.The x-z, y-z projections of a muon event are shown to-gether with the x-y projection in Figure 11.

IV.2 Alpha particles

IV.2.1 Primary light emission

Alpha particles from natural radioactivity have a rangeshorter than 1 cm in CF4 at 3 bar : 4.1 mm at 5 MeVand 8.6 mm at 8 MeV. The spatial extensions of � eventsare limited to a few x-y strips and a few time bins. Whileno primary scintillation light was seen for Compton elec-trons at a threshold as low as 30 keV electron equivalent,the light from � particles was detected. A prompt sig-nal equivalent an 153 keV electron (22 photoelectrons)is produced in the liquid scintillator by a 5.3 MeV �

particles. In taking into account the light collection andphotocathode quantum eÆciency, we �nd 1100 photonsper � particle or 207 � 30 photons/MeV. This is com-patible with values of 220 photons per Mev in the rangeof 360-600 nm measured at lower pressures [14].

The time di�erence between primary light andavalanche light signals of alpha particles emitted fromthe cathode provides an other measurement of the driftvelocity.

IV.2.2 Suppression of charge collection

The total charge (or avalanche light) collected for �

particles is much smaller than for electrons of the sameenergy. A suppression factor as high as 28 has been mea-sured for � particle emitted close to the anode plane.This factor however decreases when the drift time (i.e.distance from the anode) increases (Figure 12). For al-pha particles emitted from the cathode, the suppressionfactor is only 5.6. The charge suppression is very likelydue to an screening e�ect in the avalanche region whilethe change of the suppression factor may be a conse-quence of electron di�usions along transverse(T) as wellas longitudinal(L) directions. In the MUNU TPC, for adrift length of 1.62 m from the anode to the cathode, thedispersion �L or �T is 2 mm, about half of the range of� particles at 5 MeV. The smearing of the charge den-sity along the drift path could than attenuate the chargespace screening e�ect in the avalanche region.

IV.2.3 Radon identi�cation and detection

Radon was introduced accidentally into the TPC byone of the Oxysorb [9] puri�ers used to clean the CF4from oxygen and water. A maximun rate of 37 s�1 wasmeasured. Alpha particles detected in the TPC camefrom the decay of 222Rn (E�=5.48 MeV) and its radioac-tive daughters. In the radon 222Rn decay chain, the betafrom 214Bi is followed by the � from 214Po(half life 160�s). Such delayed coincidence events can be observed

4

easily within the 80 � s range of our ash ADC's. A cor-related � � � event attributed to the 214Bi- 214Po decaycascade is shown in Figure 13. When the Oxysorb wasremoved from the gas circuit, both the rates of � eventsand of correlated �-� events from 214Bi-214Po decay cas-cade decreases with a period of 3.2 days compatible withthat of 222Rn decay period (Figure 14). After the CF4gas change, residual � events were still observed at asteady level of � 1 mBq/m3 of 222Rn in the gas and anactivity of 17 �Bq/cm2 from the cathode. For these lastevents, a well de�ned drift time and a well de�ned en-ergy peak at 940 keV electron equivalent were observed.These events can be attributed to � particles from 210Podecay, a radioactive daughter of 210Pb (half life 22 years).Radioactive daughters of 222Rn were very likely driftedto the surface of the copper cathode but apparently notdeeply implanted into the metal. The cathode copper foilhas now been replaced by a new one, the surface of whichwas chemicaly etched with radiopure chemicals before in-stallment. The rate of � from the cathode was reducedby a factor of 10.

IV.3 Electrons

IV.3.1 Compton electrons

Compton electron events generated within the TPChave in most cases an associated scattered ray detectedin the light collection vessel. The sequence for a typicalevent, shown in Figure 15 is the following:- �rst, scintillation light due to the absorption of the scat-tered ray in the liquid scintillator is detected,- then, when ionisation charges of the Compton electronhave drifted to the anode plane, avalanche light is mea-sured with a time delay proportional to the distance fromthe anode,- almost simultaneously electric current signals are mea-sured on the anode and x-y strips.The display of the event is completed by the reconstruc-tion of the projections in the x-z/y-z and x-y planes.The localization of the gamma along the axis of theanti-Compton can be evaluated from the assymetry ofthe light collection(see section III.1). The distributionof time di�erences of Compton electrons and associatedscattered 's is at along the TPC which indicates thatthe background is homogeneous along the TPC. A sharpdrop corresponds to the end of the chamber. The driftvelocity extracted this way is compatible with the muonresult.

IV.3.2 Electrons from the anode plane

For such events, charges are created next to the anodewires. Thanks to a larger electric �eld and thus faster

drift time between the grid and the anode the chargescreated in this region are collected much faster. The re-sult is a charge blob at the end of the track on the anodeside. If the electron stops inside the gas volume, thenormal blob due to increased ionization before stoppingwill be seen in addition at the true end, resulting in a"double blob" event. Normal blobs have a higher inte-grated charge than anode blobs while anode blobs havecharges accumulated in the �rst or the second time binsat the start of the track. The avalanche light pulses areused to detect the sharp rise of pulse height and identifytracks originating from the anode plane. A typical "dou-ble blob" event is shown in Figure 16.A distribution of pulse heights for an unselected set ofelectron events is shown in Figure 17 together with a dis-tribution of Compton electron events which do not crossthe anode. This last criteria is full�lled by requiring anassociated scattered ray signal well separated from theavalanche signal. For a well ajusted cut (1500 mV or �20 keV), only 1.5% of normal Compton electron will bewrongly identi�ed as originating from the anode. Thedirection distribution of "double blob" events, projectedin the yz plane(Figure 18) shows that they are indeeddominantly originated from the anode side.

IV.3.3 Electrons from lateral walls

Electrons emitted from lateral walls can be identi�edfrom their xy projections. In practice, a �ducial radiusis de�ned to reject events having tracks touching theoutermost x or y strips. Only electrons coming fromthe cathode wall are not distinguishable from containedelectrons of the same direction.

IV.4 Contained single electron events

The MUNU set-up is optimized for the detection oflow energy single electron events in particular those fromneutrino electron scattering. The intense electron anti-neutrino ux from the 2800MW power reactor at 18.6m amounts to 1.2 1013 � /cm2.s. The direction of neu-trinos to the detector is well de�ned.(� 5o) A neutrinocandidate is a contained single recoil electron in the for-ward solid angle. In data taking, we take dvantage ofthe properties of the light signals to apply hardware andonline cuts to reject undesired types of events: muon as-sociated events, "micro electric ashes", alpha particlesand Compton electrons. Electron tracks down to 300keV have been measured. The gross counting rate inthe TPC (70 s�1) is dominated by cosmic muons, Comp-ton electron events amounting to 0.14 s�1. After succes-sive cuts: muon rejection, �ducial volume cuts (events

5

touching the lateral walls or the anode plane), then anti-Compton rejection( 100 keV threshold), the counting rateis 780 day�1. The counting rates are shown in Table I.The recoil energy spectrum of contained electron eventsis shown in Figure 19.For neutrino electron scattering events induced by faraway reactor, a solid angle cut can further be applied.For a forward angle con�ned within (�/4), the residualcounting rate is 97 day�1 per 11kg of CF4 at an energythreshold of 300 keV or 8.8 (kg.MeV.day)�1.The recoil energy spectrum of contained electron eventsdoes not show any precise structure. Very likely, the re-maining background does not have a dominant originebut may have di�erent components: contaminants ofthe walls and the electrodes of the TPC or single elec-tron from the gas itself (cosmic activation or presence ofradioactive gas). This is under further investigations.

V. ANGULAR AND ENERGY RESOLUTION

V.1 Energy resolution

The energy resolution of a wire chamber in propor-tional regime usually depends on the primary ionizationstatistics and mainly on the single electron avalancheresponse spectrum, which dominates in the present caseof pure CF4 through a Polya parameter � = 0:21 alreadymeasured in ref [19].Moreover this gas reveals a strong attachment for valuesof the reduced �eld between 40 and 140 V.cm2. While noattachment occurs during the drift phase due to the lowvalue of drift �eld, nearly 98% of the drifted electronsare attached in the early stage of the avalanche, a�ectingthe resolution [20].For a primary deposition of 300 keV in the TPC only2% of the primary ionization electrons will survive andreach the multiplication phase of the avalanche, the samenumber as for a 6 keV deposition and no attachment; the uctuations should then be of the same order of magni-tude.However the energy of a track is not measured throughone single avalanche, but through successive ones (upto 100 independent avalanches for a 300 keV track, de-pending on track length, energy loss per unit length andangle in respect to the drift direction). The overall uc-tuations are then reduced when compared to a singlemeasurement. A complete calculation including the pre-vious parameters and simulated tracks with energy losstables predicts a resolution of � = 10:1% at 640 keV inagreement with measurements.Due to the di�erent contributions, the resolution is notexpected to follow a simple

pE law; a �t to the simula-

tions at di�erent energies rather shows an empirical E0:7

law, in better agreement with the low energy part of the

Compton spectra.

V.2 Angular resolution and incident energy

reconstruction

The angular resolution has been measured through � e� Compton scattering from various sources. Thedi�used is measured in the scintillator, the recoilingelectron track in the TPC. The time di�erence betweenthe and the drifted electron gives the absolute local-ization along the drift direction, allowing (together withthe XY track projection) a full 3D determination of thevertex of the interaction and then the exact incident ray direction.The electron scattering angle is obtained by �tting thedirection of the �rst centimeters of the track since theangular resolution is mainly limited by multiple scat-tering in the �rst few centimeters of sampled track. Inthe present case the electron track recognition, the ver-tex determination and the direction were obtained via afully automatic image processing which will be describedelsewhere.

Figure 20 shows the reconstructed 835 keV ray froma54Mn source measured at 3 bars with a recoiling electronenergy threshold of 300 keV. The � of the reconstructedenergy peak at 835 keV is 220 keV, although the shapeis far from a gaussian shape. The overall angular reso-lution averaged over the Compton recoil spectrum above300 keV threshold is �� = 39.7o� 2.2. To our knowledgethis is the �rst time that a � 1 MeV photopeak is re-constructed by measuring the Compton scattering in agaseous detector.This result is encouraging in view of using a gas TPC

to detect solar neutrinos. It demonstrates that the 7Besolar neutrino peak at 862 keV could be reconstructedeven at 3 bars by measuring the neutrino-electron scat-tering with a detector based on ths same principe as ours.By lowering the pressure to 1 bar, we shoud be able tomeasure the pp solar neutrino which has the end-pointat 420 keV.

VI. CONCLUSIONS

The MUNU detector is a 1 m3 TPC enclosed within a10 m3 light collection vessel. It has been built to study of�ee

� scattering at the Bugey reactor and to set a lowerexperimental limit on the neutrino magnetic moment. Itis particulary dedicated to detect low energy electronsand to reconstruct their energy and direction.One of the originalities of MUNU is the presence of

the external light collection vessel. It sees the light pro-duced by ionizing particles in its scintillator. This allows

6

to eliminate cosmics related events as well as Comptonevents occuring in the TPC. In addition the primary scin-tillation light produced by alphas is well resolved, whichallows to identify these particles. For all these events themeasurement of the primary light provides an absolutedetermination of the track position along the TPC axis.

The primary light from minimum ionizing particlesis too weak to be seen, but the light emitted at theanode during the ampli�cation process gives a strongsignal. It complements that from the anode signal, hav-ing even lower noise. It is very useful in characterisingthe topology of tracks, which will allow for a powerfulevent selection when analysing the data in terms of neu-trino scattering. Installed within low activity shieldings,the detector is now collecting neutrino induced electronevents at the Bugey reactor at a recoil electron thresholdof 300 keV.The result of the reconstruction of the incident energyobtained with the 835 keV 54Mn source is encouragingand it shows that the spectroscopy of low energy neutrinois attainable. The background has to be however furthersuppressed. The ultimate goal is to use an upgradedMUNU detector underground, where the backgroundconditions are optimal, and to look for solar neutrinos.

AknowledgementThis work has been supported by the Institut Nationalde Physique Nucl�eaire et de Physique des Particules(IN2P3/CNRS), INFN, le Fond National Suisse.We are grateful to the sta� of the Bugey nuclear powerplant (EDF) for their hospitality and help.The support of the technical sta� of the participatinglaboratories, in particular that of C. Barnoux, B. Guerre-Chaley, L. Eraud, M. Marton, R. Blanc, D. Schenker,

J.M. Vuilleumier, D. Filippi, D. Maniero, M. Negrello isgreatly aknowledgedG.Bagieu, R. Brissot, J.M. Laborie, G. Gervasio tookpart in the early phase of the experiment, we thank fortheir contributions.

[1] J. Baechler, Nucl. Inst. and Meth. A, Vol. 409(1-3)(1998)9-13

[2] V. Hlinka et al. Nucl. Inst. and Meth. A, Vol. 419(2-3)(1998)503

[3] D. A. Bryman et al. Phys. Rev. LETT. 55 (1985) 465[4] S. R. Elliott et al.J. Phys. G: Nucl. Part. Phys. 17 Sup-

plement (December 1991) S145-S153[5] J.C. Vuilleumier et al. Phys. Rev. D 48-3(1993)1009[6] K.N. Buckland et al. Phys. Rev. Let. 73-8 (1994)1067[7] U. Titt et al. Nucl. Inst. and Meth. A, 416-1 (1998)85[8] C.Amsler et al. Nucl. Inst. and Meth. A 396(1997)115[9] Oxisorb �lter by Messer Griesheim Gmbh,47805 Krefeld,

Germany[10] SAES Pure Gas, Inc. San Luis Obispo, California(USA)

Monotorr Puri�ers[11] J.M. Laborie, thesis Univ. de Grenoble 1998[12] C. Cerna, thesis Univ. de Grenoble 2000[13] H.A. Van Sprang et al., Chemical Physics 35 (1978)51[14] A. Pansky et al. Nucl. Inst. and Meth. A 354(1995)262[15] C. Broggini, Nucl. Inst. Meth. A 361(1995)543[16] L. G. Christophorou et al., J. Phys. Chem. Ref. Data, vol

25(5) (1996)1341-1375[17] J. Lamblin, thesis Univ. de Grenoble (in process)[18] Written by R. Veenhof at CERN[19] J.Va'vra et al NIM A324 (1993) 113[20] W.S.Anderson et al. NIM A323 (1992) 273

7

TABLE I. Counting rates in the MUNU detector

cuts ratesT > 150 keV 70/s

T >300 keV and no muon veto 0.35/swithin �ducial volume 0.15/sno anti-Compton veto 778/day

forward angle within �/4 97/day

FIG. 1. Isometric sketch view of the MUNU detector inside its shieldings.

8

FIG. 2. The electric �eld con�guration in the vicinity of the read-out planes. The horizontal axis corresponds to the z driftaxis, the vertical axis is the bisecting line between the x and y pick-up strips. The ticks on both axis are every 2mm. Theequipotentials are shown with the corresponding voltages. The grid is at -2000 V, the anode wires at 3540 V, the potentialwires at 390 V.

FIG. 3. A cosmic muon producing two high energy delta electrons. The one near the center with an energy of 1 MeV iscontained, and the increased charge density at its end is clearly visible.

9

keV

Hz

low threshold

high threshold

10-1

1

10

10 2

0 500 1000 1500 2000 2500 3000 3500 4000

FIG. 4. Total absorption spectra obtained with di�erent radioactive sources measured in the external detector, and comparedto simulations (solid lines).

keV

137Cs = 662 keV

54Mn = 835 keV 22Na = 2332 keV

103

FIG. 5. Energy spectra collected in the 10 m3 external detector at two hardward energy thresholds.

10

FIG

.6.

Fraction

ofCom

pton

electroneven

tsnot

detected

bytheextern

aldetector

asafunction

oftherecoil

electronenergy.

Data

measu

redwith

a54M

nsou

rce(E

=

835keV

)are

show

nwith

errorbars

andare

compared

tosim

ulation

sat

two

anti-C

ompton

energy

thresh

olds:

100keV

(dash

edcross)

and125

keV(dotted

cross).

Anode charge signals (arbitrary units)

Phototubes light signals (arbitrary units)15000

20000

25000

30000

35000

40000

45000

50000

55000

60000

600800

10001200

14001600

18002000

22002400

FIG

.7.

Eventper

eventcom

parison

oftheam

plitu

des

ofthecurren

tandligh

tsign

als.

11

MCMCMCMCMC

MCMCData

keV

arbi

trar

y un

its

450 500 550 600 650 700 750

FIG. 8. A Compton spectrum of 54Mn in the TPC, compared to Monte Carlo simulation.

Gain Monitoring

FIG. 9. Gain monitoring with radioactive sources(diamonds) and muons( squares).

12

50

100

150

200

250

50 100 150 200 250

0.93

0.94

0.95

0.96

0.97

0.98

0.99

1

Y strip

X s

trip

FIG. 10. Gain varaiation on the anode plane. The vertical scale on the right indicates percentage of gain variation.

FIG. 11. A muon event: the xz, yz and xy projections are shown.

13

µs

keV

200

400

600

800

1000

1200

0 10 20 30 40 50 60 70

FIG. 12. Electron equivalent energy .vs. drift time (for � particles of 5.5 and 6 MeV).

FIG. 13. A delayed �-� coincidence event

14

days

Hz

of α α

βα

10-4

10-3

10-2

10-1

Hz of βα

Removal of,oxysorb, filter

τ = 3.2+/-0.5 days

10-3

10-2

10-1

1

10

-30 -20 -10 0 10 20 30

FIG. 14. Radon decay rates after removal of Oxysorb �lter(see text).

15

Light Signalµs

arbi

trar

y un

its avalanche gas scincillation

liquid scincillation light

Current Signalµs

arbi

trar

y un

its

anode charge

10 3

0 8 16 24 32 40 48

0

200

400

600

800

1000

1200

1400

0 8 16 24 32 40 48

FIG. 15. Waveforms of a Compton electron event

16

ANODE CURRENT SIGNAL

time in channels (1 channel = 80 ns)

ampl

itude

PHOTOTUBES LIGHT SIGNAL

time in channels (1 channel = 80 ns)

ampl

itude

0200400600800

1000120014001600

320 340 360 380 400 420 440 460

0

1000

2000

3000

4000

5000

6000

320 340 360 380 400 420 440 460

FIG. 16. A double blob event. The current and light signals are shown together with the xz and yz projections.

17

number of events .vs. track end pulse height

All electrons (contained or not)

number of events .vs. track end pulse height

Electrons from Compton scattering (totaly contained)

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

0 500 1000 1500 2000 2500 3000 3500 4000 4500 5000

FIG. 17. Track end pulse height distributions (see text).

Z

Y

Θprojected

from the

cathode side

from the

anode side

FIG. 18. Direction distribution of double blob events.

18

keV

coun

t/day

0

20

40

60

80

100

120

140

160

400 600 800 1000 1200 1400 1600 1800

FIG. 19. Energy spectrum of contained electron events.

54Mn source Eγ=835 keV

0

20

40

60

80

100

120

140

400 600 800Electron recoil energy (keV)

Cou

nts

05

101520253035404550

0.5 0.6 0.7 0.8 0.9 1Scattering angle (cosΘ)

Cou

nts

0

5

10

15

20

25

30

35

0 1000 2000 3000Reconstructed γ energy (keV)

Cou

nts

FIG. 20. Initial gamma energy reconstruction(see text).

19