Liquid noble gas and warm liquid detectors

Liquid noble gas and warm liquid detectors

Nuclear Instruments and Methods in Physics Research A283 (1989) 375-386 North-Holland, Amsterdam 375 LIQUID NOBLE GAS AND WARM LIQUID DETECTORS J. F...

914KB Sizes 11 Downloads 391 Views

Nuclear Instruments and Methods in Physics Research A283 (1989) 375-386 North-Holland, Amsterdam



The fundamental properties of liquid noble gases and room temperature hydrocarbons of interest for calorimeters are summarized . The application of liquid dielectrics to existing detectors is reviewed . Emphasis is put on the recently growing family of hadron calorimeters . Some future plans for pure liquid electromagnetic calorimeters as well as liquid imaging detectors are considered .

1. Introduction The high mobility of electrons in liquefied noble gases was discovered in 1948 [1]. A few years later J.H . Marshall constructed the first liquid argon chamber to detect ß-rays [2]. It is, however, after suggestions by T. Doke in Japan [3] and L.W . Alvarez at LBL [4] that the use of liquefied Ar and Xe as particle detector media did really get impetus, leading to high spatial resolution LXe chambers [5] and to the first use in an experiment of a LAr calorimeter by Willis and Radeka [6]. Charge amplification has turned out to be very hard to control m cryogenic liquids while calorimeters based on simple collection of deposited charge in the liquid have superseded other types of calorimeters in many applications . The basic advantages of sampling LAr calorimeters are well established : - reproducibility of charge collection, and thus possibility of an accurate calibration, - uniformity of response, - small granularity is easy to obtain, - operation in a magnetic field, - radiation hardness, - easy handling . In some applications, several drawbacks have shown up recently : - a cryostat with a small fraction of dead material is a technical challenge in a 41T detector, - the slow charge collection time (200 ns/mm) makes the calorimeter difficult to use at high rates, - the absence of free protons prevents equalizing directly the electromagnetic and hadronic responses in a hadronic shower . The discovery in 1968 of the high mobility of excess electrons in liquid hydrocarbons at room temperature ("warm liquids") has opened a new field of research for detectors where some (or all) of the above drawbacks can be overcome [7].' 0168-9002/89/$03 .50 © Elsevier Science Publishers B.V . (North-Holland Physics Publishing Division)

In this review, the fundamental properties of dielectric liquids are summarized, with emphasis on the not so well known mechanisms of detection of heavily ionizing particles. Then various applications are considered and especially the growing field of sampling hadron calorimeters and pure liquid detectors. 2. Physical properties The physical properties of liquid noble gases have been presented in detail in excellent reviews about 10 years ago [8,3], and more recently for liquid hydrocarbons [9]. We consider below the most important parameters for detector operation : electron drift velocity, yield of free electrons and saturation characteristics, sensitivity to electron scavengers, scintillation and recombination, and radiation hardness . 2.1 . Drift oelocities

The mobility of excess electrons in an insulating liquid has been the subject of intensive studies since the beginning of this century. The velocity depends strongly on the applied electric field. In liquid noble gases the behavior is always the same. The velocity rises with the field (v = p,E) and then saturates at a field of about 10 kV/cm (fig . 1). In liquid hydrocarbons, in contrast, the velocity remains proportional or nearly proportional to the electric field up to the highest fields reached so far (about 100 kV/cm) . A detailed theory of drift velocity has been developed for monoatomic liquids [13] but is still controversial [14] . Qualitatively, this behavior in an electric field can be summarized as follows [15] . If the mean free path of electrons of agitation velocity v a is X, the average time between collisions is Vv a and the acceleration in the electric field direction is eE/m . Hence the drift velocity is (eE/m) (IN/Ua) = yE and the mobility p is inversely proportional to vd . At low field, I. PLENARY TALKS


J Fellesse / Liquid noble gas and warm liquid detectors 70 60 50

G Ef1

40 30

20 10 0














Fig 3. Saturation characteristics for liquid argon and xenon [3] Solid lines are fits of the experimental data with the model of Onsager [25] Fig. 1. Dependence of electron drift velocity m liquid noble gases and liquid hydrocarbons on the electric field [10-12]. when the energy gain is small, there is thermalization : the average energy is unchanged, the mobility remains constant and the drift velocity rises with the field. In

10 9



monoatomic liquids at sufficiently high field, the electrons gain more than kT from the electric field between collisions, va increases, the mobility decreases and the drift velocity saturates. In contrast, in liquid hydrocarbons or m liquid noble gases doped with molecular solutes there are enough scattering centers to keep the electrons cool and the agitation velocity does not increase . For example, it is possible to triple the drift velocity in liquid argon by adding 3% of methane (fig . 2) . Another way to increase the drift velocity is to increase the liquid temperature by operating at high pressure [19,20]. This does not seem very practical for cryogenic and warm liquids. 2.2. Ionization yield





3 2




(k V/cm )

Fig. 2. Dependence of electron drift velocity m argon-methane and argon-N2 mixtures x o v ref. [16], 0 ref. [17], O ref. [18], 0 ref. [111 .

The operation of ionization detectors depends on the mean number of ion pairs produced by the ionizing particle and on the fraction which can be collected by the field. The yield of free electrons is usually expressed as Gf  the number of free ions for 100 eV deposited m the liquid . Gt, depends on both the electric field and the ionization power of the particle entering the liquid . As for the drift velocity, the field dependence is different for liquid noble gases and liquid hydrocarbons . In a liquid noble gas, the number of free electrons produced by a minimum ionizing particle (mip) is already sizeable at zero field, then rises linearly with the field and finally saturates for fields greater than 10 kV/cm (fig. 3) . G frtma', the maximum value of charges, is related to the W value, the average energy to produce an ion pair : Gfmax = 1001W (table 1) . For heavily ionizing particles, the free electron yield is an order of magnitude smaller but keeps rising up to the highest field which can be applied. In liquid hydrocarbons, m contrast, no saturation has been observed so far, for mips or heavily ionizing particles (fig . 4 and table 1) . In the absence of


J. Feltesse / Liquid noble gas and warm liquid detectors

Table 1 Electron yields


G0 a)


4 .7 `) Xe Kr 2 .3 `~ Ar 0 .75 d) CH 4 0 .75 e~ TMP TMS -0 .6 a> Extrapolated using

G f, 15 kV/cm

W [eV]

Yield/mm 15 kV/cm

6 .4 4.9 4.2 3 .7 1 .37 1 .42

15 .6 9) 20.5 `) 23 .6 h) 27 d)

2.4 X 104 b) 1 .6 X 10 4 8.9 X103 3.8X10 3 2.1 X 103,) 1.9X10 3 1 t

Onsager's law . At saturation, before amplification . `) Ref. [3] . d) Ref. [11] . `) Ref . [10] . r) Ref . [21] . 9) Ref . [22] . h) Ref. [23] . ') Ref . [24] . b)





~f }


any impurities, the charge collection is affected by two competing processes : the gemellar recombination and the columnar recombination . In liquids, the electron produced by ionization might not escape the parent ion but be driven back to recombine with it . This is the initial or germinate recombination . By considering this sole process, Onsager has calculated the fraction of electrons which escapes at low field [25] : 1 exp( Gf,(E)= W

kT ro




where rkT =e z/(ekT) is the Onsager length, E is the dielectric constant in the liquid, Eo = 2k 2 T Z /e 3 and ro is the thermalization length (28 rim in LAr) [28] . The linear rise of the electron yield at low field appeared for many years as a reasonable description of the data in liquid noble gases and liquid hydrocarbons for both mip and heavily ionizing particles (figs . 3 and 4 and ref. [26]) . However, more recent data in extremely pure LAr [28,27], where field values as low as a few V/cm were reached, have shown that the yield at low field and specifically the slope-to-intercept ratio were below the




w cD









20 30 40 E (kV/cm ) Fig . 4 . Relative charge carrier yield in liquid TMP exposed to a 60 CO -y-source [21] .








Fig . 5 . Q O (E) deposited in liquid argon by the passage of a cosmic ray as a function of the electric field [28] . The dashed line is calculated from the data of ref. [27] .

Onsager prediction (fig . 5) . This is not surprising because in a liquid the ion density is so large along the track that electrons which drift toward the cathode can meet positive ions and recombine . This is the columnar recombination, first proposed by Jaffe [29] to describe ion tracks . The higher the field, the smaller the recombination. The higher the ion density (d'//x) along the track, the higher the recombination . These two dependences are expressed in Jaffe's relation :

Gf' _ -

1 1 + (k/E) do/dx

Gmax f'

where k is a parameter given by the model [29] . This relation is often used at a constant electric field in an expression first introduced to describe light quenching in scintillators : Gf' -_




1 G max 1 + k, dg'/dx f'

where k B is called the Birks constant [30] . For mips in LAr, the saturation curves at high field are well described by Jaffe's relation [16,28] . The attenuation of the collected charge in TMP for heavily ionizing particles as X-rays is well reproduced by Jaffe's or Birks' relations [9,31] (fig . 6) . No systematic studies of the dependence of the charge collection on dd'/dx in liquid noble gases exist so far . However, by comparing collected charges for a- and ß-particles in liquid argon at a given field, it has been possible to extract k B and then to check that k B is inversely proportional to the field as predicted by the model of Jaffe [321 . In fig . 7 the variation of k B with the electric field is shown for pure LAr and LAr mixed with 10% of CH, It is worth noticing that the Birks constant in LAr is increased by a factor 5 when adding 10% of CH, We shall come back 1 . PLENARY TALKS

37 8

J. Feltesse / Liquid noble gas and warm liquid detectors

Table 2 Attenuation length at 10 kV/cm for 1 ppm of impurity




Ar [331 Xe [361 CH 4 1111


Ar [36] Ar [331

N2 0 N2

0.015 3000

TMS [24] TMS [24]


0.5 0.125

TMS [241

0.01 1 .01

i 0.1

dE/dx (eV/Á )

i 1 .0



Fig. 6 Dependence of the electron yields at 10 kV/cm on the rate of energy loss of the ionizing particle (do/dx). Solid points from ref. [12], open points from ref. [31] . Solid line : k B = 0.036 g/MeV cm2. to this point in the third section. Another track model has been proposed for LAr and LXe [33] . The predicted field dependence is : Gf,=


f' kln l+ ( .This where k is a parameter which depends on d'//x relation is also a good description of the field dependence of the data of refs . [26, 28]. The variation of k with d&/dx has not yet been studied.




Attenuation length [mm] 15 20



2.3. Attachment to impurities

The problem of liquid purification is one of the major technical difficulties in connection with the operation of liquid dielectric detectors. The effect of electronegative impurities is to trap electrons to form negative ions which will then drift at a velocity about 10 5 lower than free electrons and hence will not contribute to the collected charge . For oxygen in liquid noble gases, an empirical expression of the attenuation length a gives a fair description of the data for concentrations at the ppm level [34] : X = aE/p, where E is the electric field, p is the concentration of impurities and a the trapping constant which depends on the liquid and on the chemical nature of the impurity . At the ppb level, some deviations of this expression have been observed [35] . As a method of comparison the attenuation lengths per ppm of impurity are given in table 2 for a field of 1 kV/mm. The enhanced sensitivity of liquid hydrocarbons to attachment of 02, together with the difficulty of purifying a liquid at room temperature shows how complicated it is to handle the warm liquids used so far. However, in the last years purification techniques based on combinations of oxygen absorber and molecular sieves have made significant progress, purities of a few ppb in liquid TMP [37] and of about 1/10 ppb in LAr [38,39,27] have been obtained for tens of liters of liquid . 2.4. Hardness to radiation



Fig. 7. k B as a function of electric field for pure LAr and LAr doped with 10% methane and 15 ppm allene [32] .

In contrast to liquid noble gases, liquid hydrocarbons are decomposed by radiation. In TMP or TMS, per 100 eV of absorbed energy several molecules of mainly hydrogen, methane, neopentane and isobutilene are formed [401 . As far as these products stay in the liquid, no real deterioration of the performances is expected up to exposures of 10 7 rad. The main problem, however, is that the pressure built up in the gaseous phase will reach 5 to 10 atm [411 . This poses a serious safety problem.

J. Feltesse / Liquid noble gas and warm hquid detectors Table 3 Scintillation m liquids [43] Liquid Ar Kr Xe



t0` A l /A z






1300 1500 1750

6 .5 2 3

1100 85 22

14.6 0.49 25




2.5. Scintillation in liquid noble gases Ultraviolet luminescence in dielectric liquids irradiated by ionizing particles has been studied for more than 30 years [42] . A narrow band of intrinsic UV light is observed in LAr, LXe and LKr . It is explained as the radiative decay of an excited molecule which emits ultraviolet photons in transitions from the lowest excited molecular state to the dissociative ground state : Ar * -> h P + Ar . The decay has two components with different lifetimes and amplitudes but the same wavelengths [43] . Their characteristics are given in table 3 . Clearly, in LAr and LXe, the fast component of a few ns is the dominant one and could be used for trigger purposes . Another important feature is the variation of the emitted light with the electric field . Kubota et al . [44] have demonstrated that the luminescence intensity decreases when the field increases and also when the recombination decreases (fig. 8) . This is called recombination luminescence and its physical mechanism is well understood [3] . When an ionizing particle enters a liquid noble gas, the atoms are excited or ionized . Excited atoms collide with other atoms and produce excited molecules which decay radiatively. A fraction of the




E (KV/cm) Fig . 8 . Vanations of relative luminescence intensity L and collected charge Q in liquid argon and in liquid xenon against applied electric field strength for 0.976 and 1 .05 MeV electrons [44]-


Fig . 9. Relation between the energy (40)expt obtained from the linear combination of the ionization I and the scintillation S, I + a S, and the deposited energy '0 in liquid argon. The incident particles are 613 MeV/n Ne ions, 705 MeV/n Fe ions and 0.976 MeV electrons [45] .

positive ions recombine with escaped electrons . The recombined atoms also produce, after collisions, excited molecules which then decay radiatively . The scintillation light is not quenched in heavily dense tracks . The loss of free electrons due to recombinations or excitations can be recovered by detecting the scintillation light . The application of this property is that a linear combination of the signals of scintillation light and ionization charges has been measured [45] to be remarkably proportional to the absorbed energy up to heavy ions of Fe which deposit 10 4 more energy than a minimum ionizing particle (fig . 9) . In practical applications, to detect simultaneously the light and the charge is a serious complication . The UV light is poorly reflected on any wall and the attenuation length is a few mm [46] . A more promising approach has been suggested by Policarpo [47] . The UV photons could be converted to electrons by adding photosensitive dopants to the liquid . D.E. Anderson made systematic studies of many dopants in LAr and LXe [48] . The goal was to find a molecule with a photoionizing potential below the emitted UV light and with a high solubility. The best results so far have been obtained with TEA, TMA and allene . For each dopant, there exists an optimal concentration which gives the largest charge. A spectacular application of these dopants has been the improvement of the energy measurement of a-particles absorbed in the liquids . By adding TEA in LXe, the energy resolution (FWHM) has decreased from 15% to 4% [49] . By adding allene to LAr, the resolution improves from 20% to 2 .9% [50] . 1 . PLENARY TALKS

38 0

J. Feltesse / Liquid noble gas and warm liquid detectors

3. Application to detectors

the absorber plate [54], and (cos 0 ) is the average angle of electrons in the electromagnetic shower with respect

3.1 . Sampling electromagnetic detectors

to the average direction of the shower .

Until very recently the only detectors which used a dielectric liquid as readout medium in real experiments

were LAr sampling electromagnetic calorimeters . The qualities of these detectors have been reviewed several times [51] . Let us just mention two important performances : the linearity and the resolution of the energy

response . In the energy range 2 [52] to 200 GeV [53] the collected charge has been checked to be proportional to

the total deposited energy at the 1 % level. The energy

resolution is dominated by sampling fluctuations and is well approximated by the following formula: a - ~

3.2 . Sampling hadronic calorimeters The first devices to measure hadronic energies using dielectric liquid as a readout medium were iron liquid argon calorimeters [55,56] with large sampling ratios (1 .5 mm Fe). The energy resolution was parametrized by a large constant term (8 .5%) plus a modest term in 1/r (table 4) . The large constant term is also present

in a calorimeter with thick iron plates (19 mm) where the 1/r term is much larger [61] . These large constant terms should dominate the energy resolution already in the 100 GeV region . What is even worse, the energy resolution has large tails at large energy and deviations

F(cos 0)9

where 0S is the energy absorbed in a cell, F is a reduction factor which depends on the ratio of the energy threshold m the readout to the critical energy in

of 7% from linearity in energy response are observed between 10 and 170 GeV. A new class of detectors, using


as absorber material, has been built in the

last years. The measured performances are still pre-

Table 4 Energy resolution m hadronnc calorimeters Ref.


Active layer

Energy range





[GeV] 1 .5 mm Fe

Fablan et al . [55]

Sessoms et al [56]

3 .0 mm Fe

2 mm LAi

4 mm LAi





1- 38 61

Hl [53]


mm Fe

2.4 mm Pb

5 mm LAi

10- 80

mm Fe



45 ® 0.8


57 -®63 15-170


1 .34-1.21

2.8 mm LAi

Hl [61]

+8 .5


5 mm LAi

44 ® 0.9


DO [57]

4-8 mm U

3.2 mm LAi



48 ® 5.0


DO [58]


4.4 mm LAi



48 ® 5.1


Helios [59]

1 .7-3 .6 mm U

5 mm LAi


1 .12-1 .07

45 ®1 .5

UAl [60]

2-5 mm U

3.3 mm TMP

1 .03

54 ®8.0

" After weighting.


mm U

After noise subtraction. `) ® = quadratic sum.


38 1

J. Feltesse / Liquid noble gas and warm liquid detectors liminary but demonstrate the possibility of decreasing the constant term in the case of liquid argon (table 4) . An alternative approach keeping standard absorber materials such as lead or iron but using local energy deposition in a fine grain LAr calorimeter also makes it possible to reduce the constant term and to decrease the energy tails . The understanding of the mechanisms which govern the fluctuations of the visible energy, specially its dependence on both the absorber and readout media, has made spectacular progress after extended Monte Carlo simulations [62-64] . A detailed review of these mechanisms is beyond the scope of this paper . Let us only discuss the most relevant components of the energy resolution . The experimental energy resolution is frequently parametrized as :

It is, however, possible to break up the resolution into more terms [63] :


z- (


Al I z+ (



+ A2( h



Bz + 6,2

The B terms represent the detector imperfections . B t holds for the wrong intercalibration between electronic channels and for dispersions in mechanical dimensions . It is usually close to 2% and is hard to improve in large detectors. Bz is due to the electronic noise and corresponds typically to only a fraction of the signal of a single minimum ionizing particle [65] . The A o term represents the sampling fluctuations and can be parametrized as 0 .09 A&(MeV) [66], where 0& is the energy absorbed in one sampling layer. The A 1 and A Z terms are related to the physics process in the shower development . The A 1 term can be interpreted as the contribution to the energy resolution of a hadronic cascade where no m ° 's are produced ; it depends mainly on the nuclear properties of the absorber medium . The A z term holds only for noncompensating calorimeters where the energy responses of hadrons and electrons are not equal . This latter feature has been identified as the principal limitation to the energy resolution of hadronic calorimeters . A considerable part of the secondaries in the hadron cascade are m ° 's . The size of this Tr o component can differ from event to event, depending on the nature of the first interaction . These fluctuations are not Gaussian and generate tails in the energy resolution . The fraction of energy spent on m ° 's varies like the total absorbed energy and produces deviations from linearity in noncompensating calorimeters . The A 2 ((e/h) - 1)2 contribution to the measured energy resolution is usually included in the constant term. For example, a value of 1 .2 for e/h, the ratio of energy response for electrons and hadrons, corresponds to a 4% constant term [63] .

The e/h ratio depends on many parameters : - Z of the absorber material : a large fraction of the absorbed hadronic energy is spent in binding energy of the nucleus and is therefore invisible. The electron signal has therefore to be attenuated in high-Z absorber plates to compensate the invisible energy . - Nuclear properties of the absorber material : the amount of slow neutrons generated mainly by spallation depends on the nuclei of the absorber . - Detection of slow neutrons in the readout medium : about 30% of the absorbed energy is spent on production of slow neutrons which are mainly detected by the ionization electrons of recoil protons . It is therefore better for the readout medium to contain free protons . The recoil protons have an energy of about 4 MeV and hence are heavily ionizing . We have seen in the first section that a large fraction of the electrons recombine with the ions and that this recombination property is expressed by the Birks constant kB. With a k B value close to the one to the scintillator the amount of free electrons is already attenuated by a factor 5 and the attenuation varies approximately linearly with kB [64] . - Integration time of the' electric signal : a large fraction of the neutrons are captured in the absorber

U calorimeters

Fig. 10 . The signal ratios e/h, at about 10 GeV, for uranium calorimeters as a function of R d , the ratio of the thicknesses of absorber and readout layers . v : LAr [67[, o : LAr [57], A : LAr [58], o : LAr, 135 ns integration time [59], ": LAr, 500 ns integration time [59], O : TMP [37] . Solid lines are from ref. [63] . Lines 1 and 2 : LAr readout with 0 .1 ixs and 1 ~is gate respectively . Lines 3 and 4: TMP readout for k B = 0.0098 g/MeV cmz and k B = 0.0045 g/MeV cmz respectively . 1 . PLENARY TALKS

38 2

J Feltesse / Liquid noble gas and warm liquid detectors

material and release y-rays . In


it takes about 1

Ps before all neutrons are captured [64] .

- Ratio of thicknesses of the absorber and of the active layers .

The relative importance of these various parameters

is illustrated m fig. 10 where Monte Carlo estimations and data on e/h are given for 238U absorber plates . We

see that both data and predictions point to values of 1.05-1 .1 for similar thicknesses of uranium and LAr and explain why it has been possible in at least one

experiment to decrease the constant term to 1.5% [59] . The still large constant term in the DO experiment is attributed to instrumental effects not yet corrected [58] .

The preliminar result on U-TMP indicates a large value


for the Birks constant in TMP, several times higher than to the scintillator . This is in contradiction with the previously quoted value by the UAI collaboration (k B


= 0.014 cm/MeV) [24] . This old value was, however,

not a measurement of kB but rather a parameter to

tune a global fit of the energy profile. The effect of the integration time has been checked by the Helios collaboration [59] . It has the predicted behaviour but is

hardly visible within the errors (fig . 10). Attempts to further improve the e/h ratio by adding CH 4 in liquid argon have failed [58] . This is now well understood : if,

on the one hand, the slow neutrons can scatter on the free protons of CH 4, on the other hand the recombina-


Fig. 12 . The signal ratio e/h, at about 10 GeV, as a function of R d the ratio of thicknesses of absorber and active layers . x LAr [55], 0 : LAr [61] Solid lines 7 and 2: TMP readout for k B = 0 .0098 g/MeV cm 2 and k B = 0.0045 g/MeV cm z respectively [63] .

tion effects are so much enhanced by the addition of

methane that the result leads even to a deterioration of the e/h ratio. We should finally note that the e/h ratio is predicted to decrease at high energies . This is well verified by the first results of the DO collaboration [58] (fig. 11).

Concerning inexpensive absorber material such as

iron or lead, there is also a fair agreement on the e/h

An alternative to uranium for compensating calorim-

eters has been proposed at FNAL [68], used by the CDHS collaboration in an iron- scintillator calorimeter [69] and developed by the HI collaboration m lead-

iron-liquid argon calorimeters [53,61]. By considering the difference in the spatial extension of the energy

ratio between expectations and data (fig . 12). Monte Carlo calculations and data show that there is no way to

get direct compensation in iron-liquid argon calorimeters . This is also true in iron-TMP calorimeters if the

11L,,~l~i 1

d ,~~



large value of k B in TMP is confirmed.



1 .05r1 001_



100 Energy




Fig. l l The e/m signal ratio in the DO 238 Ur-LAr calorimeter [58] .

U', (PC) Fig. 13 . Average fraction of deposited charge on an individual pad, resulting from mo interactions, relative to the total de posited charge QTOF, m an individual Q bin for MC pions of 170 GeV in the iron -LAr test device in ref. [61] .

J. Feltesse / Liquid noble gas and warm liquid detectors



w w

0 (PC I Fig. 14 . Distribution of o (Q,)/Q, for data of peons at 170 GeV . The solid line represents a fit using an exponential parametrization The dashed line represents a fit using a polynominal parametrization [61] . deposited by a m ° and a charged pion, one should be able to assess the relative contribution of electrons and charged hadrons. In reality these components overlap to a large extent and can be separated only statistically (fig . 13). In a calorimeter with e/h > 1, the method consists in weighting negatively those parts of the cascades with a high local signal . Several algorithms have been tried and are still being developed [53,61]. The goal is to find weighting functions with smooth energy dependence which optimize the energy resolution and correct the deviations from linearity. Examples of weighting functions of exponential or polynomial types are given in fig. 14 . Applying these parametrizations gives the following results : the constant term in the measured energy resolution in a test beam becomes compatible with the beam resolution, the contribution of the 1/r term has decreased from 65% to 45% and the deviation of the mean reconstructed energy from the real absorbed energy is below 1% . It is remarkable that without constraining it in the optimization procedure the effective e/h ratio obtained is very close to unity after weighting [53,61].

38 3

factor in liquid noble gases [3]. However, the measured energy resolution has turned out to be an order of magnitude larger . This is nevertheless a good performance (1 .5% at 1 MeV [70]). There are some hints that this is due to heavy recombinations in the 8-ray tracks [71,72]. At high energy there exists only one measurement at 1 GeV to LAr which gives an energy resolution of 2 .5% [73] and is expected to scale as 1/V? at higher energies . This energy resolution is comparable to what can be obtained with Nal crystals but the fine granularity which is easily obtained in liquids can provide excellent additional spatial resolution . Applying these properties, several detectors are being built or planned at Novosibirsk : - KEDR . A barrel shower detector of pure liquid krypton is under construction for an experiment at VEPP4 [74] . It has an inner radius of 1 .46 m, a depth of 70 cm and is 3 m long . The expected energy resolution is 5% at 100 MeV and 1 .5% at 1 GeV. A full-size prototype of 0.4 t has been successfully tested . A spatial resolution of 0.4 mm has been achieved . - CMD-2. It is planned to replace the actual CsI calorimeter by liquid Xe to the CMD-2 detector at VEPP-3 [75] 3 .4 . Liquid imaging chamber

The idea to use liquefied noble gases in a time projection chamber exists since the first application of these liquids to calorimeters [76] . The main obstacle was to get argon pure enough to let electrons drift over large distances. We have seen in the first section that after . .SIYI'~~.11T -.S1~T'.J1YI'_STYS~.S'IJ~


3 .3 . Pure liquid electromagnetic calorimeters

The small radiation lengths of liquid Xe and Kr (2 .8 cm and 4.6 cm respectively) combined with their properties of stability and radiation resistance make them attractive media for liquid calorimeters once the difficulties of procurement are solved . The potential of pure liquid detectors to be used as calorimeters with high energy resolution has been mostly investigated so far in the MeV region where resolutions of a few keV were anticipated after calculations of a very small Fano










Fig. 15 . Front side of PC board used as readout anode in the liquid argon time projection chamber of ref. [77] . I . PLENARY TALKS

38 4

J Feltesse / Liquid noble gas and warm liquid detectors

electron trajectories -1 0


M equipotenhal curve


c 0




v c 0

itnunui Ituuw uulwwuuuau Illlllllnllllllllllll illllllllllllltlllllll !111111111111111111111


induction wire plane

Nilllllmnltllllll ~111~;t ~ii~~ tll

aU C .10


focusing/screening grid

~Imnnmlttlltlm uttuuuuuuuun ilwuuuuntnnu uuunnuuuuuu illnnnunatunn ituunuuuuuuu





screening wire




collection wire plane

sense wire


0 .4



Distance across the wire plane (cm)

Fig . 16 . Equipotential contours and electron trajectories for the collection-mode plane of the Icarus 1 detector [80].

many years of development several groups have achieved purities of the ppb level in large quantities . In liquid TPCs operating in the ionization mode, two principal ways have been tried to achieve 3-D tracking : (1) Collection of the electrons on one plane containing two perpendicular interwoven sets of coordinate strips (fig . 15) . The method has been successfully tested in 1983 with the aim to build a large LAr detector for neutrino physics [77] . This technique will be used in a LXe detector for -y-ray astronomy [78] . The goal is the construction of a liquid xenon imaging chamber of 1000 cmz area, triggered by the primary ionization light, with a few mm spatial resolution and capable of detecting -y-rays in the energy range of 0 .1 to 10 MeV with a 2% energy resolution at 1 MeV . (2) Use of induced signals by a nondestructive locali-

zation . The electrons are drifted through a plane of wires and are collected on a second plane (fig. 16) . The collected charge is used for calorimetric measurement . The validity of the technique for image reconstruction has been tested in a two-dimensional LAr chamber with a 24 cm drift gap [79] (fig . 16) . The Icarus detector, based on these principles, is designed primarily to detect solar neutrino interactions in the few MeV region m the Gran Sasso tunnel . In a first step, it consists of a 300 t LAr image chamber [80] . The trigger is provided by signals induced in thick wires located every 20 cm along the drift path in the liquid argon . After a maximum drift length of 2 .3 m, the width of the electron cloud is 1 .8 mm in a field of I kV/cm . The expected energy resolution is a/6= 2 .6%/ o'(MeV) + 0 .6% . A 2 m3 prototype is under construction.

J. Feltesse / Liquid noble gas and warm liquid detectors 4. Conclusion After a long period, during which the only applica-

tions of dielectric liquids were in LAr sampling electromagnetic calorimeters, new classes of detectors are coming to maturity :

- Hadronic sampling calorimeters where the initial poor

performances in iron-LAr detectors are improved by

either replacing the iron by uranium or by using the local energy deposition in a fine grain readout.

- Sampling calorimeters using liquid hydrocarbons as

readout media where the difficulties of handling are

being overcome. - Pure liquid krypton or xenon electromagnetic calorimeters with excellent energy and spatial resolution . - Liquid argon image detectors. After drastic improve-

ments on purification methods, liquid imaging detec-

tors with long drift paths are becoming reality with the Icarus project and detectors for y-ray astronomy.

The possibility of measuring with an unrivalled pre-

cision the energy deposited by heavily ionizing particles through the detection of both charge and scintillation has not yet been used in a real experiment. Note finally that the ideal liquid needed in high energy physics, which has the following properties :

- easy handling for purity control and safety hazards,

- operation at room temperature, - high drift velocity : less than 50 ns/mm, - high yield of escaping electrons for heavily ionizing

particles (small kB), - radiation hardness : no damage in an exposure of 10 Mrad, has yet to be discovered .

Acknowledgements It is a pleasure to acknowledge fruitful discussions with D. Anderson, A. Givernaud and B. Mansoulie. I am indebted to G. Cozzika and J. Ernwein for critical reading of the manuscript .

References [1] N. Davidson and A.E. Larsh, Phys . Rev. 74 (1948) 220; G.W . Hutchinson, Nature 162 (1948) 610. [2] J.H . Marshall, Rev . Sci. Instr. 25 (1954) 232. [31 T. Doke, Portugal Phys . 12 (1981) 9. [4] L.W . Alvarez, Lawrence Radiation Laboratory Physics Note no . 672 (1968) . [5] S.E. Derenzo et al ., Nucl . Instr. and Meth. 122 (1974) 319. [61 W.J . Willis and V. Radeka, Nucl . Instr. and Meth. 120 (1974) 221 . [71 P.H . Tewari and G.R. Freeman, J. Chem . Phys . 49 (1968) 4394 ; W.F. Schmidt and A.O . Allen, J. Chem . Phys . 50 (1969) 5037 .


[8] C. Brassard, Nucl . Instr. and Meth. 162 (1979) 29 . [9] R.A . Holroyd and D.F. Anderson, Nucl . Instr. and Meth . A236 (1985) 294. [10] K. Yoshino et al ., Phys . Rev. A14 (1976) 438. [111 A.S . Barabash et al ., Nucl . Instr. and Meth . 186 (1981) 525. [12] R. Muhoz et al ., J. Phys . Chem . 89 (1985) 2969 . [13] J. Lekner, Phys . Rev. 158 (1967) 130. [14] R.A . Holroyd and W.F . Schmidt, submitted to Ann. Rev. Phys . Chem . [15] W.F . Schmidt and A.O . Allen, J. Chem. Phys . 52 (1970)

4788 . [161 E. Shimabura et al ., Nucl . Instr. and Meth. 131 (1975) 249. [171 S. Nakamura et al ., JIEE Japan A107 (1987) 543. [18] P. Muhlemann and R. Tavano, Nucl . Instr. and Meth . 166 (1979) 583. [19] T.G. Ryan and G.R . Freeman, J. Chem . Phys . 68 (1978) 5144 . [20] K. Slnnsaka and Y. Hatano, Proc. 3rd Workshop on Radiation Detectors and Their Uses, KEK report 88 .5 (1988) .

[21] H. Jungblut and W.F. Schmidt, Nucl. Instr. and Meth. A241 (1985) 616. [22] T. Takahashi et al., Phys . Rev. A12 (1975) 1771 . [23] M. Miyajima et al., Phys . Rev. A9 (1974) 1438. [24] M.G. Albrow et al., Nucl . Instr. and Meth. A265 (1988) 303. [251 L. Onsager, Phys. Rev 54 (1938) 554. [261 C.R . Gruhn and M.D . Edmiston, Phys . Rev. Lett . 40 (1978) 407. [27] R.T. Scalettar et al ., Phys . Rev. A25 (1982) 2419 . [28] E. Buckley et al., CERN-EP 88-120, Nucl . Instr. and Meth. A275 (1989) 364. [29] G. Jaffe, Ann. Phys . (Leipzig) 42 (1913) 303. [30] J.B . Birks, Proc . Phys . Soc. (London) Sect . A64 (1951) 874. [31] R.A . Holroyd and TX Sham, J. Chem. Phys . 89 (1985) 2909 . [32] D.F. Anderson and D.C. Lamb, Nucl . Instr. and Meth . A265 (1988) 440 . [33] J. Thomas and D.A. Imel, Phys . Rev. A36 (1987) 614. [341 W. Hofmann et al ., Nucl . Instr. and Meth . 135 (1976) 151. [35] S.D. Biller et al., Nucl . Instr. and Meth. A276 (1989) 144. [36] G. Bakale et al ., J. Phys. Chem. 80 (1976) 2556 . [37] E. Radermacher et al ., Proc . 24th Conf. on High Energy Physics, Munich (1988) . [381 E. Aprile, K.L . Giboni and C. Rubbia, Nucl. Instr. and Meth . A241 (1985) 62 . [39] P.J. Doe et al ., Nucl . Instr. and Meth . A258 (1987) 170. [40] J.A . Knight and C.T. Lewis, Radiat . Res. 23 (1964) 319. [41] R. Holroyd, to be published. [42] J.A . Northrop, J.M . Gursky and A.E . Johnsrud, IRE Trans. Nucl. Sci. NS-5 (1958) 81 . [43] S. Kubota et al ., Nucl . Instr. and Meth . 196 (1982) 101. [44] S. Kubota et al ., Phys . Rev. B17 (1978) 2762 . [45] T. Doke et al., Nucl . Instr. and Meth . A235 (1985) 136. [461 I.R . Barabanov, V.N . Gavrin and A.M . Pshukov, Nucl . Instr. and Meth . A254 (1987) 355. [47] A.J.P.L. Policarpo, Proc. INS Int. Symp . on Nuclear Radiation Detectors, Tokyo (1981) . I. PLENARY TALKS


J. Feltesse / Liquid noble gas and warm liquid detectors

[48] D.F . Anderson, Nucl . Instr. and Meth . A245 (1986) 361 . [49] S. Suzuki et al ., Nucl . Instr. and Meth . A245 (1986) 78 . [501 K. Masuda et al ., Proc. 3rd Workshop on Radiation Detectors and Their Uses, KEK report 88-5 (1988). [51] C.W . Fablan, CERN-EP 85-54, and in : Concepts and Techniques in High Energy Physics III, ed . T. Ferbel (Plenum, New York, 1985) p. 281; J. Engler, Nucl. Instr . and Meth . 225 (1984) 525. [52] J.H . Cobb et al ., Nucl . Instr. and Meth. 158 (1979) 93 . [53] W. Braunschweig et al ., Nucl . Instr and Meth . A265 [54] [55] [56] [57] [58] [59] [60]

[61] [621 [63] [64] [65]

(1988) 419. U Amaldi, Phys . Scrnpta 23 (1981) 409. C W. Fabian et al ., Nucl. Instr. and Meth . 141 (1977) 61 . A.L . Sessoms et al., Nucl . Instr. and Meth . 161 (1979) 371 . S Aronson et al ., Nucl . Instr. and Meth . A269 (1988) 492. S Wimpenny, Int. Conf . on Advanced Technology and Particle Physics, Como, Italy (1988) H. Gordon et al ., Proc . 24th Conf . on High Energy Physics, Munich (1988). M. Krammer, these Proceedings (Wire Chamber Conf ., Vienna, Austria, 1989) Nucl. Instr. and Meth . A283 (1989) 630. W. Braunschweig et al ., DESY 89-022 (1989), J.E. Bran and T.A . Gabriel, Nucl . Instr. and Meth . A275 (1989) 190 and references therein . R. Wigmans, Nucl . Instr. and Meth, A259 (1987) 389. H. Brückman et al., DESY 87-064 (1987) . V Radeka and S. Rescia, Nucl. Instr. and Meth . A265 (1988) 228.

[66] C.W . Fabian, Nucl Instr. and Meth . A252 (1986) 145 . [67] D. Hithn, Proc . Workshop on Compensated Calonmetry, Pasadena (1985) CALT-68-1305 . [68] J.P Dishaw, SLAC Report-216 (1979) . [691 CDHS Collaboration, Nucl . Instr. and Meth . A180 (1981) 429. [70] E. Aprile et al ., Nucl . Instr. and Meth . A261 (1987) 519. [71] E. Apnle, W.H .M . Ku and Jun Park, IEEE Trans. Nucl . Sci. NS-35 (1988) 37 [72] K.L Giboni, Nucl . Instr. and Meth A269 (1988) 554. [73] T Doke et al ., Nucl . Instr. and Meth . A237 (1985) 475. [74] D. Hitlin, Proc . 24th Conf on High Energy Physics, Munich (1988) . [75] E.V . Anashkm, these Proceedings (Wire Chamber Conf ., Vienna, Austria, 1989) Nucl . Instr. and Meth . A283 (1989) 752. [76] C Rubbia, EP Internal Report 77-8, CERN (16 May 1977); E. Gatti et al ., IEEE Trans. Nucl . Sci. NS-26 (1979) 2910 [77] H.J Mahler, P.J. Doe and H.H. Chen, IEEE Trans. Nucl . Sci. NS-30 (1983) 86 . [78] E. Aprile and M. Susuki, AIP Conf Proc. 170, Nuclear Spectroscopy of N Astrophysical Sources, Washington, DC (1987). [79] A Ciocio, Proc. Conf . Neutrino 88, Boston (June 1988). [80] L Bassi et al ., Icarus 1 (March 21, 1988).