This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
test results given in sections 0 and 7. Finally. our detected photoelectrons N is determined by the reprogram is described in section 5 and compared to the ·,l24 for triangular detector response. The number of
The general design of the Monte Carlo simulation energy bandwidth UE = AE/ i/I2 for square. JE = AE/tronics and the test facility are described in section 4. strongly dependent on E. The last term is the rmsters of the tested photodetector. The detector, its elec second term (dn / dE ). the radiator dispersion. isfactory RICH counter which determined the parame 1/n tan 8. which is only weakly dependent on E. The
3 contains :1 description of the proposed B meson the particle velocity). It then follows that 88/Bn =
detectors with single photoelectron sensitivity. Section »zB cos 8 = 1 (rt is the radiator refractive index and Bgaseous or solid photoconverters and surveys the gas (2) mav be found from the Cherenkov relation
fast photon detectors. briefly analyzes the various T = transmission. R = reflection). The first tertn of eq.
Section 2 sets out some characteristic features of (between Eh and E,) of efficiencies (Q=quantum.diverse pixel sizes and photoconverters. where e =(1/AE)f(QTR)dE is the energy averagewas then used to predict detector characteristics for
tr,(E) =(80/6n)(dn/dE)0E, (Z)Carlo program simulating its operation. This programits geometry and to develop and test a general Monte N,,=(a/hc)s AE=(370eV" cm")e AE. (1)pixel resolution. reconstruction efficiency). to optimise
error 0·,(E )formed to find its response (detection efficiency, speed,temtines its response parameter N0 and its chromaticDetailed tests of the photodetector have been per
i) Its energy acceptance (AE = E, — Eb) which des", tg = 10.6 GeV) operating at 10 MHz.ized by:luminosity B factory (e ’-c` collider. L 21033 cm'
The photodetector for a RICH counter is charactertodetectors, undertaken originally for use at a highThis paper reports an experimental study of such pho
2. The fast RICH methodmultiplicity environment at 100 MHz interaction rates._1 A fast photon detectors. capable of operating in at high
liders (LHC, SSC and ELOISATRON) will require(RICH) counters at future high-luminosity hadron col measuring method.
Use of panicle identifying Ring Imaging CHerenkov to the granularity of the detector and the coordinate
the reconstruction error limits are discussed in relation
I. Introduction conclusions about design parameters are outlined and
Carlo simulation program in order to predict the response of a photodetector with various photoconverters and pad sizes.
error (ar < 1.5 mm) over the full surface of the detector. lts performance was used to tune the parameters of a general Monteconstruction parameters to fulfill the requirements of speed (a, < IO ns), cathode pad detection efficiency (ec > 0.98) and pixelcollider. may also be used at hadron colliders as well as at fixed target facilities. These tests allowed the optimization of relevant
TEA as the photosensitive converter. This detector. designed as part of a fast RICH counter for a B factory electron-positronaltemative fast and efficient photosensor. A detailed experimental investigation has been made of a multiwire photon detector withcompared to the known standard gases (TEA and TMAE). A solid Csl/TMAE reflective photocathode is also considered as an
high luminosity environment. The quantum efficiency of several new photosensitive gases (TMA and DMA) are measured andFast ring imagmg Chcrcnkov counters. with pad readout for unamuiguous tmage reconstruction, are considered for use in a
Recenved 7 October 1991
d/ C CERN, Geneva, Switzerland( Ad 2Ccllége de France. Paru, France
$2 3 O 0 2;* “ CRN. INZP3-CNRS / Louu Pasteur Uniuersny, Szrasbourg, mnceR_ Arnold “_ Y, Giomatans "‘°, J.L, Guyonnet “‘, A. RaCZ “, J. S€gLll1’10l b Hfld T. YDSIIHHIIS··‘ @**"°Cx.JQu? A fast—cath0dc pad-photon detector for Cherenkov rmg xmagmg
2 Q32 ru
$ec¤·Q¤ARESEARCH{ _ _ \ C} I1 N {V|N pgygggNorm-Holland
&ME·|·"°¤sNuclear Instruments and Methods m Phvsncs Rcscarcn A_%14(1992)4b5—-I94INSTRUMENTS
Name Formula State [KI IKI lKl {Torr] IK) l¤Vl l¢Vl Wl lMbIlMoleculei in(NTP) Tm T, Ilii AH / R Eth
Physical properties of selected photoeonverters
Table l
hence use of these gases will require a single crystalinp=p..¤x¤[.xH/R(h7‘ -1**)]higher than the 7.6 eV cutoff energy of fused quartz
sius-Clapeyron equation vlamine (DMA). Their thresholds of 8.0 and 8.4 eV areture T. determined by the ratio AH/R in the Clau the NTP gases trimethylamine (TMA) and dimeth
ii) The vapour pressure p of the liquid at tempera We have recently measured the quantum yields otil The ionisation threshold E,,,. molecules virtuallv useless for RICH counters.
some of their characteristics [4]: methanol and ethanol. listed in table I. renders these.~\ list of photoconvertcrs is shown in table l with energy thresholds of acetylene. benzene. acetone.
eounter`s particle identification capability. The high2.1. The phazoconrerters reduces chromatic aberrations hence improves the
permits a larger choice of windows and radiators anddigital or analog. A molecule with a low photoionisation thresholdtector and the readout method which mav be either 2.1.1. The photosensitive gasesgeometry of the radiator. the pixel size of the ph0tode—
solid [3]. The spatial resolution is govemed by thethe list of usable gaseous photoconverters.
choice of photoconverter which may be gaseous [2] or hence. {ph $.0.5 mm. This condition severely restrictsThe parameter N,, and the speed 0, depend on the
Ars, I5 ns (to separate successive beam crossings)the detector and readout electronics.
mm/ns). To operate a detector at LHC will requirev) lts maximum data acquisition rate is defined by
3Inh/ty, where tg, is the electron drift velocity (gillrate photons from different images.
absorption in a gaseous photoconverter is .\: == Sa, =iv) Its pixel size determines its capability to sepa
sorption length [nh. The time spread for 95% photonthe photoeicctron arrival times.
time spread .\1 is determined solely by the photoab·iii) lts speed ir, is determined by the variance ot
ln a last photodetector the photoelectron collectionHug) is the reconstruction relation [ll.
Torr Mb K` ' ).tained from the equation u,,(i·,) = M9/Hr·,>¢r, wheremole " K` ') and BoItzmann`s constant k = I.036 (mm
I', =¢.r.. :,.. x. y. :). The unguiur errors are thcn obpressure (I mole "). the gas constant R=8.3I ll
and the detection point (x. _i·. :> errors arc ir , whereHere. .\H is the liquid heat of evaporation at constant
error wnh the error <r,,(c,). The emrssion point (xc. sc}
is determined by the quudruture sum 0f the chromanciv) The quantum efficiency QI E) at photon energy
mines the photon mean free pathri) hs total angular crroriii) The photoahsorption cross section ir, deter
\'=N,,L >1n·r/. 1})atures are Tm and Te. respectively.
qurdmz 10 thc rclurion perature respectively. The melting and boiling temper»punsc parameter _\‘,, And mc rudiunor lengm L rm where p., and TZ, are the reterence pressure and tem
carrier gas. The data are from this work. OCR Output elevated temperature would pose considerable techniphoton wavelength A (energy E) with either He or CH; as the ' Dl‘€V€nl condensation). Operation of a detector at thisFig ]_ The TMA and DMA quayuurn efficiency Q Versus the same lm (the detector must be heated to l05°C to
X (nm) hubbler. however. must be heated to l()0°C to obtain130 11.0 150 160 gives lm =().56 mm at T= 300 K [eq. (5)]. A TMAE
at 20°C. The TEA cross section of 9.7 Mb (see fig. 4)Torr) have sufficient vapour pressure for fast operation
(see figs. l—3) but only TMA (760 Torr) and TEA (52
interesting molecules are TMA. TEA and TMAE [5.6.7]Due to their high quantum efficiencies. the most0.1 l
nantlv chromatic).H¢•DHA
than for TEA (if the total angular errors are predomi
alent particle identification is ll% higher for TMA0.2equal Qm then. from eq. (6), the momentum for equivUB/B; =/<, ll]. For triangular quantum response and9 S 9 0 8.5 8.0The B resolution of a RICH counter is then simply
130 1LO `|50 160and 0,, is the total angular error per detected photon.
kr=n(’11/\,lNtiL) ~ (7)
where the RICH constant kr is defined as¤.1 |- cHt~ wi
The data are from this work and ref. [5].0.2
length A (energy E) with either He or CH , as the carrier gas.
Fig. 2. The TEA quantum efficiency Q versus photon wave
X (nm)
m i+¤·rMA 130 1L0 150 160 1700.3 l- °
95 9.0 8.5 8.0
E leVl
0.1
This work02(6). e p=l(mE—1ny)B/211`,kr
1 V:
rtw, :1.5 mmstions is given ll] by relation 0.3 l" cm · TEAund m, can be discriminated with 11,, standard devia
9 S 9 0 8 5 8 0 7 5The momentum p at which particles of mass mi130 1L0 150 160 170chromatic error will he smaller bv = l.5.
much smaller (().6 versus l.l eV) hence. the TMA
larger (average energy 8.35 versus 8.2 eV) but .\E isM A TEA [via eq. (2)] we note that the dn /dE is slightly
0.133"E). Comparing thc chromatic errors of TMA withefficiency (Qm = 28%) near to that of TEA (Qm =detector operates stably and has a maximum quantum
We have shown (see section 7) that a methane-TMA 0.2 l- _will have a sufficiently large No to be useful for RICH.
This workdepends on AE [see eq. l ll] we surmise that only TMAll.3 cV. rcspcctivcly. Since the detector response factor
0.3 l- He · reaenergy ucccptuncc bandwidth of thcsc gases utc lm or
whnsc transmission limit is ubout 8,7 u\/ thus. thccounting rcquircs un amplifying gus such us mcthunc 9 5 9 O 8 5 S 0 7 S
mndow such us CuF~. Efficient single photoclcctmn E (ev)
R. .-{rm;/d cr ul. , P/zomn uerecznr mr (`1tt·rt·n1k<11 rmx: zmuwmz :67
- .' WT
are from this work and ref. [5]. because both have a single amplifying gap and both are OCR Outputwavelength A (energy E) with He as the carrier gas. The data tom UD) has much the Sami? C0¤f[BUYa¥[O¤ as inc MWPCrig. 4. The TEA photoabsorption mss section versus photon (operating inversely with amplification frnni inc bot
chamber avalanche chamber (PPAC. fig. 7) or a multiproportional chamber (MWPC. fig. 6). a parallel platetoelectron amplifier element. This may be a multiwiretween the conversion volume (or surface) and the pho
$-1Fast photodetectors have a direct connection be\\ "
2.2. Fast photodetectors{ in -l •3 *;\
later.w \ “ \ /\ xi \ \use and its effect on detector operation are discussedts-1photosensor in a fast RICH counter. Conditions for its‘/ E • . This workreal alternative to gaseous TEA (or TMA) as the•· Belt-uyd es al. Sl l
RICH counters. This cathode therefore represents a
I Y B feedback problems which, up to now. are endemic toE (•vI 5
gas mixture. This will reduce considerably the photonwithout a photosensitive component in the amplifyinghence. a detector with such a cathode can operateciency (Q(7 eV)= 34%) but usually operated in vacsurface even when the flow of TMAE gas is stoppedlow threshold (Em =5.9 eV) and high quantum effieV). The adsorbed layer remains adherent to the Cslted isochronously. Reflective CsI photocathodes have athreshold energy is reduced to that of TMAE ( Em = 5.3RICH detectors because the photoelectrons are emitadsorbed on the Csl cathode surface (fig. 5) while thePhotocathodes are of great interest for use in fastTMAE gas is attained when a thin film of TMAE is2.1.2. Phomcarhodesquantum efficiency comparable to. or better than.
methane at modest electric fields (= I kV cm"). Ator constitute the main subject of this paper.efficiency even with electron extraction into NTPRICH counter. Tests of a TEA + methane filled detecposited, Csl reflective photocathode has high quantumimportance of TEA as the photosensitive gas in a fast
It was shown that a 500 nm thick. vacuum deN,,=85 cm". This large value shows the potentialadsorbed thin film reflective photocathodes.R= T= 1) an upper limit for the detector responseus [3] have made a systematic study of solid. liquid andtum efficiency (fig. 2) between 7.5 and 9 eV gives (forextracted into a gaseous medium. To this end. some ofIntegrating eq. (I) with the TEA + methane quantive multi-wire counters. the photoelectrons must bephotodctectors.uum [8]. For a cathode to be usable with p0sition-sensinow the only photosensitive gases suitable for fast
cal difficulties. For this reason. TEA (and TMA) are
shown for comparison.
sorbed TMAE [3]. The TMAE gas quantum efficiency isare trom this work and refs. [5] and [6]. of a 500 nm thick reflective Csl film without and with ad
wavelength A (energy E) with He as the carrier gas. The data Fig. 5. The quantum efficiency Q versus photon wavelength AFig. 3. The TMAE quantum ctficicncy Q versus phomn
tr G g ~y\as;g-¢§`»g\ _. _\ T \ML \ akeun, rm H°°THAEas E` \- 0CsIt y ATMAE
9 a 7 ICs| + Gdsorbed TMAE°·7 E"E (ev)
468 R. Arnold er ul. x Phmon detector mr Claerenkoc nm: Amaszmz
. ·;».·2..#r·"*tJ;° 2
.. ( .- QamtkT ‘¥*§&?Z* #
ton detector [12.13]. version point is improved by detection of feedback OCR OutputFig. 7. A parallel-plate-avalanche-chamber lPPAC) fast pho The tests show that determination ot the photocon
ethane the gain was limited to o >< l0°
gain up to 2 >< l()° whereas with TMAE + methaneEfilled with TEA-+- methane has operated stably with1.5 HQ .
carbon line to create feedback. A MWPC detector" _ 7
isobutane absorbs above 7 eV leaving only the 6.43 eVwith csi-was [ 002 F . u A mixture of TMAE + isobutane is also good becauseHV ‘Cathode Pads —--,-—— -—- —- —
the TMA concentration is usually s 7% of the total).1.5 MQ at 8.3 eV can contribute (nitrogen is 28% of TMA but
feedback whereas in TMA only the weak nitrogen line
Quartz wmdnw 7.5 eV (fig. 2) only the 7.94 eV carbon line can cause
gain limited to g 5 (uQ)` '. Since the TEA threshold isa PPAC in the presence of a photosensitive vapour has
photoelectrons are not amplified in both gaps. This tector becomes unstable if nQ 2 l hence. a MWPC orhas much reduced feedback because the secondary with relative probabilities of (13. 57. 30)% [10]. A desame photoconversion gas thickness as a MWPC but 7.7>< l0`° [7], averaged over the three carbon lineshas been accumulated. In general. a MSAC has the proportional to the detector gain g with constant cz =in experiments [9] but. as yet, little operational data surfaces. The number of feedback photons n =¤g isa MWPC. Detectors of this type are now being tested phase photoionization or by the photoeffect on solidput signal (after electronic filtering) is greater than for lines are free to create secondary electrons by gasof the total developed charge contributing to the out eV only one nitrogen line (8.3 eV) and all the carbondrift of positive ions towards the cathode). The fraction Since methane strongly absorbs all photons above 9faster than the MWPC positive pad signal (due to the and at 8.3. l0.() and 10.3 eV from nitrogen [l0].trons are collected giving a negative pad signal which is gen (Lyman ct). at 6.43. 7.46 and 7.94 eV from carbontions and sparking. ln a MSAC, the avalanche elec predominant photon lines are at l0.2 eV from hydrobetween the mesh planes cause distortions, gain varia rapidly de-excite by isotropic photon emission. Thetric fields needed over large areas. Electrostatic forces collisions during the photoelectron avalanche process.feedback but. it is difficult to maintain the large elec gen, produced in the detector gas by electron-atomseparate gain stages and so largely eliminates photon Excited atomic levels of carbon. nitrogen or hydroanode pad plane is superficially attractive since it has
An atmospheric pressure MSAC (fig. 8) with an 2.3. Secondary photoelectrons: limitation ofderector gainreadout channels.
images but. requires a large number of detection and eration of a reflective photocathode.method ensures unambiguous detection of Cherenkov gain separation. however. is incompatible with the opthe two photon separation ability. This pad readout
Cherenkov angle reconstruction error and determines
angular pads (fig. 0). The pad size contributes to the
reading out thc lower cathode plane. divided into rectdetector.
mensional position of the image point is obtained by Fig. 8. A multi·step-avalanche-chamber (MSAC) fast photona retlective cathode. ln all cases the projected two·di ’1mm SDGEIDQ)
EO um wire plcineMSAC. with two amplifying gaps. is incompatible with Anode buds
compatible with zi rctlcctivc photocathodc. whereas the
ton detector.1 T
Fig. h. A muiu·w1rc-proporuonubchzxmbcr KMWPC) fast pho
FQQGVYWGJQHJ.*5 »m¤ ,1mm spncuqr / C°°;’¤°E'
Sense/ware / y\ /
·/»·/» Mn w 7 1+4 »/»§§€§§‘§ 3/% §<it ••¤••, ......••»•••»•.•.·•
nw wml meshesCathode stripsT\
»\ M y :~;·c2»+G·rv¤e(·o¤-¤¤¤reH&~ #a:.<e:·c:quartz lC¤FZl;_qry, yg5_ Nm, Czvsonz mcs
LM)R. Arnold er ul. , Plmmn rletecmr mr ( `/zervnkoz mu: rmuumu
.`i$:"m??1}:,* 2_';_;1 IQ; .1. :l’\n_
vessel. before exposure to air. The original work. how surface made up of a rectangular array (5.08 >< 6.604 OCR Outputwhich was sensitized with TMAE inside the vacuum asymmetric MWPC with the radially outer cathode[3]. They evapored a 2 um thick Csl photocathode The proposed photodetector. shown in fig. 9. is anever. was produced differently than the original work ( = 130 mm) by the needed angular resolution oy,.effects may be accentuated. Their photocathode. how number of image points and the lever arm distancetor sealed at low pressure (20 Torr) where charging thickness (= 10 mm) is determined by the requiredcharging ot the cathode [13]. They operate their detec ring) containing 10-20 image points. The radiatorbut. showed evidence that this enhancement is due to detected photons form a parabola-like image (not aquantum response of the Csl/TMAE photocathode He) and are absorbed in a MWPC photodetector. The[12]. An even newer work has confirmed the large lever arm region filled with a transparent gas (N: orand shown to give good single photoelectron response passing through the NaF crystal. propagate through theCsl/TMAE cathode and pad readout has been tested Cherenkov photons. produced by a charged particle
Recently, a low pressure PPAC detector with a tend an azimuthal angle of 12°.being tested [ll]. polygonal barrel staves (30.9 m in length) which subphotocathode has been built and is now (summer 1991) region extends between radii of 0.65 to 0.85 m and hasA MWPC pad detector with a reflective Csl/TMAE end cap counter longitudinal view in fig. 11. The barreldetector (the TEA detector is stable for g 5 2 >< l0°). is shown in fig. 9, the transverse view in fig. 10 and theeffect of feedback will be to limit the gain of the 1.36). The cylindrical barrel counter longitudinal viewdetected (correctly) as a single signal. The only adverse (n= 1.47 at 8.2 eV. threshold Lorentz factor ylh=um/ns) hence. these signals will tend to merge and be focusing [18] with a 10 mm thick NaF crystal radiatoranode-cathode distance d = 0.5 mm and ty, = 50 The proposed RICH detector system has proximityafter the initiating photoelectron signal (td = 10 ns for
tion. the photocathode feedback signal will occur soon 3.1. General constderarionshave only 72% of the TEA photon feedback. ln addi
30% and 13% hence. the Csl/TMAE detector shouldwas undertaken for the B factory project.
2 and 5). The respective C* emission probabilities areThe experimental study of a TEA photodetector
Csl/TMAE photocathode at 6.43 eV is 30% (see figs.same fast photon detectors [1.18].
efficiency of TEA at 7.94 eV is about 18% whereas theaccomodated by gaseous Cherenkov radiators using thc
only sensitive to the line at 7.94 eV. The quantum200-300 GeV/c. This higher momentum range can be
than that of a TEA + methane photodetector. which iscounters for particle identification at momenta up to
instability. The feedback probability will be even lesstarget facilities [17]) which will require fast RICH
the 6.43 eV carbon line as the only source ot` feedback[OTS have been proposed (at colliders [16] or fixed
sponse energy of the detector (El == 7 eV) and leavesGeV/c. In addition. B experiments at hadron accelera
the quencher. The quencher determines the top rewill require er/K particle identification up to about 4
as the carrier gas and isobutane, pcntane or hexane as(still at i/s = 10.6 GeV) have been proposed [15] which
and 7.46 eV. This may be achieved by using methaneGeV/c. More recently. asymmetric B factor designs
with a gas which is opaque to the carbon lines at 7.94cation of kaons from pions at momenta up to 2.5
cathode pads. a MWPC detector would be operated10.6 GeV [14]. These experiments will require identifi
With a Csl/TMAE photocathode deposited on thee “e` collider B factory at center of mass energy vs
wires as needed for TMAE at T < 40°C [7].high-luminosity (3 >< 10-10cm s) symmetric
3357 ; i`tor operates stably. without blinds (cloisons) betweensection. lt was initially proposed for an experiment at a
signal. A thin TEA (or TMA) A- methane filled dctec—The fast RICH project is described briefly in this
thin (= 2 mm) thus minimizing the direct ionization
he attained at NTP hence. the detector can be made3. The B meson factory fast RICH project(or TMA) since an absorption length lnh = 0.5 mm can
per cm of NTP methane). This is possible with TEA
ionization is so large (i.e. 200 primary and secondary e
gation of feedback photoelectrons because the direct NTP).
tector it is essential to limit the production and propa with a Csl/TMAE cathode pad detector operating at
When the charged particle passes through the de then (or now during the current photon counting tests
constructed Cherenkov angle. 47%. No charging effects whatsoever were observed
photoelectrons and degrade the resolution of the re sensitized with TMAE. increasing the efficiency to
pads, these ure wrongly interpreted as independent efficiency rose to 35% (at 170 nm). Only then was it
feedback photoelectmns are detected on nonadiacent with a flow of clean methane gas until the quantum
lating mc baryccmcr uf thc hit pads. However. when transferred through air to the detector and then rinsedphowclccxrons on udjaccm pads of 11 cluster. by calcu ever. used a thinner Csl cathode (0.5 um) which was
470 R. Arnoid er ul. / P/mum dewcrur mr Cixercnkm rmx; amuuznq
,` — -·¤$·‘•htay.:-*¤=·:.¤&»·-z
· 1 "‘ ; ‘ 3,-A
back photoelectrons. (dn/dE =().074 eV") at the center of the TEA cnOCR Outputcharged particles and the resultant production of feed— i) The chromatic dispersion of NaF [19] is largegreatly reduces the direct ionization signal due to The drawbacks are:
iii) The thinness of the photoionizing gas layer fast reflective CsI/TMAE photocathode.CC ii) The detector runs at ambient temperature. photosensitive gas operation (TEA or TMA) or with a
of cosmic rav, machine and electronics noise. viii) This detector geometry is compatible with fastt)I' ated with a given beam crossing. which allows rejection 4 >< 1()° total detector pads in under 5 p.s.
ns) is possible. An interaction may therefore be associ— vii) Fast readout (with zero suppression) of the
sive beam crossing ( = 100 ns). fast data sampling ( = 50 particle multiplicity trigger.
= 36 ns) is shorter than the time between two succes photoelectrons and allow. for example. a secondary
i) As the photoelectron collection time spread (.\r tion) can discriminate between charged particles and
The advantages of these choices are: vi) Fast wire readout (with pulse height discriminaphotons with a time spread (At = 36 ns. 0-, = 12 ns). may be still further reduced (see iii) below).
modular installation of the elements. The total masstotal gas thickness (2.5 mm) absorbs 99.3% of the UV
mm) with a 3.5 mm thick CaF» entrance window. The rial to about = 20% X., (radiation length) and allows
v) Its simple construction minimizes the total matesaturated with TEA at 2U°C (p == 52 Torr. [ph = 0.5The detector operates with a mixture of methane quadrupoles. not by the RICH counter.
thickness) i.e. a,(x), ¢r,,(,v) s ¤·,,(zc). A12/4rr=O.9S. This angle is limited by the machinedesign covers Bp 2 l8° (w.r.t. beam direction) hence.tion from the emission point error (due to radiatorthe barrel-end cap boundary without dead space. Theerror contribution to 0,, is much less than the c0ntribu~
The pad size was chosen so that the detection point radiator support structure and radiator interfacing at
arc individually readout to define thc image points. ators allows full solid angle coverage without inactive
mm·> of indcpcndcnt mctal pads {scc also fig. 6} which iv) The proximity focusing method with crystal radi
photon detector with VLSI electronics.
Fig. 9. A longitudinal view of thc B factory RICH counter with :1 NuF crystal radiator. a CaF` crystal window and thc MPWC fam
MZTTW` \ Q\
2 I /· 5 K \ /// M/ ml \% , % wNapvomml HOHEYCOMD (ammlRadnaror / /
.17}R. Arnafd cz ul. , /’/wma uezyczrzr mr L /zvrw1km rmx: mmumu
..: ‘n...?»~*
electrons in a solenoidal magnetic field) it is best to total counter mass of about 10% X,,. This option will OCR OutputTo minimize 1·r" loss (due to curl up ot low energy crystal LiF may be used (7.5 mm = 5% X.,) to obtain acalorimeter. placed radially outside the RICH counter. detector mass to about 5% XU. As radiator. the low Znot be optimal with regard to an electromagnetic suppress the quartz entrance window thus reducing thethe inner radius and detector at the outer radius may image points. With this cathode it may be possible to
iii) The chosen geometry with the crystal radiator at additional pathlength does not cause a serious loss olsive fused quartz windows. in the less absorptive (6 to 7 eV) interval where thecathode would permit the use of robust and inexpen a Csl/TMAE cathode detector has its energy responsesive CaF. windows. A Csl/TMAE reflective photo in the absorbtive TEA range (7.5 to 8.9 eV)]. However.
ii) Operation with TEA requires fragile and expen which reduces the number ot image points [especially4.5 GeV/c and even with quartz to 3.7 GeV/c [18]. absorption (due to increased pathlength in the crystal)tion with LiF goes to 5 GeV/c. with NaF or Cal:] to leads to photon loss (due to reflectivity) and additionalTMAE energy region (6 to 7 eV). Then. wr/K separa detector by mirroring the exit side of the crystal. Thischromatic dispersion of all crystals is much less in the Cherenkov photons must then be reflected back to thewith a Csl/TMAE reflective photocathode because the detector is placed at the inner radius. the
detector at the outer radius. as in the present design. IfGeV/c [IB]. This drawback is considerably alleviated·i·r/K separation to 3 GeV/c and LiF (or MgFz) to 4 tor has 8.6% X,, hence it is preferable to place theOther. less dispersive. crystals such as CaF2 permit has about I0% X0 whereas a IO mm thick NaF radiaever, is a property of the radiator not the detector. TEA detector (with entrance window and electronics)
separation (3 a) to 2.5 GeV/c. This limitation. how detector) at outer radius near the EM calorimeter. The
crgy region (7.5 to 8.9 eV) which Iimits the rr/K place the largest mass RICH component (radiator or
detectors and VLSI electronics.
Fig. IU. .~\ transverse vicw oi thc barrel and and can sections ot the B factory RICH counter with its radiaune crystals. fast photon
0 100 200 300 b00rnm
`—""®H
&0°
12·
v` `x Ir:30O
He ur NZ
/,/' NaF(10mm)
/ \Radiat¤r
,rq650 fm,
x w § xHe or N1 {\ g 1 »
nn u ¤¤»¤¤ u u n
{VLSI} . , (cathode nad readout)| *4aF wir•¤¤vs Hmm!
Section AA
R. A~lrnu1u` er ul. P/101011 durecror mr C/1crcnk0L· rmv tmuvmu472
each column) thus minimizing the clock pulses needed and the cathode surface (4) was adjustable between 0.4 OCR Outputthe shift logic compacts the data upward (to the top of The distance d between the anode wire plane (1)frequency of 20 MI-Iz. At the start of a read sequence. adjacent pads was measured to be = I pF.word RAM and then shifted along the pad column at a this method of fabrication. the capacitance betweenaddress of the hit pad. The data are loaded into a lo pins (5) to which the preamplifier is connected. Withpriority encoder to suppress zeroes and provide the linked to the outer surface of the detector by metalhas an input register and each chip contains a 4-bit the residual brass thickness is 0.5 mm. Each pad iswith test results is being prepared [20)). Each channel face has been trued and polished (using diamond tools)beyond the scope of this paper ta complete description pads is filled with epoxy resin. After the cathode surchip. Its operation is complex and its description goes independent pads. The 0.2 mm wide division betweenThe readout circuit proper is integrated on the digital away crossed strips of metal to define the electricallyexternal logic) to select the corresponding RICH event. to an 8 mm thick vetronite substrate. then machiningdence with a 60 ns wide write gate (generated by (0.2 >< 0.3 in?). lt is made by gluing a thin brass platetrigger. The delayed trigger pulse is used in coinci— 6X 8 rectangular pads each of area 5.08 >< 7.62 mm;circuit is to provide the time delay for a first level sponds to 4.5 {ph. The cathode (2) is segmented intoregister operating at 50 MHz. The purpose of this which. for methane saturated with TEA at 20°C. corredelay pipeline at the input realized by a 64 step shift 2.5 mm between the window (3) and the cathode (4)
The second is a digital CMOS chip with a 1.28 its The test photodetector (fig. I3) has gas thickness ofDAO system via the second VLSI chip.
ke) between l—l6 ke by a 4~bit address supplied by the 4. I. T/te photon detectorThe discriminator threshold is adjustable (modulo 1
4. Experimental test setupThe mean rms noise level at the input is (aq) = l0’ e.of a current preamplifier. amplifier and discriminator.
The first is an analog bipolar chip with 8 channelsschematically in fig. l2. several microseconds.tord Laboratory [20]. The readout circuit is shown the lZ° sectors. the total data acquisition time will betion with the microelectronics division of the Ruther For 30 hit pads per sector and with parallel readout of
Two VLSI chips are being developed in collabora of clock pulses needed for essential shift operations.3840 channels. number of hit pads and nu = I6 is the average numberbus lines (optical fibres) per l2° half-sector containing readout time is about 50(nn + nn) ns, where np is thetion system are limitcd to the 5 data monitoring and 6 Simulations of a l2° sector show that the averageformed in situ so that connections to the data acquisi column address (generated by internal logic).data selection. memorization und rcudout are per quentially, giving the hit pad row address and thetributed 0vcr thc dctcctor buck surface. Dctcction. for data extraction. Each column is then readout se
474 OCR OutputR. elnznizl ur ui. f /J/mwzx uezecwr mr ( `lrcrcrr/my run: mzzusrrrg
: ~·
i(L4 the position sensitivity of the photodetector. Transmis and pad positions. OCR Output; (1) cathode is uniform over the surface. as needed to test function of the photon entry point relative to the wire
acceptable for these test). The transmissivity. of this allow accurate mapping of the detector response as axc€¤ eV (unacceptable for a working RICH detector but rr\=rr,'=150 ttm. The beam is sufficiently fine towith transmissivitv of this coating is only about -11)% at 8.2 by a two-dimensional Gaussian with rms widths olictul window to define the upper cathode potential. The at the PM output. The fx. y) profile is well described1Ll iS was vacuum deposited on the inner surface of the intensity at each point (x. y) was measured by an ADC¤t)lS) resistive (R = 25 kl!) tungsten coating (1.5 nm thick) mounted at the normal detector position (1). The lightsul' crystal disk 76 mm in diameter and 5 mm thick. A r or v) relative to a XP2020 photomultiplier (PM) tube.ccn The detector entrance window is a CaF; single ured by displacing the assembly (in 50 um steps along;allv the wire hence. would not be amplified and detected. The photon beam intensity distribution was measning Photoelectrons produced in this region do not drift to of the two transverse directions (y is along the wires).ilate dead region at the edges of the electrostatic cell. move the lamp. filter and collimator assembly in eachnin` twecn the upper and lower cathodes thus creating a adjustable by means of micrometer screws (4. 5) which1ntO spacing 5 > 1.27, direct field lines are established be mal to the detector window at a point lx. y) which is»rTC— and tcsts with the d=().5 mm show that for wire photoeleetron counting. The photon beam enters nor(4) be 5 0.5 mm (sec section 6). Electrostatic calculations flash was less than 0.3%. thus ensuring only single
rs ol (hence the pad efficiency) it was found that tl should the probability of converting more than one photon per
ated by metal mesh filters fb) to such a low level thatmaximize the amplitude ol the induced pad signal
ensuring uniform rcsponsc ovcr thc full surfacc. To GmbH). The intensity of the photon beam was attenu
width so that wire positions scalc lor cach pad column. l`ied through an Oxisorb cartridge (Messer Griesham
mm pitch. This spacing is a sub—multiple (§) of thc pad the photodetcctor is swept by flow of argon gas. puri
strung parallel to thc long pad dimcnsion with a L27 15-20 ns (FWHM). The region between the lamp to
Thc I5 ttm diamctcr gold plated tungstcn wircs wurc the lamp is emitted at lbf) nm with a time width of
induccd pad signal thus. to optimisc this paramctcr. separated metal discs (3). The maximum intensity of{ he and 0.7 mm in ordcr to mcusurc thc variation of thc 200 um diameter holes (collimators) drilled in two
.t ot quency of 100 Hz. A light beam is defined by means ofans. self-triggering relaxation mode and an average fre
(4) the vctronitc: chumbcr body and (5) thc pud connectors.ibcr charge lamp with a CaF2 window (7) operating in a48 cathode pads. (3) the CuF; window 45 mm thick. 76 mm rb).(hi? source is a low pressure (190 Torr) hydrogen are disFig. 13. The test dctccwr wuh (l> thc 32 anode wires. (2) {hcVugc position sensitivity of the photodetector. The UV light
xlmml
The test bench (fig. 14) was used to study of the-60 .40 40 0 20 40 60
thc
4.2. T/ic rest bench. \`“ L s -1n•toroids..75 Lbetween the various planes (fig. 13) are made of Vitonare connected via high frequency filters. The ()—ringsthe anode wires LQ (see fig. 41). The power suppliesdetection electronics. the potential window was lj.; and
The cathode pad plane L" was earthed by the
trolled bath.
was stabilized to ;0.1°C by a thermostaticallv conor TMAE to saturate the carrier gas. lts temperature
isobutane) with TMAE. A bubbler was used with TEA
with TEA. TMA or DMA and (75% methane + 25%
A carrier gas of pure methane was use in these tests,r/ ML5 r transmission loss.
able from the flash lamp to compensate the additional1 \:\however. more than sufficient light intensity was avail
50 P MWPC electrostatic field configuration). ln these tests.
wire direction (to minimize the modulation of the
1l()l) um traces with 1.5 mm pitch) transverse to thetype detector by vacuum deposition of a metalic gridsivity of (JB'? has been attained in thc full scale proto
nation level is adjusted via a variable external voltage several representative channels. OCR Outputinjected charge at the preamplifier input. The curves are forhysteresis of li) mV for shaping (fig. 16). The discrimiFig. 17. Detection inefficiency ot the discrimtnator versusa amplifier-receiver (A) and discriminator (D) with a
q(fC)channel of this CAMAC board (MECL 10 K logic) hasm long twisted pairs with lll) fl impedance. Each 0 0.4 0.21 1.2 1.6 Z :.4
are connected for readout to a CAMAC module via 5
The differential outputs of 16 channels of TRA4()3s
was measured to be S 50 dB (0.3%).
rear face of the detector. Crosstalk between channels0.2 Lon printed circuit boards which are plugged into the
32 wire outputs are fed into preamplificrs (2) mounted
0.4the preamplifier. as shown in fig. 15. The 48 pad andon 120 Q [21]. A discharge protection circuit precedes
with 5 ns risetime and a transresistance of l00 mV/(LA 0.6 ifintegrated bi·po\ar current preamplifier (4 per chip)
0 A LEach pad and wire is amplified by a LeCroy TRA403
A °o08 '
4.3. The electronics and data acquisitionl °¤5g. r O rt
supply, (9) trigger signal from lump.i><0iscrews. (6) metal mesh filters. (7) Hs utc lamp. (8) lamp power
47h R. Arnold or ul. , P/10mn detector tar Chcrunkvt rmx; zmautm:
»·—»·—.•¤:
°‘V * . :·"S==:L. {.:4 <»$:’$;,*\·#T&¤l~i£
avalanche is characterised by the mean value (q) = eu OCR Outputvia Ethernet. to a VAX 750 minicomputer for analysis.The charge tj induced on anode wire by thestored temporarily on a hard disc and then transferred.
sources of the MAC system. The acquired data are
munication and histogramming use the available re productionhistogram drawing) are written in PASCAL [22]. Com 5.2. r*ll`tl[UH('/ll} distribution and secondary phoroelectronacquisition program andthe various libraries (CAMAC.
CRON" interface and a Mac·CC crate controller. Thethis point is then taken as g =gmx (E/Emu).computer with access to the CAMAC bus via a "Mlat the arrival point (xl,. ji. sn) and hence. the gain at
The data were acquired in a Macintosh ll micro(from the relations of appendix B) the electric field Etected signal charge in each ADC channel.the wire facing the window). The program calculatesoth degree polynomials and used to correct the de(i.e. the gain for electrons which arrive at the top ofpreamplifier input. The calibration data was fit withwith the experimentally determined maximum gain gmx
ing known charges into a 1 pF capacitor at the eachof UW and (L,) is associated by the simulation programinput electrons). All channels were calibrated by injectThis maximum electric field Emx (which is a functioncharge on the wires and pads (after filtering in thealong the generatrix of the wires facing the window.0.25 pC/channel) for measurement of the inducedand B. For the potentials generally used. it is maximumpulse. without loss. to a LeCr0y 2249A ADC (10 bits.anode-cathode distance d. as shown in appendices Adelivers a signal proportional to the area of the inputstant. lts variation is strongly dependent on theand the amplifier-receiver on the CAMAC board. Itdeveloped in the avalanches. is not necessarily contransmission line between the preamplifier (TRA4U3)
The electric field at the wire surface. hence the gainand shaping circuit (NIM standard) is interposed in thegiven in appendix A.bility of about 0.1% per 00 ns gate. An analog filtertor and the applied voltages (JW and U3. from relations( = 3.5 >< (U" e) corresponds to a noise detection probamade. given the geometrical parameters of the detec
For the noise levels (= l0e` e). the chosen thresholdsurface at point (xn. y, :.,). These calculations are
tating the adjustment of the discriminator thresholds.MWPC electric field until it arrives at the anode wireshaped signals) is available per lb channel board faciliprogram follows the electron drift trajectory in the
A fast "OR" function (of the discriminated andFrom the generated conversion point (x. y, :). the
tions.photon mean free path. obtained from eq. (5).
readout and initialized via the usual CAMAC funcwhere I is the absorption length and l¤,,(E) is the
(D flip-Flop) using TTL logic. The registers are thencoincident signal is then loaded into the data register (9)P(I)=exp(—l/[ph)/[ph,the trailing edge of the monostable tlip·tl0pl. This
a detector signal (we require a time coincidence with absorbed. The absorption probability isa 60 ns write gate which defines the time acceptance of the photosensitive gas volume of the detector until it is
When the lamp flashes. the current pulse generates tal beam profile. Each photon is propagated throughadjustment of time constants. ances rrx =oj. = 0.15 mm. reproducing the experimenthe moriostable flip—tlop output is about 5 ns after window with a Gaussian position distribution with varispread in all channels. defined by the trailing edge of (energy E) enter the detector normal to the entrance
tributed via the P5 line of the CAMAC bus. The delav The Monte Carlo generated single UV photonsjustable (between l(l() and 500 ns} by a voltage disof the monostable flip-llops are simultaneously ad 5. I. Generation and collection of 1/ie phoroelectronsnately switched to reduce dead time effects. The delay
by two nonresetable monostable Hip-Hops (M) alterbe estimated precisely enough by calculation.Each shaped and discriminated pulse was delayedti "gain of the MWPC avalanches as this quantity cannotthreshold was set to = 3.5<0·qy.process. Nevertheless. it requires as input, the averagean average noise (rrd) = l()·‘ e. The discriminatormental knowledge concerning the various stages of thebution has a spread of about 0.3 fC (L9 >< l0} e) andelectron detection. using theoretical and/or experialent charge (NEC) at the input. The threshold distrilates all the processes from photon emission to photoficiencyl and whose rms width qi gives the noise equivresolution. etc.) can be assessed. The program simumean KQ? gives the diseriminator threshold (50% inet·ous geometries so that their performance (efficiency.old. The derivative of this curve is a gaussian whoseand subsequently to simulate photodetectors with variinput calibration signal. at lixed discriminator threshtor response was developed to interpret the test resultsciency (for several channels) is plotted versus a varizible
.—\ Monte Carlo program simulating the photodetecnrc shown in fig. 17. The observed detection inerti
thc CAMAC bus. Examples 0i ihc discrimination curvcs5. The Monte Carlo simulation programdistributed in purullcl io all the modules via lim: P4 of
477R. .·lrrmA·J ez ul. /’}mmn c/crecr<1r mr ( `/wmr:/<m· rrmz rmuuznu
:,:4\ ·»_ Q ·**—‘·
‘ -°,' x ··— ' ` `
the straight line slope determines the cathode-anode OCR Outputi(:)=.-4-exp(—1.»r.,), (12) 865 single anode cluster events. The least-square fit of
wire and cathode pad clusters is shown in fig. 18, for
delta function current input is The charge correlation measured between anode
The measured response of the preamplifier for a
6. I. Cut/t0de—anode coupling
5.3. Electronics response
lated events.
The same selection is. ofcourse. applied to the simuand then summed per pad.with production of a separate secondary photoelectron.primary and all the secondary avalanches of the chainfrom two incident photons or a single incident photonduced signals on the pads are thus generated for thetwo clusters were rejected because they arise eitherexperimentally distribution (see subsection 6.1). Incluster of hit wires were kept for analysis. Events withinduced charge onto one or more pads according to and=().4. 0.5 and 0.67 mm. Only events with a singleavalanche around the wire. The program allots theto this quantity. Tests were therefore performed foranode—cathode distance d and on the location of thesince. the anode—cathode coupling (C) is very sensitiveproportional to the anode signal. to depend on theespecially the anode wire-cathode pad distance ld)The induced pad signal has been measured to besecond purpose was to optimise the design parameters.mained (the process is quenched).Carlo simulations. in order to fine tune the program. Atinue to be generated until no further photons re
The data were continually compared with the Montechain) the charge signals are summed. Avalanches conpotential U._ = -1.4 kV.falls in the same pad cell (as previous ones ofthe sameotherwise stated. all tests were made with a windowavalanche charge q and it`s position. lf an avalancheof detector gain and photon position ( t. y). Unlesslates. as before. the electron drift trajectory. it`spad multiplicity. the position resolution as a function(1 ·- Q(E)). In the first case, the program then calcuof the detector response: the detection efficiency. the __(quantum) efficiency Q(E) or is lost with probability(UW, LQ). The second part concerns the determinationabsorbed in the gas. it creates a photoelectron withdependance of the detector gain on the voltage pairsorbed with probability (1—Rw). lf the photon isinduced charge on the cathode pad array and theflected with probability RW (wall reflectance) or abthe charge ratio C =qC/qa), the spatial spread of thelength l¤hlE)] or strike a wall whereupon. it is resimulation program: the anode—cathode coupling (orthe gas with probability given by eq. (9) [absorptiondetermination of the parameters of the Monte Carlosults. The remaining photons are either absorbed inTEA at 20°C. The first part of this section concerns theinclude this effect to reproduce the experimental rc
These tests were made with methane saturated withlabsorbedi with probability ll - R). lt was necessary to
photon is then reflected with probability R or lost6. The TEA test resultslar reflection coefficient Rl8) is calculated and the
volume. lf the photon encounters the wire. the specuprogram then propagates these photons into the gas
electron.and critical radius lsee appendix Bl. respectively. Thecalculated to define the position of the primary photoradius lp + rc)/2, where p and rc are thc wire radiusgrouped into clusters and the digital barycenter isemitted isotropically about the avalanche center at(per event). The adjacent hit wires and pads arebilities ol (13. 57, 30)%. respectively [lll]. They areare the hit wire and pad addresses and their charge[7]) wnth energies E = (6.43. 7.46. 7.94) eV with probatalk of 50 dB (0.3%). The end result of the simulationback photons ln =ag. g = gain =q/e. tz = 7.7 >< ll)`coupling between adjacent pads by simulating a crossof the detector. The program generates n carbon t`eed—
The program accounts for the measured capacitivedepending on the gaseous mixtures used and the gainfiltering is seen, not the true avalanche charge.The choice between these two distributions is madepreamplifier output only the charge after electroniczero the the ?’0lya approaches the Furry distribution.17). Since all the charge measurements are made at thewith the mean charge (q). Note that as 8 approachescompared with the known discriminator thresholds (fig.where b = l ·*- 9. and 8 is at parameter which variesthe simulated anode and cathode charges which are
P(q)=[b/F(b)][bq“]¤¤p<—b<i/<¤1>)/<<i>· lll) lation program. this factor is calculated and applied toavalanche charge at the output maximum. ln the simuor. more generally. thc Polya typcent [23]. This detector developed = 22% of the total
<10>F(q)=¤¤¤t·°1/viz}/<q» reduced by a factor which is detector and gain dependthe charge developed at the discriminator input isFurry type
which has its peak output at t = rj, = 10 ns. ln a MWPCof us diszribuuon. This distribution muv hc of thc
478 R. Arno/cl et ul. ,’ P/mmn zlezecmr Mr (,`/reren/wz rmx: zmawng
__~· ,- ·- —·.c·? .·· ;A?L*`..I1$*¤»·%¤..$¤‘ I . so =·i‘¢J.z
the average. OCR Outputhence. to obtain the best cathode efficiencv d must be(b) the anode-cathode distance d for two beam positions andThe large slope shows the strong d dependenceversus (a) the anode potential Un for three listed values ol tl.Fig. 19. The cathode-anode coupling constant C = (](. /7 qll(13)titt (C) =0.99—0.43d [mm].
squares fit gives the mean coupling (C> (averaged over
wire (eb = 0°> or between wires (tb = 90°). A linear least0.65fig. l9b. for the cases when the photon beam crosses a
The coupling C =qC/qa varies with d as shown ih0.7additional argument for a solid photocathode.
the wire and the cathode (90 < d>< 180°). providing ancr0.75
coupling reaches unity for electrons produced betweenbetter coupling to the cathode than the latter. The
arrives at ¢ = 9l° (fig. Zia). the former having much
arrives at tb = l14° (fig. 21b) whereas for d = 0.4 mm it0.85
for d = 0.67 mm the line of force from the cell edge
The electrostatic field map of the unit cell shows that0.9
0 facingawir:has been previously observed [24] and calculated [25].rk meanamplitude modulation. however, increases with d as 0.95
<`> bctweentwuvtiruand minimum when it crosses a wire (d>=0°). The
maximum when the beam is between wires (da = 90°)C = qt / qlits x dependence is shown in fig. 20. The coupling is
U_ ( kV )..·-` (fig. 19a). lt decreases with increasing d (fig. 19b) andelectrostatic configuration and very little on the gain 1.28 1.32 1.36 1.4 1.44 1.48 1.52
fig. 41).0.65 A é¢with the photon beam axis crossing a wire (d> = l]° in
analog electronics. These measurements were made0.7
and 29 fC. corresponding to the linearity interval ofthecoupling. The anode charges were selected between 4
0.75
0.8
0.85(¢ = 0:,).
cathode pad charge qc for the photon beam crossing u wire0.9 A d = 0.67 mmFig. 18. A scatter plot ofthe anode wire charge q__ versus me
O d = 053 mmq,& IC)0.95
o 4 s 11 16 zo :4 za D d = 0.40 mm (8)
i C : qc) q.4 +__ .·r?`:,Ir e
. ‘ :;;*5 ,
pI`Ogl’HlTl.;4,·~‘;;·’12 P
Eqs. (13) and (14) are included in the simulation< x < Ax/2).
16 r_~¥ T: the photon beam in the unit electrostatic cell ( — Ax/2
where AC=Cmx — Cmm/2, and x is the position ofzo r
(14)C=<C)—.\C cos(2·n·x/.\x).
sinusoidal functionas +
cell of width Ax there Ax = 1.27 mm!. is well fit with a
q(<fC> minimized. The variation of the coupling C. within a
479R. Arnold et ul. Phoum dcrecmr {br (4`/wrenkov rmx: zrnuymg
·,. .». .,.•.';*
tl = ll.67 mm. OCR Outputi :1.17 mm pitch. for L/‘ = l.~1 kV. and la} ti = 0.4 mm or tbl n = 0Af;8_ B = (mq [mm .t] Q3 (L3 lmFig. 21. The electrostatic field configuration in a unit cell with “pl_(,_ d),,,L,,,] i
ttl.} 3.1 0.3exp[—<r.t1li/2]xtmml Hmm)
0.4 0.53 0.67.o.s -0.4 u oa as an .0.4 it ii.4 its
.n.4 i .0.4 f`<r.dl dlm. m]
F/DOF for charge distributions ftr. dl vs anode distance al
Table 2
0.4 l [K Q- 0,1 r- t M l l lThe data and the fit curve are shown in fig. 22.l l l
zu l 5 0.02 and B = 0.99 j 0.03 mmii.; Etaneous fits (X‘/DOF = 0.3) with constants ex = 0.68
i.z the three measured d values we find very good simuli.; iexp{—(r/d)“*“"] where a and B are constants. ForI illll l
it V more general function of the form flr, dl =
(Xl/DOF = 0.3. DOF = degree of freedom). We use a: i
our data except at the greatest distance d = 0.67 mm
z E i.s iever. table 2 shows that this function does not describe
deduced from tests on the DELPHI TPC [26]. HowA distribution f(r, d)=exp{-(r/dl:/2] has been
umn! ntmml (I5)q,=lf(r,d)ds.
position. then:charge on the preceeding li — l)th pad is negligibly
unit cathode area ds at distance r from ofthe avalancheproaching the boundary between pads t and i 1- l. theIf ffr, d) is the distribution of charge density per
charge on next (i + l)th pad. Generally when apwithin a circle of about l mm radius.Here. q, is the charge on the ith pad and qw, is theshows that 95% of the induced charge is confinedy is shown in fig. 22 for a d =tl.4 mm. Un = I.4 kV.small. The intra-pad coordinate is at yn = 0. The dataof a pad. The variation of thc ratio q,/(q, + q, H) with
mm steps along a wire (direction y) near the .r centreat the pad edge). The solid line shows the fit to the data.function ofthe distance d. The beam was moved in 0.1
qw i) versus the light spot position y., along at wire ty., = U isdistribution of the induced cathode charge density as aFig. 22. The charge fraction seen on adjacent pad q,/(q, +
Measurements were made to determine the radial
0.2. Spatial distribution ofthe induced cathode charge y,t mmm
-2.s ·z .t.s -1 -0.5 0 0.s i 1.5 1 :.5
positions.
.md normal :0 the wire direction. The arrows show the wire
posnuon .r,,, for three vuluus ot 41. The Scan as along u pad uxns
Fig. 10. The cuthodc—zmudc cuuplmg C xcrsus hgh! spot0.1 L
) 4 S ¤ 7 8 Q IG il0.4
A A A A A A
06
0.6A-o A ~A¤·’2¢;¤$ AES LMA; Ali Mita Asif;. :4
2A A`Q 3$P¤“AA°=>*°f°¤’°°&_° A _A ¤,c°A2 O g’;__:,‘°¤O¤¤= ¤=_ ‘¤ or - L _;. A A ".~. 3m L> g
[mml [RV] UC] lYCl[YC]EC I It U)d LY tqny pu gnn; R 1/tyq) T/¢[q)
Parameters Monte CarloExperiment
multiplicity <.ri¥,; and event ratio R vs avalanche parameters td. U,)Comparison between experiment and Monte Carlo predictions for wire charge <q__ ,, anode charge mn). cathode efficiency ec. pad
Table 3
of q (at point xw) characterizing a Furry distributionlevel (fig. 23a) is due to a nonuniformity in d.pling. The slow decrease (:1%/mm) in the overall {gw) =g,,,a,( = (qw)/e. where (qw) is the mean value
tor gain at the point xw facing a wire (tb =0°) byphase with the variation in the cathode-anode cou
On this basis. it was decided to characterize the deteccide with wire positions (arrows) which are l80° out of
in two cases (d = 0.4 and 0.67 mm) is about the same.d =0.4 mm (see fig. 42). The maxima (fig. 23b) coinever, that the magnitude of the maximum charge seenaround the wire which is. as expected. not seen for
A comparison of the figs. 23a and 23b shows. howfor d = 0.67 mm reflects the variation of the gain
23a) and d = 0.67 mm (fig. 23b). The periodic variation
beam is moved in the x direction for d = 4).4 mm (fig.positions.d____ UJ = l.3 kV. measured between 4 and 29 fC, when thc
mm and tb) d=l).h7 mm. The arrows indicate the wirethe wire. Fig. 23 shows the mean charge (qi,) atposition x,, on a pad axis normal to the wires for fa) d = 41.4the gain varies with position of the avalanche aroundFig. 23. The mean anode charge (qa) versus the light spot
electronics and by feedback photoelectrons. Moreover.¤,(¤•)
(q> 29 fC) are distorted by the non-linearity of thes 4 s s v n o in ii
by the discriminator cutoff and the high gain events
tion. Here. however. the low gain events are rejected
measured wire charge distribution to the Polya func
detector would normally be determined by fitting the'11 l- l
(developed after electronic filtering). The gain of acharge at threshold and (qu) is the mean anode chargeea = expt —<q,,,>/<qa>), where (qu,} is the mean input
l at A * 1 i W idetection efficiency was determined from the relation lNIn these tests with TEA + methane. the anode wire 1s k
lU).
distribution tends to a simple Furry exponential (eq. I A a=o.¤·i mm 16 rciencv. At gains aq < 5.3 >< 10*. H tends to ll and thethreshold) and variance determine the detection etfi~ i1_~fC)
distribution whose position (relative to the electronicxatmmr
Hence. with increasing g a peak develops in thelc J 4 S a 7 5 9 10 ll
,-\ 6=(l
H=1.8><10"’g—1).095 forgz5.3><10 it L- 4 A { 4 A T * lcording to the relations: iz »
parameter H varies linearly with gain g= <_q)/e ac"#ianode charge q corrected for electronic filtering) the fq. ¤‘& T we #~Fi*l%¢· 5
C:H,, (25%) gas mixture [7], we established that lfor iw- T *i Wife new T ‘¤l=l’. ~ · W *~l··é¢» ?· ~ L ?+é% Q ¢¢?¢§i? ' '5i?q})>l?¢@¤$‘t`¢_¢f’¢l*bprevious work with a TMAE (0.9 Torr) + CHl (75%) ¤ T T le é»cr1bed by the Polya distribution P(8) of eq. (ll). In a IS xli
produced by 11 electron avalanche as generally de
The dibmbution of the anode or cathode charac Z Z d:0.40mm is l
q_tfC)0.3. Detector yam and tts pasmon dependence
AMR. .—lrn01`d ez ul. _, P/10:0n uerecror rOr Clwrenkar rum 1n1us;m2
Gain parameter lu. bl,) determined from data of table 3 and eq. (7)
Table 4
and Li, shown in table 4. which proves the coherence and validity of the simulafrom the data of table 3 we calculate the parameters u following paragraphs. Excellent agreement is foundwhere U,, is the value of Ul at the avalanche threshold. (after discrimination) and the simulation is given in the
A comparison between the measured digital data(W)(qw>=¢¤¤[¤(U`.·€L»)l·tion 7).
with U3 l.e. opposite to that reported for TMAE (8 ~ 1). (see secAssuming. as usual. that (gw) varies exponentially cant photon feedback (Ua > 1.3 kV). This conclusion is
shown in table 3. (6=0). especially in the voltage interval with signifiexcellent accord between experiment and simulation is that the data is best reproduced by a Furry distribution(1.15 and 1,3 kV) and d (0.4 and 0.67 mm). The the event ratio R with Un, shown in fig. 24b confirmsperformed independently for pairs of parameters (JJ reproduced by a Furry distribution. The variation oi.V(4 < q < 19 fC)/N(q > -1 l`C)]. This optimisation was ine charge distribution (for TEA + methane) is b·FC. the pad multiplicity (nn) and R [the event ratio and for the beam crossing a wire (rb =1l). shows that.;electronic filtering). the cathode detection efficiency (a. U.,) of table 4. This comparison. for d =0.4 mmjoint agreement between the mean charge (qu) (after U3 is calculated from eq. (17) with the fit parametersprogram which allowed (qw) to vary to obtain the best (Furry). 6 = l and 0 variable. The gain variation withgain was determined by the Monte Carlo simulation pared to the Polya distribution simulation for 9=0(eq. (9)). bcfnrc filtering by the electronic chain. This bution (qa) (after filtering) as function of Us is com
In fig. 24a. the measured mean anode charge distri
and (bl the event rauc R = A/(4 < q < 29 {C)/N(q > 4 (C).versus anode potential UA of la) the mean anode charge wa)
d = 0.67 mm, Un = 1.4 kV. The arrows show the wire position.Fig. 24. Companson bctwccn measured and simulated data.versus light position x,, normal to the wires (on a pad axis) foru_(nv) U_uV)
Fig. 25. Simulated efficiencies for a Furry charge distribution.
¤,tm···»2: x1C<P¤|ya.9=\> ` { xr MC¢Po4ya.9·1»
e ~1c4w¤¤y.,¤m.» l L ¤ ucu>¤¤y..am.» * “i“
I m;¤P¤•»¤,e:u» * I MCrP¤|yn.9=01 * O glcbatellielency :‘:¢_.:(g Q `Q Dan A __! g_; D Data g ! l
A Monte Carlo, d=0.67 mm ‘ Q- I A Home Carla. d=0.67 mm
A Data. d=0.67 mm A Dafa. d=0.67 mm
I Monte Carlo, d=0.40 mm I Monte Carlo. d=0.40 mm0.2 l 0.2 L.) l— AD Data. d=0.40 mm
l· l ` ‘ l l· j A? G Data. d:0.~l0 mm
(1.4 w r e0.4 L
U.6 0.6 l if Q A
0.80.8 LI U i . Q ttT L I C U , Q - i i 1
G ?*(al (b)
[CTS. efficiently detected by the threshold electronics. This
the fraction of cathode pad to single anode wire clus pads with consequently smaller signals which are less(triggers). The cathode efficiency ec, however. is simply is due to division of the induced charge onto adjacent
ured by normalizing to the number of lamp flashes variation. which is well reproduced by the simulation.The anode efficiency eu could therefore not be meas minimum (= 90%) at the y pad edge (fig. 28a). This
0.3% of the triggers gave an event (see section 4.2). ciency (= 97%) at the y pad center which drops to a
intensity of the light beam was reduced so that only wire. near the x pad axis) shows a high cathode effi
In order to count only single photoelectrons. the mm (fig. 27). A scan along a wire (v. eb = (1 facing aciency ec = *97% is measured and simulated for d = (1.4
fx. on a pad axis). a roughly constant cathode effi6.4. Measurement uf detection erficiencies
When the light beam is scanned normal to the wires
/14.2. Cathode efficiency us jtmctton ot posttton
d> = ll4°).
cb = (J°) and the anode—cathode coupling (maximum atFC 2 97%.difference between the gain variation (maximum atanode efficiency e_,;98% and a cathode efficiencyefficiency} is caused by the already mentioned phasetl = 0.4 mm and (L, 2 1.4 kV. the simulation gives anode efficiency (which is slightly lower than the anode
ments shown in fig. 23b. The modulation of the cath < 1.5 kV. Beyond 1.5 kV the detector is unstable. For
tion period and phase exactly reproduces the measure simulation with a flat plateau region between 1.35 < U,The variations with voltage are well reproduced by thewhen the UV light beam is scanned in x. The oscillabeam crossing ii wire (rb = ll. .r = (). center ofthe pad).mm. U; = l.4 kV) calculated by the simulation program
and d =(l.o7 mm. These data were obtained for theanode and cathode detection efficiencies (for d = 0.07
As illustration. fig. 25 shows he variation in the shown in fig. Zoa and 26b. respectively for d = (1.4 mm
metrical comigurutions. tirid photosensors. l at L', = l.5 kV) and cathode efficiencies versus U, areRICH detector. for diffcrcnt electrostatic and /or geo The measured and simulated anode (normalized to
tion program to predict thc pcrformancc of u fast r>.4.l. Detecttvn eyficiencv as flmcttvn of UU and tl
axis normal to the wires (1,, = 0 is the pad center) for d = 0.4sc versus the light spot posuion 1,,. The scan is along a pad versus Us for d = 1).4. 0.67 mm.Fig. 27. The measured and calculated cathode pad efficiency Fig- Z9- Th¤ 1T1¢¤Sl11’€d and €8l€¤|¤1¤d Dad m¤1(1¤H€1¢Y (#1,,}
“(mm) U_( kV)
0 0.4 0.s 1.2 1.0 2 2.4 0-9 1 1-1 1-1 13 1.4 1.5M
°”’ xt-Angniiig
O M0meCar|0
1250.8 }·· U Dana
4
cxs L
usO`, r
A M0nteCarlo,d=0.67 mm{g) $ §$· gg ?•é M k¢¢F* $ ¢ $+5 1 $T· Ollla
2-25 L A Data.d=0.67 mm
I Monte CarIo,d=0.40 mm
Ls r' D Data,d=0.40mmLOS L
us
<I\')
t,, = 2.5. yn = 3.8 mm. (a) An x., scan. (b) A yi, sean on a wire. OCR OutputFig. 30. Measured and calculated pad multiplicity win; versus the light position for il = (1.4 mm. Pad center .t,, = X , = U. pad edges
xol mm) Y,( mm)
0 0.4 0.s 1.2 1.0 : :.4 u0s11.s2zs1a.s·s
l •gB¤ggi;lgU¤Eg
1.2 l 1.2
1.4
bam¢• I ' • :55 ¤¢¤
I 9 Monte Carlo 1.s i L8 L B \1cnt¢ Carlo* U Data " D Data
lla) (bl
<lI,> <I\’>
with respect to the pad array and (x. y) is the meas resolutions Ax/1/12 = 1.47 mm and .xy/ 1/12 = 2.26lf (x,,, y,,) is the known position of the light spot values are clearly lower than the naively expected
kV) are (av) = 1.03 mm and (ar) = 1.43 mm. These6.6. Spatial resolutions rersus position of the light spot ages obtained with the above data (d = 0.4 mm. U3 = 1.4
Ay/2] (here Ax = 5.08 mm. Ay = 7.62 mm). The averunderstanding of the detector response. pad surface [—.\x/2 5x,, 3 Ax/2. —.\y/2 S Vit 2full accord with the simulation. thus confirming our 32b with a uniform distribution of the photons over thesimulation program. The y scan data (fig. 30b) are in are obtained by averaging the curves of figs. 3lb andin multiplicity (1.1 to lb) and is well reproduced by the The mean rms resolution (»r_ ,». og of a detectorwire near the pad edge which causes the observed rise 0.6.2. Standard deviation cr.. ir_
of the feedback photoelectrons are collected on thebetween a pad center and edge. At x = l.27 mm. half (fig. 32a).pad multiplicity for a beam scan along x direction tized structure is. ofcourse. not observed in the v scancharge. Fig. 30a shows the measured and simulated incremented (x,, = () is the pad center). Such a quanLQ = 1.5 kV. excessive photon feedback causes dis wire position which collects the photoeleetron as .c,, isby the production of secondary photoelectrons. Beyond The position of the maxima (fig. Bla) occurs near thewith the plateau region of fig. 26) is caused essentially tl = 0.4 mm. UX = L4 kV). The agreement is excellent.rise in the multiplicity above l.35 kV (which coincides scan. respectively and compared to the simulation (forU.67 mm. and compared with the simulation. The fast Figs. Bl and 32 show the data obtained in an x. 1shown in fig. Z9 as a function of U3 for d = (L4 mm and 6. 6.1. Hemi resolutions (x — xi,). ( y — py,)
The variation ofthe average pad multiplicity <_n¤,i is
deviations.
6.5. Pad multiplicity UX 2 Wx _X_‘]Jy.]1-/2_ (,_ = [1.f(t·—yu)*]~]‘’; nrc thc rmstion between the true and reconstructed oositions and
tuncc d. as shown in liz. 18b. thc quantities <.r — mz. U ·_\‘.i/ LHC thc m€11¤ devia
loss ns ucccmuatcd with increasing unodc-cathode dis- urcd position (buryccntrc of thc pad coordinates) thcn
QRSR. Arnold et ul` . l’/mwn aurccmr for (,`Izcrcn}<m· run: zmaumg
"L‘?·éi5*i
made it possible to develop and test a general Monte variations of the cg factors with pad dimensions .\x OCR Outputhowever these detector response measurements have with analogous relations for the y components. Thefixed for the B factor RICH project (see section 3)
The geometry of the test detector was as designed (18)
cgrd = (tru,)/( .\x/v'vl2 cgw = (am)/{ .\x/vl2TEA
6.7. Ultimate spatial resolution and pad efficiency with .\x/t/12 as:
fined. as before. relative to the naive pad resolution
cathode distance d. the center of gravity factors (cgrd and cg ra) are demm. again confirming the need for a small anode To facilitate comparison between the resolutionswith gain is shown in fig. 33b for d = 0.4 mm and 0.67 calculated.fig. 33a). The variation ofthe global efficiency eg = eaec were measured and an analog center of gravity were
(om), (aw} which could be obtained if the pad chargemm; to provide better equality for x and y errors (seelation program determined the pad size 5.08 X 6.604 aged over the detector surface) and resolutions
RICH detector project with a NaF radiator, the simu calculating the digital barycentre of the hit pads (aver
fected by the distance d. For the B meson factory Table 5 shows resolutions (ord}, (tr",) obtained byAs expected, the resolutions (ax). (cr`.) are little af i.2s X lO°.affected because it is dominated by the wire spacing. mm. d= 0.4 mm and the mean detector gain <gw>
in fig. 34. In this simulation. the wire pitch is s = 1.27lution (ag.} (fig. 33a) but. the resolution (ax) is hardlyondary photoelectrons substantially improves the reso without secondary photoelectron production is shown
detector gain rises the increased production of sec versus pad sizes (from 35.6 to 1.27 mm;). with anddue to the digital center of gravity calculation. As the The variation of the pad cluster multiplicity (np)
static configuration.cgvd = (a\,)/(Ay/ v`l2 ) = 0.63 give the improvementCarlo simulation program applicable to any electromm). The ratios gid = <'a‘.)/(.\.x/ yl] ) = 0.7 and
edge) for d = 0.4 mm. (al (x - xo) and (b) ug = <(x — x,,)):l/2Fig. 31. Measured and calculated distributions versus the true light position x,, normal to thc wires (xo = 2.5 mm is pad center.
¤'(rnm)x'(mm)
u n.4 o.s 1.z 1.s z 4 1.40 0.4 0.8 L2 1.6 2 2.4.¤.z
5I M0nl¢Carl¤D Dau
ax
as02 ‘ V ca
#9*Q » 1 OF ’ W o.vs b ,¤g¢ 0 at '*! ‘P Q 0 g' 0.4 #- H, {; * 5
across the spacer. .—\t the minimum. the charge loss is 206 eV/mm. Taking as the effective energy of ionizaOCR Outputfig. 36 as the light beam was scanned along a wire and mixture of methane (93.4%%-TEA (6.6%) is about
The minimum ionization energy loss in a l barvariation of the anode mean charge (qu) is shown innear a 0.53 mm diameter polyamide wire spacer. The ured.
Tests were made to determine the efficiency loss NWC and pad clusters npc per event were also measgain g. per cluster np. In addition. the number of wire clustersforces. This method insures good uniformity in d and the number of wire per cluster nw, the number of padsing spacers. to counteract the effect of electrostatic charged particles we used cosmic rays and measured
only if the wire is supported every = 10 cm by insulat In order to determine the detector response to
rn) with small anode—cathode spacing d is possible
Stringing of anode wires over great lengths (= ll.9 6.9. Charged particle signal, multiplicity and feedback
6.8. The spacers between anode wires and pads factory detector" now being assembled.
will use ceramic spacers for the full size "5% of B
glass spacers. For mechanical simplicity. however. we(er,) = Ax/JE for small Ax.cr, is limited by wire spacing (i.e. cgra = cgxd = l) hence a surface charge. This behaviour was not observed withphoton feedback (cgva = 0.33). In the other direction. were observed which show that this material developed
tests proved unsuitable. Operational instabilities in timemeasurement is analog and without (or suppressed)reached with pads measuring Ay =l mm unless the In practice, the polyamide wire spacer used for thebe noted that a resolution a·v=0.l mm cannot be 100 mm.
whereas for larger dimensions it is worse. It may also ciency loss (0.6%) for this detector with spacers everyof the analog barycenter improves the y resolution region is 0.6 mm hence. this represents negligible effi
threshold (see fig. 36b). The FWHM of the affectedthat for pads l.27 >< 1.905 mm* or less, measurementphotons. arc shown in fig. 35. It is interesting to note ttl this point is. in fact. zero since the charge is below
influenced by the beam size. The detection efficiencyand Ay. with or without the production of secondary
edge) for d =0.4 mm. (al (y — yu) and (b) ug = <(y — y(,)*) l/2Fig. 32. Measured and calculated distributions versus the true light spot position vu ulong a wire (yl, = 0. 3.8 mm is pad center.
y.,<mm1y,( mm h
0 0.5 I 1.5 Z 2.5 J 3.5 40 05 I 15 2 2.5 J 1.5 4-2.8
-L4 0.4
L •O M¤n1¢Carl¤ ' E [ W
l [5 W D Data “__ p 9 U
31[ •
·l.6 i P i
-1.2 |
Z O Q-0.8 E9 aQ :1-0.4 2.4
}· I M0meCarI0 la *
D Data
2.8
(b)L(a)
<y·y°>{mm\ c,(rnmr
437R. Arnold er ul. Photon deteczvr mr (.`/xercnkur rms: rmuurnu
Calculated average resolutions (ag), (er,) from digital ld) or analog (a) barycenters vs pad size Ax·Ay
Table 5
the 3 mm gas gap of the MWPC. the parallax sidewise= 80 total electrons (primary + secondary). This chargevertical axis was 9. = 18.4° (i.e. tan 6.. = é) hence. inionizing particule will create = 21 primary electrons or13). Therefore. the maximum track angle from theLion W= 30 eV/e then. in 21 3 mm gap. :1 minimumsecond was almost touching the readout cards (see fig.
120 mm above the test detector readout cards and the(b) the global dczeczion efficiency (eg = elec). tered on the test detector axis. The first scintillator was
0.67 mm of (a) the averaged rms rcscluiions (0,), (av) and of two scintillation counters 40 mm in diameter cenFig. 33. The simulation predictions versus U. for d = 0.4 and The charged particle trigger telescope was made up
U_( kV} tal determination of the multiplicities.1.18 1.32 1.36 1.4 1.44 1.48 152 charge effects hence. we rely on the direct experimen
0.6charge. It is difficult to convincingly simulate spaceelectron) to produce a large signal limited by spaceI d = 0.67 mm0.65 Eib)
I d : 0.40 mm will be amplified by the MWPC (.2 = l0° for a singl•
mm.
0.75 feedback and for gain Qgwk =1.25><1()". d = 0.4 mm. s =1.27for various pad size Ax and Ay with and without photon
0.8 Fig. 34. Simulation predictions of mean pad multiplicity (np)
Aytmml
0.85
1214567891o
0.95 • n j I 1 il 11J_( kV 1
1.:a 1.12 1.36 1.4 1.44 1.48 1.51
41<al- A ` c_ , d = 0.40 mm . Ay=6.604 mm
0.23cx, . d : 0.67 mm
c` . d = 0.67 mm
0, . d = 0.40 mm
¤_ . d : 0.40 mm6}0.75
Q U I S nP O
f‘ Z I O U O81** O without feedback photons
O Z
3 with feedback photons9}
1.75
< cl > <n’>.<¤_> umm)
R. Arnold er ul. Phomn uurecmr rbr (.`/wren/<0L· nm: zmuzmgM8
_;._,!.:*»¤—. .-.—1·£·»i gk."F,.*. EF:.··j_-.4;;
cluster however. this is somewhat misleading as each degraded. OCR Output
figures indicate = 2 wires and = 5 pads are hit per identifiable hence. the ring image resolution will not begle clusters occur on wires and pads. The first two particle hits and the single photon hits should be easily(awe = 0.60). The last figures indicate that mostly sin and its near neighbors). As consequence. the charged{ann = 7.21). nw = 1.33 (awa = 0.42) and npc = 1.27 limited number of pads per cluster (i.e. the central pad
highly localized (i.e. mostly a single cluster) with adistributions with nw = 2.27 (am, = 2.75), np = 4.96The observed multiplicities are fitted by gaussian of the charged particle passage through the detector is
trigger muons were fast. with np = 4,8 ; 2.9. These values show that the effecttween the two scintillation counters to insure that the hit, the remaining 80% have a Gaussian distributions = 1.27 mm. Lead absorber (50 mm) was placed be 1.6. For the pads 20% of the clusters have a single pad
7.62 mm but of the same order as the wire spacing ing 52% have a gaussian distribution with nw = 3.6 1mcm is small relative to the pad size Ax = 5.08. Ay 48% of the clusters have a single wire hit. the remain
distribution has two peaks. In the case of the wiresdisplacement of the track is 5 1 mm. This displace
for d = 0.4 mm and(gw>=1.25 ><10°Fig. 35. Simulation predictions ofthe digital (al. (ln) and analoglcl, (d)cg factors lcq. <11·l)|versus pad sizc Ax (zi). tc) or .\y lb). ld)
Axlmm) Ay(mm)
2 4 6 01 0 2 -1 6 80.4 " 0
0.5 %tc) 0.1 P (d)
n.c by 0.4 b A. L
0.s F—°0.7 P r
0.6 fg- O n
0.7
ns E 0.80.8
1.1 j 0.9
I.2 ‘cg!.°B"
.\x(mm» Ay(mm)
2 J 6 Ol 0 Z 4 6 80.J “ 0
0.5 " 0.3 L ""
0.4fl.6 *"
0.7E 2 0.5
0.8 *0.7 0.6
0.9 P
0.8 A E O1,1 ’" 0.9 E
Cgvnfla
A wimhout feedback photons
O winh feedback photons
489R. Arnold ei ul. Piwmn uerector mr L'/zcrenkor rmx: zmuluzng
.··.,—4g€g f*_ •~'I€s ¤‘r fr
mean charge (qa} and (b) the counting rate. OCR Output
Fig. 56. A 1*,, scan along a wire and across a 0.53 mm diameter polyamide spacer (at You = 5.5 mm) with measurement of (a) the
¥,,( mm) y°( mm I
0 2 4 s s tu 0 2 4 0 s 10
200
400
600 um
600
000
O QD.
¤OoOo¤m ¤00°o 1000 QQ; §> ? f W @0 .+0**0**(al (bl
1200
<<i.>(tc> Counts
expected value of = 100 fC. As the discriminatormonoatomic layers of TMAE [3.7], indicating that theorder of magnitude smaller than seen with adsorbed after electronic filter. is (qa) =h fC rather than theabsorption (A = (l) the value of QPE = l.3% is an 1.35 kV (see fig. 42). The measured cathode charge.
for the actual operating conditions UW = — 1.7 kV. Unnumber of photons per flash (np,. Q1 = 0.6. At zerothan that of an avalanche facing the window (o = 0°)surface. A linear least squares fit gives the average
ode (tb = l80°) is calculated to be about 3 times smallerthe photoelectric quantum efficiency of the cathodeavalanche localized on an anode wire facing the cathsity maximum of the lamp emission spectrum) and QPEare not efficiently detected. The average gain in anthis case QP. = QTMAE = 0.4 at A = l(>() nm. the intenphotoelectrons created at the photocathode surfaceflash. Qpl is the photoionization quantum efficiency (in
where (nun) is the average number of UV photons per
(19)<”w>=(”pn>[AQr·i+(l‘A)Qi=ei·50°C.
methane. The A ¢ () points are for TMAE at T = 20. 30. 40.the gas or on the cathode surface. are detected i.e.;versus UV light absorptivity A in a 3 mm gap of TMAE+A is expected if all photoelectrons. created either inFig. 37. The measured mean wire hits —in__,· per lamp flash[nh is given by eq. (3). The linear variation of(11__> with
at = r»l)°C. The absorption A = l — expf —//[ph). where0 0.1 0.1 0.5 0.4 0.s4ll and 50°C respectively. with the detector maintained
correspond to a TMAE bubbler temperature of Z0. 30.
TMAE ~ methane mixture. The experimental points 0.02 Ethe photon beam absorbed in a I = 3 mm thickness of a
l A0.04 L 'par lamp Hash (nw) as a function ofthe fraction A ofFig. 37 shows the average number of wire clusters 0.06 5
lines in photon feedback.0.0s i;
thev exhibit the effect of the various carbon excitation
section 4.1 with d =().55 mm. are important because 0.1 i;
measurements. made with the detector described in
u.1z i;methane or methane — isobutane as carrier gas). These
compared t0 TMAE at various partial pressures (with0.14 l;
Tests with TEA (amd methane as carrier gas) were
0,16 ij
7. Comparative tests with TMAE and TEA
490 R. Arnold et al. p Photon dczeczorfor (,`herenkor rms: rmwzmg
_? §;,f’w·. '. ·; . e.· j»;—»¤ 5..·»&.¤.aI
.*’.—~- - ·A `
TEA-1»CH, or TMAE + CH, + iso—C,H 111 lilled detector. OCR Outputme Fig. 39. Variation of (a) the measured mean number of hit pads (11,; and (hl the simulated mean wire charge (th} versus U. for
U_(kV> U_(kVI
1 1.1 1.1 1.1 1.4 1.s 1 1.1 1.2 1.3 1.4 1.sos
2 (blla)
10 » Aiil- Zogg1 L- A‘;£-/
1.s — s 1
0 A
zs »
.1 in · 11.s l /9 A1
4 L [A 1) l
4.s F
0 CHI 1 rm sr :0“c
1 C11, . Tania at s0"cL o C11_ . run; 2150°C
A c11_ t as¤.c_11,° + TMA1: at s0“C1 A cir. .a$¤.c,1~1,,, ~ TMA! al s0“Clm
thes.s I0 ’
gc.
< np > < q, >1fC)
11°)
ilu potential Un for a (TEA Of 'I`MAE)+ CH, filled detector.Fig. 38. Variation of (a) the measured mean number ol hit pads (n > and (b) the simulated mean wire char e versus anode p 8 (Q1.,
. 0 cu, . rm a1Z(l c _ , 1 : 0 0 cu, . rm si 20 C M15 1. e CHI . mas a15U`C c11_ . rmx; at s0“c * ·
A cn. . 111111; at wc .; cul 1 ruse a140°C
I cul . rs1A12¤1s0“clj CI CH‘ . rxuzar s0“ccit. . 111111; ut :0"c0 t;11_ . rx1,11; at 20°C
<np> r14;··l'C1
:3;HT
at 50°C+CH, + iso-C,H,,,. (c) TEA at 2l)°C+-CH,. OCR Output
Fig. 40. Measured histograms of pad hits obtained at the same gain for the detctor filled with (a) TMAE at 50°C + CH,. (b) TMAE
g-.r]· e/;·".` i/ll l»<.»>*€>Fl. V l Z "?t}5€{*~;<€"'i t. ;.. ;§Z.,\.3\_ i irq-l l fill ;l;P¢ic`i.»-tal * `¢rpY’l`i»‘?l€lil<i'éZol i ` fi-T.;iii.-€=··*‘·"i/* ’;,‘ l --.-l l *r‘.`.<Jl l *¥ `tl A -’—/E 5l··»‘-lf "l ` I/` -A j-.ire. \·‘l il i l sll lY l A tl i l lu-<>l. i-·l<l`—l
I l l fil,:_ ~ l l i ~ l-] l l li X.] i ( ..>’~. i. `»/tit
l {li
l2`
lb)isi l [iw,-I l l lc)
re ,
ficiently fast for operation at LHC or SSC.and its performance approaches that of TEA. Thereadout electronics which has been developed are suftector is greatly improved as shown in figs. 39a and 39bGeV/c. The intrinsic speed of the detector and theusing a methane T isobutane mixture). the TMAE dehadron colliders with rr/K identification up to 200the carbon excitation line at 7.48 eV is eliminated (bywith this detector (with smaller pads) could be used inof operating with TMAE in pure methane. If. however.identifying rr/K up to 4 GeV/c. A gaseous radiatorTEA. The comparisons of fig. 38a shows the difficultybased on this detector and a LiF radiator is capable tothe highest TMAE concentration does not occur with
An e°e` B factory particle identification counter(fig. 38a). The gain saturation observed in fig. 38b for
less chromatic radiators.reproduce the variation in multiplicity (np) versus Usbility, ease of operation and for RICH applications.a function of Us is simulated by the program. tuned toshould allow considerable improvement in speed. stastrated in fig. 38b where the mean wire charge {qw) asvacuum deposited onto the pads (currently being tested)secondary photoelectron production. This is demonthat the use of solid CsI or Csl/TMAE photocathodesvariation in the mean gain which is sensitive to thephotosensitive organic substance for fast detectors butTMAE concentration increases. This shift reflects the
It was shown that TEA is at present the best gaseousrise in multiplicity moves to lower values of UU as thedigital center of gravity estimator of position.toiomze TMAE but. not TEA. The onset of the fastphotoelectron production which allows an improvedtion lines of carbon at 0.42 and 7.48 eV which phoexpected from the pad pixel size due to secondaryTMAE concentrations. is caused bv thc atomic excita
struction of the photon impact point is better than thatdifference between TEA and TMAE. especially at highoptimising the anode—cathode coupling. The reconwith pure methane as carrier and amplifying gas. Theduce a uniform electric field around the anode wire byvarious temperatures) compared to TEA (at 20°C) both
·"‘pads. It was shown that this condition suffices to pro{np) as a function of the voltage Us for TMAE (atanode wire plane is at a distance of 0.4 mm from theFig. 38a shows the average cathode pad multiplicityto obtain a pad detection efficiency 2 97% if theing the threshold.photosensitive organic vapour show that it is possibleincreasing the maximum gain of the detector or reducdigital pad readout using triethylamine (TEA) as thethe anode wire surface (see appendix B). rather than
The tests made with a MWPC type detector withmm. to produce 11 more uniform electric field around
photoelectrons more efficiently by reducing d to 0.4
8. Conclusions and outlookode (discussed in section 2) could detect these cathode
A detector with a retlective Csl/TMAE ph0t0cath—
with the observauons.
compared in fig. 40.ciency ns expt—q_uh, <q_,>)=U.82 in mugh agreementmeasured pad distributions for TMAE and TEA arethreshold is quh = 1.2 fC hcncc. the detection ineffi
492 R. Arno/d er al. / Photon dezecmr rar Cherenkvv rmq zmugang
_` ·~•§——,,..w’fl¥I-.,. 4;_~___;,‘·;y¤ V.
»a<:.;.;'‘·:
values of tl. OCR Outputdistance c. anode-cathode distance tl and wire spacing is s..\E {maximum gain cb =ll° at the top of the wire! lor the listedpoteltials are U,__, Li, (jk. Wire radius 0. window-anode
incidence rb of the ayalanching electron around the wireFig. 41. The MPWC pad detctor. Window. anode and cathode
xlmml Fig. 42. The gain of an asymmetric MPWC versus the angle of
i ii i z J <1>(dtgr¤tsl.(]_.| L
0 :0 -10 so so 100 110 140 iso 180
»U0-· e ai1 llrr/2. —¢ ' ’ B "·
M - 5
i.: ir
I6 r Cd = 0.67 mm
i (MY.)
1..1 ·— L
1.x E
ZUIIM}6 L
d:0.40 mm
xg + :; = p"
determined for the initial position lx. y. s) by setting
electron on the wire are calculated from the value of ryThe coordinates of the arrival point (xg, y. ;;_) of an
then As the critical field EC is independent of tb hence.Uk)/d. and m =as/·i·rc, rz = as/·rrd. a = lnls/rrp)
g(d>)=exp{2pv//wt/E(tb)(\/n(<b) - (B.5)with the field definitions EW =(Uw — Ul)/c. Ek =(Uu
hence. eq. (B.l) becomess s
5 sin:——·- sinh: (BA)r.<<1>)=~l<!>)1¤·5 X
M
_ sinh ;cosh Z .. C (B.3)E(<b.r)=E(<.b)¥.
(AJ )E` = ll.atc vicinity of the wire it is assumed that:
v orand k a constant depending on the gas. ln the immedi
5 sins; — sinh:with N=the number of molecules per unit volume
ii~ __C sin —-cos ··* t (B.2)a(<b.r)=y'kNE(d>.r).
wx wx
mean free ionization path length.the conventional electrostatic relations are:where rC(d>) is the critical radius and l/r1(d>. rl is theThe components of the electric field derived from
(B.1)ries Cg(d>)=exp cx(d>,r)dr[ 0r(d>)
The asymmetric MWPC electric field and drifr zrajecxo
wire (because of the asymmetrical chamber geometry}Appendix A
with the angle of incidence dr of the electron on theFor the given potentials LL,. and U], the gain varies
CERN—LAA. for which we express our appreciation.
The electric Held around an anode wireThis work was supported by [N2?} of France und by
Appendix BAcknowledgement
493R. Arno/:1 er ul. /’/wron dytecwr mr ( lxcrvrzkm rmu zmuxzrng
· W`:}
15/06/88. note 83-62 Elec.[9] CERES/NA45 Experiment at CERN/SPSC 8825. [26] A. Hrisoho and K. Truong. TPC linear chain-DELPHI[8] G.R. Carruthers. Appl. Opt. 14 (1975) 1667. [25] G.A. Erskine. Nucl. Instr. and Meth. 198 (1982) 325.
289. Nucl. Sci. NS-25 (1978) 794.and T. Ypsilantis. Nucl. Instr. and Meth. A27O (1988) [24] J. Fischer. H. Okuno and A.H. Valenta. IEEE Trans.Y. Giomataris. J.L. Guyonnet. G. Passardi. J. Séguinot Nucl. Instr. and Meth. 200 (1982) 219.lon. H..I. Besch. M. Boosteels. E. Christophel. M. Dracos, Séguinot. J. Tocqueville. M. Urban and T. Ypsilantis.See also the accompanying paper by R. Arnold. P. Bail See also E. Barrelet. T. Ekelhof. B. Lund-Jensen. J.(1988) 255. [23] See F. Sauli. CERN 77-09 03/05/77.queville and T. Ypsilantis. Nucl. Instr. and Meth. A27(] avec un systeme MAC-VEE. Rapport interne CRN.G. Passardi. P. Petroff. .1. Séguinot. D. Toet. J. Toe [22] P. Guterl. Logiciel d`acquisition de données CAMACChristophe!. M. Dracos. Y. Giomataris. J.L. Guyonnet. Elec.
[7] R. Arnold. P. Baillon. HJ. Besch. M. Boosteels. E. [21] E. Christophel and Y. Kornelis. DELPHI note 83-118ogy, Stockholm (1981). (1992) 207. L z/l ML
[6) S. Ekelin. Internal Report. Royal Institute of Techno1— Egger and K. Gabathuler. Nucl. Instr. and Meth. A 306son. Nucl. Instr. and Meth. A261 (19871446. Ypsilantis. R. Arnold. E. Christophel. J.L. Guyonnet. J.
15] R.A. Holroyd. J.M. Preses. C.L. Woody and RA. Iohn— [20] M. French. M. Lovell. E. Chesi. A. Racz. J. Séguinot. IEditions de Physique} pp. 249-315. T. Ypsilantis. Nucl. Instr. and Meth. A306(199I) 207. »—Physique Nucleuire et en Physique des Particulcs (Les [19] M. Hempstead. J. Chauveau. G. Comby. J. Séguinot and
[4} J. Seguinot. CERN-EP/89-92. Instrumentation en as a preprint CERN-LAA/PI/91 /04.A297 (19901 133. published in Nucl. Instr. and Meth. A. Part I is availableTischauscr and T. Ypsilantis. Nucl. Instr. and Meth. Ring Imaging Cherenkov Counters (in preparation) to be
[3] J. Séguinot. G. Charpak. Y, Giomataris. V. Peskov. J. [18] J. Séguinot and T. Ypsilantis. An ICFA review article on(1977) 377. see also B. Cox et al., SSC-EOI0013.
[2] J. Ségumot and T. Ypsilamis. Nucl. Instr. and Meth. 142 Fidecaro. p. 288:1.G. Morfin (Fermnabl p. 133. and Y. Zolnierowski. p. 260: F. Grancagnolo. p. 277; G.liders. Fermilab. April 5-7, 1989. eds. T.], Gourlay and 90-10. Vol. 2. eds. G. Jarlskog and D. Rein: Y. LemoigneParticle Identification at High·Luminosity Hadron Col See also the Aachen LHC Worshop Proceedings. CERN
[ll T. Ypsilantis. CERN-EP/89-150. and Proc. Symp. on B. Cox et ul., SSC-EOI0013.
CERN/DRDC/91-32.References Y. Lemoigne. S. Loucatos and Ph. Rebougeard.
Perroud. M.T. Tran. R. Chipaux. J. Derré. C. Kockowski.G. Charpak. Y. Giomataris. C. Joseph. C. Morel. J.-P.
mm. with U, = l.35 kV. UQV = -1.40 kV. Instr. and Meth. A306 (19911439.variation as function of tb for d=O.4 mm and U.67 [17] G. Charpak. Y. Giomataris and L. Lederman. Nucl.distance rl and potentials. Fig. 42 illustrates the gain proposal at Fermilab dated 08/10/90.
dated 07/Il)/88 as well as N. Lockyer et al.. p.BCDcalculate thc critical radius for unv anode—cathode
Collider Detector lor the FERMILAB TEVATRONUsing the property that EC is constant. it is possible toSee also The BCD proposal N. Lockyer et al.. A Bottom
n=r1<d>=l)°)= l.l55 (B.8)P. Schlein et al., FNAL-P84507/01/91.
for UJ = 1.35 kV. (fw = -1.40 kV I6/Ol/89:(qw) for d=G.67 mm (see fig. 23bl. We ·us obtain [16] P. Schlein et al.. CERN·SPSC/88-33. SPSC/P238.
See also SLAC-372. February 1991.be determined from the amplitude of tha- variation[15] CERN /90-02. PSI PR-90-08 (30/03/90).dcpcnds upon only one single parameter n which may
Villigen. Switzerland.Thus. the variation in thc gain around a wire thus[14] PR-88-09 (July 1988). Paul Scherrer Institute CH-5234
A309(l99I) 190.·—·—··—····:··. (B.7) Emax VH — l[13] S. Kwan and D.F. Anderson. Nucl. Instr. and Meth.
and Meth. A3U2 (1992) 447.1 `Bw) V Emr E(¢) [12] B. Hoeneisen. D.F. Anderson and 5. Kwan. Nucl. Instr.
[11] R. Arnold et al.. Internal report CRN/HE 91-06.in,2(<b=<i°) In umu
4734.
[10] H.D. Morgan and J.E. Mentall. J. Chem. Phys. 60 (1974)ln H( é) ln 2( cb)
494 R. Arnold et af. Ph0!0n detector Mr Clzeren/<v4· nm; tmuuzm: