Single-particle detection of products from atomic and molecular reactions in a cryogenic ion storage ring

2.6 MB·

by  C. Krantz

Single-particle detection of products from atomic and molecular reactions
in a cryogenic ion storage ring
C.Krantza,∗,O.Novotnýa,∗∗,A.Beckera,S.Georgea,M.Griesera,R.vonHahna,C.Meyera,S.Schippersb,K.Sprucka,c,
S.Vogela,andA.Wolfa
aMax-Planck-InstitutfürKernphysik,Saupfercheckweg1,69117Heidelberg,Germany
bI.PhyskalischesInstitut,Abt.Atom-undMolekülphysik,Justus-Liebig-UniversitätGießen,Heinrich-Buff-Ring16,35390,Gießen,Germany
cInstitutfürAtom-undMolekülphysik,Justus-Liebig-UniversitätGießen,LeihgesternerWeg217,35392Gießen,Germany
7
Abstract
1
0Wehaveusedasingle-particledetectorsystem,basedonsecondaryelectronemission,forcountinglow-energetic(∼keV/u)massive
2
productsoriginating fromatomicand molecularion reactionsinthe electrostaticCryogenicStorage Ring(CSR).The detectoris

nmovable within the cryogenic vacuum chamber of CSR, and was used to measure production rates of a variety of charged and
aneutraldaughterparticles. Inoperationatatemperatureof∼6K,thedetectorischaracterisedbyahighdynamicrange,combining
J
alowdarkeventratewithgoodhigh-rateparticlecountingcapability.On-linemeasurementofthepulseheightdistributionsproved
7
tobeanimportantmonitorofthedetectorresponseatlowtemperature. Statisticalpulse-heightanalysisallowstoinfertheparticle
2
detectionefficiencyofthedetector,whichhasbeenfoundtobeclosetounityalsoincryogenicoperationat6K.

]
tKeywords: Storagering,Lowtemperature,Single-iondetection,Secondaryelectrons
e
d
-
s1. Introduction For years, medium-energy magnetic ion synchrotrons have
n
been used very successfully for these kinds of experiments—
i
s. Single-particlecountingdetectorsareimportantinstruments a remarkable development considering that the technology of
in many atomic and molecular physics experiments on fast-
c thosemachineswasoriginallyaimedatnuclearphysicsapplic-
ipropagating ion beams [1, 2]. In such experiments, an ion
s ations [5–7]. Based on that success, a new class of heavy-ion
ybeam is guided through a target medium which can consist,
storageringshasemerged, withdesignsthatareoptimisedfor
he.g., of photons, electrons, neutral atoms, or molecules. Re-
experimentsonatomicandmolecularphysics. Theyusepurely
pactions of the projectile ions with the target particles typically
[ electrostaticionoptics,matchingtheoutputenergyofrelatively
leadtoproductsofdifferentcharge-to-massratio. Thisresults simpleelectrostaticinjectorsthatcanbeflexiblyequippedwith
1in the formation of daughter beams of different ion-optical ri-
state-of-the art molecular ion sources [8–10]. The most ad-
vgiditycomparedtotheparent,whichcanbeseparatedfromthe
vanced set-ups use cryogenic cooling machines to reduce the
2
latterbyelectricormagneticanalysingfields.Atknownintens-
4 temperatureoftheirbeamguidingvacuumvesselsdowntoval-
ityoftheparentbeamandthicknessofthetarget,detectionof
0 ues near that of liquid helium [11–15]. On the one hand, this
8the daughter particles reveals the rate coefficients of the pro- resultsinavastlyimprovedresidualgaspressurecomparedto
0cesses involved in their production. Due to the typically low
conventional ultra-high vacuum (UHV) set-ups, with corres-
.ionnumbersandreactioncross-sections,theproductdetection
1 pondinglylongerionstoragetimes[16,17]. Ontheotherhand,
0needstobedoneonthesingle-particlelevel.
storageinsuchacoldenvironmentallowsinfra-red-activemo-
7 Heavy-ionstorageringsenhancesuchtargetexperimentsby
lecularionstode-excitetotheirlowestrovibrationallevelsprior
1theirabilitytostoretheprojectilesforextendedperiodsoftime.
: tostartingexperiments—asignificantimprovementoverroom-
vDuetoenergeticprocessesintheionsource,unknown,highly-
temperatureion-storagefacilities[18,19].
iexcitedquantumstatesareoftenpopulatedinatomicormolecu-
X
The advantages of these cryogenic ion storage rings come
lar ions directly after production. In may cases storage of the
r with technological challenges with respect to the particle de-
aions enables them to reach a well-understood state population
tector equipment. A restriction regarding possible detection
byspontaneousdecaybeforeundergoingtheactualexperiment.
principles arises from the low energy of the product particles.
The extended storage time also allows phase-space manipula-
Limited by available high-voltage technology, typical kinetic
tion of the ion beam, such as electron or stochastic cooling,
energies in electrostatic storage devices are of order a few
or initial-state preparation techniques as required for laser- or
keV/u or below. This rules out detection mechanisms where
collision-drivenpump-probeexperiments[3,4].
the counting volume of the detector is covered by significant
layers of passive material—as is the case for surface-barrier
∗[email protected] — present address: Marburg Ion-
semi-conductor counters [20] and, to lesser extent, for scintil-
BeamTherapyCentre,35043Marburg,Germany
∗∗[email protected] lators[21]. Openmicro-calorimetricdetectorsareapromising
PreprintsubmittedtoNucl.Instr.MethodsPhys.Res.,Sect.A 30thJanuary2017
option for product detection at cryogenic storage rings, which
1
ispresentlyunderinvestigation[22–24]. Theirfabricationand 4
operation are however extremely difficult and expensive, such 7
6
5
thattheirusemaybelimitedtoselectedexperimentsinthefore- 2 3
seeablefuture.
Suitable detectors for cryogenic storage rings, which can
be widely deployed at acceptable manufacturing and operat- 13
ing costs, are therefore based on surface secondary-electron
emission with subsequent multiplication [13, 15]. This detec- 12
9
tiontechniquehasprovenitselfalsoatparticleenergiesbelow
1keV/u[25],butthelow-temperatureenvironmentdoescome
with new challenges. Besides engineering problems related 11
to thermal expansion and embrittlement of materials, the effi-
8
ciencyofchargemultiplicationstagescommonlyusedinlow-
energyiondetectionisknowntosufferincoldoperation. Due
10
to their semi-conductor-like properties, the electric resistance
of micro-channel plates (MCPs) and single-channel electron
multipliers (CEMs) rises strongly upon cooling into the cryo-
genicregime. Thehighresistancecanleadtodecreasedgainor Figure1:Schematicviewoftheexperimentalset-up,consistingofthestorage
evencompletechargedepletion, especiallyatelevatedparticle ringCSR[15](top)andtheCOMPACTdetector[30](bottom). Thelatticeof
CSRisafour-foldsymmetric,approximatelyquadratic,andpurelyelectrostatic
hitrates. Dependingontheapplication,MCPshavebeenused
beam-line,consistingofatotalofeight39◦deflectors(1),eight6◦deflectors
near∼10Kwithvaryingdegreesofsuccess[26–28].Evenless (2),andeightfocussingquadrupoledoublets(3). Oneofthe6◦ deflectorsis
isknownaboutthelow-temperaturebehaviourofCEMs[29]. fast-switchabletoallowionbeaminjection(4). Inmanyofthehere-reported
In a recent publication, we have presented the design of measurementsalaserbeamoverlappedthestoredionsinanexperimentaltarget
section(5).The6◦deflector(6)directlyfollowingthetargetactsascharge-to-
a movable single-particle counting detector for the Cryogenic
massanalyserprecedingthemovableCOMPACTdetector(7). Thelattercan
Storage Ring (CSR) of the Max Planck Institute for Nuclear bepositionedtointercepttheproductparticles(8)fromatomicreactionsinthe
Physics(MPIK)inHeidelberg,Germany[30]. Here,wereport targetwhileallowingtheparentionbeam(9)tocirculateunhinderedinthe
CSR.Inthedetector, theproductparticleshitasecondary-electronemitting
onthefirstoperationofthisdeviceunderreal-lifeexperimental
cathode (10). These electrons (11) are accelerated towards a micro-channel
conditionsattheCSR.
platestack(12)wheretheyaremultipliedtoformthedetectorcurrentpulse.
This paper is structured as follows: In Section 2 we briefly Foroff-beamtesting,thedetectorcanbeirradiatedbyanultra-violet(UV)light
describethe instrument. In Section3wepresent themostim- emittingdiode(LED,13)installedintheoppositesectorofCSR.Fordetailssee
textandRefs.[15,30].
portantfindingsfromthefirstoperationofthedetectionsystem
with the storage ring CSR at its lowest temperature of ∼6 K.
InSection4wequantifyanddiscusstheresultsfromthatseries
system has been described extensively in a dedicated publica-
of experiments, with emphasis on the single-particle detection
tion[30],hencewelimitourselvestoabriefoverviewhere.
efficiency of the set-up. Section 5 closes with a summary and
Equipped with a 20-mm-wide entrance window for heavy
outlookontofuturedevelopments.
particles, thedetectorismovabletransverselytothebeamdir-
ection in the plane of the storage ring. It is installed 1.0 m
2. OverviewoftheExperimentalSet-Up downstream of a short (6◦) electrostatic bending dipole of the
storage ring. Product particles generated from the stored ions
TheCSRisafullyelectrostaticstorageringdesignedforpos- aredeflectedatacharacteristicangleinthedipoleelement. By
itiveornegativeionsofkineticenergiesupto300keVperunit placementatasuitablehorizontalposition,thedetectorcanin-
ofcharge[15]. Thebeamguidingvacuumvesselaswellasthe tercept products with a charge-to-mass ratio that differs from
ion optics contained therein can be cooled to temperatures of thatofthestoredparentbeambymorethan100%inbothdirec-
∼6Kbyaclosed-loopliquid-heliumrefrigerator. Forthermal tions. Specifically,itcandetectneutralproductsonaxisofthe
insulation, the beam line is enclosed in an additional isolation ionbeamintheexperimentaswellas,e.g.,ionisationproducts
vacuum vessel and protected by several layers of black-body- uptothedoublechargeofastoredatomiccationbeam[30].
radiationshields. Eventually, the detector is designed to intercept product
With an orbit circumference of 35 m, the storage ring (cf. particlesoriginatingfromion-electroninteractionsinthefuture
Fig. 1) consists of four identical ion-optical sectors which en- electron cooler of CSR—like electron recombination or elec-
close four field-free drift sections. While one of the latter is tronimpactionisation[32,33]. Incontrasttothedetectorset-
occupiedbythebeamdiagnosticinstrumentationofthestorage up, the cooler was not yet operational during the 2015 exper-
ring[31],theotherthreearefreeforinstallationofexperimental iments. Instead, an ion-photon interaction beam line was in-
equipment.Thecountingdetector(lowerpanelofFig.1)isloc- stalledintheexperimentalCSRsectionprecedingthedetector
ated downstream from an experimental section, within one of [15]. Itallowedtooverlapthestoredionsatgrazinganglewith
the ion-optics sectors of CSR. The technology of the detector laserbeamsofvariouswavelengthsthatwerecoupledintoCSR
2
usingasystemofbroadbandview-portsandmirrorsinthecryo- 1000
7 Apr 2015
genicvacuumchamber.Thisin-ringlasertargetwasusedinex-
4 Apr 2015
perimentsonphoto-inducedelectrondetachmentofstoredan- 10 ×
100
ions. In addition, without using the laser beams, experiments (80 mW)
)
Ω
onauto-detachmentandauto-fragmentationofexcitedmolecu-
G
larandclusterionswereperformedusingthesameset-up. At e ( 10
c
higherCSRoperatingtemperatures,productsofelectrontrans- n
a
fer from the residual gas to stored cations were observed. For st
testingpurposes,thedetectorcanbeirradiatedbyphotonsfrom esi 1
r
P 1 Apr 2015
anultra-violet(UV,245(5)nm)lightemittingdiode(LED)[30]. C
The UV-LED is located in a room temperature annex of the M 30 Mar 2015
0.1
CSRsectoroppositeofthedetector. Thebeamofphotonsfrom 23 Mar 2015
theLEDispracticallyuncollimatedandenterstheCSRvacuum 18 Mar 2015
chamberviaasetofUV-gradesapphireview-ports. 0.01
The detector employs a variant of the ‘Daly’ ion detection 1000 100 10 1
principle,whereincidentmassiveparticlesimpingeontoasec- CSR temperature (K)
ondary-electron emitting cathode made of aluminium [25, 30,
34]. The secondary electrons released in each hit are acceler-
Figure2: ElectricresistanceoftheMCPsetasafunctionoftheambienttem-
ated by 1.2 kV towards a small chevron micro-channel plate perature(dots). Thedatawastakenduringthe2015cool-downofthestorage
stack (cf. Fig. 1). The latter acts as secondary-electron mul- ringCSRfromroomtemperatureto∼6K,asindicatedbythetimestamps.
Thedashedlineisapower-lawfittothedataandintendedasaguidetotheeye
tiplier, while being protected from direct hits by the primary
only. TheverticalarrowindicatestherangewithinwhichtheMCPresistance
massiveions. Themultipliedelectronbunchesarethencollec- couldoptionallybevariedusingitsbuilt-inheateratapowerof∼80mW.In
tedonametalanode. thereportedsingle-particleexperiments,theheaterwasnotusedandthede-
Aftercapacitivedecouplingfromhigh-voltage,thepulsesare tectorwasleftatthe6(1)KtemperatureofthesurroundingCSRstructures(cf.
Sect.3.2).
drivenintoafastfront-endamplifierof50Ωinputimpedance
and gain factor 200 (Ortec VT120A). In most of the presented
experiments, the resulting ∼10 ns-short electric pulses were optiontowarmuptheelectronmultiplierinoperation,asmall
convertedintologicalsignalsusingalineardiscriminator, and electric heater made of a bare Constantan wire is included in
counted by a VME-based multiscaler. Simultaneously—but thesupportingframeoftheMCPstack[30].
asynchronously—sample pulses could be recorded using a di-
gital oscilloscope which served as waveform digitiser. This
simplesolutionyieldstwoindependentdatasetsforthedetector 3. LowTemperatureOperation
countrateandforthesamplewaveforms,whichcannotbecor-
relatedonthesingle-particlelevel. Inthecourseoftheexperi- The first experimental beam-times atthe CSR took place in
ments,asecond,moreadvanceddataacquisitionsystemwasset 2015,andlastedforapproximatelyfivemonths,includingcool-
up,consistingofafastanalog-to-digitalconverter(FADC,Agi- down of the storage ring by the liquid-helium refrigerator and
lentAcqirisU1084A)equippedwithalargesamplingmemory. rewarming of the facility [15]. Besides the afore mentioned
This system allows gapless recording of the pre-amplified de- measurements on electron detachment and cluster fragmenta-
tector signal. Via an on-line peak-finding routine, it yields a tion,amultitudeofexperimentswithpositiveandnegativeions
single, consistent dataset containing the amplitude and arrival wereconductedinanefforttocharacterisethestorageringand
timeofeachindividualdetectorpulse. beam diagnostic instrumentation. During most of the experi-
Much care was taken in preparing the detector to perform ments, the CSR operated at an average temperature of ∼6 K.
at cryogenic temperatures. With reference to that purpose, Duetotechnicalissuesoftheinjectionaccelerator,theionen-
the device has been called ‘COMPACT’, the ‘COld Movable ergies were limited to 80 keV, i.e. well below the CSR design
PArticle CounTer’ [30]. In order to support optional room- energy of 300 keV per unit of charge. The stored ion species
temperature operation of CSR, the design additionally needed included Ar+, N+, O−, OH−, Si−, C−, Co−, Co−, and Ag−
2 2 2 3 2
tofulfilthestandardlow-out-gassingrequirementsofbakeable [15]. TheCOMPACTdetectorsystemwasusedinalmostallof
UHVequipment. Allelectronicsiskeptontheatmosphereside theexperiments,sothatitslow-temperatureperformancecould
of CSR’s nested vacuum system, as is the rotary actuator that bestudiedinavarietyofusecases. Thissectionpresentsafew
allowshorizontalpositioningoftheparticlesensorviaathread examples of measurements that showcase the most important
driveinsidetheCSRbeamline. findingsmadeduringoperationoftheparticledetector.
The chevron MCP stack consists of two matched, circu-
lar ‘extended dynamic range (EDR)’ micro-channel plates
3.1. Cool-Down
(Photonis18/12/10/12D40:1EDR,MS)of18mmusefuldia-
meter. EDR MCPs are characterised by a significantly lower Whilecryo-adsorptionincoldoperationvastlyimprovesthe
resistanceascomparedtostandardvariantsandarethusexpec- residual gas pressure, the CSR vacuum concept does not rely
tedtoperformbetteratverylowdetectortemperatures. Asan oncryogenicsalone. Beforestartofthecool-downprocedure,
3
thebeam-guidingvacuumvesselofthestorageringwasUHV- 100 γ (245 nm) 40Ar (60 keV)
baked at 180◦C and subsequently reached a residual gas pres- 200 K 200 K
sure of ∼1×10−10 mbar already at room temperature [15]. 10 6 K 6 K
Consequently, the storage ring and the detector could already 1
operateduringthecool-downphaseofCSRfrom300Kto6K,
)
whichtookapproximatelythreeweeks. nits 100 γ (245 nm) 40Ar (60 keV)
Duringthecool-down,abeamof60-keV(1.5keV/u)40Ar+ u 110 K 110 K
ions was regularly stored in CSR. The detector was routinely rb. 10 6 K 6 K
a
switchedontodetecttheneutralArproducts,originatingfrom s ( 1
residual-gas electron capture by Ar+, in order to deduce the nt
u
o
stored-ion lifetime [15]. Additionally, the detector was irradi- C 100 γ (245 nm) 40Ar+ (60 keV)
ated by 245-nm photons from the UV-LED for comparison of 6 K 6 K
thesignals(seebelow). 10 6 K 6 K
Liketheheavyparticles,theUVphotonsdonotirradiatethe 1
MCPsdirectly,butreleaseelectronsfromthesurfaceofthecon-
0.1
verterelectrodewhicharethenacceleratedtowardstheMCPs.
0 0.25 0.5 0 0.5 1 1.5
During UV irradiation, the aluminium converter thus acts as a
Amplified pulse height h (V)
photo-cathodeoflow(∼10−4)quantumefficiency[35]. Itwas
verifiedthat,whentheelectronaccelerationpotentialsweredis-
Figure3: Pulseheightdistributionsoftheamplifieddetectorpulsesatthree
abled while keeping the MCP gain voltage enabled, the count
different temperatures of the CSR vacuum chamber (indicated by the labels
rateoftheset-updroppedtozero. ThisshowsthattheMCPin- 200 K, 110 K, and 6 K). The detector was irradiated alternatively by 245-
deeddetectssecondaryelectronsonly,andnoprimaryparticles nmUVphotons(5.1(1)eV,filledcircles)andby60-keVargonatoms(40Ar
(photonsorions)reachitinnormaloperation. Viathedriving and40Ar+,filledtriangles). Thepulseheightspectraarenormalisedtoequal
totalnumbersofcounts. Theverticaldashedlinesindicatethediscrimination
voltageoftheUV-LED,therateofphotondetectionscouldbe
thresholdof∼35mV.Toeasecomparison,afittothepulseheightdistribution
variedovermanyordersofmagnitude. Incontrasttofastions, at6Kisshownineachframeoftherespectiveparticle(solidlines,cf.Sect.
each photon can emit at most only a single electron from the 4.3). Toemphasisethedifferenthorizontalscales,theUVpulseheightfitis
converterelectrode,asthephotonenergyof5.1(1)eVislower alsoshownintheargonplots(dash-dottedlines).
thanthedoublework-functionofthecathodematerial.
After pre-amplification, the pulses were discriminated and
counted using the VME multiscaler, while sample waveforms vidualmeasurementswerenormalisedtothenumberofcounts.
wererecordedbytheoscilloscope. TheMCPbiascurrentwas Nosignificantchangesinthedistributionswerefound,neither
continuously measured by a floating nano-amperemeter. Fig- forUVphotonsnorfor60-keVargonparticles,betweenroom
ure2showsthederivedelectricresistanceofthestackedMCP temperatureandthefinaloperatingpointofCSRof6K.
set as a function of the average temperature of the relevant The intensity of the UV light source was set such that the
CSR sector. No dedicated temperature sensor is attached to averagerateofdetectedphotonswas∼600s−1 ineachmeas-
the particle detector itself. However, as the CSR cool-down urement. Forargonirradiation,thedetectedparticleratevaried
process was slow, we assume that the MCPs were in thermal strongly with temperature, as—at given intensity of the stored
equilibriumwiththeirsurroundings. ion beam—the rate of electron capture events scaled with the
Starting at the specified value of 56 MΩ at room temperat- residualgaspressureinCSR,whichimproveddrasticallydur-
ure,theresistanceofthechevronMCP-setrosebyalmostfour ingthecool-down[15]. At200Kthedetectorrecordedneutral
orders of magnitude during the cool-down, reaching values of Arproductsatratesuptoseveral10000s−1whileat110K,the
∼300 GΩ at 6 K. After switching the detector on, a gradual production rate had decreased to a few 1000 s−1. At the final
increaseoftheMCPbiascurrentbyuptoafactorofthreewas CSR temperature of 6 K, the rate of electron capture products
routinelyobservedwithinthefirsthourofoperation,especially wastoolowtobeidentifiedabovealowbackgroundeventrate
atverylowtemperatures. Itisyetunknownwhetherthiseffect of∼10s−1,inspiteofthehighstoredioncurrentof∼1 µA.
isduetooperation-inducedwarminguporwhetheritreflectsa Asthebackgroundeventswerenotlocalisedtothepositionof
purelyelectricchangeinthechannel-plateproperties. Figure2 the axis of a neutral daughter beam, they are believed to be
showsonlytheinitiallymeasuredMCPresistance,directlyafter duetostraysecondaryparticlesproducedalongthebeampipe
enablingofthehigh-voltagesupplies,whenthetemperatureof by the primary ions. In order to obtain a reliable pulse height
theplatescanbeassumedtohavebeenequaltothetemperature distribution from impact of 60-keV argon particles at 6 K, the
oftheCSRvacuumchamber. detectorwasmovedtowardstheclosedorbitinthestoragering
Figure3showsthepulseheightdistributionsoftheamplified until direct hits from parent Ar+ ions could be detected at a
detectorsignalsobtainedforUVphotonand60-keVargonirra- rateof∼500s−1. Itwasassumedthatthesecondaryelectron
diationatthreedifferenttemperaturesduringthecool-downof ejectioncoefficientof60-keVAr+ionswassufficientlysimilar
CSR.ThegainvoltageacrossbothMCPswaskeptconstantat tothatofneutralargonatomsofthesameenergy. Indeed, the
1.85kVinallmeasurements.Pulseswererecordedaboveadis- measuredpulseheightdistributionofAr+ionscorrespondedto
crimination threshold of ∼0.035 V. For comparison, the indi- that of the neutral atoms at higher temperatures, as shown in
4
Fig.3. Time (ns)
0 50 100 150 200
3.2. LocalisedHeatingoftheMCPs
0
WiththebeamguidingvacuumchamberofCSRat6K,the
electricheaterbuiltintothedetectorcanbeusedtowarmupthe V) −0.2 OH (60 keV, 6 K)
(
micro-channelplatesetabovethetemperatureofitssurround- al
n
ings. A functional test showed that, by operating the heating Sig −0.4
wire at a power of ∼80 mW, the resistance of the MCP stack
−0.6
couldbeloweredbyafactorof∼10. Anapproximatecalibra-
tion, asindicatedinFig.2, translatesthischangeinresistance OH− + γ → OH + e
toawarmingofthechannel-platesetfrom6Kto∼15K.No b.) 100
warming of the neighbouring CSR structures was observed in ar
(
the process. Previous experiments have shown that even sub- e
at 50
stantiallygreaterheatingpowerscanbeappliedwithoutdanger nt r
tothedetector[30]. u
o
C
ThepulseheightdistributionobtainedforUVirradiationdid 0
notchange duringthetestsof theheating. Thisconfirmedthe
earlier observations, as the UV-induced detector signals had 0 10 20 30 40 50 60 70 80
alsonotbeeninfluencedbythecooling-downfromroomtem- Horizontal detector position (mm)
perature (cf. Fig. 3). It is however expected that the option
of localised heating of the MCPs can improve the detector re- Figure 4: Sample amplified detector pulse obtained from a neutral product
sponse to high-rate impact of massive particles (cf. Sect. 3.3). beam(topframe, solidline)andaveragediscriminatedcountrateasafunc-
tionofthehorizontaldetectorposition(bottomframe,filledcircles)measured
In the experiments reported in the following, which focussed
atatemperatureof6K.ThedetectedparticleswereneutralOHmoleculesof
on the extreme low-temperature behaviour of the device, this 60keVtotalkineticenergy,originatingfromphoto-detachmentofstoredOH−
possibility was not yet checked, and the MCP stack was de- anions.Thearrowindicatesthediscriminationthresholdof∼37mV.Thesolid
liberately left at the 6 K temperature of the surrounding CSR lineinthelowerframeisafitofanassumedGaussiandaughterbeamenvel-
ope(dashedline,renormalised),convolvedwiththeknownhorizontaldetector
vacuumchamber.
apertureof20mm(indicatedbytheverticaldottedlines).
3.3. GeneralPerformanceat6K
WiththeCSRoperatingatitslowesttemperature,theCOM- parent. Using the horizontal beta function of the storage ring
PACT set-up was employed to detect a variety of neutral and [15] one derives a 95% horizontal transverse emittance of the
charged product beams. For single-particle experiments, the ion beam of 24(3) mm mrad. One also derives that, in this
storedbeamcurrentswerewellbelowthatoftheAr+ionsused example, the horizontal width of the daughter beam leads to
for CSR commissioning. The above mentioned secondary-ion a27(3)%geometricdetectionlossduetothenarrowsensitive
backgroundwasnotobservedinanyotherexperiment,andeven aperture.Thisisbydesignandnotconsideredcritical,asthede-
weakdaughterbeamscouldbeeasilyidentifiedbymovingthe tectorisintendedprimarilytodetectproductsoriginatingfrom
detectorhorizontallyacrosstheCSRbeamlineandmonitoring future electron-cooled ion beams [30]. Such beams are char-
theaverageparticlecountrateasafunctionoftraveldistance, acterised by a much lower transverse emittance, and in their
asshowninFig.4. Duringthefirstexperimentalcampaign,the casethenarrowdetectoraperturewillhelpidentifyingproducts
cryogenic thread drive—as described in ref. [30]—was used basedoncharge-to-massselection. Notethattheheightofthe
tomovethedetectorbyatotaldistanceofmorethan∼7mat sensitive aperture is much wider (50 mm [30]) so that no sig-
lowesttemperature,equivalentto24fullstrokesacrosstheCSR nificantverticalcutontheproductbeamisbelievedtooccurin
vacuumchamber. theexperimentsdescribedhere. VariantsoftheCOMPACTde-
tectorwithlargerhorizontalacceptanceforuseinfuturemeas-
In the example of Fig. 4, a product beam of 60-keV neut-
urementsonuncooledionbeamsarepresentlybeingdeveloped.
ralOHmolecules(3.53keV/u)wasdetected,originatingfrom
electrondetachmentofstoredOH− ionsina633-nmcwlaser Thedetectorpulseshapesandheightdistributionsfoundfor
beam in the experimental CSR section preceding the COM- atomic or molecular products from experiments at 6 K were
PACTdetector. Usingtheknownsizeofthehorizontaldetector quitesimilartothoseobservedforargonhitsduringcool-down
aperture of 20 mm and the particle count rate measured as a (cf.Fig.3). AsshowninFig.4,theamplifiedanodesignalsare
functionofdetectorposition,thehorizontaltransversedaughter 10–30nsshortpulsesoftypicallyafew100mVamplitude.
beamenvelopecanbeobtainedbydeconvolution.Thestandard Characteristic features found in all experiments conducted
deviation of the assumed Gaussian product beam profile was duringCSRcommissioningarepulseheightdistributionswhich
derivedtobe9.0(5)mmatthedetectorposition. Astheneutral arenotpeaked(cf.Figs.3and5). Theexplanationforthissig-
particles are not influenced by the ion optics, and as the mo- nature lies in the fact that the chevron MCPs detect the 1.2-
mentum transfer to the molecule during the photo-detachment keVsecondaryelectronsejectedbytheprimaryions[30]. All
is negligible, the product beam maintains the emittance of its of these secondary electrons have to be assumed to impinge
5
2 1 Co2−+γ→Co2+e Co2(60keV,6K) 21::~~12470000ss--11 100
1000
3:≤1000s-1
10
)
s
100 3 nit
-1(s) 1 rb. u
e a
Countrat 10 F(h;γ~,b) 12::~~21740000ss--11 100 Counts (
3:≤1000s-1
1 10
1
0.1
0 2000 4000 6000 0 0.5 1 1.5
StoragetimeinCSR(s) Amplifiedpulseheighth(V)
Figure5:Detectedproductratefromphoto-detachmentofCo−anionsasafunctionofstoragetimeinCSRat60keVtotalenergy(0.51keV/u)and6Koperating
2
temperature(leftframe),andpulseheightdistributionsobtainedfromthe60-keVCo2moleculesatdifferentaveragehitrates(rightframes).Thedistributions1–3
(seeannotation)weremeasuredattheindicatedaveragecountrates.Therightbottomframeshowsfitsofthemeasuredpulseheightdistributionsusingthemodel
fromSect.4.3,includingdampingoftheMCPgainathighratesinthecases1and2(seetext). Below∼1000countspersecond,nodependenceofthepulse
heightsonthedetectionratewasobserved. Theleftframeshowsthediscriminatedcountrate(dots)ofneutralCo2productsfromphoto-detachmentofasuitable
60-keVCo−beam,storedinCSRfortwohours. After7200sofstorage,theionbeamwasdeliberatelykickedoutoftheclosedorbit. Thelong-dashedlineisan
2
exponentialdecayfit,yieldinga1/elifetimeofthestoredionsof1383(5)s.Thehorizontalshort-dashedlinesindicatetheintrinsicdarkcountrateofthedetector,
whichwasfoundtobeaslowas0.3(1)s−1inallexperimentsdescribedhere.
closetodifferentMCPchannels,withinatimeshorterthanthe ∼ 2700 s−1 (case 1) the mean pulse amplitude was ∼ 25%
observed pulse width of ∼10 ns, as has been verified by nu- lowerthanbelow1000s−1(case3). Giventhesimplecounting
merical simulation of the electron trajectories in the detector. logicsbasedonafixedsignaldiscriminationthreshold(dashed
Hence,thetotalpulseheightfromheavy-ionimpactis,infact, vertical line in Fig. 5), this caused the detected particle rate
the result of pile-up of several independent pulses generated tovarynon-linearlywiththestoredintensityoftheCo− beam
2
by 1.2-keV electron impact on the MCPs. The resulting sum aboveadiscriminatedcountrateof1000s−1. Incontrast, UV
pulse height distribution tends to be monotonously decreasing photonsfromthetestLED—whichproducemuchsmallerMCP
with higher amplitudes, as will be discussed in the upcoming signals—could be detected at 6 K at significantly higher rates
Sect. 4, following the original analysis by Spruck et al. [30]. (more than ∼ 3000 s−1) without deterioration of their pulse
Duetothischaracteristicpulseheightdistribution,thedetector amplitudes. Weattributethiseffecttogainsaturationduetothe
count rate was sensitive to the signal discrimination threshold onset of charge depletion of the MCPs at simultaneously low
inthepresentwork,andalowelectricbase-linenoisewasim- temperature and elevated heavy-particle hit rate. This hypo-
perativeinordertoobtaingoodoveralldetectionefficiency. thesisisalsosupportedbytheearlierobservationthat,athigher
temperature, 60-keV Ar could be detected at much higher av-
3.4. High-AverageRateResponseat6K erage count rates with no evidence of signal degradation (c.f.
Fig. 3). It is expected that local warming of the MCP set (cf.
At6K,adependenceofthepulseheightsontheaveragede-
Sect.3.2)canmitigatethesesaturationeffects,howeverthishas
tectorcountratewasobservedforheavy-particleimpactabove
notbeenstudiedyet.
a certain critical hit rate. This is illustrated in Fig. 5: In the
experiment, a beam of Co− ions was stored in CSR at 60- In the Co− experiment the pulse height distribution—and
2 2
keV total energy (0.51 keV/u) and passed through a grazing- thusthediscriminatorefficiency—wasrate-independentalsoat
angle, 633-nm-lasertargetintheexperimentalsection. Photo- 6K,aslongastheaveragecountratewaskeptbelow1000hits
detachmentyieldedneutralCo moleculesthatreachedthede- persecond. Undertheseconditions,thedetectorcouldbeused
2
tector at the kinetic energy of the parent beam. By variation toreliablymeasuretheevolutionofthephoto-detachmentrate
oftheintensitiesoftheionorlaserbeams, theaveragerateof over very long storage times of the Co− ions (left-hand frame
2
productparticlescouldbeadjusted. of Fig. 5). A fit to the Co count rate as a function of stor-
2
It was observed that above ∼1000 discriminated Co hits agetimeyieldsa1/elifetimeoftheanionsintheexperimental
2
persecondinaverage,thepulseamplitudesstartedtodecrease set-up of 1383(5) s. After 7200 s of storage, the remaining
noticeably (cases 1 and 2 in Fig. 5). At a detection rate of ionsweredeliberatelykickedoutoftheirclosedorbitswithina
6
1 2 3
singleturn,sothatthedarkcountrateofthedetectorcouldbe
measured. Asvisibleintheleft-handframeofFig.5,eventwo 0
hoursafterioninjection, themeasuredCo− photo-detachment V)
2 (
ratewasstillmorethananorderofmagnitudegreaterthanthe al−0.5
dfoeutencdtotorbbaec0k.g3r(o1u)nsd−1le,vweli.thInnoalnlomtaebalseudreepmeenndtesn,ctheeolnattteemrpwears- dsign −1 0 Co2−→Co+Co−
e
oatfurraed.iMo-noustcolifdtehseindathrkepMuClsPessaurbesbtrealtieev[3e6d]toanbdeadrueevteoryβs-idmecilaayr mplifi−1.5 −0−.51 45 6 7
A
toactualcountingpulsesinshapeandamplitude. Thefactthat −2
0 10 20 30 40 50 60 70 80 90
thisbackgroundrateisfoundtobeverylowisimportantinthat
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0.04
rejectionofthedarkeventsbasedonpulseshapeanalysisdoes
StoragetimeinCSR(s)
notatpresentseemfeasible.
Figure6: DetectorpulsesgeneratedbyCo−anionsfromauto-dissociationof
3.5. ShortParticleBurstsat6K Co− moleculesstoredinCSR.TheMCPsetwasoperatingatthe6Ktem-
2
peratureofthesurroundingbeam-line. Thedata-setcorrespondstoasingle
In the experiments described so far the product particles ion-injectionintothestoragering.DuetotheshortlifetimeoftheexcitedCo−
2
reached the detector in quasi-steady streams, with average molecular ions, the rate of Co− hits steeply decreased after the start of the
measurement. Theevolutionofthedetectorsignalswasobservedasafunc-
fluxes that varied slowly compared to all other time constants
tionoftime(andthusaveragecountrate)byevaluatingthemeanpulseheight
of the set-up. Due to the long storage times of the CSR, this distributionsintheindicatedtimeintervals.
situation is common for parent ions that are in a stable state,
orwhentheirinternalstatepopulationvariesslowly, suchthat
therateofinteractionbetweenthestoredbeamandthetargetis
nearlyconstant.
In other cases the reaction products show a burst-like time- averagecountratewerecalculated. Thelengthofthetimewin-
structure. E.g.,interactionofthestoredionswithapulsedlaser dowsincreasedasafunctionofstoragetime,sothateachwin-
targetortheionproductionprocessitselfcanleadtopopulation dowwascharacterisedbytheapproximateaveragecountrates
of metastable levels. By timing the detection of the resulting given in Fig. 7. To easily quantify the shape of the measured
end products with respect to the time of interaction, the life- pulseheightdistributions, eachwasapproximatedbyasimple
timesofthemetastablescanbemeasureddowntothescaleof doubleexponentialdecayfitfunction.
therevolutionperiodinthestoragering. Insuchexperiments,
the detector hit rate may vary drastically within a few milli- As visible in Fig. 7, at 6 K detector temperature the pulse
seconds,withtheburstratedirectlyfollowingtheformationof heightdistributionwasconstantacrossstoragetime,evenifthe
the metastables significantly exceeding the average count rate peak rate during the first millisecond after ion injection was
intheexperiment. as high as 105 s−1, i.e., two orders of magnitude higher than
As an example, the auto-fragmentation of Co− molecular the maximum useful count rate observed in the Co− photo-
2 2
ionswasstudied. TheCo− beamwasproducedinametal-ion detachment experiments described in Sect. 3.4. The apparent
2
sputter source, accelerated to a total kinetic energy of 60 keV, changeinshapeatverylatestoragetimes(distributions6and7
andstoredinCSRfor90s.Afterthatstoragetime,anyremain- inFig.7)isdueonlytotheincreasingcontributionofdetector
ingionsweredumpedbeforethenextinjectiontookplace.Also darkcounts(distribution8)tothemeasuredsignals.
here, the storage ring operated at 6 K. The sputter ion source
naturallyproducespartoftheanionsinauto-dissociatingmeta- The different saturation threshold compared to the (steady-
stablestates. TheCOMPACTdetectorwaspositionedsuchas current) photo-detachment experiment from Sect. 3.4 is most
tocollectthe30-keVCo−fragmentsthat,inabsenceofresidual likely due to the fact that the short auto-fragmentation bursts
gas collisions or other target interactions, could only be pro- werefollowedbyextendedperiodsofnear-zerocountrate,dur-
duced from the metastable ion population. The results of the ing which the MCP channels could recharge via the low bias
experimentwillbepublishedseparately. current at 6 K, before the next ion injection would take place.
AsshowninFig.6, theinstantaneousdetectionrateofCo− For peak hit rates even larger than 105 s−1, saturation effects,
was very high directly after ion injection, but then steeply de- similartothosedepictedinFig.5,wereindeedobservedalsoin
creased on a time-scale of milliseconds. In the experiment, the burst-type auto-fragmentation experiment. In those cases,
theadvancedFADC-baseddataacquisitionsystem(cf.Sect.2) thenon-linearityinthemeasuredcountratewithrespecttothe
wasused. Incontrasttothesteady-currentexperiments,asde- truefragmentproductionratecouldnotbefullyeliminatedby
scribedinSect.3.4,itwasthuspossibletoobservechangesin data processing, in spite of the FADC-based data acquisition
the pulse-height distribution on very short time scales, limited systemallowingforadvancedpulsediscriminationtechniques.
onlybythecountingstatistics. Alsointheburstexperiments,on-linemeasurementofthepulse
Starting with the ion injection into CSR, integration time heightamplitudesalongwiththedetectionratethusturnedout
windowsweredefinedasshowninFig.6. Foreachintegration to be a crucial tool for evaluating the reliability of the experi-
window, the height distribution of the detector pulses and the mentaldata.
7
22
10−1 depends on the time the MCP set is allowed to recharge in-
Co−→Co+Co−
arb. units) 1111100000−−−−−65423 41235:::::11111000000000ss000-1002-s10s-1s-1-1 1111224680-1efficients(V) bbbtoeeeefctelTttwihnoeheveriseset‘cundDofdamtatviohleobyedbui’nebrsceayuaosbsrnclstaveeotsemlnlarosotarweegftqrie-cputeaersellomeenldnyccptuseryeicotreotidavtf.tpeeui—atrasretpw‘ibecDirletetahhuslayr,ave’a—-ironteyuldaperetfeioffivedfnecetelthsydietgshbmdnaye.tatTthlehlchaeMtseosndrCizoeiPes-t
Counts (11110000−−−−10987 867:::b10e.s2a-ms1-1off 1 2 3 4 5 6 7 46810Exponentialco sttarhecoettditvveheo.alTtaurchmeoealeblaeaosccfttkshtghetrheoseeussnunesbdcistoetivrvnaeedtnaeatrpy[re3are6ttel]ue.orceftArMoonfnCstEhPeDesjeRCnca-OtMteuMdrCafPPlrAloyoCmsfTcttahhdlaeeetstseewaclmetiotche-r
10−11 2 (20×50 mm) can be expected to have a dark count rate that
10−12 0 is an order of magnitude higher than the one measured in the
0 0.2 0.4 0.6 0.8 1 1.2 10−410−310−210−1 1 10 102
experiments reported here. Additionally, larger (and thicker)
Amplifiedpulseheight(V) StoragetimeinCSR(s)
MCPshavebeenmeasuredtoreachmuchhigherelectricresist-
ancenearliquid-heliumtemperaturethanthe∼300GΩfound
Figure 7: Left: Pulse height distributions obtained for 30-keV Co− anions
reachingthedetectorfollowingauto-fragmentationofstoredCo− molecular here[26].EvenEDRvariantsoflargechannel-plateshavebeen
2
ions. ThelabelscorrespondtothetimeintervalsasindicatedinFig.6. The found with unfavourable electric behaviour at lowest temper-
distributionsareartificiallyshiftedbymultiplicationbyfactorsoftenforbet- ature, which likely leads to earlier depletion at high detection
terreadability. Thelabelsindicatetheapproximateaveragecountrateinthe
ratesandthusworsehigh-rateacceptance[27,28].SmallMCPs
respectiveinterval. Right: Decaycoefficientsobtainedfromdoubleexponen-
arealsoproducedonalargescaleroutinely,which—besidesthe
tialdecayfitsofthepulseheightdistributionsforthedifferenttimebins. The
changeinthelaterpulseheightdistributions6and7isduetotherisingcontri- obviousadvantageoflowerprices—mightleadtomorestable
butionofdetectordarkcounts(distribution8)tothemeasuredsignal. productionprocessesandmorepredictableproperties.
4.2. DetectionLosses
4. Analysis
Knowledgeoftheparticledetectionefficiencycanbeimport-
ant in experiments seeking to measure absolute cross sections
Inthefollowingweseektoquantifytheexperiencefromthe ofionreactionsinthestoragering. Insomecasesthedetector
first operation of the COMPACT detector at 6 K temperature. canbecalibratedagainstaknownprocess,oritsefficiencycan
Themanyengineeringtopicsrelatedtothecryogenicenviron- be inferred by controlled variation of the experimental condi-
ment have been discussed in detail by Spruck et al. in a pre- tions. However, if other parameters of the experiment are un-
vious article [30]. Here we focus on the performance of the known, independent knowledge of the product detection effi-
instrumentduringthefirstatomicandmolecularphysicsexper- ciency can be the only way to interpret the measurements in
imentsusingthestorageringCSRatlowesttemperature. Fora terms of absolute numbers. Cryogenic operation of MCPs is
particle counting detector, two basic properties come to mind: consideredout-of-specificationbytheirmanufacturers,andthe
Theyarethesingleparticledetectionefficiencyandusefuldy- detectorbehaviournearliquid-heliumtemperatureisnotguar-
namicrangeofthecountrate. anteed. Although there has been some research on the topic,
open questions remain [26–28, 37]. In this situation, a simple
4.1. DynamicRange waytomonitortheabsolutedetectionefficiencyduringtheex-
perimentisimportanttoensurethereliabilityofthedatataken.
Thedynamicrangeisdefinedbytheintrinsicdetectorback- In the experiments reported here, the particle detection effi-
ground on the one hand, and by the maximum particle count ciency of the COMPACT detector is limited by three effects.
rate that can be reliably measured on the other hand. For Thefirstisthelossofproductparticlesduetolimitedgeomet-
the chevron MCP set operating at the lowest CSR temperat- ricacceptanceofthedetectorinthehorizontalplane(cf.Fig.4).
ure of 6 K, the experiment on Co− photo-detachment from Thisisnotdiscussedfurtherhere. Asnotedabove,thenarrow
2
Sect. 3.4 can be considered as a benchmark: from the dark widthofthesensitiveapertureisbydesign. Infactsignificant
eventrateof0.3(1)s−1 tothemaximumproductcountrateof effortshavebeenundertakentorealisetheverticallyelongated
∼1000 s−1 that could be reliably discriminated, the dynamic detectionwindow. Itisstillwideenoughtointerceptdaughters
rangeof∼3×103 incontinuous-ratemeasurementsallowsto ofelectroncooledbeamswith100%efficiency[30].Inallother
followproductformationfromatomic,molecularorclusterpro- casesthegeometriclossratiocanbedeterminedbyahorizontal
cessesin theCSRforup toeight1/e-lifetimesof thereaction scanofthedaughterbeamenvelopeasdescribedinSect.3.
at hand. It is expected that local heating of the MCPs can ex- Asecondlimitationofthedetectionefficiencyarisesfromthe
tendthedynamicrangeevenfurther,butthishasnotbeenstud- discrimination threshold applied to the anode signals. Pulses
iedyet. Forburstmodeoperation,theCo− auto-fragmentation of amplitudes below threshold are not recorded in the count-
2
measurement—also carried out at 6 K detector temperature— ingelectronicsusedintheexperimentspresentedhere. Thisis
shows a significantly greater dynamic range, reaching up to an issue that must be addressed technically. At high gain of
∼3×105 in the example at hand. However, that value likely the pre-amplifier and simultaneously low baseline noise level
8
oftheanodehigh-voltageline,thediscriminationthresholdcan 0 0.5 1
bsmeavlelrcyulto-owffreralatitoivecatnobtheeemaseilaynepsutilmseahteedigihftt.hTehoevereramllaisnhianpge, f1 f2 f3 f4 f5 f10 f15 f20 f25 f30 10
100
ofthepulseheightdistributionisknownorcanbeextrapolated.
FADC-based data acquisition systems, like the one presented 1 1)
-V
in Sect. 2, may not involve a fixed discrimination threshold at s) 10 y(
all, as they allow identification of particle hits using numer- nit 0.1 sit
u n
icalpulse-shapeanalysisofthetime-resolvedanodesignal. In- b. 1 UV(245nm) de
dsteopreangdee-rnitngextitmerinnaglstyrisgtgemer)sa(rferotmhe,net.ygp.,icaalplyulusseeddltaosesrtaortratnhde nts (ar 1000 F(h;γ~,b)~ΣPkfk ability
stop the FADC. Whether the FADC data can be processed in Cou 100 Co2(60keV) 1 rob
real-time, orneedstoberecordedforsubsequentanalysis, de- P
pendsontheprocessingspeedofthecomputer,theparticlerate,
0.1
andthecomplexityofthechosenpulsedetectionalgorithm. 10 Ag (60keV)
2
The third and most fundamental source of detection effi-
0.01
ciency loss lies in the stochastic nature of the electron ejec- 1
tionprocessfromtheconvertercathodeandofthedetectionof 0 0.5 1
these secondary electrons by the MCPs. Even if the average Amplifiedpulseheighth(V)
number of electrons released per impinging ion can be quite
highinsomeexperiments,thereisinfactanon-zeroprobabil- Figure8:Illustrationofthefitprocedurefordeterminationofthedetectoreffi-
itythatanioneitherreleasesnoelectronatall,orthatnoneof ciency.Topframe:Pulseheightdistribution(f1)measuredfor245-nmphoton
the ejected electrons is detected by the MCPs. In these cases irradiationofthedetectorat6Ktemperature(filledcircles)andfit(solidline).
noanodepulsecanbeobserved,nomatterhowtechnicallyad- Thenumericallycomputedconvolutions fk(fork>1)areindicatedbythefine
dashedanddash-dottedlinesasindicatedbythelabels. Bottomframe: Via
vanced the readout electronics is. In the following, we denote Eq.(3)the fksumuptothepulseheightdistributionsforheavy-particleimpact
byP0 thelikelihoodfornoMCPmultiplicationeventtooccur, (F,seetext).Best-fitmodelsofFareshownfor60-keVAg2andCo2products
(solidlines,aslabelled),reachingthedetectorataverageratesof∼100s−1
although the converter electrode did receive an impact from a
and6Koperatingtemperature.
heavyparticle.
4.3. ModellingtheDetectionEfficiency In absence of beam cooling—as in all experiments reported
In the case of the COMPACT detector, a value of P can here—orforstronglyexothermicmolecularbreakupreactions,
0
bederivedbycomparisonoftheanodepulsesobtainedforthe the product particles irradiate a large fraction of the sensitive
heavyparticleunderstudywiththosegeneratedbyUVphotons aperture of the detector. In that case a possible variation of γ˜
fromtheLEDsourceinstalledinCSR.AsdiscussedinSect.3, acrossthesensitivedetectoraperturemustbeaccountedfor.
anodepulsesforheavy-particleimpactcanbeassumedtores- A model of the secondary-electron statistics valid for non-
ultfrompile-upoftheMCPsignalsgeneratedbythesecondary uniformγ˜hasbeendevelopedwithintheframeworkofdiscrete-
electrons released from the ‘Daly’ converter cathode. In con- dynode electron multipliers, where a similar situation occurs
trast, the 245-nm photons can be assumed to release at max- [38].There,thenumbernofelectronsemittedfromonedynode
imum a single electron from the cathode material due to their towardsasecondoneisdescribedbyaPólyadistribution
lowenergy(5.1(1)eV).Comparisonofthepulseheightspectra
obtainedinbothcasesthusallowstoestimatetheaveragenum- γ˜n n−1
W (γ˜,b)= (1+bγ˜)−n−1/b ∏(1+jb). (1)
berofsecondaryelectronscontributingtotheheavy-ionsignals. n n!
j=0
InthecaseoftheexperimentonstoredAr+ showninFig.3,
comparisonofthemeanpulseheightsobtainedfor60-keVar- γ˜isnowthemeannumberofsecondaryelectronsreachingthe
gon and UV irradiation suggests that, in average, 4–5 MCP seconddynodeforeachimpactonthefirstone. bistherelative
eventsfromsecondaryelectronspileuptoformtheAr-induced varianceofγ˜.Forb=0,W isequaltothePoissondistribution,
n
signals.AtanassumedMCPdetectionefficiencyof60%forthe in the special case of b=1 it assumes the shape of the expo-
1.2-keVelectrons,thismeansthatforeachimpingingheavyAr nentially decreasinggeometric distribution. In ourapplication
atom an average number of γ˜≈7.5 secondary electrons reach weidentifythe‘Daly’convertercathodewiththeemittingdyn-
the MCP surface. Based on Poisson-statistics, one would ex- ode, whiletheroleofthecollectingdynodeisassumedbythe
pect the chance for no charge multiplication to occur in the positivelybiasedMCPinputsurface.
MCP-stackafteraheavy-ionimpacttobeaslowas∼1%. EachofthenelectronsfromEq.(1)hasachanceε togener-
ThePoissonianmodelishoweveronlytrueifthepointofim- ateachargemultiplicationavalancheintheMCPs. Hence,the
pactoftheionsonthedetector—andthustheaveragenumber totalnumberkofMCPcascadesgeneratedbyasingleprimary
γ˜ of converter electrons attracted towards the MCP surface— ionisdistributedaccordingto
is fixed [30]. This is expected to be a good approximation in
future experiments on electron-cooled atomic ions in CSR, as ∞ (cid:18)n(cid:19)
P(γ˜,b)= ∑ εk(1−ε)n−kW (γ˜,b), (2)
their product beams are characterised by very low emittances. k k n
n=k
9
which is the discrete convolution of W with a binomial dis- showing the effect of detector saturation, f was scaled by a
n 1
tribution. In the following, we assume ε =0.6 based on the factor d <1 along the h-axis, before the pile-up distributions
geometric open-area ratio of the MCPs. The principal results f werecomputed(cf.Eq.3). Thissimulatesthereducedgain
k
of the analysis are largely independent on the choice of ε as oftheMCP-detectorwithrespecttounsaturatedbehaviour. For
discussedbelow. valuesofd≈0.4(distribution1inFig.5)andd≈0.6(distri-
Let f (h)bethedistributionofpulseheightshproducedby bution 2), the subsequent fit procedure yields values for γ˜ and
k
theMCP-stackuponsimultaneousmultiplicationof(precisely) bthatarecompatiblewiththeunsaturatedcase(distribution3)
k converter electrons. If f is known for all k, the sum pulse- withintheirstatisticaluncertainties.
k
heightspectrumF forheavy-particledetectioncanbemodelled Table 1 summarises the data from a few selected experi-
as ments.Thefirstthreerowsshowtheanalysisofthepulseheight
∞ distributionsfoundintheAr+storageexperimentsduringcool-
F(h;γ˜,b)=C ∑P(γ˜,b)f (h), (3)
k k downofCSR(cf.Sect.3),asshowninFig.3. Asnotedearlier,
k=1
nochangeinthepulseheightsasafunctionoftemperaturewas
withC being a normalisation factor. In the case of the COM- observedforeitherphotonorargonirradiationofthedetector.
PACT set-up, the k-electron distributions fk can be inferred Also the fit results obtained using the model from Eq. (3) are
from the pulse height distribution f1, measured for irradiation consistent among all three operating temperatures. The num-
ofthedetectorbytheUVphotonsource. Asthephotonsnever berofsecondaryelectronsemittedbytheconvertercathodeis
emitmorethanoneconverterelectron,theirMCPpulseheight derived as γ˜∼8, in good agreement with the above estimate
distribution is equal to f1. The pulse height spectra for k-fold obtained by comparison of the mean pulse amplitudes of UV-
pile-upsignalsarethengivenbytherecursiveconvolutionfor- photons and heavy ions. The Pólya parameter b fits at a large
mula fk= f1∗fk−1(fork>1). value of 0.8–0.9, which enhances the likelihood P0 for emis-
With all fk known, γ˜ and b can be obtained from a fit of sionofnosecondaryelectronbyafactorof∼10comparedto
Eq.(3)tothedataofeachgivenCSRexperiment. ViaEq.(2), the Poisson-statistical case (b=0). The large value of b may
these parameters yield an independent experimental value for indicate that a large fraction of the sensitive detector aperture
the likelihood P0 and, hence, for the maximum possible de- wasirradiatedbytheArparticles—ascouldbeexpectedfrom
tector efficiency due to secondary electron statistics, 1−P0. an intense uncooled ion beam. The expected maximum pos-
Analytically,thenormalisationfactorCfromEq.(3)isequalto sible detection efficiency 1−P is hence reduced to 86(3)%.
0
(1−P0)−1.However,duetotheexperimentaldiscriminatorcut- Thesignalacquisitionthresholdcausesanother∼15%lossin
off, the measured pulse height distribution for heavy-particle efficiency,determinedbythefractionofthebest-fitmodeldis-
impactcannotbereliablyrenormalisedto1aslongasthebest- tribution (Eq. (3)) below discrimination level (cf. Fig. 3). The
fit distribution F is not known. For simplicity, C is therefore resultingtotaldetectionefficiencyforAratomsenteringthede-
treatedasanindependentfreeparameter. tectoristhusdeterminedtobeanaverage73(3)%. Geometric
Figure 8 illustrates the procedure: By UV-irradiation of the lossofparticlesduetothenarrowdetectoraperture—likelyto
detector,wemeasurethesingle-electronspectrum f1. Thepile- have occurred in all experiments reported here—is not taken
updistributions fk areobtainedbynumericalconvolution. The into account as it has been measured in the case of OH− only
statisticalsumspectrumFfromEq.(3)isthenfittedtothepulse (cf.Fig.4).
heightdistributionmeasuredforheavy-iondetection,asFig.8 At intermediate temperature of CSR, dissociative residual-
showsontwoexamples. gas collisions of N+ were observed. Significantly different
2
It should be noted that the method to extract b and γ˜ from pulseheightdistributionswerefoundforthechargedandneut-
the detector pulse height distribution is not new. In fact it is ralproductbeams(centraltworowsofTab.1),asreflectedby
thestandardwaytomeasuresecondaryelectronyieldsofions theverydifferentbest-fitvaluesofthesecondaryyieldγ˜.Partly,
impingingontosolids[39–44]. Normallyasecondaryelectron thedifferencemaybeduetodeflectionofthechargedfragments
detector with good energy resolution is used, so that the com- outofthestorageringplane,sothattheyhitadifferentareaof
ponents fk show up as clearly resolved peaks in the measured the converter cathode. In addition, the neutral product beam
pulse height spectrum. For the COMPACT MCPs, the pulse is believed to originate not only from dissociative collisions
heightspectrum f1 forasinglesecondaryelectron(i.e.forUV N+2 +X → N++N+X, but also from the competing disso-
irradiation of the detector) is found to be a monotonously de- ciativeelectrontransferreactionN++X →N+N+X+which
2
creasing exponential distribution. Unsurprisingly, the resolu- leadstotwoneutralNatomsinthefinalstate. Assumingakin-
tion with regard to electron multiplicity is therefore very bad. eticenergyreleaseintheorderof1eV,bothneutralfragments
Theaimofthisanalysisisnottoderivethesecondaryelectron may reach the converter cathode within a time interval in the
yieldγ˜buttoestimatetheamountofundetectedions. orderof10ns. Afastmulti-fragmentdetectordedicatedtoob-
The assumed MCP electron detection efficiency ε was kept servationofsuchreactionsinvolvingmultipleneutralproducts
fixedat0.6. Duetotheeffectofε onthemeanofthebinomial underCSRconditionsispresentlybeingset-up[45]. However,
distribution in Eq. (2), its choice correlates inversely to the fit the electronics of the COMPACT detector is not designed to
valueofγ˜.Theresultsforthedetectionefficienciesarehowever resolve such nearly-coincident double hits. In the present ex-
largelyindependentofthatchoice. periment, thetwoNatomsmaythusappearasasingle, larger
Inthefitsofthepulseheightdistributions1and2fromFig.5, anodepulse.
10