REVIEW OF SCIENTIFIC INSTRUMENTS VOLUME 70, NUMBER 5 MAY 1999

Photoelectron–multiple-photofragment coincidence spectrometerK. A. Hanold,a) A. K. Luong, T. G. Clements, and R. E. Continettib)

Department of Chemistry and Biochemistry, University of California San Diego,La Jolla, California 92093-0314

~Received 4 December 1998; accepted for publication 29 January 1999!

A new photoelectron–photofragment-coincidence spectrometer is described. Using a multiparticletime- and position-sensitive detector, this apparatus allows the study of dissociation processes ofnegative ions yielding three photofragments in coincidence with a photoelectron. The photoelectronspectrometer uses two detectors and works in time of flight mode, detecting 10% of thephotoelectrons with an energy resolution of 5% at 1.3 eV as shown in studies of thephotodetachment of O2

2 . A third detector is used for collection of multiple photofragments~up to8! in coincidence. This multiparticle detector uses a crossed-delay-line anode and fast timing signalsto encode the time- and position-of-arrival of multiple photofragments. The detector wasdemonstrated to record all three particles produced in a single three-body dissociation event,yielding an energy resolution of'15% DE/E at 0.7 eV in experiments on the three-bodydissociative photodetachment of O6

2 . © 1999 American Institute of Physics.@S0034-6748~99!02505-8#

traersa

tsnas,reuo

-incironeiqplodao

na

prsapan

-ur

is-

s ino-g aiesnce

on

by

llyionsis-o-

n ofout

inticthequeex-f

arrycoilso-

re-dy-ode-thebe

ho-s

I. INTRODUCTION

Although three-body chemical reactions among neuspecies play an important role in a number of atmosphand combustion chemistry systems, the detailed dynamicthese reactions have proven difficult to study. This is primrily due to the need to perform coincidence measuremendetermine the correlation between the distribution of interand kinetic energies among the products. In spite of thinumber of elegant experiments have characterized thbody photodissociation processes by uncorrelated measments of the momenta and internal energies of one or mof the photofragments.1–7 Significant insights into both concerted and sequential dissociation reactions have been gafrom these experiments. However, a general techniquepable of performing the coincidence measurements requus to reveal the partitioning of energy and recoil-angle crelations among the products has not been available fortral dissociation processes. We have developed a technfor the coincident detection of a photoelectron and multineutral photofragments, based on the dissociative phottachment of molecular and cluster anion precursors. Thisproach allows a new level of insight into three-body dissciation processes for molecules with a known interenergy.

This new apparatus allows experimental techniquesviously used in studies of three-body dissociation procesyielding charged atomic and molecular products to beplied to neutral species. Elegant experiments by Lutzco-workers on ion-impact-induced Coulomb explosions8 andby Eland and co-workers9,10on dissociative double photoionization have recently achieved complete kinematic meas

a!Present address: Syagen Technology, 1411 Warner Ave., Suite B, TuCA 92780.

b!Electronic mail: rcontinetti@ucsd.edu

2260034-6748/99/70(5)/2268/9/$15.00

licof-tolae-re-re

eda-edr-u-ueee-p--l

e-es-d

e-

ments of the heavy particle kinetic energy and angular dtributions in these processes. Stace and co-workers11 havealso recently carried out an interesting set of experimentwhich both positive ion and neutral fragments from the phtodissociation of argon cluster ions are detected, providindetailed view of the correlations between the recoil velocitof an ionic and three neutral products using the covariamapping technique.12 An additional level of detail is pro-vided in the kinematically complete photelectron–photoiexperiments on the double photoionization of D2 using thecold target recoil ion momentum spectroscopy pioneeredSchmidt-Bocking and co-workers.13

In our experiments, we seek to carry out kinematicacomplete studies of neutral three-body dissociation reactat low relative energies. The study of neutral three-body dsociation reactions by dissociative photodetachment of mlecular or cluster anion precursors represents an extensiothe use of negative-ion photodetachment to carrytransition-state spectroscopy. The seminal experimentsthis area relied on interpretation of the photoelectron kineenergy spectrum to reveal insights into the dynamics onneutral potential energy surface. Examples of this techniare the studies by Neumark and co-workers of hydrogenchange reactions14,15 and by Lineberger and co-workers oisomerization dynamics in small organic molecules.16,17 Wehave previously developed experimental techniques to cout coincidence measurements of the energy and reangles of a photoelectron and two photofragments in disciative photodetachment ~DPD! processes.18 Thesephotoelectron–photofragment coincidence experimentsveal the asymptotic correlation between the dissociationnamics and the Franck–Condon region accessed by phottachment. The apparatus described here allowsphotoelectron–photofragment coincidence experiment tocarried out on half-collision processes producing three ptofragments.

tin,

8 © 1999 American Institute of Physics

rodio-ufntis

swvint

eae

discfrfo

ess

le

inlti

au

de

-o

sesid

gl

an

bai

e-thdesInpctleseoa

pic

fsedforach

tial

p-n at

ofreiong-

nesTheis-rge

eed.

e oftorelayaled.lsbeeare

reu-ters

bertec-um-nldFor, able-ing

theen-

-linethein-

ad-to-r is

e-de-

rsorsed

rgyrom-

2269Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

Detailed measurements of three-body dissociation pcesses require a multiparticle detector capable of recorthe time- and position-of-arrival of all three particles in cincidence. While measuring two out of three particles sfices to determine momentum conservation if the fragmetion pattern is known, detection of all three particlesnecessary to reduce the possibility of false coincidencethe data and to reveal the fragmentation pattern in unknocases. Dissociation processes at beam energies of sekeV typically lead to all fragments arriving at the detectora range of a few hundred nanoseconds. To avoid bias indata set it is desirable to minimize any dead time or darea on the detector face. While false coincidences betwthe atomic or molecular photofragments can largely becarded by checking for conservation of momentum, coindence between the detected photoelectron and the photoments is determined by the overall detection efficiencythe apparatus. As discussed later, this requires operatingexperiment at high-repetition and low-signal rates. Thconsiderations impose significant constraints on the chodetection scheme. Nonetheless, there are several possibproaches to solving this problem.

While it is beyond the scope of this article to reviewdetail the approaches that have been taken to detect muparticles at laboratory energies of several keV,19 a brief re-view is in order. For both neutral and charged particlesthese energies, it is essential to amplify the signal by prodtion of secondary electrons. Microchannel plates~MCPs!provide the optimum means for doing this over a largetector area~4 cm diam field-of-view!, providing gain on theorder of 107 and detection efficiencies of'50%. Thus, thequestion rests on how to encode the time- and positionarrival of the electron clouds emerging from the MCP.

In our previous studies, two-body dissociation proceshave been studied using an anode split into two side-by-wedge-and-strip patterned anodes.20 This works well sincethe kinematics require that the two bodies for a sinbreakup appear on opposite sides of the center of mass~CM!.An important limitation of this setup is that the detector ha 6-mm-wide dead area that crosses the entire detector imiddle of the 40-mm-diam active area. The anode couldsegmented into three pieces, but the increased deadwould severely limit the detectable range of configurationsa three-body breakup.

A ‘‘pixellated’’ anode, composed of many discrete dtector elements, with dimensions small enough to providedesired resolution provides an ideal multiparticle anoCharge-coupled devices~CCDs! or charge-injection device~CIDs!, respectively provide one approach to this limit.general, however, CCDs and CIDs lack the data throughrates required in coincidence experiments and do not direencode the time-of-arrival information for a given particAmitay and Zajfman have recently developed a CCD-badevice that can detect the time- and position-of-arrivalseveral particles in coincidence, however, by using alternmeans of encoding the time of arrival.21 The primary limita-tion of this technique for coincidence experiments of the tyreported here is the low data readout rates characteristCCDs~'50 Hz!. X-ray astronomy22 and real-time x-ray dif-

-ng

-a-

inn

eral

heden-

i-ag-rtheeenap-

ple

tc-

-

f-

se

e

stheerean

e.

utly.dfte

eof

fraction experiments23 have motivated the development otrue time- and position-sensitive pixellated detectors baon application-specific-integrated circuits developedhigh-energy particle physics detectors. In these devices, epixel will be ‘‘fully instrumented,’’ capable of recordingcharge and time information. However, at the desired sparesolution~,100mm over a 4 cmfield-of-view!, the numberof pixels required and the complexities of circuit develoment and data acquisition make this an expensive solutiothe present time.

An alternate approach involves the use of a numberparallel wires forming a wire plane. Two wire planes that aorthogonal to each other can be used to give informatabout the two-dimensional positions of the particles impining on the detector. Position measurement with wire plahas been accomplished by at least three techniques.charge division technique joins neighboring wires with restors, and each end of the resistor chain is coupled to chasensitive electronics.24 By measuring the fraction of chargat each end of the resistor chain, the position is determinThis technique requires charge integration and is capabldetermining the position of only one particle per resischain. The second technique couples each wire to a dline.25 By measuring the relative time of arrival of the signat each end of the delay line, the position is determinMultiple hits in one delay line result in a number of signafrom each end of the delay line, however, which maydifficult to disentangle.26 Both of these techniques have thadvantage that a small number of electronics channelsrequired for the position encoding~two channels per resistochain or delay line!. The third technique of reading out thwire planes is to instrument each wire. A number of Colomb explosion experiments on small molecules and clushave used detectors of this type.27–29 Since the time andcharge of the signal on each wire is measured a large numof data acquisition channels are required. This type of detor has the advantage of being able to measure, a large nber of particles arriving virtually simultaneously, but oftehas resolution limited by the spacing of the wires. It shoube noted that the wire planes need not be physical wires.example, in one of the Coulomb explosion experimentsthree-wire-plane anode was fabricated on a single dousided circuit board by photoetching with interconnects usthrough-plated holes in the circuit board.27

The multiparticle detector we have built is based ondelay-line approach. The anode is composed of four indepdent quadrants, each composed of a crossed-delayanode.30,31 This represents a compromise with respect tonumber of particles detectable from a given event, but matains high spatial and temporal resolution. With each qurant instrumented to detect two particles, up to eight phofragments can be recorded for each event. This detectodiscussed in more detail in Sec. II B.

In the following sections, the photoelectron-multiplphotofragment coincidence spectrometer is described intail. In these experiments, photodetachment of a precunegative ion and energy analysis of the photoelectron is uto prepare a neutral complex with a known internal enethat subsequently dissociates. The photoelectron spect

smheM

heticexat

crighnauaioP

thuab

aassle

erearoexnt

ef 4ets

rcarrgioedd

hte

c

sin

I A.re-atic-

mhis

serec-be

reerids

ten-ummul-ns,m-

tec-

lity

tack

eter-erely

een

ctedcor-p-

two-

inec-and

oni-

lti-

2270 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

eter consists of two equivalent detectors on opposite sidethe laser–ion beam interaction region. Each detector is tiand position-sensitive which allows for determination of tphotoelectron kinetic energy and recoil angle in the Cframe. Atomic or molecular photofragments recoil out of tfast ion beam, and are measured using the new multiparphotofragment detector. Following a discussion of theperimental technique, photoelectron and photofragment cbration experiments are presented, along with results onthree-body DPD of O6

2 at 532 nm.

II. EXPERIMENT

Construction of an apparatus capable of detecting, inincidence, a photoelectron and three photofragments onating from a single dissociation event requires higefficiency detection schemes. Carrying out photoelectrophotofragment coincidence experiments in a fast ion beprovides a practical route to coincidence detection of mtiple photofragments due to the kinematic constraint ofphotofragments to a small cone about the ion beam directThis, coupled with the high detection efficiencies of MCbased detectors for fast neutrals,32 allows all the photofrag-ments produced in a dissociation event to be collected wisingle detector. Fast neutral beam methods have previobeen applied to the study of dissociation dynamics, with fneutral beams and photofragments producedcharge-exchange33–35 or photodetachment processes.20,36 Ef-ficient detection of photoelectrons with high-resolution mesurements of photoelectron kinetic energy in a fast beposes a significant challenge, however, due to the neceof correcting for the electron-recoil-angle-dependent Doppshift induced by the fast ion beam.37 We have previouslydescribed a fast-ion beam photoelectron spectromet38

demonstrating that this limitation can be overcome by msuring the energy and recoil angle for each photoelectThis allowed photoelectron–photofragment coincidenceperiments on the DPD of molecular or cluster anions itwo photofragments and a free electron to be studied.39 Theapparatus described here has been optimized to increasdetection efficiency of the photoelectrons by a factor oand has a new multiparticle detector allowing measuremof the time- and position-of-arrival of three photofragmenin coincidence.

The fast-ion beam apparatus is composed of ion souacceleration, and time-of-flight regions in addition to the pticle detectors comprising the coupled translational eneand photoelectron spectrometers shown in Fig. 1. Thebeam line is similar to that described in Ref. 40. A pulsmolecular beam of pure O2 operating at 1 kHz is crossewith a 1 keV electron beam, forming O2

2 , O42 , O6

2 , andother clusters. After acceleration to 4 keV, a time-of-flig~TOF! mass spectrometer is used to select the anion of inest. The ion beam is defined by two 1 mm apertures~1!spaced 41 cm apart, with the first one at a distance of 192from the ion source. The fast, mass-selected ion beamintersected with a pulsed laser~;100 ps pulse widthNd:YAG! in an interaction volume;0.5 mm diameter by 1mm long ~2!. Photodetached electrons are measured u

ofe-

le-li-he

o-i--–ml-lln.-

aslysty

-mityr

,-

n.-

o

the

nt

e,-yn

tr-

mis

g

the photoelectron detectors discussed in detail in Sec. IAfter the ion beam exits the photoelectron spectrometer,sidual ions are deflected out of the beam with an electrostdeflector~not shown!, ensuring that only neutral photofragments reach the multiparticle detector~7!, described in detailin Sec. II B.

A. Photoelectron detector

The laser–ion beam interaction region is 50 mm frothe entrance grid of the two opposed electron detectors. Tshort flight path dictates that the temporal width of the lapulse must be less than 0.5 ns for high-resolution TOF sptra and that the laser–ion beam interaction region mustsmall. Photoelectrons ejected along the axis of the field-fdrift regions pass through 81% transmission tungsten g~3!. A triple layer magnetic shield~4, Eagle-AA! enclosesthe entire electron spectrometer. Electrostatic patch potials are reduced by practicing standard ultrahigh vacutechniques and coating exposed surfaces with graphite esion. The first element of the detector is an accelerating lewhich focuses the electrons that pass through the 100-mdiam entrance aperture onto the 40-mm-diam electron detor. In this lens, electrons are accelerated to;500 V yieldinga detection efficiency of about 50%. Three image-quaMCPs ~Philips G12-46DT! are used in a Z-stackconfiguration41 ~5! providing a gain of.107 for detectedphotoelectrons. The secondary electrons from the MCP simpinge on a wedge-and-strip-patterned anode~6!.42,43 Thisanode uses charge division among three conductors to dmine thex,y position for a single photoelectron per lasshot. The time of arrival is recorded by using a capacitivcoupled 350 MHz amplifier~Ortec VT120! attached to thewedge conductor. The wedge-and-strip anode has bshown under laboratory conditions to be capable of,100mm local resolution,,500 mm global resolution, and,0.5ns timing resolution.38

Each photoelectron detector accepts electrons ejeinto a 110° cone. This large acceptance angle requiresrection for both the flight path of each electron and the Do

FIG. 1. Schematic diagram~to scale! of the photoelectron–multiple-photofragment spectrometer. The ion beam enters from the left through1-mm-diam defining apertures~1!. The triple-layer magnetic shield is indicated by~4!. The laser is directed out of the page, crossing the ion beamthe interaction region~2!. Photoelectrons cross the entrance grid of the eltron spectrometer~3! and are accelerated and focused onto the time-position-sensitive photoelectron detectors~5,6!. After exiting the magneticshields, residual ions are electrostatically removed from the beam and mtored using a microchannel-plate ion detector~not shown!. Photofragmentsrecoil out of the beam over a 104 cm flight path and impinge on the muparticle detector~7!. The multiparticle detector is mounted on a bellows~8!so it can be translated in the plane perpendicular to the page.

on

sah

thecleo

at.

theldee.erag

ivao

anoxgythatbyoao

ith

dth

ecsnas

seflo

onhele

stoThid

d

to

atn.

to-ilystop.theps

nis

ing-

r-thetorhot

or-nt isarged by

d inf

d-

ulti-ayersers 2

ordi-

2271Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

pler correction required to remove the angle-dependent ctribution of the ion beam velocity to obtain the CM electrokinetic energy~eKE!.38 The flight path for each electron igeometrically calculated from the measured position ofrival at the detector using the focusing factor of the lens. Tlaboratory velocity is calculated from this path length andmeasured TOF. The field-free flight path of the photoeltrons varies from 50 mm for electrons arriving at the middof the detector to 85 mm for electrons arriving at the edgethe detector.

The focusing lens was designed using the electrostlens simulation programSIMION44 to be as simple as possibleAs shown in Fig. 1, the lens consists of two elements,ground planes of the inside surface of the magnetic shi~4! and the entrance grid~3!, and the 500 V surface of thcylindrical MCP mount~5!. Note that the plate behind thMCP mount and the anode~6! is also at ground potentialThe radial size of the front plate of the MCP mount detmines the focal properties of the lens. This dimension woptimized usingSIMION, showing that a one-to-one mappinof electron ejection angle to the measured position-of-arrfor each electron energy is preserved. The lens does, hever, exhibit a chromatic aberration in the form ofelectron-energy dependent focusing factor, which is apprmated by a linear function of the laboratory electron enerThe calculated laboratory position of the photoelectron at100-mm-diam entrance grid determines both the flight pand laboratory velocity. The CM velocity is then obtainedvector subtraction of the ion beam velocity from the electrlaboratory velocity. Each photoelectron detector subtendsproximately twice the solid angle of the fast-ion beam phtoelectron detector previously reported.38 Thus, the photo-electron detection efficiency is a factor of four higher wthis new apparatus.

B. Multiparticle detector

The multiparticle detector used in these experimentstermines the position of a charge cloud emerging fromMCPs by time division on a delay line.30,31This represents acompromise with respect to the number of particles detable per shot, but maintains high spatial and temporal relution. In Fig. 2, the layout of the multilayer anode is showschematically. There are four independent quadrants, ewith an active area of 22322 mm. The serpentine delay lineare positioned outside the field of view. Figure 3 showcross section of the multilayer anode, which has four layand is constructed from a copper-clad ceramic-doped Tesubstrate~RT-Duroid, Rogers Corp.!. Layer 1 is a groundplane behind the entire anode. Layer 2 has fingers indirection while layers 3 and 4 are divided into fingers in tperpendicular direction. The fingers on layers 2 and 4 colcharge for measurement of thex and y positions, and areconnected to equally spaced points along two separatepentine delay lines. The fingers on layer 3 are connectedground plane that electronically isolates layers 2 and 4.charge cloud that hits the anode is large relative to the wof each individual finger~0.47 mm! and is distributed amongseveral neighboring fingers. The signal is then introduce

n-

r-ee-

f

ic

es

-s

lw-

i-.eh

np--

e-e

t-o-

ch

arsn

e

ct

er-ae

th

at

neighboring points along the delay line and propagateseach end, producing pulses of'3 ns full width half maxi-mum ~FWHM!. The time difference between the signalseach end of the delay line is linearly related to positioUsing constant-fraction discriminators and a time-amplitude converter allows the time difference to be easmeasured, using one signal as the start and one as theGiven the 20–25 ns propagation time of the signal acrossserpentine delay lines, timing resolution on the order of 35is required to achieve the spatial resolution of'70 mm.

An advantage of carrying out high-resolution timing oshort delay lines is that the dead time for the detectorsmall. We have taken advantage of this by instrumenteach delay line with a custom-built two-hit time-toamplitude converter,45 allowing the measurement of two paticles per quadrant. The minimum dead time imposed byelectronics is 7–10 ns. This gives the multiparticle detecthe ability to record up to eight photofragments per laser sin a case of ideal kinematics~two particles per quadrant, witha 10 ns temporal separation!. This multiplicity is required toenable detection of three particles arriving in nearly anyder and orientation. The charge generated by each evealso measured on each quadrant. Particles whose chclouds straddle the border of two quadrants are accepte

FIG. 2. Schematic layout of the four-quadrant multiparticle anode usethese experiments. Four independent 22322 mm quadrants composed ocrossed charge-collection fingers compose the active area. Thex,y delaylines, indicated byx1 ,y1 for quadrant 1, are positioned outside of the fielof-view.

FIG. 3. Schematic view of a cross section of the active area of the mparticle detector anode. Labels 1, 2, 3, and 4 are photoetched copper lseparated by layers of insulator. Layers 1 and 3 are ground planes. Layand 4 consist of copper fingers oriented along thex andy axes. These fingerscollect and conduct the charge to the serpentine delay line for each conate.

pfoiv

wevihayntchri

r-anital

e

Comt ue

ate

hen,of-ergeithach

per

r-ternede

wles

fentsaskle.

ad-uad-the

eto

theway. 5ap-

temn o

adse

t

de-dataer aholesm,

2272 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

electronics for both quadrants. Particles whose time andsition of arrival are nearly identical on adjacent quadrantsa given event can be handled by analyzing the charge dsion between the two quadrants.

C. Data acquisition

The data acquisition system for the experiment is shoin Fig. 4. The central clock for the experiment is the pulsphotodetachment laser, with the previous laser shot proing the start time for production of the pulsed ion beam. Tpulsed ion beam switching is controlled by a digital delgenerator~SRS DG535!. Photoelectron and photofragmedata acquisition and timing is initiated by the photodetament laser pulse as recorded by a fast photodiode triggeone channel of a quad constant fraction discriminator,~CFD,Tennelec TC454!. This provides the logic signal for measuing the TOF of the photoelectron and the photofragmentsgenerating the gates required for the single time-to-digconverter~TDC LeCroy 3371! and the four eight-channeanalog-to-digital converters~ADC LeCroy 3351! used indata acquisition. Custom-built charge-sensitive preamplifishaping amplifier~PA/SA! combinations and fast~300 MHz!timing amplifiers~FA! are used to amplify the signal fromthe detector anodes.45 The gates for these ADC and TDmodules are generated by a second SRS DG535 and a hbuilt gate generator. All the data are transferred to a PC ato 1 kHz acquisition rates using a computer-automated msurement and control~CAMAC! interface. The multiparticledetector has a total of 16 delay-line signal outputs. Theseprocessed by eight two-hit CFD/time-to-amplitude conver

FIG. 4. Schematic diagram of the CAMAC-based data acquisition sysThe laser provides the clock for generation of the ion beam and initiatiodata collection. The preamplifiers~PA!, shaping amplifiers~SA!, and fast-timing amplifiers~FA! for one of the photoelectron detectors and one qurant of the multiparticle~QXDL! detector are shown. The signals from theamplifiers are processed by constant-fraction discriminators~CFDs!, time-to-amplitude converters~TACs! and analog-to-digital converters~ADCs! asdiscussed in the text prior to transfer to a personal computer throughCAMAC interface.

o-ri-

ndd-e

-ng

dl

r/

e-p

a-

rer

~TAC! modules followed by 16 ADC channels to encode tposition-of-arrival of two particles per quadrant. In additiothere are eight TDC channels for recording the time-arrival of two particles per quadrant, four ADCs for thcharge in each quadrant, and one ADC for the total chafor that event. This is a total of 29 channels. Coupled wthe two wedge-and-strip anode photoelectron detectors, erecording three charges and one electron time-of-flightevent, the experiment has 37 data acquisition channels.

The imaging properties and resolution of the multipaticle detector have been determined using a pinhole-patresolution mask placed directly on top of the MCPs followby illumination with ultraviolet photons. The width of thpeaks in the resulting image show a localx,y position reso-lution of '70 mm. The top frame of Fig. 5 shows the raimage of the pinhole pattern, which consists of rows of hothat are separated by 4 mm with a 2 mmspacing within eachrow. The fourth row from the bottom has 25-mm-diam holes,while all other rows have 50-mm-diam holes. The borders othe quadrants are shown on Fig. 5, as well as line segmdrawn between the actual position of the pinhole in the mand the detected position of the events from that pinhoThere are a few pinholes that lie on the border of two qurants and the events from these pinholes appear in both qrants. The root-mean-square position error away fromquadrant borders is,100 mm. A simple power law correc-tion is applied within 3 mm of the quadrant border. Thencoded positions in this region are compressed duecharge spilling into the neighboring quadrant, causingapparent centroid of the charge cloud to be pushed afrom the border of the quadrant. The bottom frame of Figshows the pinhole pattern after this correction has been

.f

-

he

FIG. 5. Pinhole test pattern showing performance of the multiparticletector. The upper frame shows the raw data with lines that connect theto the real position of the pinhole. The lower frame shows the data aftnumerical correction near the quadrant borders has been applied. Theare 50mm in diameter with the exception of the fourth row from the bottowith 25 mm holes.

he

teenerin

thgl

w

t9

conrdsehoalsoicc

nc

-rafcogcenaheraisli

een

ecie

rea

actenksumar-sershesnlyKE

-esofe-thede-

gy

torpytion

tictively

glesionTheeri-

cameel torum

2273Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

plied. The root-mean-square position error is now,50 mmand is.150 mm for only a few cases near the edge of tfield-of-view and the quadrant borders.

D. Coincidence count rates

The acceptable count rate for this experiment is limiby the product of the detection efficiencies for the coincidparticles. The multiparticle detector is an essentially zdead area and a minimum dead time device. The nomdetection efficiency for each of the neutral fragmentsf n andthe photoelectronf e is '50% with an MCP-based detector.32

If we assume that 100% of the neutral fragments hitmultiparticle detector and 20% of the electron solid anDVe is collected with a transmission efficiencyTe of 81%,the average number of true coincidences per laser shotbe

ntrue5P~1!• f n3f eDVeTe . ~1!

If we take P(1), thePoisson probability for a single evenper shot to be 0.09, Eq.~1! gives a signal count rate of 0.Hz when the experiment is run at 1 kHz.

A consideration of sources of background and falseincidences is also important in the design of this experimeMultiple-event false coincidences can be eliminated by haware and software charge-discrimination techniques. Convation of momentum between the atomic or molecular ptofragments provides an important means of discarding fcoincidences. The most fundamental type of noise in a ccidence experiment of this type, however, is false coindences between an uncorrelated photoelectron and threerelated photofragments. The rate of these false coincideis given by

nFalse5P~2!•2•~ f n!3•~12 f n!3

•~ f eDVeTe!

•~12 f eDVeTe!, ~2!

where P(2)50.016 is the Poisson probability for two detachment events occurring per laser shot when the avenumber of events per shot is 0.1. The factor of 2 accountsthe two indistinguishable ways of getting a false coincidenof this type when two DPD events occur. The terms proptional to f n take into account the probability for detectinthree correlated fragments out of the six fragments produin two dissociative events. Likewise, the terms proportioto f e relate to the probability of detecting only one out of ttwo photoelectrons. This gives a false coincidence countof 0.04 Hz when the experiment is run at 1 kHz. This'4% of the true coincidence count rate and will be neggible for most purposes.

III. RESULTS

A. Photoelectron spectra

The operating characteristics of the photoelectron sptrometer were experimentally verified by photodetachmof a 4 keV beam of O2

2 . Figure 6, frame A, shows theelectron laboratory kinetic energy spectrum for photoeltrons that recoil perpendicular to the ion beam directionthe laboratory frame. For these electrons, Doppler broad

dtoal

ee

ill

-t.-r--e

n-i-or-es

georer-

dl

te

-

c-t

-nn-

ing from the large ion beam velocity is not significant. Theis, however, a noted shift of the laboratory energies tolower value, relative to the combs on frame C, due to the fthat the correction for the ion beam velocity has not bemade to yield the CM eKE in this plot. Clearly defined peaare visible in this frame. Frame B shows the eKE spectrfor an identical width gate on electron detector positions pallel to the ion beam direction and perpendicular to the labeam. The Doppler broadening present in these data waout the structure in the photoelectron spectrum so that oone broad feature is seen. Frame C shows the final espectrum obtained for O2

2 at 532 nm for all electrons independent of position. Determination of this spectrum involvremoval of both Doppler broadening and the distributionfield-free flight paths for the photoelectrons in the largsolid-angle photoelectron detectors. Two combs showexpected positions of the photoelectron peaks for phototachment of O2

2(2Pg)1hn→O2(3Sg2),O2(1Dg)1e2 at 532

nm.46 Examination of the peak widths shows the enerresolutionDE/E'5% at eKE51.3 eV.

B. Three-body photofragment translational spectra

To evaluate the performance of the multiparticle detecfor carrying out photofragment translational spectroscomeasurements, initial experiments on the photodissociaand dissociative photodetachment of O4

2 and O62 at 532 nm

were performed. The photoelectron–photofragment kineenergy correlations observed in the three-body dissociaphotodetachment of O6

2 at 532 nm have been previousreported.47

As discussed by DeBruijn and Los,33 it is straightfor-ward to determine the mass, kinetic energy, and recoil anof the two photofragments produced in a binary dissociatusing a time- and position-sensitive detection scheme.equations for calculating these quantities from the expmental observables will not be duplicated here.20,33Knowing

FIG. 6. Photoelectron spectra of O22 at 532 nm. Frame A shows the kineti

energy spectrum for electrons recoiling perpendicular to the beam. In frB, the spectrum for photoelectrons that have velocity components parallthe beam is shown. Finally, in frame C, the CM photoelectron spectafter Doppler correction is shown.

tim, irgfra

hosenetioto

tiereioroaciclnt

omre

he

oin

oum

ta

ntta

besianwns

ned.sion

hisoilM

.olu-eul-ing

thetionom

dlo

e

so--two

ho-

2274 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

the parent particle mass and velocity and measuring theand position-of-impact of two coincident photofragmentsis straightforward to determine the CM translational enerelease, recoil angles, and relative masses of the photoments.

In the case of three-body dissociation, where all the ptofragments are recorded in coincidence as in the preexperiment, a similar analysis can be carried out. GardGraff, and Kohl discussed fast-beam three-body dissociaexperiments and presented equations for determiningphotofragment mass from the two-dimensional projectionthe CM velocities onto a photofragment detector.36 In thepresent experiment, the three-dimensional recoil velociare recorded. This makes the kinematic analysis of a thbody dissociation in terms of linear momentum conservatin the CM frame straightforward. Figure 7 illustrates the pcess of a three-body dissociation originating in the intertion region, the subsequent paths of travel of each partand the final projection of the three-dimensional event othe time- and position-sensitive detector. In the analysisthe data, the positions of the particles in the laboratory fraat the time of arrival of the first particle on the detector acalculated using the known velocity of the CM and tflight-time differences of the detected particles.

Each three-body dissociation process is checked for cservation of linear momentum. A CM reference frame usthe largest particle velocity vector as the positivez axis isadopted, as shown in Fig. 8. Rotation of the three recvectors into this frame allows conservation of momentand mass to be easily applied to calculate the mass ofthree photofragments. The principles of conservation of mand momentum in the CM frame@Eq. ~3!# are then applied todetermine the photofragment masses

m11m21m35M, p11p21p350. ~3!

The variablesmi are the masses of the three photofragmeM is the parent mass, andpi are the photofragment momen

FIG. 7. Kinematic diagram showing a three-body dissociation and thetected position of particle impact on the detector. The incident beam veity is the velocity of the CM,VCM , and the CM recoil velocities of the threphotofragments are shown asV1 , V2 , andV3 .

etyg-

-ntr,n

hef

se-n--e,ofe

n-g

il

hess

s,

in the CM frame. The momentum of each particle candecomposed into components along each of the Carteaxes, and a system of three equations in the three unknom1 , m2 , andm3 can be solved to yield, for example,

m35MS V1zV2y

V3yV2z2V3zV2y2V3yV1z1V1zV2yD . ~4!

The masses of the other particles can then be determiThe mass spectrum of O6

2 obtained from this technique ishown in Fig. 9, and shows a nominal mass resolutm/Dm'2.5. In the dissociation of O6

2 , two fragments arefast and one is slow, barely recoiling out of the beam. Tlimits the mass resolution, since it is based on the recdistance of the particles relative to the distribution of Cvectors. Fortunately, in the DPD of O6

2 at 532 nm, energeticconsiderations dictate that three O2 molecules are producedSignificant improvements in the photofragment mass restion will require reducing the distribution of velocities of thCM or, in other words, the size of the ion beam at the mtiparticle detector. This may be achieved in the future usappropriate collimating ion optics.

Given the photofragment masses and CM velocities,translational energy released in the three-body dissociacan be calculated by simply adding up the contributions freach particle

e-c-

FIG. 8. Kinematic diagram of the CM frame chosen for three-body disciation processes. The longest CM velocity vectorV1 is chosen as the positive z axis. The polar angles between this velocity vector and the otherfragments in the plane of the breakup are shown asu12 andu13 .

FIG. 9. Photofragment mass distribution recorded for the dissociative ptodetachment of O6

2 at 532 nm. The FWHM of the peak is'13 amu.

asosl

defr

at

eenn

mdirenO

ly

epu

esrabyths

Thddeaensns

aym-

them-ri-se-

dsyn-notcanplehent isdyn-mss.

so-mo-ent,ki-tralso-rag-ndex-to-idetionmea-pon-

nd

ula

s inarein

tedar-

2275Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

ET5 12m1V1

21 12m2V2

21 12m3V3

2. ~5!

An illustration of the translational energy distributions mesured with this detector for both two- and three-body disciation processes is shown in Fig. 10. This shows trantional energy distributions for both O4

2 and O62 recorded at

532 nm. The broad feature at 0.4 eV for O42 and 0.35 eV for

O62 is due to the two- and three-body dissociative photo

tachment processes in these two species leading to aelectron and separated O2 molecules. The narrow peaks0.8 eV for O4

2 and 0.7 eV for O62 result from the photodis-

sociation processes O421hn→O2(1Dg)1O2

2(2Pg ,v50)and O6

21hn→O2(1Dg)1O22(2Pg ,v50)1O2(3Sg

2), re-spectively. These processes are observable with the ntively biased multiparticle detector due to photodetachmof some of the nascent O2

2 by a second photon. Examinatioof the width of the feature at 0.7 eV in the O6

2 spectrumindicates an energy resolutionDE/E'15% at 0.7 eV.

The product angular distributions can also provide iportant insights into the dissociation mechanism. For thesociative photodetachment of O6

2 , an analysis of the TOF okinetic energy distributions for each particle in a given evshows that in general there are two fast and one slow2produced. Thus, it appears that this dissociation is neartwo-body dissociation between two O2 molecules with thethird playing the role of a spectator. Following Fig. 8, thCM recoil angles between the photofragments can be simcalculated from the vectors. In Fig. 11, the angular distribtions of the photofragments in the CM frame of the largrecoil velocity are shown. These are generated frombinned angular distributions by weighting each bin1/sinu. As shown in Fig. 11, the second fastest particle inCM frame peaks sharply in the opposite direction of the faest particle, constituting a pseudo two-body dissociation.third ‘‘spectator’’ particle, while also distributed beyonu5p/2 in the CM frame, is seen to have a much broaangular distribution. Given that the fastest particle wassumed to lie along thez axis, and that all three particles havthe same mass, this is a consequence of momentum covation. A detailed analysis of these angular distributio

FIG. 10. Photofragment translational energy distributions for the molecphotofragments produced in the dissociative photodetachment of O4

2 andO6

2 at 532 nm.

--

a-

-ee

ga-t

-s-

t

a

ly-tw

et-e

rs-

er-,

perhaps in terms of a classical trajectory simulation, mprovide some insights into not only the dissociation dynaics of O6 , but perhaps even the structure of O6 .

There is considerably more detail to be obtained instudy of three-body dissociation dynamics. In a recent coprehensive review article, Maul and Gericke define two pmary three-body dissociation mechanisms: concerted andquential processes.1 Further classifications of concerteprocesses into the subcategories of synchronous and achronous concerted reactions, according to whether ormultiple bond-breaking processes occur simultaneously,also be made. These distinctions are not made in the simdata analysis we have carried out to date. In the case wthe breakup is sequential, a more complex treatmenneeded to fully describe the kinematics of the three-bodissociation process.48 As more systems are studied, the sesitivity of the data to the various dissociation mechaniswill be assessed with the help of Monte Carlo simulation

As shown by experiments on the DPD of O62 , this ap-

paratus allows studies of the dynamics of three-body disciation processes induced by photodetachment of stablelecular and cluster anion precursors. Photodetachmcoupled with coincident detection of the photoelectronnetic energy, provides a novel route to production of neumolecules or clusters with a known internal energy. Disciation of these energy-selected species into three photofments can then be studied using a multiparticle time- aposition-sensitive detector described above. In the future,amination of photoelectron–photofragment and phofragment–photofragment angular correlations should prova direct measure of the time scale for three-body dissociaprocesses. The next system we hope to extend thesesurements to is the three-body association reaction ressible for the formation of ozone: O1O21M→O31M. Theinsights into three-body dynamics of neutral molecules a

r

FIG. 11. Photofragment angular distributions for the two slower particlethe CM frame of the fastest particle recoil velocity. These distributionsobtained from the raw histogrammed data by weighting each bin by 1/suand are normalized to an area of unity. The solid curve,N(u12), correspondsto the angular distribution of the intermediate velocity particle. The dotcurve, N(u13), corresponds to the angular distribution of the slowest pticle.

aac

ofanor.ekaDndane-oro

ul

A

es

s,

er,

u-

.

cl.

th-

ov-

hys.e,

ev.

E.

ci.

v.

E

2276 Rev. Sci. Instrum., Vol. 70, No. 5, May 1999 Hanold et al.

clusters gained from these experiments should provideimportant test of the theories of three-body chemical retions that are currently being developed.49

ACKNOWLEDGMENTS

This work was supported by the Air Force OfficeScientific Research under Grant No. F49620-96-1-0220DURIP Grant No. F49620-97-1-0255. One of the auth~A.K.L.! is supported by AFOSR AASERT Grant NoF49620-97-1-0387. R.E.C. is a Camille Dreyfus TeachScholar, an Alfred P. Sloan Research Fellow, and a PacFellow in Science and Engineering. The authors thankO.H.W. Siegmund, R. Raffanti, and J. Hull of SiegmuScientific Inc. for assistance with the manufacture of theode and design and assembly of the amplifiers and doubltime-to-amplitude converters for the multiparticle detectThe contributions of Jacqueline Kessler to constructioncircuitry used in the ion beam apparatus are also gratefacknowledged.

1C. Maul and K. H. Gericke, Int. Rev. Phys. Chem.16, 1 ~1997!.2P. M. Kroger and S. J. Riley, J. Chem. Phys.67, 4483~1977!.3X. Zhao, W. B. Miller, E. J. Hintsa, and Y. T. Lee, J. Chem. Phys.90,5527 ~1989!.

4C. E. M. Strauss and P. L. Houston, J. Phys. Chem.94, 8751~1990!.5T. Gejo, J. A. Harrison, and J. R. Huber, J. Phys. Chem.100, 13941~1996!.

6S. W. North, A. J. Marr, A. Furland, and G. E. Hall, J. Phys. Chem.101, 9224~1997!.

7D. Stranges, X. Yang, J. D. Chesko, and A. G. Suits, J. Chem. Phys.102,6067 ~1995!.

8U. Werner, K. Beckord, J. Becker, and H. O. Lutz, Phys. Rev. Lett.74,1962 ~1995!.

9J. H. D. Eland,VUV Photoionization and Photodissociation of Moleculand Clusters, edited by C. Y. Ng~World Scientific, Singapore, 1991!, pp.297–343.

10S. Hsieh and J. H. D. Eland, J. Phys. B30, 4515~1997!.11P. Jukes, A. Buxey, A. B. Jones, and A. J. Stace, J. Chem. Phys.14, 5803

~1998!.12L. J. Frasinski, K. Codling, and P. A. Hatherly, Science240, 1029~1989!.13R. Dorner et al., Phys. Rev. Lett.81, 5776~1998!.14R. B. Metz, S. E. Bradforth, and D. M. Neumark, Adv. Chem. Phys.81, 1

~1992!.15S. E. Bradforth, D. W. Arnold, D. M. Neumark, and D. E. Manolopoulo

J. Chem. Phys.99, 6345~1993!.16K. M. Ervin, J. Ho, and W. C. Lineberger, J. Chem. Phys.91, 5974

~1989!.17P. G. Wenthold, D. A. Hrovat, W. T. Borden, and W. C. Lineberg

Science272, 1456~1996!.

n-

ds

r-rdr.

-hit.fly

18R. E. Continetti, Int. Rev. Phys. Chem.17, 227 ~1998!.19J.-C. Brenot and M. Durup-Ferguson, Adv. Chem. Phys.82, 309 ~1992!.20R. E. Continetti, D. R. Cyr, D. L. Osborn, D. J. Leahy, and D. M. Ne

mark, J. Chem. Phys.99, 2616~1993!.21Z. Amitay and D. Zajfman, Rev. Sci. Instrum.68, 1387~1997!.22J. G. Timothy, Opt. Eng.~Bellingham! 24, 1066~1985!.23P. Datteet al., Nucl. Instrum. Methods Phys. Res. A391, 471 ~1997!.24G. Knapp, Rev. Sci. Instrum.49, 982 ~1978!.25M. B. Williams and S. E. Sobottka, IEEE Trans. Nucl. Sci.36, 227

~1989!.26J. H. D. Eland and A. H. Pearson, Meas. Sci. Technol.1, 36 ~1990!.27A. Belkacem, A. Faibis, E. P. Kanter, W. Koenig, R. E. Mitchell, Z

Vager, and B. J. Zabransky, Rev. Sci. Instrum.61, 945 ~1990!.28W. Koenig, A. Faibis, E. P. Kanter, Z. Vager, and B. J. Zabransky, Nu

Instrum. Methods Phys. Res. B10/11, 259 ~1985!.29J. Becker, K. Beckord, U. Werner, and H. O. Lutz, Nucl. Instrum. Me

ods Phys. Res. A337, 409 ~1994!.30M. Lampton, O. Siegmund, and R. Raffanti, Rev. Sci. Instrum.58, 2298

~1987!.31P. G. Friedman, R. A. Cuza, J. R. Fleischman, C. Martin, D. Schimin

ich, and D. J. Doyle, Rev. Sci. Instrum.67, 596 ~1996!.32A. Muller, N. Djuric’, G. H. Dunn, and D. S. Belic’, Rev. Sci. Instrum.57,

349 ~1986!.33D. P. DeBruijn and J. Los, Rev. Sci. Instrum.53, 1020~1982!.34W. J. van der Zande, W. Koot, J. R. Peterson, and J. Los, Chem. P

126, 169 ~1988!; J. H. M. Beijersbergen, S. Kornig, W. J. van der ZandP. G. Kistemaker, and J. Los, J. Phys. Chem.95, 9059~1991!.

35H. Helm and P. C. Cosby, J. Chem. Phys.86, 6813~1987!.36L. D. Gardner, M. M. Graff, and J. L. Kohl, Rev. Sci. Instrum.57, 177

~1986!.37H. Hotop and W. C. Lineberger, J. Phys. Chem. Ref. Data14, 731~1985!.38K. A. Hanold, C. R. Sherwood, M. C. Garner, and R. E. Continetti, R

Sci. Instrum.66, 5507~1995!.39K. A. Hanold, M. C. Garner, and R. E. Continetti, Phys. Rev. Lett.77,

3335 ~1996!.40C. R. Sherwood, K. A. Hanold, M. C. Garner, K. M. Strong, and R.

Continetti, J. Chem. Phys.105, 10803~1996!.41O. H. W. Siegmund, K. Coburn, and R. F. Malina, IEEE Trans. Nucl. S

32, 443 ~1985!.42C. Martin, P. Jelinsky, M. Lampton, R. F. Malina, and H. O. Anger, Re

Sci. Instrum.52, 1067~1981!.43O. H. W. Siegmund, R. F. Malina, K. Coburn, and D. Werthimer, IEE

Trans. Nucl. Sci.NS-31, 776 ~1984!.44D. A. Dahl, J. E. Delmore, and A. D. Appelhans, Rev. Sci. Instrum.61,

607 ~1990!.45Siegmund Scientific, 15 Sequoia Lane, Walnut Creek, CA 94595.46M. J. Travers, D. C. Cowles, and G. B. Ellison, Chem. Phys. Lett.164,

449 ~1989!.47K. A. Hanold, A. K. Luong, and R. E. Continetti, J. Chem. Phys.109,

9215 ~1998!.48M. Lavollee and H. Bergeron, J. Phys. B25, 3101~1992!.49R. T. Pack, R. B. Walker, and B. Kendrick, J. Chem. Phys.109, 6701

~1998!.