IEEE Transactions on Nuclear Science, Vol. NS-32, No. 1, February 1985

PRELIMINARY EVALUATION OF RCA C83036EPROTOTYPE PHOTOMULTIPI rER

Chu-Chung Lo and Branko Leskovar

University of CaliforniaLawrence Berkeley Laboratory

Berkeley, California 94720 U.S.A.

PEAK OUTPUT CURRENT MEASUREMENTS

Characteristics of the prototype RCA C83036E 52 mmdiameter photomultiplier have been measured. The re-sults of the measurements of the gain, dark current,photocathode quantum efficiency, peak output current,electron transit time, and output pulse rise time as afunction of voltage between anode and cathode are giv-en. Furthermore, single photoelectron time spread,multiphotoelectron time resolution, pulse-height spec-trum and afterpulse time spectrum were measured andare discussed.

INTRODUCT ION

The RCA C83036E 52 mm-diameter photomultiplier isa modified version of the RCA 8575. B3th types havesimilar electron multiplier structures, consisting of12 in-l ine electrostatically-focussed copper beryll iumdynodes, the input electron optics and a semitranspar-ent cesium-potassium-antimony photocathode depositedon the entrance window. The structure is contained ina pyrex envelope. The difference between C83036E and8575 is that C83036E has improved characteristics,particularly with respect to its peak output currentcapability. The spectral response of the photocathodeextends from 300 nm to 650 nm and peaks at 390 nm.Measurements made on three photomultipliers gave anaverage quantum efficiency of 28 %.

GAIN AND DARK CURRENT MEASUREMENTS

Gain and dark current measurements were made withthe system described in Ref. 1. The voltage dividernetworks used for these measurements are shown inFigs. 1 and 2. Figure 1 shows the voltage divider forhigh gain applications while that for high peak cur-rents is shown in Fig. 2. Figure 3 shows the gain anddark current characteristics of C83036E photomultipli-er as a function of voltage applied between the anodeand cathode using the high gain voltage divider. With2000 V applied between the anode and cathode, the gainwas 4.0 x 107, while the dark current was approxi-mately 1.6 x 10-8A. The gain and dark current as afunction of voltage between the anode and cathode us-ing the high-peak current divider is given in Fig. 4.With this divider and with 2000 V applied between theanode and cathode the gain and dark current was 107and 2 x 10-9A, respectively.

QUANTUM EFFICIENCY MEASUREMENTS

A calibrated RCA 8850 photomultiplier with bialka-li photocathode was used as the standard for the quan-tum efficiency measurements. The light source wasadjusted to yield an output signal of 10 nA from the8850 with 500 V between the photocathode and anode.With the same light level setting, the C83036E wasconnected as a diode and positioned with the photo-cathode exactly at the same distance from the lightsource as the photocathode of the 8850, and the outputsignal was measured. The average quantum efficiencywas found to be 28 % at 410 nm.

The peak output current of the C83036E photomulti-plier was measured with a pulsed mercury light sourcecapable of emitting enough photons per pulse to satu-rate the photomultiplier.

Calibrated neutral density filters were used toattenuate the light pulse intensity in steps for themeasurements. The light pulses were 2.6 ns wide, witha repetition rate of 60 pulses per second. The re-sults are plotted in Fig. 5. The peak linear outputcurrent was 250 mA and 160 mA for the operating volt-age of 2400 V and 2000 V, respectively. The voltagedivider for high current applications was used forthis measurement.

ELECTRON TRANSIT TIME MEASUREMENT

A description of the measuring system has beenpublished elsewhere by the authors2. An LED lightsource served to produce the light pulses and theelectrical pulses driving the LED were utilized asreference pulses. An adjustable air line was used tobring the LED driving pulse and reference pulse intocoincidence on the oscilloscope, thus establishing azero time reference. The C83036E photomultiplier wasthen put in place and the delay of the output signalmeasured.

The electron transit time as a function of anode-cathode voltage is plotted in Fig. 6. At 2250 V ofapplied voltage, the electron transit time was approx-imately 30 ns for three tubes measured.

SINGLE PHOTOELECTRON PULSE RESPONSE

The true characteristics of the pulse response ofa photomultiplier can best be obtained from singlephotoelectron pulses. Since dark noise pulses aremostly due to single photoelectrons, they can be usedto show the single photoelectron pulse response. How-ever, wh re a higher pulsing frequency is needed,either a low-level dc light source, or a very-shortlight pulser can be used to produce more single photo-electron pulses. The true pulse response of a photo-multiplier can be observed only if the divider socketassembly is properly designed to reliably transmit theoutput pulse to an oscilloscope or other monitoringdevice. Improper matching of the output circuit ofthe photomultiplier may result in reflections whichwill distort the results of an experiment.

To ensure proper matching in the socket assembly,two 5 nF capacitors, one from each of the last twodynodes were connected to the output coaxial cableshield with the short -t leads possible. By doingthis, the last two dynoaes look l ike a ground plane tothe fast pulses. This lowers the impedance of the out-put structure of the photomultiplier, thus achieving abetter match to the 50 Ohm external load. The pulseresponse under this condition is shown in Fig. 7. The10-90 % risetime of a single photoelectron pulse was

0018-9499/85/0002-0360$01.00 ©1985 IEEE

ABSTRACT

360

361

found to be 2 ns with 2000 V on high gain voltagedivider. Figure 8 shows the pulse risetime as a func-tion of anode-cathode voltage using divider for photo-multiplier high current applications.

SINGLE PHOTOELECTRON TIME SPREAD MEASUREMENT

The system described in Ref. 3 was used for thetime spread measurement. The light pulses were gener-ated by an avalanche li ght emitting diode and had awidth of 200 ps or less, FWHM. The system resolutionwas in the order of 30 ps, FWHM.

Figure 9 shows two single photoelectron time dis-tributions spaced 4 ns apart. With full photocathodeillumination and with 200 ps light pulses the averagesingle photoelectron time spread using either the highgain or the high current voltage divider with a gainof 1 x 108 was 860 ps, FWHM.

MIJLTI PHOTOELECTRON TIME RESOLUT ION

Measurement of the multiphotoelectron time resolu-tion was made with a mercury light pulse generatorcapable of producing thousands of photoelectrons perpulse from the photocathode of the photomultiplier.The number of photoelectrons per pulse was calculatedfrom the output pulse width and amplitude and knowingthe gain of the photomultipliers at the operatingvoltage. Figure 10 shows the plot of the time resolu-tion as a function of the number of photoelectrons perpulse from one to ninety photoelectrons. The timeresolution of the single photoelectron pulses was2.35 ns, FWHM. This number is the result of the lightpulse width and the single photoelectron time resolu-tion (i.e., 860 ps). With 90 photoelectrons per pulse,the time resolution of the C83036E is tapered down toapproximately 0.2 ns, FWHM.

PULSE-HEIGHT SPECTRUM MEASUREMENT

Pulse-height spectrum was measured by illuminatingthe full photocathode with a light emitting diodewhose light output intensity was controlled by varyingthe amplitude of the driving pulse. The photomulti-plier output pulses were stretched before they wereprocessed by the pulse-height analyzer. The photo-electron peaks were calibrated with a RCA 8850 photo-multiplier which could resolve 1, 2, 3 or 4 photoelec-tron peaks. The pulse height spectrum of the C83036Eis shown in Fig. 11. With an operating voltage of2250 V which gives a gain of 108 the dark pulsecount was found to be:

16 photoelectrons= 134 counts per second

1/8 photoelectron

Both the pulse-height spectrum and dark pulse cur-

rent were very similar using high gain or high pulssecurrent divider at the same gain.

AFTERPULSE MEASUREMENT

Although contemporary fabrication and activationtechniques have reduced afterpulses from most photo-multipliers to a point where they are rarely import-ant, for some applications afterpulses still may in-troduce serious error. For example, in photon count-ing systems for subnanosecond fluorescence lifetime

measurements5, photomultipl i er afterpulses can gen-erate small amplitude late spurious peaks in the sam-ple fluorescence profile6. Also, in Deep UnderwaterMuon and Neutrino Detection (DUMAND) systems which usea large number of photomultipliers in a three-dimen-sional array, a significant amount of afterpulsingcould introduce serious error in measuring particletrajectories.7, 8

In general, the afterpulses are produced as a re-sult of the ionization of residual gases, such asHe+, Ht, Nt and CO+, in the volume between the photo-cathode and the first dynode. The positive ions form-ed are accelerated toward the photocathode by thefocusing electric field. On impact these ions liber-ate up to five secondary electrons from the cathodewhich produce an afterpulse signal. The afterpulsesgenerally occur from 20 ns to several microsecondsafter the main pulse. The time of occurrence of theafterpulses can be closely correlated with the atomicmass-to-charge ratio of the residual gas inside theglass envelope and the potential distribution in thevicinity of the cathode. This phenomena was studiedsystematically by several authors9 who introducedtrace amounts of various gases into experimentalphotomultipliers. Subsequently, in further work1lO-3,the physical origins of afterpulses were investigated,particularly with respect to the afterpulses whichresult from the diffusion of helium through the photo-multiplier glass envelope. Although the helium ispresent in small concentration in ambient air, it issufficient so that its atoms, which can permiate read-ily through the glass envelope and cause afterpulses.Other phenomena may also cause afterpulsing, such asdynode fluorescence, electrical fields over the expos-ed glass of the envelope, etc. In order to measurethis characteristic, a time spectrum must be taken ofthe anode output pulses. Measurements were made usingthe photomultiplier high gain voltage divider networksuggested by RCA. With 2250 V applied between theanode and cathode, the average gain was approximately108, while the dark current was 7 x 10-8A.

The system shown in Fig. 12 was used to measureafterpulses. A pulse generator was used to drive alight-emitting diode, type XP 21, which produced lightpulses for the photomultiplier. The trigger pulsefrom the pulse generator was delayed and shaped andthen used as a start pulse for a time-to-amplitudeconverter (TAC) . The output pulse of the photomulti-plier was used as the stop pulse for the TAC afterbeing processed by a constant fraction discriminator.In order to count the afterpulses which came immedi-ately after the main photomultiplier pulse, the mainoutput pulse (marked by symbol B in Fig. 12) was de-layed after the start pulse. The time at which themain photomultiplier pulse occurred was taken as timezero. However, in order to count the afterpulseswhich occurred significantly later in time and with avery low count rate the trigger pulse from the pulsegenerator was purposely delayed to come after the pho-tomultiplier main pulse so that an output pulse fromthe TAC would only occur when the photomultiplier gen-erated an afterpulse. The output of the TAC was thenrecorded and displayed on a pulse-height analyzer. Inorder to look for pulses mnany microseconds after themain pulse (the timing range of the TAC being set ac-cordingly) the operating frequency of the test systeinwas quite low.

Under single photoelectron counting at a rate of100 KHz, afterpulses were detected in the time rangeof 19-20 ns at a count rate of 140-280 cps after themain output pulses in all three tubes, in other wordsabout one afterpulse for 500 main pulses. Afterpulses

362

were also detected in the time ranges of 620-650 nsand a 2-4 ps. The afterpulsing count rate in thesetime ranges was below 100 cps, or less than one after-pulse per thousand main pulses.

The same behavior was observed when three photo-electron pulses were used at a rate of 10 KHz. Oneout of the three photomultipliers was excessivelynoisy that afterpulsing could not be measurable beyondthe 150 ns time range. No afterpulse was detectedbeyond 4.5 ps up to 80 ps.

Figure 13 shows a time distribution of outputpulses in the time interval 0 - 150 ns after themain photomultiplier pulse under a 100 KHz single pho-toelectron counting rate. The first distribution atthe beginning of the spectrum is the main output pulseof the photomul ti pl i er. The second distributionrepresents the afterpulses which occur 19.8 ns afterthe main pulse at a count rate of 254 cps.

Figure 14 shows a time distribution of outputpulses over the range 0.2ps-4.5vs under 100 KHz singlephotoelectron counting rate. The second distributionat the beginning of the spectrum represents the after-pulses which occurred at about 60 ns after the mainpulse. They have a count rate less than 100 cps. Thebroad peak of afterpulses from 2.3ps to 3.8vs afterthe main photomultiplier pulse had a count rate alsoless than 100 cps.

CONCLUS IONS

Important characteristics of prototype photomulti-plier C83036E have been measured and are discussed.Generally, measurements were made using voltage divid-er networks suggested by RCA for photomultiplier highgain and for high output current operation. The meas-

urement technique and a description of the system usedfor afterpulse time spectrum measurements are given indetail. Afterpulse measurements were made using thehigh gain voltage divider circuit.

The afterpulses in the 19-20 ns-range were mostprobably generated by He+ ions formed between thefirst and second dynode, striking the first dynode andcausing the secondary emission. This conclusion isbased on calculations using the geometry and fields inthe first dynode region of the photomultiplier. The19-20 ns time range increased when the voltage betweenthe photocathode and the first dynode/focusing elec-trode was decreased because of the change in the elec-tric field intensity and distribution.

The afterpulses occurring in the 620-650 ns rangewere attributed to He+ ions which are created in thespace between the photocathode and the first dynodeand the first and second dynode. This conclusion is

based on the time-of-flight of the helium ion from its

point of origin to the photocathode.

The afterpulses occurring in the 2.3-3.8ps rangecould be attributed to O+ ions or N+ ions with

large mass-to-charge ratio, which may be present in-

side the glass envelope as a result of outgassing of

the photomultiplier components during processing and

operation.

ACKNOWLEDGEMENTS

The authors would like to express their apprecia-tion to RCA Corporation, Solid State Division, ElectroOptics and Devices, Lancaster, Pennsylvannia, for the

loan of the photomultipliers.

This work was performed as part of the program ofthe Electronics Research and Development Group of theDepartment of Instrument Science and Engineering andwas supported by Lawrence Berkeley Laboratory and theU.S. Department of Energy under Contract No. DE-ACO3-76SFOO098. Reference to a company or product namedoes not imply approval or recommendation of the pro-duct by the University of California or the U.S.Department of Energy to the exclusion of the othersthat may be suitable.

References

1. C.C. Lo, P. Lecomte and B. Leskovar, PerformanceStudies of Prototype Microchannel Plate Photomul-tipliers, IEEE Trans. Nucl. Sci. NS-24, No. 1,pp. 302-311, February 1977.

2. C.C. Lo and B. Leskovar, A Measuring System forStudying the Time Resolution Capabilities of FastPhotomultipliers, IEEE Trans. Nucl. Sci., NS-21,No. 1, pp. 93-104, 1974.

3. B. Leskovar and C.C. Lo, Single Photoelectron TimeSpread Measurement of Fast Photomultipliers, Nucl.Instr. and Methods, 123, No. 1, pp. 145-160, 1975.

4. C.C. Lo and B. Leskovar, Afterpulse Time SpectrumMeasurement of RCA 8850 Photomultiplier, IEEETrans. Nucl. Sci., NS-30, No. 1, pp. 445-450, 1983.

5. B. Leskovar, C.C. Lo, P.R. Hartig, K.H. Sauer,Photon Counting System for Subnanosecond Fluores-cence Lifetime Measurements, Rev. of Sci. Instr.,47, No. 9, pp. 1113-1121, 1976.

6. P.R. Hartig, K.H. Sauer, C.C. Lo and B. Leskovar,Measurement of Very Short Fluorescence Lifetimesby Single Photon Counting, Rev. of Sci. Instr.,47, No. 9, pp. 1112-1129, 1976.

7. B. Leskovar, Photomultiplier Characteristics Con-siderations for the Deep Underwater Muon and Neu-trino Detection System. Proceedings of the 1980Deep Underwater Muon and Neutrino Detection SignalProcessing Workshop, A. Roberts, Ed. Published byHawaii DUMAND Center, University of Hawaii, Hono-lulu, pp. 21-40, 1980.

8. C.C. Lo and B. Leskovar, Evaluation of the NewGeneration RCA 8854 Photomultiplier, IEEE Trans.Nucl. Sci., NS-29, No. 1, pp. 184-190, 1982.

9. G.A. Morton, H.M. Smith, R. Wasserman, Afterpulsesin Photomultipliers, IEEE Trans. Nucl. Sci.,NS-14, No. 1, pp. 443-448, 1967.

10. P.B. Coates, The Origins of Afterpulses in Photo-multipliers, J. Physics D: Applied Physics, 6,pp. 1159-1166, 1973.

11. W.C. Paske, He+Afterpulse in Photomultipl ier:Their Effect in Atomic and Molecular LifetimeDeterminations, Rev. Sci. Instr., 45, No. 8,pp. 1001-1003, 1974.

12. D.F. Barllett, A.L. Duncan, J.R. Elliott, After-pulses in a Photomultiplier Tube Poisoned withHelium, Rev. Sci. Instr., 52, No. 2, pp. 265-267,1981.

13. C.C. Lo and B. Leskovar, Afterpulse- Time SpectrumMeasurement of RCA 8850 Photomultipl ier, IEEETrans. Nucl. Sci., NS-30, No. 1, pp. 445-450, 1983.

363

OUTPUTNOTES: UNLESS OTHERWISE INDICATED

1. RESISTORS 1W 5% CARBON COMPOSITION2. CAPACITORS 1KV VOLTAGE RATING

XBL 842-634

Fig. 1. Schematic diagram of voltage divider networkfor photomultiplier high gain applications.

OUTPUT

NOTES: UNLESS OTHERWISE INDICATED1. RESISTORS 1W 5% CARBON COMPOSITION2. CAPACITORS 1KV VOLTAGE RATING

XBL 842-635

Fig. 2. Schematic diagram of voltage divider forphotomultiplier high current applications.

364

10 - - -.z~~~~~~~~

T jir o8 X

Lo210: 10 --- io-l10 lo25

0z

10~~~~~~~~~~~~1

10 l-500 1000 1500 2000 2500SUPPLY VOLTAGE BETWEENANODE AND CATHODE (V)

XBL 842-636

Fig. 3. Gain and dark current as a function of volt-age between photocathode and anode usinghigh gain voltage divider.

w

'-3'cnz

a:

O O~~~~~~~~~~~~~~~~~~~-)00 1500 2000 2500VOLTAGE BETWEEN CATHODE AND ANODE (V)

XBL 842-639

Fig. 6. Electron transit time as a function of volt-age between photocathode and anode usingdivider network for photomultiplier highgain applications.

C.0

* il

1000 1500 2000 25CSUPPLY VOLTAGE BETWEENANODE AND CATHODE (V)

a.

I10-8z

LU

.i10-9 Q

; lo-loLL

0z

0

lo-12

XBL 842-637

Fig. 4. Gain and dark current as function of voltagebetween photocathode and anode using highcurrent divider network.

RCA C83036E

2

E0It)

2ns/divXBB 842-986

Fig. 7. Single photoelectron pulse response usinghigh gain voltage divider.

365

E

zw

0

24

M0

w

0

0

z

w 1I

LIGHT INTENSITY (ARBITRARY UNIT)XBL 842-638

Fig. 5. Peak linear anode output current as a func-tion of light pulse intensity.

-;, 2.5

w2

wcncr

-J0 2.

ICJw00z< 1.5.

U

2000 2500 3000ANODE-PHOTOCATHODE VOLTAGE (V)

XBL 842-640

Fig. 8. Single photoelectron pulse risetime as afunction of anode-cathode voltage dsingphotomultiplier high current applicationsdivider.

-j

wUz

z

= 2x103-

a-

I-

z

0

UI.0

In 0-

z I-o4ns S0SINGLE PHOTOELECTRON

TIME SPREAD = 0.86ns, FWHMXBB 842-987

Fi g. 9. Single photoelectron time spread of C83036Ewith full photocathode illumination.

I I I I- I I I I I

RCIA Cr23n3arF

2 46

NUMBER OF

8 2 4 6 810 1O(

PHOTOELECTRONSXBL 842-64 1

Fig. 10. Multiphotoelectron timeC83036E as a function ofphotoelectrons per pulse.

-lwzz

o 1x103-LU.a.U)I.-z

0C.)U-0

m 0-Z

PULS

resolution ofthe number of

DrA C3R2na3E

i 2 345E HEIGHT PHOTOELECTRON

Fig. 11. Pulse height spectrum ofmultipli er.

EQUIVALENTSXBB 842-988

C83036E photo-

_J START PULSE

1J T 2r,-,r-STOP PULSEA B

Fig. 12. System block diagram for measuring time spectrum of afterpulse.

aRA R23nfl

TIME (ns)

-j

36E ui22

Ulx* c 1X 103-

*I=L

Ca

2ml-z0

m1m

20:2

XBB 842-989

XBL 81-7657A

TIME (jus) 4.5XBB 842-990

Fig. 13. Afterpulse spectrum of anode output pulse in Fig. 14. Afterpulse spectrum of anode output pulse inthe 0-150 ns time range. the 200 ns - 4.5 pIs time range.

366

-p

-SZ20

-j0w

n

11w

LIGHT PULSE WIDTH=2.2nsPULSE REPETITION FREQUENCY= 6OHz

I I I I I

1

-jwzz

0 7x1O5-'U

a-1

z

0U.)00: 1-

z 0

12

RC'A C83036ER