ARTICLE IN PRESS
Nuclear Instruments and Methods in Physics Research A 595 (2008) 169–172
Contents lists available at ScienceDirect
Nuclear Instruments and Methods inPhysics Research A
0168-90
doi:10.1
� Corr
Univers
E-m
journal homepage: www.elsevier.com/locate/nima
Timing and cross-talk properties of BURLE multi-channel MCP PMTs
S. Korpar a,b, P. Krizan c,b,�, R. Pestotnik b, A. Stanovnik d,b
a Faculty of Chemistry and Chemical Engineering, University of Maribor, Maribor, Sloveniab J. Stefan Institute, Ljubljana, Sloveniac Faculty of Mathematics and Physics, University of Ljubljana, Ljubljana, Sloveniad Faculty of Electrical Engineering, University of Ljubljana, Ljubljana, Slovenia
a r t i c l e i n f o
Available online 9 July 2008
Keywords:
Microchannel plate photomultipliers
Cherenkov detectors
02/$ - see front matter & 2008 Elsevier B.V. A
016/j.nima.2008.07.022
esponding author at: Faculty of Chemistry
ity of Maribor, Maribor, Slovenia.
ail address: [email protected] (P. Krizan).
a b s t r a c t
We report on studies of the two types of BURLE multi-anode micro-channel plate (MCP) PMTs; one with
64 channels and 25mm pores and the other with four channels and 10mm pores. Possible applications of
these tubes are for RICH detectors and time-of-flight counters. We have investigated their timing
properties and cross-talk. The origin of the cross-talk between the channels was investigated by
studying the correlations between signals on neighbouring pads, their amplitudes and time of arrival.
We have also studied the influence of cross-talk on timing and spatial resolution.
& 2008 Elsevier B.V. All rights reserved.
d0
γγ
α
β
d1
e-
1. Introduction
For the upgrade of particle identification systems at B-factoriesthree different types of Cherenkov detectors are proposed: aTime-Of-Propagation (TOP) counter, a proximity focusing aerogelRICH, both at the Belle experiment [1] and a focusing DIRC as anupgrade of BaBar DIRC [2]. All these counters require photondetectors that work in high magnetic fields of about 1.5 T, havepad sizes of a few mm and timing resolution around 50 ps.
A most promising candidate for the photon detector is amicrochannel plate photomultiplier (MCP-PMT). In this device,the dynode structure is replaced by MCP amplification. One of thephoton detector candidates is the 64 channel Burle 85011 MCP-PMT tube [3]. Two types of the tube are being tested, one with 64channels ð8� 8Þ and 25mm pores, and the other with fourchannels ð2� 2Þ and 10mm pores. The 64 channel tube hasalready been tested as photon detector in aerogel RICH counter[4]. It was also shown that it can serve for time-of-flightmeasurement [5]. The group at SLAC, which is working on theupgrade of the DIRC detector [2], has also investigated the timingproperties of the BURLE/Photonis MCP PMT and has obtained verypromising results [6]. In the present contribution we discuss ourmeasurements of timing and cross-talk properties of 64 and fourchannel versions.
In order to better understand the parameters that determinethe position resolution and the timing resolution, we show inFig. 1 the physical processes that lead to a signal on the anodepads of a MCP-PMT.
ll rights reserved.
and Chemical Engineering,
Different angles a at which the photoelectron exits thephotocathode, as well as possible backscattering of the photo-electron on the microchannel plate, contribute to the broadeningof both position and timing distributions. In addition, the signalcan be detected by more than one pad due to a charge sharingeffect.
Fig. 1. Physical processes that lead to a signal on the anode of a MCP-PMT. The gap
between the photocathode and MCP is around 6 mm; the corresponding potential
difference applied during the tests was 200 V.
ARTICLE IN PRESS
Fig. 2. Apparatus for the measurement of single photons with the MCP-PMT. Left: laser head with filters and light guide. Right: exit end of light guide with focusing lens,
beam splitter, reference PMT and MCP-PMT detector in a dark box.
533.2 / 65
P1
P2
P3
P4
P5
P6
coun
tsco
unts
0
1000
2000
3000
4000
5000
0
0
1000
2000
3000
4000
5000
0
TDC [ps]
TC max ch. 1
TC max ch. 3
TDC [ps]
2000 4000
146.7
230.2
549.2
40.40
-2.509
4605.
χ2 / ndf
620.6 / 65
P1
P2
P3
P4
P5
P6 145.0
234.3
547.9
38.95
-1.841
4678.
χ2 / ndf
2000 4000
TC max ch. 2
632.8 / 65
P1
P2
P3
P4
P5
P6
coun
ts
792.4 / 65
P1
P2
P3
P4
P5
P6
coun
ts
0
1000
2000
3000
4000
5000
0
0
1000
2000
3000
4000
5000
2000 4000
0 2000 4000
TDC [ps]
TC max ch. 2
TDC [ps]
χ2 / ndf
χ2 / ndf
144.9
233.8
4890.0
-1.525
37.17
553.4
4685.0
-1.389
38.04
564.4
232.2
137.1
Fig. 3. Corrected TDC distributions of all four channels of the four-channel MCP-PMT. Distributions are fitted with a sum of two Gaussian functions.
S. Korpar et al. / Nuclear Instruments and Methods in Physics Research A 595 (2008) 169–172170
ARTICLE IN PRESS
104
S. Korpar et al. / Nuclear Instruments and Methods in Physics Research A 595 (2008) 169–172 171
2. Experimental set-up
As a photon source we used a picosecond PiLas laser with ablue (406 nm) or a red (635 nm) head. The laser light was firstattenuated by neutral density filters and guided into a light tightbox along an optical fibre, with the far end attached to a computercontrolled 2D stage (Fig. 2). The light was focused by a biconvexlens onto the photocathode of a MCP-PMT to a spot size ofs � 10mm. The repetition rate of light pulses was 2 kHz and theintensity was reduced to detect a single photon in more than 90%of signal events.
Signals from the anodes were amplified by an Ortec FTA820amplifier. After amplification, each signal was split into twobranches; one for the timing and the other for the signalamplitude measurement. The timing signal was discriminatedby a Phillips 708 module (NIM) and used as a stop signal. The timeof the pulse was measured by the Kaizu Works KC3781A TDC(CAMAC) with 25 ps per channel. A common start signal wasdetermined by the laser control unit. The signal pulse hight wasregistered in a CAEN V965 charge sensitive ADC (VME).
1
10
102
103
x [mm]
−500
0
500
1000
1500
2000
2500
3000
3500
20
1
10
102
103
x [mm]
TD
C [
ps]
−500
0
500
1000
1500
2000
2500
3000
3500
20
TD
C [
ps]
30 40 50 60 70 80
30 40 50 60 70 80
Fig. 4. Position ðxÞ dependence of the time distribution of single photon signals for
a 2� 2 channel MCP-PMT (top) and for an 8� 8 channel MCP-PMT (bottom).
3. Results
From a plot of the pulse time versus pulse height, we haveobtained corrections for the pulse arrival time. After applyingthese corrections, the timing distribution has a main peak withrms of about 40 ps for the four-channel tube, and about 45 ps forthe 64-channel version. In both cases there is a characteristicallylong tail due to the backscattering of photoelectrons (Fig. 3).Contributions of electronics and laser to the width of the mainpeak are s � 15 and 12 ps, respectively.
The position dependence of the time distributions of signalsfrom both the four-channel and the 64-channel MCP-PMT areshown in Fig. 4. It is seen that signals at pad boundaries and
x [mm]
coun
ts
1
10
102
103
20
1
10
102
103
20x [mm]
coun
ts
30 40 50 60 70 80
30 40 50 60 70 80
Fig. 5. Individual pad count rates as a function of position ðxÞ across two channels
of the four channel MCP-PMT (top) and across eight channels of the 64 channel
MCP-PMT (bottom).
ARTICLE IN PRESS
1
10
102
103
x [mm]
AD
C1/
(A
DC
1+A
DC
2)
0
0.2
0.4
0.6
0.8
1
49
225.1 / 157
Constant
Mean
Sigma
0
20
40
60
80
100
120
140
160
50
x [mm]
CO
UN
TS
50 51 52 53 54 55 56
52 54
χ2 / ndf138.5
52.09
0.3203
Fig. 6. Position ðxÞ dependence of the signal fraction appearing on the left pad with
the boundary between the pads seen at x ¼ 52 mm (top figure). Distribution of
events for which the pulse is equally distributed between the two neighbouring
pads (bottom figure).
S. Korpar et al. / Nuclear Instruments and Methods in Physics Research A 595 (2008) 169–172172
especially at the edges of the sensitive surface tend to besomewhat delayed.
The origin of the cross-talk between the channels wasinvestigated by studying the correlations between signals on
neighbouring pads, their amplitudes and time of arrival. Fig. 5represents the position dependence of the count rate of individualpads for both types of MCP-PMT (four-channel and 64-channel).The contributions of cross-talk are seen to be relatively small butto extend about 12 mm beyond the pad boundary in accordancewith the estimated maximum range of photoelectrons back-scattered on the multichannel plate.
Fig. 6 shows the position dependence of the fraction of thetotal signal that appears on one anode pad for the four-channelMCP-PMT. In the boundary region that is about 3 mm wide chargesharing can be used to improve the position resolution. Thedistribution on the bottom of Fig. 6 shows the position distribu-tion of the number of events for which the pulse is equally dividedbetween the two neighbouring pads. The width of this distribu-tion mainly comes from the spread of photoelectron impact pointon the MCP due to its initial energy (� 1 eV for blue light). Thiscould be improved by reducing the distance between thephotocathode and MCP, and by increasing the potential differencebetween them.
To conclude we note that the observed properties of the PMTare in very good agreement with a simple model based on Fig. 1. Adetailed discussion of the model and some further measurementsare beyond the scope of the present contribution, and will be thesubject of a forthcoming publication.
References
[1] K. Abe, et al., in: S. Hashimoto, M. Hazumi, J. Haba, J.W. Flanagan, Y. Ohnishi(Eds.), Letter of Intent for KEK Super B Factory, KEK report 2004-04 hhttp://belle.kek.jp/superbi.
[2] C. Field, et al., Nucl. Instr. and Meth. A 553 (2005) 96.[3] BURLE 85011 data sheet: hhttp://www.burle.com/cgi-bin/byteserver.pl/pdf/
85011-501.pdfi.[4] S. Korpar, et al., in: Proceedings of the IEEE Nuclear Science Symposium, Rome,
Italy, October 17–22, 2004;P. Krizan, et al., Nucl. Instr. and Meth. A 567 (2006) 124.
[5] P. Krizan, et al., Nucl. Phys. B Proc. Suppl. 172 (2007) 227.[6] C. Field, et al., Nucl. Instr. and Meth. A 518 (2004) 565;
J. Va’vra, et al., Nucl. Instr. and Meth. A 572 (2007) 459.