Transcript Study and characterization of a multianode photomultiplier

Characterization of the
Hamamatsu R8900-M16
Multianode Photomultiplier Tube (PMT)
Paul Mekhedjian
Department of Physics
University of California, Santa Cruz
Department of Energy – INFN Summer 2007 Studentship
Why do we care?
High granularity Pb-Sci Fibers
calorimeter
KLOE readout
4.4 x 4.4 cm2 read by photomultipliers
1 module 52x25 cm2 read by 60
photomultipliers.
Increasing granularity by a factor 16
using HAMAMATSU photomultipliers.
 Better particle identification
 Less merging probability for pair of
clusters
 Useful for neutron detection.
Device Aspect
Hamamatsu R8900-M16
Window material: Borosilicate glass
Arrangement and Type: 4 x 4 grid
Number of channels: 16 (each 5.7x5.7mm2)
Effective Window Area: 23.5x23.5mm2
Photocathode material: Bialkali
Spectral response range: 300 to 650 nm
Compact form and design practical
for assembly and use in calorimeters!
Device scheme: 1. Photons strike photocathode
Signal formation process:
16 anodes signal
2. Electrons are produced via photoelectric
effect and directed to the first dynode past
the focusing mesh
3. The dynodes are made of materials with very
low bandgap energies, which produce
additional electrons upon collision
4. Electrons are directed and oriented from the
photocathode to the multianode by a simple
electric field generated by the dynodes
5. The signal is finally collected at the anode
and its gain is dependent on the total number
of dynodes and the applied voltage.
Connector
Socket
12 (last dynode signal output)
Analog Circuit Amplification
Each anode/DY12 output can be connected
by a LEMO cable to the oscilloscope
Amplifier Circuit:
Connector socket
Summary of Talk







Quality of Lunch?!
Linearity of response
Signal shape(Rise Time, Fall Time, Full Width at
Half Maximum “FWHM”)
Transit Time & Transit Time Spread
Relative Gain vs. High Voltage
Response in Channel Homogeneity
Channel-Channel Crosstalk
Apparatus & Setup
Oscilloscope
50Ω Terminator
Low Voltage Power
Supply for Analog
Amplifiers
2D Micrometer Slide
High Voltage Power Supply
Laser / Photon Source
Laser Pulse
PMT Input
Laser Control Unit
Socket
Apparatus Continued…
A Typical Signal
• This is what a typical signal would look like on the
oscilloscope once the high voltage power was turned on.
Signal on the oscilloscope
From a single anode…
Rise time: Time from 10% to 90% of the signal amplitude
Fall time: Time from 90% to 10% of the signal amplitude
Area (integral) of the signal is proportional to the collected charge at a particular anode
What does this dial do?
Prior to crosstalk measurements, we discovered that
varying the dial amplitude gives interesting results…
Amplitude Response
The signal response of the photomultiplier tube is linear for a certain range of
the laser’s dial amplitude. It then enters a breakdown region (past ~900) where
the trend follows an exponential or asymptotic behaviour.
Charge Response
The response for the channel’s charge on the anode follows a nearly identical
trend!
Linearity of Both
Amplitude and Charge
a = 2.73 +/- 0.01mV
a = 0.0093 +/- 0.0001 nV*s
b = -13 +/- 8mV
b = 0.751 +/- 0.075 nV*s
• After investigating only the dial range of 500 to 850, we have found the linear
response region.
Charge Offset
a = 0.0093 +/- 0.0001 nV*s

It should be noted that the
parameters of the linear fit
produce a constant which has
physical meaning.
b = 0.751 +/- 0.075 nV*s
 y = 0.0093*x + 0.751
(i.e. Charge = 0.0093*Amplitude + 0.751)
• We know that if a light source has no amplitude,
the photomultiplier tube cannot produce a value of
charge.
•Thus, based on the fit, there is a charge offset
(given by the parameter b. This parameter will be
useful in future slides…
Signal shape vs amplitude
Slope: 0.026ns/mV
It should be noted that some signal
properties vary more than others
with an increase in amplitude.
Slope: 0.008ns/mV
Slope: 0.003ns/mV
It is important for the uniformity of the calorimeter response with respect to
particles of different energy that the signal shape is independent (more or less)
from the signal amplitude.
Transit Time Shift
So honestly… How long does it take from a photon to leave the laser, hit
the photocathode, photomultiply, and then leave the PMT from the anode?
The time transit is what this quantity is known as and we wished to see
how much this quantity differed from channel to channel…
This plot illustrates how other
channels deviate from a
reference channel (shown as a
white box without a number).
This is crucial because it
affects the time resolution
of a potential KLOE
calorimeter upgrade with
multianode PMTs.
Charge Characteristics
• Since charge is collected at the anode, it is interesting to see how this charge
fluctuates.
• We can also use this information to study a quantity known as TTS (time transit
spread) as a function of the number of photoelectrons (sQ/Q). This is where one may
use the charge offset previously mentioned to further analyze raw values of
photoelectrons which hit the photocathode in the first place.
Relative Gain of PMT Channels
• Hamamatsu documentation suggests gain might not be completely
homogenous from channel to channel
• We wished to verify this premise experimentally with our apparatus.
• Here, relative gain means that we are normalizing to the data
obtained at 500V for each channel and does not represent an absolute
quantity.
Total Channel Relative Gain
• In this measurement, we had the laser incident on a particular channel and took the
charge collected by the total channel (DY12).
• This process was repeated for each channel and its gain represented by a slope in a
linear model, the same process shown in the previous slide.
Channel Homogeneity for Socrates
• We also wanted to test how homogeneous channels were.
• Below are plots of Socrates, the first PMT we tested this summer.
Amplitude Homogeneity:
Charge Homoegeneity :
Channel Homogeneity for Nietzsche
• Below are plots of Nietzsche, the eighth PMT we tested this summer.
Amplitude Homogeneity:
Charge Homoegeneity :
Making Sense of Chaos




Fact: After many days without
food and water, you can be
convinced that 2+2=5. Similarly,
we can do the same things with
these plots.
What we need in our data is a
definitive pattern so we can say
that “Yes, this happens due to this
or that. “
Then we made the FWHM
histogram for Socrates and
discovered some order…
We hope that we can discover
additional photomultiplier
characteristics to make similar
conclusions.
FWHM*10-7 s
Crosstalk Measurements
• The crosstalk is the response of a given channel when a different channel is fired
upon by the laser beam.
• In the ideal case the channels are completely decoupled but in reality a small
correlation is observed. High crosstalk could potentially spoil the resolution power of
the device.
Amplitude Crosstalk:
Charge Crosstalk:
Alternate representation…
• Three dimensions help to illustrate how much more profound an effect
amplitude has in interchannel communication.
• Charge collection is a more important quantity for the KLOE calorimeter
because as an integral quantity, it has the ability to cancel out positive and
negative noise to leave only real information, whereas amplitude includes
background noise.
Charge Crosstalk:
Amplitude Crosstalk:
Results and Conclusions



Crosstalk measurements verify that only
channels juxtaposed with the incident channel
express crosstalk that could be dangerous in
experimental settings.
Presented in the former slide are the results for
only one PMT. We measured eleven in total, but
would need a full day to show them all in depth.
The data we have collected will be very useful in
preparation for the initialization of the
experiment in Frascati.
E poi?



Adesso…We hope to see how well the KLOE
calorimeter will work with its array of PMTs.
The granularity of the multi-anode PMTs should
invaluably assist in more precise results with
respect to determining position of photons
incident on the tubes.
More investigation on definite correlation and
studying of FWHM, the most predictable reading
we have seen so far.
Thanks!


Gratitude goes to INFN, Universita di
Roma III and also to everyone who has
welcomed me here!
Special thanks to Filippo Ceradini, Paolo
Branchini, and Biagio Di Micco for assisting
me with my work.
Grazie milioni!