SiPM Manual
SiPM Manual
Department Physik
Universität Siegen, 57072 Siegen, Germany
Contents
1 Introduction – the Silicon-Photomultiplier (SiPM) 2
1.1 Semiconductor Detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
The pn-junction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
The reverse-biased diode . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Photodiodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
PIN Diode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Avalanche Photodiodes (APD) . . . . . . . . . . . . . . . . . . . . . 5
1.3 Silicon Photomultiplier (SiPM) . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.1 Working Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.2 Characteristics and Properties . . . . . . . . . . . . . . . . . . . . . . 6
Breakdown voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Photon detection efficiency (PDE) . . . . . . . . . . . . . . . . . . . . 7
Gain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Dark count rate (DCR) . . . . . . . . . . . . . . . . . . . . . . . . . . 8
Optical cross talk (OCT) . . . . . . . . . . . . . . . . . . . . . . . . . 8
After pulse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.3.3 SiPM Hamamatsu S13360-1350CS . . . . . . . . . . . . . . . . . . . . 10
2 Experimental Set-Up 11
3 Tasks 13
3.1 Getting to Know the Silicon-Photomultiplier Signals . . . . . . . . . . . . . . 13
Signal Amplitude . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Width of the Signal . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Silicon-Photomultiplier Characteristics in the Absence of Light . . . . . . . . 15
3.2.1 General Measurement Techniques . . . . . . . . . . . . . . . . . . . . 15
3.2.2 Break-Down Voltage . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.2.3 Dark Count Rate (DCR) . . . . . . . . . . . . . . . . . . . . . . . . . 18
3.2.4 Cross Talk Probability . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3 Characterization of the Silicon-Photomultiplier with an Ultra-Fast Pulsed LED 20
3.3.1 General Measurement Techniques . . . . . . . . . . . . . . . . . . . . 20
Internal and external trigger . . . . . . . . . . . . . . . . . . . . . . . 20
Variable Gate Width . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3.2 Gain versus Bias Voltage . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.3.3 Photon Number Resolution . . . . . . . . . . . . . . . . . . . . . . . 23
1
3.3.4 Number of Detected Photons . . . . . . . . . . . . . . . . . . . . . . 23
3.3.5 Influence of the Bias on the Photon Detection Efficiency . . . . . . . 24
The electron can move freely in the conduction band. Holes move as well, as the vacancy
is filled up by an electron from a neighboring atom. They can be seen as positive charge
carriers, but their speed is much lower than that of electrons in the conduction band. This
movement of the e-h pair can be guided by an external electric field. Thus, semiconductors
act as an ionization chamber and the minimum detectable quantum of energy is determined
by the band gap energy between valence and conduction band. [Spi05]
The pn-junction
For successful radiation detection, the created e-h pairs need to be drifted towards electrodes
with a sufficiently high field. To form a high field region with low thermally induced leakage
current, a reverse-biased pn-junction can be used. To that end, impurities are brought into
the material to control the conductivity. This procedure is called doping.
• n-doping: atoms that have one valence electron more than the semiconductor ele-
ments are implanted into the crystal lattice during the manufacturing process. Typi-
cally used elements with 5 valence electrons are phosphor (P), arsenic (As) or antimony
(Sb) and are also referred to as donator. The bound level of the unpaired additional
electron from these impurity atoms is just below the conduction band so that at room
temperature electrons are introduced into the conduction band.
2
Energy
electron
Conduction band
Ionization through
incoming radiation Band gap energy
vacancy (“hole”)
Valence band
Core electrons
Figure 1: Band structure of a semiconductor and electron-hole creation. Valence and conduction band
are separated by a forbidden band gap. Due to incident radiation an electron from the valence band can
receive enough energy to reach the conduction band leaving a vacancy (hole) behind. Adapted from [Spi05].
• p-doping: in this case atoms with one less valence electron are used for doping, which
are amongst others boron (B), aluminum (Al), gallium (Ga) or indium (In). These
impurities borrow an electron from the semiconductor and are therefore also called
acceptor. The electron is bound by the acceptor in an energy level just above the
valence band in the forbidden band.
When p- and n-doped semiconductor materials are brought together to form a pn-
junction, charge separation occurs. Figure 2 demonstrates the processes happening at a
pn-junction.
Figure 2: Visualization of the processes occurring at a pn-junction. When the two doped regions are
joined together, electrons move towards the p-doped regions via thermal diffusion leaving positive space
charges behind, while holes move to the n-type region. A potential is formed at the junction. Electrons
lifted into the conduction band can now be drifted towards the n-region. [Ind07]
3
Both materials are actually neutral when separated, but when joined together, electrons
and holes will cross the junction via thermal diffusion. Electrons leaving the n-doped area
are accepted in the p-doped region and will leave the donors behind as uncovered positive
space charges. A positive space charge acts on the electrons that moved to the p-region
and, thus, a potential is formed. The movement stops when the energy required for thermal
diffusion becomes smaller than the space charge potential that has evolved between the two
regions. Right at the junction between the two doped areas, a region free of mobile charge
carriers is formed. It is called depletion region.
1.2 Photodiodes
In contrast to charged particle and X-ray detectors, photodiodes have only a small absorption
length. The devices investigated in this experiment are solely silicon-based photon detectors.
The absorption length of photons in silicon depends strongly on the wavelength. At a photon
wavelength of λ = 400 nm the absorption length is of order 100 nm, whereas at λ = 700 nm
it is about 5 µm [HB19]. In silicon photodiodes, quantum efficiencies1 for the creation of an
e-h pair are on the order of 80 % and sensitivity down to photon wavelength of λ = 200 nm
is possible [Spi05].
PIN Diode
The PIN diode is in principle the realization of the reverse-biased pn-junction that has been
explained in the previous section. It has an additional intrinsic, high-ohmic piece of semicon-
ductor between highly doped (n+ and p+ ) regions. The basic design is schematically shown
in figure 3. A field is produced across the intrinsic layer (i-layer) to separate charge carriers
and drift them towards the electrodes for detection [GH20]. An incoming photon creates an
e-h pair with the electron reaching the conduction band to be drifted towards the cathode
at the end of the n-doped layer. The signal is, therefore, proportional to the intensity of
the incoming light. Most incoming photons are absorbed in the i-layer and can contribute
to the photo current at the electrodes. Speed and spectral response are determined by the
doping-concentration of the pn-junction. [Din13]
The undepleted p-layer on top must be transparent to incoming light. Its thickness is on
the order of 1 µm [Din13]. The active area of the diode is often covered with an anti-reflective
coating to reduce reflection of light of a specific wavelength. Passive areas are usually coated
with silicon oxide. Sometimes protective layers are used, which cut off detection of incoming
light at 290 nm - 320 nm.
1
The quantum efficiency describes the probability for an incoming particle to create an electron hole pair
in the depletion region.
4
Anti-reflective coating
Anode
Silicon oxide layer
p⁺ region
incoming photon
hole
electron-hole pair
n⁺ region
Cathode
Figure 3: PIN diode structure. An intrinsic layer is sandwiched between two highly doped regions.
Ionizing radiation creates charge carriers that are drifted towards the electrodes. Adapted from [RL09].
The PIN diode shows efficiency for the absorption of photons in the optical wavelength range
up to 1100 nm, which corresponds to the band gap energy in silicon (1.1 eV).
The PIN diode has no internal gain. Therefore, PIN diodes are only sensitive to a min-
imum photon number of 200-300. Complex read-out electronics including charge-sensitive
amplifiers and low band filters are needed. Due to filter time constants on the order of 1 µs,
signals tend to become slow and high signal rates cannot be processed.
5
Figure 4: The three operational regimes of a photodiode depending on the applied reverse-bias. PIN
diodes work without gain (left regime), while APDs have linear gain (center regime). Above a certain
voltage, breakdown occurs and the gain response is exponential forming a Geiger-mode avalanche (right
regime). [GH20].
6
Figure 5: Equivalent circuit of Geiger-mode APD Figure 6: Schematic drawing of an SiPM. The
connected in parallel, which constitutes the basic de- pn-junctions and quenching resistors are intercon-
sign of an SiPM. [Oto+08] nected via a silicon substrate and a metal grid. Anti-
reflective coating increases the transmission. [Din13]
Breakdown voltage
The voltage above, which Geiger-mode avalanche multiplication occurs, is called the break-
down point or breakdown voltage Vbd . Since the required bias voltage Vbias for SiPM operation
needs to be above this value, it has direct impact on the properties of the SiPM. It defines
the applied overvoltage as Vover = Vbias − Vbd , which, in turn, influences other quantities like
dark count rate, probability for optical cross talk (OCT) and the detection efficiency, all of
which are explained in the scope of this chapter. [GH20]
where QE(λ) describes the wavelength dependent quantum efficiency. It is the probability
for a photon to be transmitted into the sensitive volume through the entrance window or
anti-reflective coating and then to be absorbed in the valence band of the silicon with the
subsequent creation of an e-h pair. pA (Vover ) denotes the avalanche trigger probability, which
depends on the overvoltage. This value quantifies the chance of an e-h pair to trigger an
avalanche and, thus, a measurable signal. Finally, the geometric efficiency or fill factor geo
determines the fraction of the detector’s front face that is sensitive to incoming light. The
metal frame and quenching resistors as well as trenches in between micro-cells limit the area
where photons can be detected. Depending on the cell size, this value can reach up to 80 %.
Smaller cells limit the geometric efficiency due to larger dead space, but a larger number
of cells is possible, which increases the dynamic range of the SiPM. The PDE increases for
higher overvoltage.
7
Figure 7: Oscilloscope graph showing waveforms of Figure 8: Single photo electron spectrum
an SiPM of type S13360-3050 by Hamamatsu. The recorded using charge integration with a Hama-
discrete signal height represents the number of de- matsu SiPM. Each peak corresponds to a certain
tected photons [Ham16]. number of photons. [Eck+10]
Gain
The number of charge carriers that are created during an avalanche discharge process of a
micro-cell and that contribute to the signal, defines the gain of the detector [Din13]. Using
the capacity of a micro-cell CD , the electron charge e and the applied overvoltage, the gain
can be calculated [Oto+08]:
CD · Vover
G= (2)
e
The gain increases with the applied overvoltage Vover . Just like the PDE it is independent
of the temperature as long as Vover is constant [GH20]. The gain strongly depends on the
capacitance CD and therefore varies especially for different cell sizes [Ham16]. Typical values
are on the order of 106 [Eck+10].
The following two characteristics are commonly referred to as correlated noise sources.
The term indicates that this type of noise is a follow-up product of a primary event – may it
be a detected photon or a dark event. Figure 9 gives an impression of these different types
of signals as they will be discussed in the following.
8
Figure 9: Waveform sketches of Silicon-Photomultiplier signals together with explanations on the different
types of correlated noise signals. [GH20].
(OCT) and it can occur both for dark signals and actual photon induced signals. The
likelihood for an avalanche to create another signal via emission of a secondary photon
into a neighboring cell is called cross talk probability and is on the order of a few percent.
Exact values mostly depend on the applied overvoltage. A quadratic dependency on the
overvoltage has been reported by Soto et al. [Sot+13] and Eckert et al. [Eck+10]. For
constant overvoltage, cross talk is independent of temperature [Din13].
Possible explanations for the light emission in an avalanche are bremsstrahlung or a
multi-mechanism scenario including indirect and direct inter-band and intra-band transitions
[RL09]. Since this process takes place on time scales of not more than a few hundred
picoseconds [GH20], the two signals appear as one larger signal twice the size.
OCT can be via direct emission of a photon into a neighboring micro cell or as external
cross talk, where the photon leaves the cell through the active side of the SiPM and can then
be reflected by the glass window or epoxy resin covering the surface. Therefore, OCT de-
pends on the architecture of the Silicon-Photomultiplier as its intensity increases for smaller
distance between the cells. Introducing trenches in between the individual micro-cells can
help block the photon’s way to the next neighboring cell. That way cross talk can be reduced
but not entirely prevented.
After pulse
An electron that is released in an avalanche can get trapped in material defects or impurities
and released again after some time. This electron can then create a second signal in that same
cell. The time delay between primary signal and after pulse is on the order of nanoseconds
up to microseconds. The charge QAP of the pulses depends on the recovery state of the cell
and can be expressed in terms of a single photon equivalent charge signal:
t
QAP = Q1 pe · 1 − exp − (3)
τ
Here, τ denotes the recovery time constant of the Silicon-Photomultiplier . For small times t,
the height of the after pulse is smaller than that of a 1 pe signal. After pulses are illustrated
9
in figure 9 as signals that are smaller than the 1 pe peak. For large time delays between
the primary signal and the after pulse, the standard avalanche signal is triggered. Then, no
distinction from delayed cross talk or a dark or single photon event is possible.
10
2 Experimental Set-Up
This part contains info on the experiment and its components. A description is provided on
how to operate the system and read out data with the graphical user interface (GUI).
Figure 10 shows the set-up of the experiment. The detector sits in its sensor holder, which
in turn is mounted to the device named SP5600. It provides the bias voltage and sends the
analog detector output to the digitizer. The digitizer (DT5720A) serves as communication
unit between detector and read-out software and converts the analog signals into digitized
values to be further analyzed with the GUI. On the left in figure 10, the LED driver is
shown (device number SP5601), which enables to send ultra-fast pulsed LED light towards
the SiPM via an optical fibre.
Figure 10: Schematics of the experimental set-up: the SiPM sits on the SP5600 and the DT5720A serves
as communication and data taking unit. For the second part of the measurements, a fast pulsed LED can
be coupled to the SiPM holder via an optical fibre.
Figure 11 shows the GUI. The GUI main panel is structured as a “virtual instrument”: its
appearance and operations imitate a real central system. Three sub-panels may be identified:
• The Power Supply and Amplification Unit (PSAU) sets the bias voltage. It becomes
active after the communication with the PSAU is started, through the “START PSAU”
button. This button opens the communication with the PSAU by the selected COMM
Port.
• The digitizer contains the parameters to tune the trigger threshold and the size of the
integration window for the charge measurement. The digitizer has a special firmware
dedicated to SiPM. In the external triggering mode, the only parameters the user has
to set are the “Gate parameters” in order to allow the firmware to integrate all the
digitized signals. This tab includes also information about connection and errors.
• The right hand side shows the visualization tabs – different panels that host the display
of various measurements. These are most importantly the charge measurements in
the tab named Histogram, the dark count measurements in PSAU staircase and a
waveform display under the tab named wave.
11
Figure 11: The graphical user interface (GUI): power supply unit at the top left, digitizer below and on
the right hand side the measurement results can be displayed graphically.
Further explanations can be found in the educational kit’s user guide [SpA16]. Instruc-
tions on how to set the measurement parameters are explained together with the measure-
ment tasks in the next section.
12
3 Tasks
3.1 Getting to Know the Silicon-Photomultiplier Signals
WARNING
The bias voltage must NOT exceed 58 V to prevent damage at the
Silicon-Photomultiplier !
In the very first part of the experiment, the signals of the Silicon-Photomultiplier will be
investigated. To that end, the output of the SP5600 of channel 0 has to be connected to one
input of the oscilloscope, instead of the digitizer. The signals will at first not be examined
using the GUI in the software but ”by hand” on the oscilloscope.
The SP5600 and the digitizer must be switched on before a connection to the GUI can be
established. After checking the communication port of the PSAU, the PSAU itself must be
started by clicking on “START PSAU” (the green light ‘ON’ switches on). In the PSAU the
SiPM nominal operation voltage can be set as well as the gain of the amplifier of the PSAU.
Only after connecting, the bias voltage can be switched on at the PSAU. At the beginning
a value well below the break-down point must be chosen (like 50 V), and then ramped up
in small steps (1 V for instance) until signals can be seen on the oscilloscope screen. This
should happen at around 53 V. In order to trigger signals at the oscilloscope one must set
the trigger threshold relatively low (below 10 mV) and chose a small voltage division (like
5 mV).
With the oscilloscope several SiPM characteristics can easily be investigated quantita-
tively: for example the peak height (given in millivolts) and an estimation of the width of
the signal (in nanoseconds). Especially the latter of these quantities plays an important role
for charge integration in the later experiment.
Signal Amplitude
The amplitude can be measured using the cursor function of the oscilloscope in the voltage
type. To be able to estimate the mean peak height of the signal, the averaging function of
the oscilloscope may be used. Averaging over 16 waveforms should be sufficient in this case.
Then the cursor can be moved to the tip of the wave form and the position of the cursor is
shown at the right side of the screen.
13
Figure 12: Screen shot of an oscilloscope measurement of the signals width (time-over-threshold TOT)
using vertical time-based cursors and averaging over 16 waveforms.
will help setting the parameters right for performing a measurement of the deposited charge
in an SiPM signal.
MEASUREMENT TASK
1. Amplitude measurement:
(a) Set the voltage between 53 V and 57 V. Play with the trigger until you have
identified the single photon peak.
(b) Using the waveform averaging function of the oscilloscope as well as the
voltage cursor as described above, measure the peak amplitude of the sig-
nal. Repeat this measurement for at least 5 different bias voltages in the
range between 53 V and 57 V.
Hint: do not modify the voltage division at the oscilloscope in between dif-
ferent measurements as it might change the input impedance of the channel
and distort your result!
14
ANALYSIS
1. (a) Describe your measurement. How did you identify the 1 photon peak?
What happened when you moved the trigger above the level of the single
photon amplitude?
(b) Plot the measured peak amplitude versus bias voltage.
2. Plot the measured widths versus bias voltage. How does the trigger level in-
fluence your result? What would have happened if you had chosen a higher or
lower trigger level?
For the rest of the experiment, the output from the SiPM at SP5600 must be connected to
the digitizer. The digitizer must be switched on at the back of its housing. It is then started
at the GUI by clicking on “START DIGITIZER” (the related green light ‘ON’ switches on).
Switch on channel 0 at the digitizer afterwards. The trigger should be set to internal for
the first part of the measurements. Now the system is up and ready to run.
For all measurements of this section, it is important to note the temperature! Only then
reliable comparison of the results to the data sheet or reference measurements is possible.
The temperature can be found in the PSAU window in the tab T monitor.
15
Figure 13: Digitizer unit of the read-out software. The two parameters to be investigated in this section
are the trigger treshold and the gate width.
visualization window must be selected, where signal and gate are displayed. The charge is in
principle the integral between waveform and baseline (corresponding to the noise level with
zero amplitude). The gate must be set such that it easily captures the entire waveform. A
pre-gate must be defined, which shifts the gate to the left to start the integration a little
earlier and to prevent cutting off the rising edge of the signal. A reasonable value for the
pre-gate for internally triggered signals is 16 ns.
In the histogram tab, the charge of the captured signals is displayed.2 Whenever one
of the digitizer parameters is changed, the histogram resets and is filled from scratch au-
tomatically. Thus, one can directly observe the influence of the trigger threshold and gate
width.
All data are saved to the following folder location by default:
Downloads/SiPMkit ControlSoftware 00.29/data/.
2
In darkness, the 1 photon peak should appear prominently and any higher photon peaks should be
significantly smaller.
16
MEASUREMENT TASK
1. Investigate the influence of the trigger threshold. Choose a bias voltage of 54-
55V. Set the gate width to 208 ns. Record charge histograms with different
trigger thresholds and observe the changes. Save meaningful histograms by giv-
ing them a unique name in the Histo file name section at the bottom right
of the visualization tab. When you are done, set the trigger threshold to a level
where the 1 photon peak can be seen.
2. Investigate the influence of the gate width. Now, leave the trigger threshold
fixed but change the gate width in a range between about 60 ns and 500 ns and
observe the change. Also, save some histograms for different gate widths for
later analysis. When you are done, set the gate to a reasonable value such that
the entire Silicon-Photomultiplier signal can be captured.
ANALYSIS
1. Describe the influence of the trigger on the histogram. What happens for extreme
trigger values? Explain your findings with 2 or 3 meaningful histograms.
2. Describe the influence of the gate width on the charge histograms. How does
the peak form and position change? What happens for very narrow and wide
gate widths? Explain your findings with 2 or 3 meaningful histograms.
MEASUREMENT TASK
Take a charge histogram for at least 4 different bias voltages between 53.5 V and 57 V.
Leave all other parameters constant.
Hint: set the bin width of the histogram to a value of 4 QDC channels or less to get
smooth curves! This is called bin size [ADC] in the software.
17
ANALYSIS
Determine the peak position of the 1 photon peak in all the recorded histograms. You
can use the mean value of a gauss fit for example. For this analysis, you do not need
to convert the QDC value into a value charge, yet.
Create a plot with the applied voltage versus the peak positions and use a linear
extrapolation to determine the break-down point at which the charge is zero.
Compare your result with the data sheet (see table 1).
MEASUREMENT TASK
Take a stair case plot for 5 different voltages between 54 V and 58 V. You can start
a measurement by clicking the REFRESH button in the PSAU staircase tab. You can
save the data by entering a file name in PSAU name file user prior to starting the
run and switching the save staircase button on.
18
Figure 14: Staircase plot showing the dark count rate versus trigger threshold.
ANALYSIS
2. For every voltage calculate the dark count rate. To that end, take the mean
value of the data points of the first plateau in the graph. This is the dark count
rate on single photon level. Find a reliable and reproducible method to define
the end of the plateau (e.g. the position of greatest slope etc.).
3. Create a plot with the dark count rate versus bias voltage.
4. Compare your results with the values from the data sheet (see table 1). Try to
find an explanation for potential deviations.
19
The probability for the occurrence of cross talk can be calculated by using the dark rate
measured on the second plateau of the staircase plot (see figure 14 as example) and dividing
it by the dark rate measured on the first plateau:
DCR2 pe
pCT = , (4)
DCR1 pe
ANALYSIS
1. (a) Take the data from task 3.2.3 and determine the mean value of the data
points on the second plateau.
(b) Calculate the cross talk probability according to formula 4.
(c) Create a plot with the cross talk probability versus bias voltage.
20
MEASUREMENT TASK
Set the pre-gate and gate so that the whole waveform of the signal can be recorded
and used for integration.
Record a charge histogram once with internal trigger on 1 photon level and once with
external trigger.
ANALYSIS
Explain the difference between the two measurements. Why is there an additional
peak around QDC=0 (called pedestal ), when using the external trigger? What has
been recorded in this case?
MEASUREMENT TASK
Record three charge histograms: one for an extremely narrow gate, then for a very
wide gate and finally for an intermediate gate width.
The same voltage as before should be used and signals should be triggered externally.
ANALYSIS
Explain the influence of the gate width on the shape of the peaks and their position
and distance from one another.
where qe is the charge of an electron 1.6 · 10−19 C. The index pp means from peak to peak.
Therefore, qpp denotes the charge difference between two consecutive peaks in the charge
21
histogram. The distance between two peaks is simply calculated as the difference between
the QDC values of neighboring peaks:
Since the charge is given in terms of QDC channels, this value must be converted to Coulomb
charge. The required QDC channel conversion factor (here denoted with FQDC ) can be
calculated according the Application Note AN2502 by CAEN [SpA11]:
For example, for an amplification gain of GP SAU = 30, the conversion factor would be
FQDC = 1.235 fC/QDC. Now, the gain G can be calculated::
∆QDCpp · FQDC
G= (8)
qe
MEASUREMENT TASK
Chose an intermediate LED intensity so that the pedestal and at least 3 more peaks
can be seen in the charge spectrum.
Now, vary the bias voltage between 53 V and 57 V in at least 5 steps and record a
charge spectrum for each voltage.
ANALYSIS
(a) Determine the position of the pedestal and the 1 pe peak in every spectrum and
calculate the distance.
(c) Show your results: plot a graph with gain versus bias voltage. What is your
suspected relation between gain and voltage?
22
3.3.3 Photon Number Resolution
In photon counting experiments, the resolution of the photon number is important. This
quantity describes how well a measured charge value can be attributed to a specific photon
number. The resolution of the system can be evaluated by plotting the width of each peak
σ versus the photon number n that created the peak. σ is the width that can be obtained
using a gauss fit to a peak.
ANALYSIS
(a) Use the data set from the 54 V measurement from the previous task 3.3.2 and
perform a gauss fit to every peak in the histogram that can be clearly identified.
(b) Plot the width versus photon number and discuss your findings: how does the
width behave for growing photon number? What does this tell you about the
resolution for higher photon numbers?
(c) Above which photon number has the resolution decreased such that no peaks
can be identified any more? To determine that, perform a linear fit to your data
and extrapolate to the point where the width exceeds half the distance between
two peaks.
MEASUREMENT TASK
1. Set the bias voltage to 54 V and use the external trigger. Observe the changes
in the charge spectrum, while tuning the LED intensity in a range between 150
and 500.
2. Record charge spectra for at least 5 different LED settings between 250 and 450.
Make sure that at least 4 peaks can be seen even for low LED intensity! In case
of high intensity the pedestal must still be visible!
23
ANALYSIS
1. Explain why higher photon peaks occur, when the intensity is ramped up. Why
do peak width and distance not change with LED intensity?
2. (a) Extract the peak heights from the spectrum with the corresponding photon
number. Create a graph with these peak heights.
(b) Perform a Poisson fit to the data points as shown in the example in figure
15. Obtain the mean number of detected photons from that fit.a
(c) Plot all mean values versus LED intensity. Describe your observations.
a
On the Desktop of this PC you can find a C++ script that performs such a Poisson fit. It
utilizes the ROOT data analysis toolkit. The script can be executed using the terminal command
root -l Poisson.cpp++.
Charge spectrum
Counts
5000
4000
3000
2000
1000
0
0 2 4 6 8
Peak Number
Figure 15: Example of a Poisson fit to the peak heights of the charge spectrum. From the fit, the mean
number of detected photons per LED pulse can be extracted.
24
ANALYSIS
(a) Use the data sets from task 3.3.2, where the voltage was scanned through a
certain range with fixed LED intensity.
(b) Again, perform Poisson fits to obtain the mean value of the number of detected
photons.
(c) Plot all mean values versus bias voltage and interpret your result.
(d) Considering the results from the other tasks, make a recommendation for an
operating voltage to obtain best results. Explain which quantities you took into
account for your decision.
25
[Ind07] Inductiveload via wikibooks.org. Pn Junction Diffusion and Drift. visited online:
March 19, 2020. 2007. url: https : / / en . wikibooks . org / wiki / File : Pn _
Junction_Diffusion_and_Drift.svg.
[Oto+08] H Otono et al. “Study of the internal mechanisms of Pixelized Photon Detectors
operated in Geiger-mode”. In: arXiv preprint arXiv:0808.2541 (2008).
[Ram08] Marco Ramilli. “Characterization of SiPM: temperature dependencies”. In: 2008
IEEE Nuclear Science Symposium Conference Record. IEEE. 2008, pp. 2467–
2470. doi: 10.1109/NSSMIC.2008.4774854.
[RL09] D Renker and E Lorenz. “Advances in solid state photon detectors”. In: Journal
of Instrumentation 4.04 (2009), P04004. doi: 10 . 1088 / 1748 - 0221 / 4 / 04 /
P04004.
[Sot+13] Orlando Soto et al. “Characterization of novel hamamatsu multi pixel photon
counter (MPPC) arrays for the GlueX experiment”. In: Nuclear Instruments and
Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors
and Associated Equipment 732 (2013), pp. 431–436. doi: 10 . 1016 / j . nima .
2013.06.071.
[SpA11] CAEN SpA. Application Note AN2502. SiPM characterization. Version rev.0.
2011.
[SpA16] CAEN SpA. SP5600AN – Educational Kit – Premium Version. Guide GD5484.
Version rev. 0-01/09/2016. 2016.
[Spi05] Helmuth Spieler. Semiconductor detector systems. Vol. 12. Oxford university
press, 2005.
26