Si APD, MPPC CHAPTER 03

1 Si APD
1-1 Features
1-2 Principle of avalanche multiplication
1-3 Dark current
1-4 Gain vs. reverse voltage characteristics
1-5 Noise characteristics
1-6 Spectral response
1-7 Response characteristics
1-8 Multi-element type
1-9 Connection to peripheral circuits
1-10 New approaches

2 MPPC
2-1 Operating principle
2-2 Features
2-3 Characteristics
2-4 How to use
2-5 Measurement examples
2-6 Selecting digital mode or analog mode

3 Applications
3-1 Optical rangefinders
3-2 Obstacle detection
3-3 LIDAR (light detection and ranging)
3-4 Scintillation measurement
3-5 Fluorescence measurement
3-6 High energy physics experiment

1
 Si APD, MPPC

The APD (avalanche photodiode) is a high-speed, high-sensitivity photodiode that internally multiplies photocurrent
when reverse voltage is applied. The internal multiplication function referred to as avalanche multiplication features high
photosensitivity that enables measurement of low-level light signals. The APD’s ability to multiply signals reduces the effect
of noise and achieves higher S/N than the PIN photodiode. The APD also has excellent linearity.
The MPPC (multi-pixel photon counter) is an opto-semiconductor made up of multiple APD pixels operating in Geiger
mode. The MPPC provides significantly higher gain than the APD and has photon-counting capability. It also features low
voltage operation.
Utilizing our unique technologies, we offer numerous types of Si APDs and MPPCs for various applications. We also offer
custom-designed devices to meet special needs.

Hamamatsu Si APDs

Type Features Applications

Short Low-bias operation

Enhanced sensitivity in the UV to visible region
•detection
Low-light-level
wavelength type Low terminal capacitance •Analytical instruments
•FSO
High sensitivity in near infrared region and low bias voltage •Optical rangefinders
(reverse voltage) operation •communications
Optical fiber
Low-bias
operation
Low cost and high reliability APD using surface-mount ceramic •Optical rangefinders
packages with the same wide operating temperature range (-20 •Laser radars
Near infrared
to +85 °C) as metal package types •FSO
type
Low •FSO
temperature
Low temperature coefficient of the reverse voltage, easy gain •Optical rangefinders
coefficient
adjustment •communications
Optical fiber

900 nm band Enhanced sensitivity in the 900 nm band •Optical rangefinders

•Laser radars
1000 nm band Enhanced sensitivity in the 1000 nm band •YAG laser detection
Hamamatsu MPPCs

Type Features Applications

•Fluorescence measurement
For general measurement Suitable for general low-light-level detection •Flow cytometry
•DNA sequencer
•Environmental analysis
High-speed measurement, Features numerous pixels that are well suited to conditions •PET
wide dynamic range where background light is present and is prone to saturation •experiment
High energy physics

For very-low-light-level
measurement
Cooling allows measurement with even further reduced dark count. •Fluorescence measurement
Reduced crosstalk suppresses erroneous counting during low
For precision measurement
count rate measurement •Fluorescence measurement
Employs a structure in which the dead area in the periphery of
the photosensitive area has been eliminated. •PET
Buttable type (semi custom)
Its four-side buttable structure enables elements to be arranged •experiment
High energy physics
two-dimensionally with narrow gaps.

Monolithic array with multiple 3 × 3 mm MPPCs mounted on a •PET

Large-area array
single chip •experiment
High energy physics

2
 pairs, which generate a chain reaction of ionization. This
1. Si APD is a phenomenon known as avalanche multiplication.
The number of electron-hole pairs generated during the
time that a carrier moves a unit distance is referred to as
The APD is a high-speed, high-sensitivity photodiode
the ionization rate. Usually, the ionization rate of electrons
that internally multiplies photocurrent when a specific
is defined as “α” and that of holes as “β.” These ionization
reverse voltage is applied.
rates are important factors in determining the multiplication
The APD, having a signal multiplication function inside
mechanism. In the case of silicon, the ionization rate of
its element, achieves higher S/N than the PIN photodiode
electrons is larger than that of holes (α > β), so the ratio at
and can be used in a wide range of applications such as
which electrons contribute to multiplication increases. As
high-accuracy rangefinders and low-level light detection
such, the structure of Hamamatsu APDs is designed so that
that use scintillators. Though the APD can detect lower level
electrons from electron-hole pairs generated by the incident
light than the PIN photodiode, it does require special care
light can easily enter the avalanche layer. The depth at which
and handling such as the need for higher reverse voltage
carriers are generated depends on the wavelength of the
and consideration of its temperature-dependent gain
incident light. Hamamatsu provides APDs with different
characteristics.
structures according to the wavelength to be detected.
This describes Si APD features and characteristics so that
users can extract maximum performance from Si APDs. [Figure 1-1] Schematic diagram of avalanche
multiplication (near infrared type)

1-1 Features Electric field strength E

• High sensitivity: built-in internal multiplication Avalanche

layer
function
• High-speed response
• High reliability High voltage

KAPDC0006EC

1-2 Principle of avalanche multiplication

The photocurrent generation mechanism of the APD 1-3 Dark current
is the same as that of a normal photodiode. When light
enters a photodiode, electron-hole pairs are generated The APD dark current consists of surface leakage current
if the light energy is higher than the band gap energy. (Ids) that flows through the PN junction or oxide film
The ratio of the number of generated electron-hole interface and generated current (Idg) inside the substrate
pairs to the number of incident photons is defined as [Figure 1-2].
the quantum efficiency (QE), commonly expressed in
[Figure 1-2] APD dark current
percent (%). The mechanism by which carriers are
Carriers that are not multiplied
generated inside an APD is the same as in a photodiode,
but the APD is different from a photodiode in that it has a
-
function to multiply the generated carriers. Ids PN junction

When electron-hole pairs are generated in the depletion Avalanche region

layer of an APD with a reverse voltage applied to the PN
junction, the electric field created across the PN junction Idg

causes the electrons to drift toward the N+ side and the - -

holes to drift toward the P+ side. The higher the electric

field strength, the higher the drift speed of these carriers. Multiplied carriers
KAPDC0011EA
However, when the electric field reaches a certain level,
the carriers are more likely to collide with the crystal
The surface leakage current is not multiplied because
lattice so that the drift speed becomes saturated at
it does not pass through the avalanche layer, but the
a certain speed. If the electric field is increased even
generated current is because it does pass through. Thus,
further, carriers that escaped the collision with the
the total dark current (ID) is expressed by equation (1).
crystal lattice will have a great deal of energy. When these
carriers collide with the crystal lattice, a phenomenon ID = Ids + M Idg ............ (1)
takes place in which new electron-hole pairs are
M: gain
generated. This phenomenon is called ionization. These
electron-hole pairs then create additional electron-hole

3
 Idg, the dark current component that is multiplied, greatly [Figure 1-4] Temperature characteristics of gain
affects the noise characteristics. (S12023-05)

1-4 Gain vs. reverse voltage characteristics

The APD gain is determined by the ionization rate, and
the ionization rate depends strongly on the electric field

Gain
across the depletion layer. In the normal operating range,
the APD gain increases as reverse voltage increases. If
the reverse voltage is increased even higher, the reverse
voltage across the APD PN junction decreases due to the
voltage drop caused by the series resistance component
including the APD and circuit, and the gain begins to
decrease.
When an appropriate reverse voltage is applied to the Reverse voltage (V)
KAPDB0089EA
PN junction, the electric field in the depletion layer
increases so avalanche multiplication occurs. As the When an APD is used near the breakdown voltage, a
reverse voltage is increased, the gain increases and the phenomenon occurs in which the output photocurrent is
APD eventually reaches the breakdown voltage. Figure 1-3 not proportional to the incident light level. This is because
shows the relation between the gain and reverse voltage for as the photocurrent increases a voltage drop occurs due
Hamamatsu Si APD S12023-05. to current flowing through the series resistance and load
resistance in the APD, reducing the voltage applied to the
[Figure 1-3] Gain vs. reverse voltage (S12023-05)
avalanche layer.

1-5 Noise characteristics

As long as the reverse voltage is constant, the APD gain is the
average of each carrier’s multiplication. The ionization rate
Gain

of each carrier is not uniform and has statistical fluctuations.

Multiplication noise known as excess noise is therefore
added during the multiplication process. The APD shot noise
(In) becomes larger than the PIN photodiode shot noise and
is expressed by equation (2).

In2 =2q (IL + Idg) B M2 F + 2q Ids B ............ (2)

Reverse voltage (V)
KAPDB0088EA q : electron charge
IL : photocurrent at M=1
Idg : current generated inside the substrate
The APD gain also has temperature-dependent characteristics. (dark current component multiplied)
As the temperature rises, the crystal lattice vibrates more B : bandwidth
M : gain
heavily, increasing the possibility that the accelerated carriers F : excess noise factor
Ids : surface leakage current
may collide with the lattice before reaching a sufficiently
(dark current component not multiplied)
large energy level and making it difficult for ionization to
take place. Therefore, the gain at a certain reverse voltage The ratio of the ionization rate of electrons (α) to the ionization
becomes small as the temperature rises. To obtain a constant
ratio of holes (β) is called the ionization rate ratio [k (=β/α)].
output, the reverse voltage must be adjusted to match changes
The excess noise factor (F) can be expressed in terms of k
in temperature or the element temperature must be kept
as in equation (3).
constant.

1
F = M k + (2 - ) (1 - k) ............ (3)
M

Equation (3) shows the excess noise factor when electrons

are injected into the avalanche layer. To evaluate the excess
noise factor when holes are injected into the avalanche
layer, k in equation (3) should be substituted by 1/k.

4
As described in section 1-6, “Spectral response,” the gain NEP = In/(M S) ....... (5)
is wavelength dependent. Likewise, the excess noise also
M: gain
has wavelength dependence. Some APDs exhibit low noise S: photosensitivity [A/W]
at short wavelengths while others at long wavelengths.
Figure 1-5 shows excess noise characteristics. In PIN photodiode operation, using a larger load resistance
reduces thermal noise, but this also slows the response speed.
[Figure 1-5] Excess noise factor vs. gain Therefore, it is not practical to reduce thermal noise and, in
(a) Short wavelength type (low-bias operation) most cases, the lower limit of light detection is determined
by thermal noise. In APD operation, the signal can be
multiplied without increasing the total noise until the
shot noise reaches a level equal to the thermal noise, thus
λ=800 nm
resulting in an improved S/N while maintaining the high-
speed response. This behavior is shown in Figure 1-6.
Excess noise factor

[Figure 1-6] APD noise characteristics

Output voltage
L
Shot noise = 2q IL M2 F B∙RL
λ=650 nm )R
M
(I
L
=
al
gn S/N max.
Si

Thermal noise = 4k T B RL

Gain
KAPDB0013EA
1 10 100 1000 Gain
Mopt
(b) Near infrared type (low-bias operation)
RL: load resistance
k : Boltzmann's constant
T : absolute temperature
KAPDB0033EC

In this case, the optimum gain (Mopt) is obtained under the

Excess noise factor

λ=650 nm conditions that maximize the S/N described in equation

(4). If Ids can be ignored, the optimum gain is given by
equation (6).
1
2+x
4k T ............ (6)
Mopt =
q (IL + Idg) × RL

λ=800 nm

1-6 Spectral response

Gain
KAPDB0022EA Spectral response characteristics of APDs are almost the
same as those of normal photodiodes if a reverse voltage is
The excess noise factor (F) can also be approximated as not applied. When a reverse voltage is applied, the spectral
F=Mx (x: excess noise index) because the equation for shot response curve will change.
noise can be expressed in the form of In2=2q IL B M2+X. The depth to which light penetrates in the silicon depends
As explained, APDs generate noise due to the multiplication on the wavelength. The depth to which short-wavelength
process, so excess noise increases as the gain becomes higher. light can reach is shallow, so carriers are generated near the
On the other hand, the signal is also increased according to surface. In contrast, long-wavelength light generates carriers
the gain, so there is a gain at which the S/N is maximized. even at deeper positions. The avalanche multiplication occurs
The S/N for an APD can be expressed by equation (4). when the carriers pass through the high electric field near
the PN junction. In the case of silicon, the ionization rate of
I L2 M2 ....... (4)
S/N = electrons is high, so multiplication can be achieved efficiently
2q (IL + Idg) B M2 F + 2q B Ids + 4k T B
RL when electrons are injected into the avalanche layer. For
2q (IL + Idg) B M2 F + 2q B Ids: shot noise example, in the case of the APD type shown in figure 1-7,
4k T B the avalanche layer is in the PN junction region on the front
: thermal noise
RL
k : Boltzmann's constant side. With this APD type, satisfactory gain characteristics can
T : absolute temperature
RL : load resistance
be obtained when long-wavelength light that reaches deeper
than the avalanche layer is incident. The APD structure
The noise equivalent power (NEP) of APDs is given by determines whether short- or long-wavelength light is
equation (5). multiplied efficiently.
5
 [Figure 1-7] Schematic of cross section (near infrared type)
Long Short
1-7 Response characteristics
wavelength wavelength
Carrier multiplication Carrier multiplication
is high. is low. The factors that determine the response speed of photodiodes
are the CR time constant, the carrier transit time (drift time)
in the depletion layer, the time needed for multiplication
(multiplication time), and the time delay which is caused by
Avalanche layer
diffusion current of carriers from outside the depletion layer.
The cutoff frequency fc(CR) determined by the CR time
constant is given by equation (7).

1 ............ (7)
KAPDC0014EB fc(CR) =
2π Ct RL

The spectral response and wavelength dependency of Ct: terminal capacitance

RL: load resistance
gain for the short wavelength type and near infrared type
Si APDs are provided below.
To improve photodiode response speeds, the terminal
[Figure 1-8] Spectral response capacitance should be reduced, for example by making
the photosensitive area smaller and the depletion layer
(Typ. Ta=25 ˚C, M=50, λ=650 nm)
40 thicker. The relation between the cutoff frequency fc(CR)
Near infrared type
S12023 series and the rise time tr is expressed by equation (8).

30 Short wavelength type 0 .35 ............

Photosensitivity (A/W)

S12053 series tr = (8)

fc(CR)

20 If the depletion layer is widened, the drift time cannot be

ignored. The transit speed (drift speed) in the depletion layer
begins to saturate when the electric field strength reaches
10
the vicinity of 104 V/cm, and the saturated drift speed at this
point will be approx. 107 cm/s. Ionization occurs when the
0 carriers that have moved to the avalanche layer generate
200 400 600 800 1000 1200
electron-hole pairs. However, since the holes move in the
Wavelength (nm) direction opposite to that of the electrons, the drift time
KAPDB0090EB in the APD becomes longer than that in PIN photodiodes.
If we let the drift time be trd, the cutoff frequency fc(trd)
[Figure 1-9] Gain vs. wavelength determined by the drift time is given by equation (9).
(Typ. Ta=25 ˚C)
100 0.44 ............
fc(trd) = (9)
trd

80 Short wavelength type

S12053 series Making the depletion layer thicker to reduce the capacitance
Near infrared type also lengthens the drift time, so it is essential to consider
S12023 series
60 both cutoff frequencies, fc(CR) determined by the CR time
Gain

constant and fc(trd) determined by the transit time.

40 The carriers passing through the avalanche layer repeatedly
collide with the crystal lattice, so a longer time is required
20 to move a unit distance in the avalanche layer than the
time required to move a unit distance in areas outside the
0 avalanche layer. The time required to pass through the
200 400 600 800 1000 1200
avalanche layer becomes longer as the gain is increased.
Wavelength (nm) If an APD is used at a gain of several hundred times, the
KAPDB0091EB multiplication time might be a problem.
This time delay caused by the diffusion current of carriers
from outside the depletion layer is sometimes as large as a
few microseconds and appears more remarkably in cases
where the depletion layer is not extended enough with
respect to the penetration depth of the incident light into the
silicon. To ensure high-speed response, it is also necessary to
take the wavelength to be used into account and to apply a
reverse voltage that sufficiently widens the depletion layer.

6
When the incident light level is high and the resulting  APD power consumption is the product of the incident
photocurrent is large, the attractive power of electrons light level × sensitivity (M=1) × gain × reverse voltage,
and holes in the depletion layer serves to cancel out the and it is considerably larger than that of PIN photodiodes.
electric field, making the carrier drift speed slower and So there is a need to add a protective resistor between
impairing the time response. This phenomenon is called the APD and bias power supply and then install a current
the space charge effect and tends to occur especially when limiting circuit. Note that when the output current is large,
the incident light is interrupted. the voltage drop across the protective resistor increases
and the APD reverse voltage declines. In that case, the
protective resistor value must be decreased.
1-8 Multi-element type  A low-noise readout circuit may damage the first stage in
response to excess voltage. To prevent this, a protective
Multi-element Si APDs have an array of photosensitive areas.
circuit should be connected to divert any excess input
The avalanche layer formed just below each photosensitive
voltage to the power supply voltage line.
area on the APD array multiplies the light incident on
the photosensitive areas. However, carriers generated ‘ APD gain changes with temperature. To use an APD over
outside these photosensitive areas cannot pass through the a wide temperature range, measures must be taken such
avalanche layer so their signal is small. This means that APD as incorporating temperature compensation, which controls
arrays have lower crosstalk than photodiode arrays because the reverse voltage to match the temperature changes, or
of their gain. temperature control, which maintains the APD temperature
at a constant level. In temperature compensation, a
[Figure 1-10] Internal structure (multi-element type) temperature sensor is installed near the APD to control
Photosensitive area the reverse voltage according the APD’s temperature
coefficient. In temperature control, a TE-cooler is used
to maintain a constant APD temperature.
Avalanche layer
’ When detecting low-level light signals, if background light
Electron not multiplied enters the APD, then the S/N may decrease due to shot
Electron to be multiplied
noise from background light. In this case, effects from
the background light must be minimized by using optical
KAPDC0015EA filters, improving laser modulation, and/or restricting the
angle of view.
[Figure 1-11] Crosstalk
(S8550-02, element gap: 0.7 µm, typical example) [Figure 1-12] Connection example
(λ=410 nm, M=50)
Bias power supply
100 (temperature compensation)

Current limiting resistance

Approx. 10 kΩ to 100 kΩ
Relative output (%)

ch 1 ch 2
10 0.1 μF or more
(as close to APD as possible)

Excess voltage
protection circuit
1

0
-1.5 -1.0 -0.5 0 0.5 1.0 1.5
High-speed op amp
Readout circuit OPA846, AD744, etc.
Position (mm) KAPDC0005EB
KAPDC0039EA

1-9 Connection to peripheral circuits

APDs can be handled in the same manner as normal
photodiodes except that a high reverse voltage is required.
However, the following precautions should be taken
because APDs are operated at a high voltage, their gain
changes depending on the ambient temperature, and so on.

7
 APD modules 1 - 10 New approaches
APD modules are high-speed, high-sensitivity photodetectors APDs are not so easy to use because they need a high reverse
using an APD. APD modules consist of an APD, a low noise voltage and their gain is temperature dependent. Hamamatsu
I/V amplifier circuit, and a bias power supply assembled in is developing APDs that operate on a low reverse voltage and
a compact configuration. By simply connecting to a low- other types to make them easier to use. Hamamatsu is also
voltage DC power supply, APD modules can detect light working on a surface-mount chip size package (CSP) type
with a good S/N which is dozens of times higher than PIN and a type in which a filter is directly mounted on the chip
photodiodes. APD modules help users evaluate and fabricate to suppress effects of background light.
their high-performance system using an APD. Moreover, since the APD gain is inconsistent within the
Figure 1-13 shows the block diagram of the C12702 series photosensitive area, developing large-area APD arrays
APD module. This module is designed with the precautions requires advanced technology. Hamamatsu is developing
described in section 1-9, “Connection to peripheral circuits,” large-area APD arrays with excellent gain uniformity
thus allowing highly accurate photometry. within the photosensitive area.
For more detailed information about APD modules, refer
to “3. APD modules” in Chapter 11, “Module products.”

[Figure 1-13] Block diagram (C12702 series)

High voltage generator
+200 V +5 V +5 V
BNC
connector
Voltage controller

Temperature monitor

APD

Incident light High-speed current-to-voltage

amplifier circuit

KACCC0013EB

[Figure 1-14] APD modules

(a) Standard type C12702 series (b) TE-cooled type C4777-01

(c) High-speed type C5658

[Table 1-1] Hamamatsu APD modules

Type Features

Standard type Contains near infrared type or short wavelength type APD. FC/SMA fiber adapters are also available.
High sensitivity type High gain type for low-light-level detection

High-speed type Can be used in a wide-band frequency range (up to 1 GHz)

TE-cooled type High-sensitivity type for low-light-level detection. Greatly improved stability through thermoelectric cooling.

8
 To halt the Geiger discharge and detect the next photon, an
2. MPPC external circuit outside the APD must lower the operating
voltage. One specific example for halting the Geiger discharge
is a technique using a so-called quenching resistor connected
The MPPC (multi-pixel photon counter) is one of the devices
in series with the APD to quickly stop avalanche multiplication
called Si-PM (silicon photomultiplier). It is a new type of
in the APD. In this method, a drop in voltage occurs when the
photon-counting device using multiple APD (avalanche
output current caused by the Geiger discharge flows in the
photodiode) pixels operating in Geiger mode. Although the
quenching resistor, reducing the operating voltage of the APD
MPPC is essentially an opto-semiconductor device, it has
connected in series. The output current caused by the Geiger
an excellent photon-counting capability and can be used in
discharge is a pulse waveform with a short rise time, while
various applications for detecting extremely weak light at the
the output current when the Geiger discharge is halted by
photon counting level.
the quenching resistor is a pulse waveform with a relatively
The MPPC operates on a low voltage and features a high
slow fall time [Figure 2-4].
multiplication ratio (gain), high photon detection efficiency,
fast response, excellent time resolution, and wide spectral
response range, so it delivers the high-performance level Structure
needed for photon counting. The MPPC is also immune to
magnetic fields, highly resistant to mechanical shocks, and Figure 2-1 shows a structure of an MPPC. The basic element
will not suffer from “burn-in” by incident light saturation, (one pixel) of an MPPC is a combination of the Geiger
which are advantages unique to solid-state devices. The mode APD and quenching resistor, and a large number of
MPPC therefore has a potential for replacing conventional these pixels are electrically connected and arranged in two
detectors used in photon counting up to now. The MPPC dimensions.
is a high performance, easy-to-operate detector that is
proving itself useful in a wide range of applications and [Figure 2-1] Structure
fields including medical diagnosis, academic research, and
measurements.1) 3)
Geiger-mode
APD pixel

2-1 Operating principle

Quenching resistor
Photon counting

Light has a property in both a particle and a wave. When the

KAPDC0029EA
light level becomes extremely low, light behaves as discrete
particles (photons) allowing us to count the number of [Figure 2-2] Image of MPPC’s photon counting
photons. Photon counting is a technique for measuring the
number of individual photons.
The MPPC is suitable for photon counting since it offers
an excellent time resolution and a multiplication function
having a high gain and low noise. Compared to ordinary
light measurement techniques that measure the output
current as analog signals, photon counting delivers a higher
S/N and higher stability even in measurements at very low
light levels.

Geiger mode and quenching resistor

When the reverse voltage applied to an APD is set higher

than the breakdown voltage, saturation output (Geiger
KAPDC0049EA
discharge) specific to the element is produced regardless of
the input light level. The condition where an APD operates
at this voltage level is called Geiger mode. The Geiger mode
allows obtaining a large output by way of the discharge even
when detecting a single photon. Once the Geiger discharge
begins, it continues as long as the electric field in the APD is
maintained.

9
 [Figure 2-3] Block diagram for MPPC evaluation [Figure 2-4] Pulse waveforms when using a linear amplifier
(with an oscilloscope) (120 times) (S12571-050C, M=1.25 × 106)
Trigger
(light output timing)
Pulse
light source 3 p.e.
Oscilloscope
2 p.e.

50 mV
Attenuator
1 p.e.
Optical fiber
Multiplied
Pulsed light MPPC signal
Amplifier

MPPC

10 ns
MPPC Amplifier
power supply power supply
(2) Integrating the output charge
KAPDC0028EB

The distribution of the number of photons detected during

a particular period can be estimated by measuring the MPPC
Basic operation
output charge using a charge amplifier or similar device.
Figure 2-5 shows a distribution obtained by discriminating
Each pixel in the MPPC outputs a pulse at the same amplitude the accumulated charge amount. Each peak from the left
when it detects a photon. Pulses generated by multiple pixels corresponds to the pedestal, one photon, two photons, three
are output while superimposed onto each other. For example, photons and so on. Since the MPPC gain is high enough to
if four photons are incident on different pixels and detected produce a large amount of output charge, the distribution
at the same time, then the MPPC outputs a signal whose can show discrete peaks according to the number of detected
amplitude equals the height of the four superimposed pulses. photons.
Each pixel outputs only one pulse and this does not vary with
the number of incident photons. So the number of output [Figure 2-5] Pulse height spectrum when using charge amplifier
pulses is always one regardless of whether one photon or two (S12571-050C, M=1.25 × 106)
or more photons enter a pixel at the same time. This means (M=1.25 × 106)
3000
that MPPC output linearity gets worse as more photons are Pedestal

incident on the MPPC such as when two or more photons 2500

Frequency (number of events)

enter one pixel. This makes it essential to select an MPPC

having enough pixels to match the number of incident 2000
photons.
The following two methods are used to estimate the number 1500

of photons detected by the MPPC.

1000
· Observing the pulse
· Measuring the output charge 500

(1) Observing pulses 0

0 1 2 3 4 5 6 7 8 9 10
When light enters an MPPC at a particular timing, its output
pulse height varies depending on the number of photons Number of detected photons
KAPDB0133EA
detected. Figure 2-4 shows output pulses from the MPPC
obtained when it was illuminated with the pulsed light at
photon counting levels and then amplified with a linear
amplifier and observed on an oscilloscope. As can be seen
2-2 Features
from the figure, the pulses are separated from each other
according to the number of detected photons such as one, Low afterpulses
two, three photons and so on. Measuring the height of each
pulse allows estimating the number of detected photons. When detecting photons with an MPPC, signals delayed from
the output signal may appear again. These signals are called
afterpulses. Compared to our previously marketed products,
new MPPCs have drastically reduced afterpulses due to
use of improved materials and wafer process technologies.
Reducing afterpulses brings various benefits such as a better
S/N, a wider operating voltage range, and improved time
resolution and photon detection efficiency in high voltage
regions.

10
[Figure 2-6] Pulse waveforms [Figure 2-8] Photon detection efficiency vs. overvoltage
(a) S10362-11-050C (previous product) (Typ. Ta=25 °C, λ=408 nm)
50
S12571-100C
(M=1.25 × 106)
100 μm pitch
Previous product

Photon detection efficiency (%)

40 100 μm pitch

S12571-025C
50 mV

25 μm pitch
30
S12571-050C
50 μm pitch
Previous product
20 50 μm pitch

10
Previous product
25 μm pitch
10 ns
0
0 1 2 3 4 5
(b) S12571-050C (improved product)
(M=1.25 × 106) Overvoltage (V)

Photon detection efficiency does not include crosstalk and afterpulses.

KAPDB0217EB
50 mV

[Table 2-1] Recommended overvoltage

Recommended overvoltage
Pixel pitch (V)
(µm)
Previous product S12571 series
25 2.3 3.5
10 ns 50 1.5 2.6
100 1.0 1.4
[Figure 2-7] Afterpulses vs. overvoltage
(Typ. Ta=25 °C)
40 Vov = Vop - VBR ............ (1)

Vov: overvoltage
Vop: operating voltage
30 VBR : breakdown voltage
Afterpulses (%)

Previous product
20
Wide dynamic range

The MPPC dynamic range is determined by the number

10 Improved product of pixels and the pixel recovery time. Hamamatsu has
developed the MPPC with the smallest pixel pitch of 10
µm, which increases the number of pixels per unit area
0
0 1 2 3 4 and shortens the recovery time. This drastically extends
the MPPC dynamic range.
Overvoltage (V)
KAPDB0256EA
[Figure 2-9] Pulse waveforms
(a) High-speed, wide dynamic range type
High photon detection efficiency S12571-010C (pixel pitch: 10 µm)
(M=5.15 × 105)
The MPPC has a peak sensitivity at a wavelength around
400 to 500 nm. The MPPC sensitivity is referred to as photon
detection efficiency (PDE) and is calculated by the product of
30 mV

the quantum efficiency, fill factor, and avalanche probability.

Among these, the avalanche probability is dependent on
the operating voltage. Our 25 µm pitch MPPC is designed
for a high fill factor that vastly improves photon detection
efficiency compared to our previous types. Using this same
design, we also developed 10 µm and 15 µm pitch MPPCs
10 ns
that deliver a high-speed response and wide dynamic range
as well as high photon detection efficiency. The fill factor of
50 µm and 100 µm pitch MPPC is the same as that of previous
types, but increasing the overvoltage improves photon
detection efficiency.
11
 (b) General measurement type S12571-050C (pixel pitch: 50 µm) [Figure 2-11] Crosstalk vs. overvoltage
(M=1.25 × 106) (Typ. Ta=25 °C)
50

40 Previous product
50 mV

50 μm pitch

Crosstalk (%)
30

20
Improved product
S13081-050CS
50 μm pitch
10
10 ns

0
Low crosstalk 0 1 2 3 4 5

Overvoltage (V)
KAPDB0257EB
The pixel that detects photons may affect other pixels,
making them produce pulses other than output pulses.
This phenomenon is called crosstalk. Hamamatsu has Metal quenching resistor
drastically reduced the crosstalk in precision measurement
MPPC by creating barriers between pixels. Due to the use of a metal quenching resistor, the temperature
coefficient of the resistance is reduced to 1/5 of the previous
[Figure 2-10] Pulse waveforms
type. This suppresses changes in the falling pulse edge
(a) Precision measurement type S12571-050C (pixel pitch: 50 µm) especially at low temperatures and so improves the output
(M=1.25 × 106) waveform.
For information on the usable temperature range, refer to
the datasheets.
70 mV

[Figure 2-12] Pulse fall time vs. temperature

(photosensitive area: 1 mm sq, pixel pitch: 50 µm)
(Typ.)
100

80
Previous product
Pulse fall time (ns)

10 ns (Previous quenching resistor)

60
(b) General measurement type S13081-050CS (pixel pitch: 50 µm)
(M=1.25 × 106) 40

Improved product
20 (Metal quenching resistor)
70 mV

0
200 250 300

Temperature (K)
KAPDB0258EA

10 ns
2-3 Characteristics
MPPC lineup and characteristics

To meet a diverse range of applications and needs, Hamamatsu

provides a full lineup of MPPC types in different pixel sizes
and photosensitive areas. The MPPC packages include metal,
ceramic, PWB (printed wiring boards), and CSP (chip size
packages). As multichannel array detectors, Hamamatsu also
provides MPPC arrays having uniform characteristics on
each channel and narrow dead space between the channels.

12
MPPC types with a larger pixel size are suitable for applications The charge Q depends on the reverse voltage (VR) and
where a high gain and high photon detection efficiency are breakdown voltage (VBR) and is expressed by equation
required, while types with a smaller pixel size are suitable (3).
for applications requiring high-speed response and a wide
dynamic range. Types with a larger photosensitive area Q = C × (VR - VBR) ……… (3)
are suitable for a wide-dynamic-range measurement or
C: capacitance of one pixel
detection of light incident on a large area, while types with
smaller photosensitive area are suitable for applications Equations (2) and (3) indicate that the larger the pixel
where a high speed and low dark count are needed. capacitance or the higher the reverse voltage, the higher
The MPPC characteristics vary with the operating voltage. the gain will be. On the other hand, increasing the reverse
To deal with various applications, the MPPC operating voltage voltage also increases the dark and afterpulses. So the reverse
can be adjusted as desired over a wide setting range. To voltage must be carefully set to match the application.
obtain an optimum MPPC performance, the operating
voltage should be set higher in applications requiring a (2) Linearity
high gain, high photon detection efficiency, and superior
As the reverse voltage is increased, the MPPC gain also
time resolution, while it should be set lower in applications
increases almost linearly. Figure 2-13 shows a typical
requiring low noise (low dark , low crosstalk , and low
example.
afterpulses).
[Figure 2-13] Gain vs. reverse voltage
[Table 2-2] MPPC characteristics versus pixel size (photosensitive area: 1 mm sq, pixel pitch: 10 µm)

Pixel size Small Large (Typ. Ta=25 °C)

2.0 × 105

Gain 1.8 × 105

Photon detection efficiency 1.6 × 105

1.4 × 105
Dynamic range
1.2 × 105
High-speed response
Gain

1.0 × 105

8.0 × 104
[Table 2-3] MPPC characteristics versus photosensitive area
6.0 × 104

Photosensitive area Small Large 4.0 × 104

Dynamic range 2.0 × 104

0
Light detection over a large area 68 69 70 71 72 73 74

High-speed response Reverse voltage (V)

KAPDB0226EA
Dark
(3) Temperature characteristics
[Table 2-4] MPPC characteristics versus operating voltage
As with the APD, the MPPC gain is also temperature
Operating voltage Low High dependent. As the temperature rises, the crystal lattice
vibrations become stronger. This increases the probability
Gain
that carriers may strike the crystal before the accelerated
Photon detection efficiency carrier energy has become large enough, making it difficult
Time resolution for ionization to continue. To make ionization easier to
occur, the reverse voltage should be increased to enlarge the
Noise
internal electric field. To keep the gain constant, the reverse
voltage must be adjusted to match the ambient temperature
Gain or the element temperature must be kept constant.
Figure 2-14 shows the reverse voltage adjustment needed
to keep the gain constant when the ambient temperature
(1) Definition
varies.
The MPPC gain is defined as the charge (Q) of the pulse
generated from one pixel when it detects one photon,
divided by the charge per electron (q: 1.602 × 10-19 C).

Q ………
M= (2)
q

M: gain

13
 [Figure 2-14] Reverse voltage vs. ambient temperature [Figure 2-16] Dark pulses
(photosensitive area: 1 mm sq, pixel pitch: 10 µm)
5) Incident light
(Typ. (Typ.
Ta=25M=1.35 × 10V)
°C, VR=69
73 timing

72 Dark pulses
Reverse voltage (V)

71
MPPC
output

70
Time
KAPDC0043EA
69

The number of dark pulses observed is referred to as the

68
dark count, and the number of dark pulses per second is
67 termed as the dark count rate [unit: cps (counts per second)].
-20 -10 0 10 20 30 40 50
The dark count rate of Hamamatsu MPPC is defined as the
Ambient temperature (°C) number of pulses that are generated in a dark state and
KAPDB0227EA
exceed a threshold of 0.5 p.e. This is expressed as N0.5 p.e..

Figure 2-15 shows the relation between gain and ambient (2) Temperature characteristics
temperature when the reverse voltage is a fixed value.
Since dark pulses are produced by thermally-generated
[Figure 2-15] Gain vs. ambient temperature carriers, the dark count rate varies with the ambient
(photosensitive area: 1 mm sq, pixel pitch: 10 µm) temperature. The dark count rate is given by equation (4)
(Typ. VR=Vop at 25 °C) within the operating temperature range.
2.0 × 105

3
Eg ……… (4)
N 0.5 p.e.(T) ≈ AT 2 exp
2k T

T : absolute temperature [K]

A : arbitrary constant
Gain

1.5 × 105 Eg : band gap energy [eV]

k : Boltzmann’s constant [eV/K]

Figure 2-17 shows a relation between the dark count rate

and the ambient temperature when the gain is set to a
constant value.
1.0 × 105
0 5 10 15 20 25 30 35 40
[Figure 2-17] Dark count rate vs. ambient temperature
Ambient temperature (°C)
(photosensitive area: 3 mm sq, pixel pitch: 50 µm)
KAPDB0228EA (Typ. M=1.25 × 106)
10000

Dark count
1000
Dark count rate (kcps)

(1) Definition
In the MPPC operation just the same as with APD, pulses 100

are produced not only by photon-generated carriers but

also by thermally-generated carriers. The pulses produced
10
by the latter are called the dark pulses. The dark pulses are
observed along with the signal pulses and so cause detection
errors. Thermally-generated carriers are also multiplied to 1
a constant signal level (1 p.e.). These dark pulses are not -25 -20 -15 -10 -5 0 5 10 15 20 25

distinguishable by the shape from photon-generated pulses Ambient temperature (°C)

[Figure 2-16]. KAPD0218EA

14
 Crosstalk Afterpulses

When light enters one MPPC pixel, there may be cases where a During the avalanche multiplication process in MPPC pixels,
pulse of 2 p.e. or higher is observed. This is because secondary the generated carriers may be trapped by lattice defects.
photons are generated in the avalanche multiplication When these carriers are released, they are multiplied by the
process of the MPPC pixel and those photons are detected avalanche process along with photon-generated carriers
by other pixels. This phenomenon is called the optical and are then observed as afterpulses. The afterpulses are not
crosstalk. distinguishable by shape from photon-generated pulses.

[Figure 2-18] Crosstalk example [Figure 2-20] Afterpulse observed

3 p.e. 3 p.e.

2 p.e. 2 p.e.

Pulse height
Pulse height

1 p.e. 1 p.e.

Time Time

We define the crosstalk probability (Pcrosstalk) as equation

(5). Dark current

N 1.5 p.e. ……… Output current produced even when operated in a dark
Pcrosstalk = (5)
N 0.5 p.e. state is called the dark current. The MPPC dark current
(ID) is expressed by equation (6).
The crosstalk probability has almost no dependence on
the temperature within the rated operating temperature I D = Is + Ij + Ib ……… (6)
range. The probability that the crosstalk will occur increases
as the reverse voltage is increased [Figure 2-19]. Is: surface leakage current
Ij: recombination current
Ib: bulk current
[Figure 2-19] Crosstalk probability vs. reverse voltage
(photosensitive area: 1 mm sq, pixel pitch: 50 µm)
When the MPPC is operated in Geiger mode, the bulk
(Typ. Ta=25 °C,Ta=25
(Typ. VR=69°C)
V) current is expressed by equation (7), assuming that the
0.2
number of pixels in which avalanche multiplication occurs
per unit time is Nfired.
Crosstalk probability

Ib = q M Nfired ……… (7)

0.1 q: electron charge

M: gain

Since the MPPC gain is usually 105 to 106, the bulk current
Ib is dominant in equation (6) and equation (7) can then
be approximated to equation (8).
0
68 69 70 71 72 73 74 75
I D ≈ Ib = q M Nfired ……… (8)
Reverse voltage (V)
KAPDB0229EA In a dark state, the number of pixels where avalanche
multiplication occurred equals the dark count rate, so the
dark current ID can be approximated to equation (9) using
N0.5 p.e. and Pcrosstalk. If the gain and crosstalk probability
at a particular reverse voltage are known, then the dark
current can be roughly estimated from the dark count rate
and vice versa.

1 ……… (9)
I D ≈ q M N 0.5 p.e.
1 - Pcrosstalk
15
 The PDEcurrent, which is determined from photosensitivity,
Photosensitivity and photon detection efficiency
is expressed by equation (13).

The photosensitivity and the photon detection efficiency 1240 S ………

PDEcurrent = × (13)
are used to express the MPPC light detection sensitivity. The λ M
photosensitivity is expressed as the ratio of the MPPC output
λ: incident light wavelength [nm]
current (analog value) to the amount of continuous light
incident on the MPPC. The photon detection efficiency is a
PDEcurrent includes crosstalk and afterpulses, and so
ratio of the number of detected photons to the number of
PDEcurrent becomes higher than the PDE.
incident photons during photon counting where the pulsed
light enters the MPPC. Both photosensitivity and photon [Figure 2-21] Spectral response (pixel pitch: 25 µm)
detection efficiency relate to parameters such as fill factor,
(Typ. Ta=25 °C, VR=69 V)
quantum efficiency, and avalanche probability. 1.4 × 105

The fill factor is the ratio of the light detectable area to

1.2 × 105
the entire pixel area of an MPPC. Unlike photodiodes and

Photosensitivity (A/W)
APD, the MPPC photosensitive area contains sections 1.0 × 105

such as the inter-pixel wiring that cannot detect light, so

8.0 × 104
some photons incident on the photosensitive area are not
detected. Generally, the smaller the pixel size, the lower 6.0 × 104
the fill factor.
4.0 × 104
The quantum efficiency is defined as probability that
carriers will be generated by light incident on a pixel. As in 2.0 × 104
other types of opto-semiconductors, the MPPC quantum
0
efficiency is dependent on the incident light wavelength. 300 400 500 600 700 800 900

The avalanche probability is the probability that the carriers

Wavelength (nm)
generated in a pixel may cause avalanche multiplication. KAPDB0230EA

The higher reverse voltage applied to the MPPC, the higher

avalanche probability. [Figure 2-22] Photosensitivity vs. reverse voltage
(pixel pitch: 25 µm)
(1) Photosensitivity
(Typ. Ta=25 °C, λ=408 nm)
2.0 × 105
Photosensitivity (S; unit: A/W ) is a ratio of the MPPC
1.8 × 105
photocurrent to the light level (unit: W ) incident on the
1.6 × 105
MPPC, as expressed by equation (10).
Photosensitivity (A/W)

1.4 × 105

I MPPC 1.2 × 105

S= ……… (10)
Incident light level 1.0 × 105

8.0 × 104
IMPPC: photocurrent [A]
6.0 × 104

The photosensitivity is proportional to the gain, so the 4.0 × 104

higher the reverse voltage applied to the MPPC, the higher 2.0 × 104

the photosensitivity. Note that the photosensitivity includes 0

66 67 68 69 70 71
a crosstalk and afterpulses.
Reverse voltage (V)
(2) Photon detection efficiency KAPDB0231EA

The photon detection efficiency (PDE) is an indication of

what percent of the incident photons is detected, and is
given by equation (11).

Number of detected photons ………

PDE = (11)
Number of incident photons

The PDE can be expressed by the product of a fill factor,

quantum efficiency, and avalanche probability.

PDE = Fg × QE × Pa ……… (12)

Fg : fill factor
QE: quantum efficiency
Pa : avalanche probability

16
[Figure 2-23] Photon detection efficiency vs. wavelength [Figure 2-25] TTS vs. overvoltage
(pixel pitch: 25 µm) (photosensitive area: 1 mm sq, pixel pitch: 50 µm)
(Typ. Ta=25 °C, VR=69 V) (Typ. Ta=25 °C)
50 0.7

0.6
Photon detection efficiency (%)

40

0.5

TTS [FWHM] (ns)

30
0.4

0.3
20

0.2
10
0.1

0 0
300 400 500 600 700 800 900 0 0.5 1.0 1.5 2.0 2.5

Wavelength (nm) Overvoltage (V)

KAPDB0232EA

Photon detection efficiency does not include crosstalk and afterpulses.

KAPDB0259EA

[Figure 2-24] Photon detection efficiency vs. reverse voltage 2-4 How to use
(pixel pitch: 25 µm)
(Typ. Ta=25 °C, λ=408 nm)
The MPPC characteristics greatly vary depending on the
50
operating voltage and ambient temperature. In general,
raising the operating voltage increases the electric field
Photon detection efficiency (%)

40 inside the MPPC and so improves the gain, photon detection

efficiency, and time resolution. On the other hand, this
30 also increases unwanted components such as dark count,
afterpulses, and crosstalk which lower the S/N. The operating
20
voltage must be carefully set in order to obtain the desired
characteristics.
The MPPC can be used by various methods according to
10
the application. Here we introduce a typical method for
observing light pulses. Using a wide-band amplifier and
0
66 67 68 69 70 71 72 73 oscilloscope makes this measurement easy. Figure 2-26 shows
one example of a connection to a wide-band amplifier. The
Reverse voltage (V)
1 kΩ resistor and 0.1 µF capacitor on the power supply portion
Photon detection efficiency does not include crosstalk and afterpulses. serve as a low-pass filter that eliminates high-frequency noise
KAPDB0233EA
of the power supply. The 1 kΩ resistor is also a protective
resistor against excessive current. The MPPC itself is a low-
Time resolution light-level detector, however, in cases where a large amount
of light enters the MPPC, for example, when it is coupled to a
The time required for each pixel of the MPPC to output scintillator to detect radiation, a large current flows into the
a signal after the incidence of light varies depending on MPPC. This may cause a significant voltage drop across the
the wiring length, etc. This variation is called TTS (transit protective resistor, so the protective resistor value must be
time spread). Increasing the reverse voltage applied to carefully selected according to the application. The amplifier
the MPPC reduces and improves the TTS. should be connected as close to the MPPC as possible.

[Figure 2-26] Connection example

+V

1 kΩ
0.1 μF

MPPC

Signal
Amplifier
KAPDC0024EB

17
 In measurements utilizing the MPPC output pulse having corresponds to the amount of the charge produced by
a sharp rising edge, an appropriate wide-band amplifier detecting one photon. The gain (M) is given by equation
and oscilloscope must be selected. Since the MPPC output (14).
pulses usually rise within a few nanoseconds, it is strongly
recommended to use an amplifier and oscilloscope capable Charge difference between adjacent peaks ………
M= (14)
q
of sampling at about 1 GHz. Using a narrow-band amplifier
and oscilloscope might dull or blunt the output pulse q: electron charge
making it impossible to obtain accurate values.
Furthermore, equation (14) can be used to create and
extrapolate a gain vs. reverse voltage graph as shown in
2-5 Measurement examples Figure 2-13 to determine the reverse voltage for gain of 1,
or namely the breakdown voltage VBR.
Examples of measuring MPPC characteristics are described
(2) dI/dV measurement
below.
Figure 2-29 shows the output current vs. reverse voltage
characteristics of the MPPC. If the voltage of Vpeak
Gain
maximizes the value to the function [equation (15)]
obtained by differentiating the output current by the
(1) Measurement using a charge amplifier
reverse voltage, Vpeak - VBR is approximately constant for
The gain can be estimated from the output charge of each type no., but the individual values Vpeak and VBR
the MPPC that detected photons. Figure 2-27 shows a are different between elements even with the same type
connection setup example for the gain measurement no. By determining Vpeak - VBR for a given type no. in
using a charge amplifier. advance, you will be able to estimate VBR for a particular
element by measuring Vpeak.
[Figure 2-27] Gain measurement connection example
(using charge amplifier) [Figure 2-29] Output current vs. reverse voltage
(photosensitive area: 1 mm sq, pixel pitch: 50 µm)
Optical fiber Photons
(Typ. Ta=25 °C)
10-5

Attenuator
Pulse light MPPC Charge A/D MCA PC 10-6
source amp converter
KAPDC0046EB
Output current (A)

10-7
When the MPPC is illuminated with pulsed light whose
light level is sufficiently reduced by an attenuator and 10-8

the number of the output charges is plotted, a frequency

10-9
distribution like that shown in Figure 2-28 is obtained.

[Figure 2-28] Frequency distribution example of 10-10

output charge
10-11
1600 64 65 66 67 68 69 70
1 p.e.
1400 2 p.e. Reverse voltage (V)
Frequency (number of events)

KAPDB0235EA
1200

1000 d dI 1 ………
log(I) = × (15)
dV R dV R I
800
I : output current [A]
600 VR: reverse voltage [V]

400

200

0
1 131 261 391 521 651 781 911 1041 1171 1301 1431 1561 1691

Channel
KAPDB0136EA

In Figure 2-28, each peak on the curve from the left

indicates the pedestal, one photon, two photons and so
forth. The pedestal is a basis of the output. This example
shows that the MPPC has mainly detected one photon
and two photons. The interval between adjacent peaks
18
 dI 1 If the threshold is swept, the dark count rate will be plotted
[Figure 2-30] Reverse voltage characteristics of ×
dVR I
as shown in Figure 2-32. The threshold voltages at which
(Typ. Ta=25 °C)
10 the dark count rate abruptly decreases correspond to the
9 levels of one photon, two photons, and so on from left. The
8
dark count rates N0.5 p.e., N1.5 p.e., N2.5 p.e. and so on can be
7
obtained from this graph.

6
[Figure 2-32] Dark count rate vs. threshold voltage
dI × 1
I

5
(Typ. Ta=25 °C)
dVR

106
4

3
105
2

Dark count rate (cps)

1 104
Vpeak
0
64 65 66 67 68 69 70
103

Reverse voltage (V)

KAPDB0236EA 102

The gain (M) is given by equation (16). 101

C × (VR - VBR) ……… (16)

100
M= 0 0.1 0.2 0.3
q
Threshold voltage (V)
C : pixel capacitance [F] KAPDB0237EA
VBR : breakdown voltage [V]
q : electron charge [C]

Afterpulses
Since the pixel capacitance is constant, the gain can be
obtained from the breakdown voltage and reverse voltage
The dark pulses are generated randomly and the time
that are obtained by dI/dV measurement. However, if the
interval of the dark pulse generation follows an exponential
operating voltage applied to the MPPC is significantly
distribution. The dark pulse generation time interval Δtdark
higher than the recommended operating voltage, noise
(unit: seconds) is expressed by equation (17).
components such as afterpulses and crosstalk will increase
and make accurate measurement impossible. Δt
Δt dark ∝ exp ( )
τ dark
……… (17)

Dark τdark: time constant of dark pulse generation [s]

The MPPC is installed and operated in a dark box and the The time interval during afterpulse generation is expressed
output pulse is input to a pulse counter. The number of by the sum of several exponential distributions. The
events where the output pulse exceeds the predetermined afterpulse generation time interval ΔtAP (unit: seconds) is
threshold (0.5 p.e., etc.) is counted to determine the dark given by equation (18).
count rate. In this case, a wide-band amplifier must be
Δt
used because the MPPC output pulse width is very short,
down to a few dozen nanoseconds.
Δt AP ∝ Σk Ak ×exp ( )
τk
……… (18)

k : number of time constants for Δt AP

[Figure 2-31] Block diagram of dark measurement Ak : constant
τk : time constant of afterpulse generation [s]

Wide-bandwidth
MPPC
amplifier
Comparator Counter Here, τdark differs greatly from τk (τdark >> τk), so it
is necessary to create a histogram of the elapsed time Δt
Dark box after the generation of a given pulse until the next pulse
KAPDC0044EA
is observed and then estimate dark pulse components in
the time region that does not include afterpulses. Then,
Crosstalk subtracting the fitted components from the entire histogram
gives the afterpulse components.
When the threshold is set, for example, to 0.5 p.e. and 1.5 During measurement, a discriminator, TAC, and MCA are
p.e., to measure the count rate of dark pulses exceeding used to create the above mentioned histogram. The photons
the threshold, the dark count rates N0.5 p.e. and N1.5 p.e. at enter the MPPC and the detected signals are amplified
each threshold can be measured. The crosstalk probability by the amplifier and sent to the discriminator. When the
Pcrosstalk is calculated by equation (5). discriminator receives a signal with an amplitude exceeding
the threshold for photon detection, it sends the signal to
19
 the TAC. When the next signal is output from the MPPC, [Figure 2-35] Connection example of photon detection
that signal is also sent to the TAC. The TAC then outputs a efficiency measurement
pulse whose amplitude is proportional to the time interval Calibrated photodiode
between the first MPPC signal and the next MPPC signal. Ammeter
The MCA sorts the pulses received from the TAC into different Optical fiber MPPC

channels according to pulse height. The data stored in the Pulse light
source
MCA displays a histogram of Δt. Amplifier
Oscilloscope PC
Attenuator Integrating sphere
[Figure 2-33] Connection example of afterpulse Trigger signal
measurement
KAPDC0051EA

Wide-bandwidth
MPPC Discriminator TAC MCA
amplifier The MPPC output signal is fed to an oscilloscope in
synchronization with the trigger signal from the pulsed
TAC : time-to-amplitude converter light source to measure the MPPC output waveform in
MCA: multichannel analyzer
KAPDC0045EB response to the pulsed light. The MPPC output charge is
then obtained from the response waveform. This output
charge is obtained for many events to create a frequency
Photosensitivity
distribution of the output charge like that shown in Figure
2-28. In an ideal case, when the pulsed light is so weak that
To measure the photosensitivity of an MPPC, the incident
only a few photons are emitted per pulse, this frequency
light from a monochromatic light source is first detected
distribution follows a Poisson distribution with a mean
by a calibrated photodetector in a dark box and the light
value of the number of photons detected by the MPPC.
level (unit: W) incident on the photodetector is found from
However, part of the events contains dark pulses and the
the output. Then, the MPPC is set in the dark box in place
events at 1 p.e. or higher are affected by crosstalk and
of the photodetector to make the same measurement and
afterpulses, distorting the actually measured distribution
the MPPC photocurrent (unit: A) is measured. Based on
from the Poisson distribution. On the other hand, since
these measurement results, the photosensitivity (S) of the
the event at pedestal is not affected by crosstalk and
MPPC is calculated as in equation (19).
afterpulses, the effects of dark pulses can be corrected on
I MPPC the basis of the number of these events and so the mean
S= ……… (19)
Incident light level value of the Poisson distribution can be found.
The Poisson distribution is defined by equation (20).
IMPPC: photocurrent [A]

n x e -n ………
[Figure 2-34] Connection example of photosensitivity P(n, x) = (20)
x!
measurement
Calibrated photodetector or MPPC n: average number of photons detected by MPPC
Dark box x: number of photons detected by MPPC

If x=0 in equation (20), then the Poisson distribution is

White Monochromatic expressed by equation (21).
Light source light Monochromator light Ammeter

KAPDC0050EA P(n, 0) = e -n ……… (21)

The left side of equation (21) is expressed by equation

Photon detection efficiency
(22) when the correction of dark pulses is included.

To measure the photon detection efficiency of an MPPC,

a pulsed light source is used as shown in Figure 2-35. The
P(n, 0) =
( )
N ped
N tot
……… (22)
monochromatic pulsed light emitted from the pulsed dark

light source is passed through an attenuator to reduce ( )

N ped
dark
N tot
the light level and is guided into an integrating sphere
Nped : number of events at 0 p.e. during pulsed light measurement
where the light is scattered and distributed equally in all
Ntot : number of all events during pulsed light measurement
directions. And then it enters a calibrated photodiode dark
Nped : number of events at 0 p.e. in dark state
dark
and the MPPC. The output current from the calibrated Ntot : number of all events in dark state

photodiode is measured with an ammeter and, based on

that value, the number of photons incident on the MPPC The average number of photons detected by MPPC, n, is
is found.2) given by equation (23). Photon detection efficiency can
then be found by dividing n by the number of incident
photons.

20
 Dynamic range

( )
N ped
dark
n =-ln
N tot
dark
N ped
=-ln ( ) ( )
N ped
N tot
+ ln
N ped
dark
N tot
……… (23)
(1) Dynamic range for simultaneously incident photons
dark
N tot
The dynamic range for simultaneously incident photons is
determined by the number of pixels and photon detection
Time resolution efficiency of the MPPC. As the number of incident photons
increases, two or more photons begin to enter one pixel.
Figure 2-36 is an example of connection for time resolution Even when two or more photons enter one pixel, each pixel
measurement using the TTS method. The pulse light source can only detect whether or not the photons entered the
emits photons and simultaneously sends a start signal to MPPC. This means that the output linearity degrades as the
the TAC. The TAC starts measuring the time upon receiving number of incident photons increases.
the start signal. Meanwhile, the photons enter the MPPC
and the detected signals are amplified by the amplifier and
sent to the discriminator. When the discriminator receives
Nfired = Ntotal × 1 - exp ( -Nphoton × PDE
Ntotal
) …… (24)

a signal with an amplitude exceeding the threshold for Nfired : number of excited pixels
photon detection, it sends the signal to the TAC. The TAC Ntotal : total number of pixels
Nphoton : number of incident photons
receives the signal from the discriminator as a stop signal PDE : photon detection efficiency
for time measurement. At this point, the TAC also provides
a pulse output proportional to the time from when photons Widening the dynamic range requires using an MPPC
entered the MPPC until the signal is output. The MCA sorts having a sufficiently large number of pixels compared to
the pulses received from the TAC into different channels the number of simultaneously incident photons (namely,
according to pulse height. The data stored in the MCA is a an MPPC with a large photosensitive area or a narrow
histogram of MPPC responses, and the time resolution is pixel pitch).
expressed as the full width at half maximum (FWHM) of
this histogram. [Figure 2-38] Dynamic range for simultaneously incident
photons (pixel pitch: 50 µm)
[Figure 2-36] Connection example of time resolution (Typ. Ta=25 °C)
105
measurement
Photons Stop signal
Photosensitive area 3 × 3 mm
104
Number of excited pixels

MPPC 103
Pulse light Amplifier Discriminator TAC MCA
source

Photosensitive area
102 1 × 1 mm

Start signal
KAPDC0030EA 101

[Figure 2-37] TTS (typical example) 100

100 101 102 103 104 105 106
(1 ch=2.6 ps)
1400
Number of simultaneously incident photons
1200 KAPDB0238EA
Frequency (number of events)

1000 (2) Dynamic range in photon counting

800 The number of MPPC excited pixels is given by equation

FWHM (25).
600

400 Nfired = Nphoton × PDE ……… (25)

200
As the number of incident photons becomes larger, two or
0
more output pulses overlap each other causing counting
4100 4200 4300 4400 4500 4600 4700
errors and degrading the output linearity. This linearity is
Channel determined by a parameter called the pulse-pair resolution.
KAPDB0137EA
The pulse-pair resolution is determined by the MPPC recovery
time (refer to “Recovery time” in section 2-5, “Measurement
examples”) and the readout circuit characteristics. Equation
(26) expresses the number of MPPC excited pixels that takes
into account the pulse-pair resolution.

21
 Nphoton × PDE ……… (26) resistor used. So a protective resistor having the right
Nfired =
1 + Nphoton × PDE × Tresolution value must be selected to prevent this problem.
Tresolution: pulse-pair resolution
[Figure 2-40] Output current vs. incident light level
(pixel pitch: 15 µm)
To widen the dynamic range, an MPPC with a short recovery
(Typ. Ta=25 °C)
time should be selected. 10-1

10-2
[Figure 2-39] Dynamic range in photon counting 10-3
(pixel pitch: 15 µm)
10-4

Output current (A)

(Typ. Ta=25 °C,Ta=25
(Typ. VR=69°C)
V)
109 10-5

10-6
Number of detected photons (cps)

108
10-7

10-8
107
10-9
106 10-10 Ideal value
S12751-015C
-11
10 S12752-015C
105
10-12
10-18 10-15 10-12 10-9 10-6
104
Incident light level (W)
103 Ideal value KAPDB0240EA

Pixel pitch 15 μm
102
103 104 105 106 107 108 109
Recovery time
Number of incident photons (cps)
KAPDB0239EA
The time (recovery time) required for pixels to restore
(3) Dynamic range in current measurement 100% of the gain depends on the photosensitive area and
pixel size. In the case of the MPPC having a photosensitive
The MPPC photocurrent (IMPPC) is expressed by equation area of 1 mm sq, the recovery time will be approximately
(27). 20 ns for 25 µm pixel pitch, 50 ns for 50 µm pixel pitch,
and 100 to 200 ns for 100 µm pixel pitch. Figure 2-41 shows
I MPPC = Nphoton × PDEcurrent × M × q ……… (27)
an output measured when light enters a pixel of the MPPC
PDEcurrent : PDE determined from photosensitivity with a photosensitive area of 1 mm sq, and a pixel pitch of
M : gain
q : electron charge
50 µm, at a period equal to the pulse recovery time. It can
be seen that the pulse is restored to a height equal to 100%
The number of incident photons is expressed by equation of output.
(28) using the incident light level (unit: W).
[Figure 2-41] Pulse level recovery
(photosensitive area: 1 mm sq, pixel pitch: 50 µm)
Incident light level × λ ………
Nphoton = (28)
h ×c

λ: wavelength [m]
Voltage

h: Planck’s constant
c: speed of light
50 ns (20 MHz)

As the incident light level increases, two or more photons

tend to enter one pixel, also the next photon tends to Time
enter the same pixel within its recovery time. These KAPDB0158EA

actions degrades the linearity. Equation (29) expresses

If the next input pulse enters before the output pulse is
the MPPC output current IMPPC while taking these actions
completely restored, then a small pulse is output, which
into account.
does not reach the gain set by the operating voltage. In
Nphoton × PDEcurrent Figure 2-41, the rising region of the pulse indicates the
I MPPC = × M × q …… (29)
Nphoton × PDEcurrent × TR process for charging the pixel. When the next photon
1+
Ntotal is detected before the pixel is fully charged, the output
pulse will have an amplitude that varies according to the
TR: recovery time [s]
charged level.
Figure 2-42 shows output pulse shapes obtained when
When a large amount of light enters an MPPC, the output
light at different frequencies was input to a pixel. It
linearity may deteriorate because the MPPC element
can be seen that as the frequency of the incident light
heats up and the gain lowers. A large output current
increases, the pulse height decreases because the pixel is
then flows which might lower the reverse voltage applied
not fully charged.
to the MPPC depending on the value of the protective
22
[Figure 2-42] Output pulses obtained when light at [Figure 2-43] Output waveforms
different frequencies was input
(photosensitive area: 1 mm sq, pixel pitch: 50 µm) (a) Light level is low (very low light level)

[Pulse length 50 ns max. (1/50 ns=20 MHz)]

Voltage
Voltage

100 MHz
50 MHz
20 MHz
Time

(b) Light level is moderate

Time
KAPDB0163EA

2-6 Selecting digital mode or

Voltage
analog mode
The readout mode (digital mode or analog mode) should
be selected according to the light level incident on the
MPPC.
Figures 2-43 (a), (b), and (c) show the MPPC output
waveforms measured at different incident light levels Time
and observed on an oscilloscope. The incident light level
was increased in the order of (a), (b), and (c), starting (c) Light level is high
from (a) at very low light levels. The output signal of (a)
as seen here consists of discrete pulses. In this state,
selecting the digital mode allows measuring at a higher
S/N, where the signals are binarized and the number of
Voltage

pulses is digitally counted. Since the digital mode can

subtract the dark count from the signal, the detection
limit is determined by dark count fluctuations.
As the light level increases, the output waveform consists
of pulses overlapping each other [(b), (c)]. In this state,
the number of pulses cannot be counted and the analog
mode should be selected to measure the analog output
Time
and find the average value. The detection limit in the
analog mode is determined by the dark current shot
noise and the cutoff frequency of the readout circuit.
Figure 2-44 shows the incident light levels suitable for
the digital mode and analog mode (MPPC photosensitive
area: 3 × 3 mm, pixel pitch: 50 µm).

[Figure 2-44] Incident light levels suitable for the digital mode and analog mode (photosensitive area: 3 × 3 mm, pixel pitch: 50 µm)

Analog mode

Digital mode
Number of incident
photons (cps) 100 101 102 103 104 105 106 107 108 109 1010 1011 1012

Incident light level (W) 10-18 10-15 10-12 10-9 10-6

KAPDC0688EA

23
 3. Applications

3-1 Optical rangefinders

The distance to an object can be determined by directing
laser light onto an object and then the APD measuring
the time required for the reflected light to return or the
phase difference of the light.

KACCC0748EA

3-4 Scintillation measurement

APDs or MPPCs arranged around 360° detect pair
annihilation gamma-rays to capture the target position
such as cancer tissue. APDs and MPPCs can be used with
MRI because they are not affected by magnetic fields.

Cancer position
information
KACCC0746EA

Pair annihilation
gamma-rays

3-2 Obstacle detection

Subject

The APD can be used in unmanned robots and the like to

detect obstacles. It can also be used to detect movement
of people in a particular area. APD, MPPC
KACCC0745EA

3-5 Fluorescence measurement

The MPPC can detect minute fluorescence emission of
reagents.

Reagent

MPPC
KACCC0747EA

Fluorescence

3-3 LIDAR (light detection and ranging) Sample

KACCC0749EB

The condition of the earth’s surface, particles in the

air, and cloud can be measured by directing laser onto
an object and then the APD detecting the reflected or 3-6 High energy physics experiment
scattered light.
The MPPC is a promising candidate for high-energy
accelerator experiments to discover the ultimate constituents
of matter. The European Organization for Nuclear Research
(called CERN) is presently assessing the MPPC for use
in calorimeter units needed to detect particle energy in
24
its next-generation International Linear Collider (ILC).
Moreover, in Japan, the High Energy Accelerator Research
Organization (KEK) and the Japan Atomic Energy Agency
(JAEA) are conducting a joint experiment at the Japan Proton
Accelerator Research Complex (called J-PARC) being built
in Tokai-mura (Ibaraki Prefecture). This experiment called
“T2K” (Tokai to Kamioka) will verify whether or not the
neutrino has mass, by sending neutrino beams to Super-
Kamiokande (Gifu Prefecture, about 300 km away from
Tokai-mura). A large number of MPPCs (62000 pieces) are
used in monitoring the neutrino beams in this experiment.

Reference
1) K. Sato, K. Yamamoto, K. Yamamura, S. Kamakura, S. Ohsuka et al., Applica-
tion Oriented Development of Multi-Pixel Photon Counter (MPPC), 2010 IEEE
Nuclear Science Symposium Conference Record (2010)

2) Patrick Eckert, Hans-Christian Schultz-Coulon, Wei Shen, Rainer Stamen,

Alexander Tadday et al., Characterisation Studies of Silicon Photomultipliers,
http://arxiv.org/abs/1003.6071v2

3) T. Nagano, K. Yamamoto, K. Sato, N. Hosokawa, A. Ishida, T. Baba et al., Im-

provement of Multi-Pixel Photon Counter (MPPC), 2011 IEEE Nuclear Science
Symposium Conference Record (2011)

25