New Materials for Photon Counting Avalanche Photodiodes
Josef Blazej
Czech Technical University Prague, Brehova 7, 115 19 Prague, Czech Republic
Keywords: Solid State Single Photon Detector, Gallium, Compound Semiconductors, Avalanche Photodiode (APD).
Abstract: The experimental results acquired on avalanche photodiodes based on III-V semiconductor materials and
operated as single photon counters with picosecond timing resolution are reported. The semiconductor
structures fabricated on the basis of GaAs, GaP and GaAsP have been operated in a Geiger mode and
employed in a photon counting experiment at the wavelengths from near ultraviolet to near infrared. The
dark count rates, photon counting sensitivity and timing resolution have been measured for the experimental
diode samples.
1 INTRODUCTION
The solid state photon counters with high timing
resolution are of interest of numerous electro-optical
techniques: laser range finding, optical time domain
reflectometry, time resolved spectroscopy, quantum
cryptography and others. At present, the most
promising technique to detect single photons by use
of a solid state detector is an Avalanche Photodiode
(APD) op-erated in the Geiger mode. In this
operating mode the diode is pulse biased above its
breakdown voltage; no current is flowing until an
avalanche is triggered by an incoming photon or a
thermally generated carrier. The current pulse rise
time marks the photon’s arrival time. An external
electrical circuit, either passive or active (Cova et
al., 1983), is used to quench the avalanche and to re-
apply the bias to the diode.
Our previous research and development in the
field of single photon avalanche diodes resulted in a
large aperture silicon APD based detector package
with and active quenching circuit well adopted for
applications listed above (Prochazka et al., 1996).
The quantum efficiency corresponds to silicon, it
drops for the wavelengths longer than 1.1 μm and
shorter than 0.35 μm. In an attempt to increase the
quantum efficiency in the near infrared, the
structures based on germanium doped silicon have
been tested. For individual photon detection in the
near infrared, the germanium APDs in a cryogenic
environment has been em-ployed. This package has
been used as an echo signal detector in the satellite
laser ranging system operating at so-called eye safe
wavelength 1.54 μm in the Communication
Research Labs., Tokyo, Japan (Prochazka et al.,
1996, Kunimori et al, 2000). For the
telecommunication applications at the 1.55 μm
wavelength the thermoelectrically cooled photon
counter based on an InGaAs/InP avalanche
photodiode has been developed (Prochazka, 2001).
For the applications in satellite laser ranging, the
increase of a timing resolution of the detector is
required. Recent developments resulted in a custom
designed APD on silicon having a diameter of
0.2 millimeters in diameter exhibiting a dark count
rate below 60 kHz. The APD chip is cooled by a
three stage thermoelectrical cooler and enclosed in a
miniature evacuated package. The detector is
equipped with the electronic compensation of the
detection delay dependence on optical signal
strength. This way, the detector may be employed
for signals of single up to several thousands of
photons (Kirchner et al., 1997). The timing
resolution of 60 picoseconds FWHM is achieved on
single photon signal level. It is the value three times
worse than necessary for one millimetre precision
required in this application. Additionally, the non-
standard data distribution when detecting individual
photons at the wavelength above 0.6 μm represents a
serious limitation. The timing resolution can be
compensated by data post-processing utilizing
famous detector delay stability. But the problems
with response distribution are inherent and for some
wavelength crucial. That is why, new materials for
the photon counting detection structures have been
investigated. The III-V materials have been selected
in an attempt to reach higher timing resolution and
238
Blazej, J.
New Materials for Photon Counting Avalanche Photodiodes.
DOI: 10.5220/0005656802360240
In Proceedings of the 4th International Conference on Photonics, Optics and Laser Technology (PHOTOPTICS 2016), pages 238-242
ISBN: 978-989-758-174-8
Copyright
c
2016 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
standard data distribution. The perspective
applications in the ultraviolet are another reason for
the III-V materials experiments.
2 III-V PHOTON COUNTING
AVALANCHE PHOTODIODES
The spectral sensitivity of the GaAs, GaP and
GaAsP are plotted on Figure 1. The diodes on GaP
are of special interest due to their sensitivity in the
ultraviolet. All the diodes have been manufactured
using a conventional planar technology. The active
area of the diode is octagonal with the “diameter” of
350 μm. The GaAs and GaAsP diodes are
constructed to be illuminated on an active area, due
to the technology reason, the GaP diode is of “reach
through” construction.
100 200 300 400 500 600 700 800 900 1000
0.1
1
10
100
GaAs
GaAsP
GaP
Absolute sensitivity [mA/W]
Wavelength [nm]
Figure 1: The spectral sensitivity of GaAs, GaP and
GaAsP according to (Van Stryland et al., 1996).
The diodes have been tested in a passive
quenching circuit consisting of a serial resistor
1 MegaOhm blocked by a 20 picoFarad capacitor.
The example of the output pulse of the GaAsP diode
is on Figure 2. The break voltage at a room
temperature is 24 Volts with a drift of 15 mV/K. The
break voltage of the successful GaAs and GaP
diodes was 40 Volts and 24 Volts, respectively. The
amplitude of the pulse is one half of the value of bias
above the break voltage. It indicates relatively low
serial differential resistance of the structure above its
break, 50 Ohms typically. It makes the diodes
attractive for application as photon counters
operated in an active quenching and gating mode.
The output pulses parameters were similar for all
three diode materials.
3 EXPERIMENTAL SETUP
In all the following experiments, the diodes have
been operated in an active quenching and gating
mode, using a circuit developed for silicon photon
counting detectors (Prochazka et al., 1992). The
block scheme of the test setup is on Figure 3.
Figure 2: GaAsP diode output pulse in a passive
quenching scheme, diode biased 2 V above break voltage,
bandwidth 400 MHz, fall time 1.5 ns, width 5.2 ns.
The time-correlated photon counting scheme has
been used. The laser diode (LD) provided optical
pulses 32 picoseconds long at 0.757 μm. The signal
has been attenuated using the neutral density filters
(ND). The Active Quenching and Gating Circuit
generated the NIM timing pulses. The Time to
Amplitude Converter (TAC) provided the timing
resolution of 20 picoseconds per channel. The data
from the TAC have been processed in the
Multichannel Analyzer Card in a personal computer.
The gate signal has been generated and the repetition
rate of the experiment has been controlled by the
Pulse Generator. The useful signal detection rate on
the single photon level has been used for a rough
estimate of the relative detection efficiency in
comparison to the silicon based diode.
PULSE GENERATOR
LASER PULSER
HAMAMATSU PLP01
757 nm, 5 mW
ACTIVE
QUENCHING
AND GATING
APD
in THERMOSTAT
0.3 m
ND
TIME TO
AMPLITUDE
CONVERTER
MULTICHANNEL
ANALYZER
PC
Figure 3: Block scheme of the detector test setup, time
correlated photon counting experiment.
4 EXPERIMENT RESULTS
AND DISCUSSION
Characteristics of photon counters have been
New Materials for Photon Counting Avalanche Photodiodes
239
measured for selected materials in visible range due
to available picosecond laser source. The
experiments with high energy photons have been
completed, its review is in paper (Blazej et al.,
2007). Some specific characteristics of detectors
under study like mutual interaction in visible
wavelength range of closely running detector have
been published in other paper (Blazej et al, 2006).
The luminescence of these type structures are also
disscused in reference (Gontaruk et al., 2014).
Figure 4: GaAsP structure photo.
4.1 GaAsP
The dark count rate of the GaAsP avalanche
photodiodes as a function of the bias above the break
is plotted on Figure 5, the measurement has been
carried out with the repetition rate 100 Hz.
012345
0.1
1
10
100
Dark count [MHz]
Bias voltage [V]
Figure 5: Dark count rate of the GaAsP diode as a function
of bias above break, diode operated in an active quenching
and gating circuit.
Considering the diode active area, the dark count
rate is in principle comparable to the silicon based
SPADs, which is of the order of 1 MHz, 5 V above
the break, at the same temperature and active area
100 μm in diameter. The dark count rate can be
reduced by diode cooling. On the Figure 6 see the
plot of the dark count rate versus temperature. The
decrease of the dark count rate with temperature is
rather low in comparison to silicon or germanium
diodes, one order of magnitude per 60 K.
The timing resolution has been measured in an
arrangement described on Figure 3. For comparison
purposes, the reference APD on silicon (SPAD)
(Prochazka et al., 1992) has been employed. The
results are summarized on Figure 7, the photon
counting time response of the GaAsP diode (upper
curve) and silicon SPAD (lower curve). Note equal
width of both the distributions and a highly non-
symmetrical data distribution (tail) of the silicon
SPAD in contrast to a nearly ideal normal
distribution of the GaAsP diode.
-100 -80 -60 -40 -20 0 20
0.1
1
10
Dark count [MHz]
Chip temperature [°C]
Figure 6: Dark count rate of the GaAsP diode as a function
of temperature, diode operated in an active quenching and
gating circuit, 5 V above the break.
The timing resolution 112 ps FWHM for both
samples has been limited by the experimental chain.
The single photon detection efficiency has been
found to be in the range 0.1 to 1 % at the testing
wavelength. The detection probability is a product of
two probabilities: probability of generating a carrier
and probability of the avalanche triggering.
According to Figure 1, the first probability should
exceed 10 %, thus the limiting factor in our detector
photon counting sensitivity was the low probability
14 16 18 20 22 24
1
10
100
1000
Si
GaAsP
Number of counts
Time [ns]
Figure 7: Timing resolution of the GaAsP and silicon
photon counters operated 5 V above their break voltage,
FWHM = 112 ps for both diodes. Note high non-
symmetry of the silicon diode data distribution.
PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology
240
of avalanche triggering. This parameter may be
improved by the diode structure design and
optimization specifically for photon counting
purposes.
4.2 GaAs
The example of the timing resolution data of the
GaAs photon counter is on Figure 8. The diode has
been biased 5 Volts above the break. The dark count
rate of 20 MHz has been observed at +25 °C.
25 30 35 40
0
200
400
600
800
1000
Number of counts
Time [ns]
Figure 8: Timing resolution of the GaAs diode operated as
a photon counter operated 5 V above the break voltage,
FWHM = 230 picoseconds.
Comparing to the GaAsP diode, the timing
resolution is more than twice worse and the data
distribution exhibits a significant non-symmetric, the
percentage of the data points in the tail of the
distribution exceeds 70 %. This fact inhibits the
application of GaAs diodes for laser ranging
purposes. The dependence of the timing resolution
on the diode biasing is plotted on Figure 9.
0123456
0
200
400
600
800
1000
FWHM [ps]
Bias voltage [V]
Figure 9: Timing resolution of GaAs diode operated as a
photon counter versus bias above break voltage.
4.3 GaP
The GaP diode timing properties are demonstrated
on Figure 10, where the time correlated photon
counting results are plotted.
16 17 18
0
100
200
300
400
500
600
Number of counts
Time [ns]
Figure 10: Timing resolution of the GaP diode operated
5 V above the break voltage as a photon counter,
FWHM = 110 ps. Note good symmetry of the data
distribution.
Although the test wavelength has been outside the
main sensitivity of the material, see Figure 1, the
residual sensitivity permitted to carry out the
measurements. The photon counting probability was
of the order of 0.01 % in this case. The timing
resolution FWHM of 110 picoseconds has been
obtained, the data distribution is quite well
symmetric and close to normal one. The diode has
been biased 5 Volts above the break, the dark count
rate of 6 MHz has been measured. The relative
sensitivity and Signal to Noise Ratio (SNR) versus
voltage above the break is plotted on Figure 11.
0123456
1.0
1.5
2.0
2.5
3.0
3.5
SNR
SNR [a.u.]
Bias voltage [V]
0
10
20
30
Sensitivity
Sensitivity [a.u]
Figure 11: Sensitivity and SNR of the GaP diode based
photon counter versus bias above the break.
5 CONCLUSIONS
The avalanche diode structures operational as single
photon detectors with picosecond resolution on the
basis of the GaAs, GaP and GaAsP materials have
been designed, developed and tested. This is to our
knowledge the first published attempt to develop a
photon counting avalanche photodiode on the basis
of these materials. The timing resolution FHWM of
112, 275 and 110 picoseconds for the diodes on
GaAsP, GaAs and GaP, respectively, has been
obtained. The dark count rate comparable to existing
New Materials for Photon Counting Avalanche Photodiodes
241
silicon avalanche photodiodes has been measured.
The GaAsP and GaP photon counters exhibit time
correlated photon counting data distribution very
close to ideal normal distribution in the near infrared
wavelength. It is expected, that further tuning of the
diode structure and its optimization for photon
counting will result in a detector with timing
resolution better than silicon detectors and will
provide a nearly ideal data distribution in the near
infrared with the detection efficiency reaching 10 %.
ACKNOWLEDGEMENTS
This work has been carried out at the Czech
Technical University in Prague. Numerous grants
were provided by the Czech Grant Agency, Czech
Ministry of Education and by international agencies.
The recent publication has been supported by
MSMT CR grant RVO68407700.
REFERENCES
Blazej, J., Prochazka, I., Hamal, K., Sopko, B., & Chren,
D. (2006). Gallium-based avalanche photodiode
optical crosstalk. Nuclear Instruments and Methods in
Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment,
567(1):239-241.
Blazej, J., Prochazka, I., 2007. Gallium-based avalanche
photon counter with picosecond timing resolution for
X to visible range. In Proc. SPIE vol. 6702 - Soft X-
Ray Lasers and Applications VII, SPIE.
Cova, S., Longoni, A., Andreoni, A., and Cubeddu. R.
1983. A semiconductor detector for measuring ultra-
weak fluorescence decays with 70 ps. IEEE J. of
Quantum Electronics 19:630-634.
Gontaruk, O. N., Kovalenko, A. V., Konoreva, O. V.,
Malyj, E. V., Petrenko, I. V., Pinkovs’ka, M. B., and
Tartachnyk, V. P. (2014). Electroluminescence of
Commercial Gap Green-Light-Emitting Diodes.
Journal of Applied Spectroscopy, 80(6):851-854.
Kirchner, G., Koidl, F., Blazej, J., Hamal, K., and
Prochazka, I. (1997). Time-walk-compensated SPAD:
multiple-photon versus single-photon operation. In
Proc. SPIE 3218 - Aerospace Remote Sensing '97.
SPIE.
Kunimori, H., Greene, B., Hamal, K., and Prochazka, I.
(2000). Centimetre precision eye-safe satellite laser
ranging using a Raman-shifted Nd: YAG laser and
germanium photon counter. Journal of Optics A: Pure
and Applied Optics 2(1):1.
Prochazka, I., Hamal, K., Sopko, B., and Mácha, I. (1992).
Novel contribution in branch of ultra-fast condensed
matter spectroscopic photon counting system.
Microelectronic Engineering 19(1):643-645.
Prochazka, I., Greene, B., Kunimori, H., and Hamal, K.
(1996). Large-aperture germanium detector package
for picosecond photon counting in the 0.5–1.6-µm
range. Optics letters 21(17):1375-1377.
Prochazka, I. (2001). Peltier-cooled and actively quenched
operation of InGaAs/InP avalanche photodiodes as
photon counters at a 1.55-µm wavelength. Applied
optics, 40(33):6012-6018.
Van Stryland, E., Williams, D., & Wolfe, W. (1996).
Handbook of optics. M. Bass (Ed.). McGraw-Hill.
PHOTOPTICS 2016 - 4th International Conference on Photonics, Optics and Laser Technology
242