1. SPIE Defense and Commercial Sensing 2016, Baltimore, Maryland, USA, Apr 21, 2016
Advanced active quenching circuits for single-photon avalanche photodiodes
Mario Stipčević
Photonics and Quantum optics Research Group
Center of excellence for advanced materials and sensing devices,
Ruder Boskovic Institute, Zagreb, Croatia
URL:

2. Si SPAD based photon detector
Single photon detector – a device that produces one standardized
logical pulse upon each successful detection of a photon.

3. We use commercial “thick reach-through” Si Single-Photon-capable
Avalanche photo Diode (SPADs) SAP500-T8 (Laser Components)
operated in Geiger mode
+ home-made electronics comprising:
Active Quenching Circuit (AQC)
TEC temperature controller
Low voltage generator (~25V)
High voltage generator (100~500V)
Detection imperfections come from: SAP500 SPAD
Diode physics
Electronics

4. Active quenching circuit (AQC)
A realistic active quenching loop circuit with a double feedback (left);
and its timing diagram (right).
M. Stipcevic, Appl.Opt. 48, (2009)

5. Detector imperfections
APD:
Detect. efficiency <1
Afterpulsing
Timing jitter
Super-linear behavior
Electronics:
Dead time
Max. count rate
Variable dead t. & eff.
Twilighting
Blindability
Distribution of pulse interval times of a realistic detector
Input: CW random light (LED)
Shown is histogram of time intervals between subsequent output
pulses (note: not every pulse corresponds to a real photon detection).

6. Jitter and detection delay vs detection frequency
Jitter and shift as funct. of detection frequency, at 635nm
Used 225ps FWHM pulsed laser triggered at 10MHz. Attenuation adjusted to set the desired detection frequency in the range MHz.
Comparison of a custom-made detector and a few commercial
Excelitas SPCM-AQRH IdQuantique ID Home-made τ-SAP
Method as described in Opt. Express. 18 (2010) 6

7. Excelitas SPCM-AQR-12 single photon timing performance
Excelitas SPCM-AQR-12 single photon timing performance. At about 500kHz a secondary
peak appears and at 1M there is on big blob with FWHM ~1ns + 1.6ns shift. Dead time ~50 ns.
(Plots show result for Gaussian sigma of the fit. Laser and detector are convoluted.)

8. IdQuantique’s ID100 is a low-efficiency small diameter (20um) APD specialized
for best timing of 50ps FWHM. Good: Peak stability is excellent. Bad: (1) long tail,
(2) resolution becomes worse with detection frequency. Dead time ~50 ns.

9. Home made -SAP (fast version)
Home made -SAP (fast version). Excellent time resolution, excellent resolution stability even at
highest tested detection frequency, excellent stability of delay, lowest detection delay
all at high detection efficiency => 4 improvements vs. commercial solutions. Dead time ~24 ns.

10. Comparison of 3 detectors
regarding time resolution (jitter)
and peak stability
as functions of the
detection frequency.
(Laser pulse width subtracted.)
The stabilities of resolution
and delay of tau-SAP are
better than stability
any major brand
of detector.

11. Twilighting
Twilighting is an effect of sensitivity of detector during the dead time
It is a period of bias voltage recovery when the SPAD is biased above Geiger breakdown and can generate an avalanche but it will generate an output pulse only after the dead time => detection propagation delay time shift.
This interval is named
the “twilight zone”
(yellow shaded) 11

12. Twilighting (detection of photons during dead time)
Time shift of photon detection vs. detection frequency
Micro Photon Devices SPD-050
Distributions of detection waiting time for MPD50 (dead time 78.1 ns) when light pulses are apart by: 40 ns (second photon not observed = noise) (left), 60 ns (second photon arrived in the twilight zone but observed after the dead time) (middle), and 80 ns (second photon arrived and observed after the dead time). Fit parameter Sigma is one Gaussian sigma of the fitted curve.

13. Twilighting (detection of photons during dead time)
Time shift of photon detection vs. detection frequency
PerkinElmer SPCM AQR
Distributions of detection waiting time for PerkinElmer SPCM-AQR (dead time 29.5 ns) when light pulses are apart by: 23 ns {second photon in the twilight zone) (left), 30 ns (right). Jitter in the twilight zone seems to be improved far beyond possible limits for the SPAD – it is an effect of a very precise dead time of the SPCM.

14. Dead time proximity detection delay
Time shift between the true and measured photon arrival time for the second photon in a pair (if both photons have been detected), as a function of the time interval between the two incoming photons (left).
Time resolution (jitter) of the second photon in a pair if both photons have been detected (right). 14

15. Electronics artifacts
Twilighting is imposed intentionally in order to avoid atomic race condition between end of quench and start of amplifier sensitivity
What if twilight zone is too narrow and amplyfier becomes sensitive while SDAD is still generating signal⇒ re-triggering
In older Perkin Elmer SPCM AQR detectors (Rev. F, ~year 2003) we
see strong retriggering (we do not see this in newer versions):
We can reproduce this in our circuit by tightening the twilight zone. 15

16. Other electronics issues
High voltage SPAD bias stability upon sudden supply current jump:
0 mA → 1 mA (left); 1 mA → 0 mA (right).
A major manufacturer
Home-made power supply 16

17. In this study, we differ “standard” imperfections (widely accepted) :
non-unity detection efficiency
dead time
dark counts
afterpulsing
jitter
And “hidden” imperfections:
variation of jitter with detection freqency (peak width)
variation of detection delay with frequency (peak position)
variation of dead time with detection effiniency
Dead time proximity effects
Retriggering and other electronics issues.
We illistrate that commercial detectors are plagued with the hidden
imperfections (not specified in the datasheets nor widely recognized).
Hidden imperfections cannot be neglected.

18. In quantum information and communication experiments performed by photons:
afterpulsing and twilighting may create false events, false correlations in data, information leakage;
unstable jitter and time shifts may cause loss of data, oss of coincidences or false coincidences.
That is why experimentalists in quantum information often resort to their own devices in building of detectors that are optimized for the given experiment.

19. Custom (home) made detector
An advanced AQ circuit with significant improvement in hidden imperfections without sacrifising performance in standard imperfections.
AC coupling of quench signal replaced by galvanic coupl.
Peak shift >100ps for photons >28 ns apart
Twilight zone <1.5 ns
Jitter virtually constant 160 ps FWHM

20. Standard imperfections example
Polarization qubit analysis in orthogonal basis (polarizing beamsplitter):
(highest entropy α2=β2=½)
PBS
Depending on which detector “clicks”, 0 or 1 is generated per each detected photon.
The same configuration used in: a receiver station of QKD and in a quantum random number generator (QRNG).
Polarization analysis setup comprises polarizing beamsplitter (PBS) and two single-photon detectors

21. We found that correlations are solely due to detector imperfections:
Serial autocorrelation coefficient as a function of photon detection rate
We found that correlations are solely due to detector imperfections:
Afterpulsing → positive, dead time → negative autocorrelation.
In this example hidden imerfections are significantly less important.
M. Stipcevic, D. J. Gauthier, Proc. SPIE DSS, paper 87270K, 29 April - 3 May 2013,
Baltimore, Maryland, USA

22. Hidden imperfections example
Application: Ultra-fast QKD with hyper-entangled photons, entangled simultaneously in: photon number, polarization, and time bin.
One “frame” consists of 1024 time bins (slots) of ~260 ps (+) two-photon entanglement
Pump power is set such that Alice and Bob receive about 1 photon per time frame
For a successful communication instance Alice and Bob must receive photons from an entangled pair in the same time bin
Twilighting and other detection time shifts greater than ~ ps cause direct errors (BER) in time-bin entanglement readout

23. 1. Autocorrelation
Probability of Alice and Bob detecting a photon in the same bin
(distinguishability)

24. Longer dead time promotes losses, larger twilighting promotes errors
Finaly, secret key channel capacity (after error correction):
2.4 qubits/photon with SPCM AQRH-12
3.6 qubits/photon estimated with the custom-made after error correction
In this example, due to tight coincidence, dark counts and less so afterpulses are supressed but hidden imperfections play a major role.

25. Custom-made detector,
Under DARPA InPho program
Detection efficiency at 635nm (InPho = 75%, SPCM-AQR = 65%, ID100 = 23%, SPD-050 = 40%)
Short, fixed dead time (24 ns)
Total visible afterpulsing probability = 3.2%
Jitter 156 ps FWHM at a rate < 100 kcps
164 ps FWHM at a rate 1 Mcps
184 ps FWHM at a rate 4 Mcps
Peak position stability 0 – 4 Mcps < 20 ps
Uses blanking circuit to shrink twilight zone to <1.5 ns
The shortest detection delay (11ns faster than SPCM or Id100)
The largest diameter of the flat top of the active region (InPho =500um, SPCM-AQR =180um, ID100 ≤50um, SPD-050 ≤100um)
Dark counts at the level of 1-2 kHz at -25 oC, while <25 cps have been observed on selected APDs.