High-fidelity conversion of photonic quantum information to telecommunication

wavelength with superconducting single-photon detectors

Rikizo Ikuta,1 Hiroshi Kato,1 Yoshiaki Kusaka,1 Shigehito Miki,2 Taro Yamashita,2 Hirotaka Terai,2
Mikio Fujiwara,3 Takashi Yamamoto,1 Masato Koashi,4 Masahide Sasaki,3 Zhen Wang,2 and Nobuyuki Imoto1
1
Graduate School of Engineering Science, Osaka University, Toyonaka, Osaka 560-8531, Japan
2
Advanced ICT Research Institute, National Institute of Information and
Communications Technology (NICT), 588-2 Iwaoka, Kobe 651-2492, Japan
3
Advanced ICT Research Institute, National Institute of Information and Communications Technology (NICT),
4-2-1 Nukuikawa, Koganei, Tokyo 184-8795, Japan
4
Photon Science Center, The University of Tokyo, Bunkyo-ku, 113-8656, Japan
We experimentally demonstrate a high-fidelity visible-to-telecommunication wavelength conver-
sion of a photon by using a solid-state-based difference frequency generation. In the experiment,
arXiv:1207.1585v1 [quant-ph] 6 Jul 2012

one half of a pico-second visible entangled photon pair at 780 nm is converted to a 1522-nm pho-
ton, resulting in the entangled photon pair between 780 nm and 1522 nm. Using superconducting
single-photon detectors with low dark count rates and small timing jitters, we selectively observed
well-defined temporal modes containing the two photons. We achieved a fidelity of 0.93 ± 0.04 after
the wavelength conversion, indicating that our solid-state-based scheme can be used for faithful
frequency down-conversion of visible photons emitted from quantum memories composed of various
media.

Wavelength conversion of photons in a quantum caused by Raman scattering of a strong cw pump light
regime [1] has been actively studied [2–10] as a quantum and relatively high dark count rate of an InGaAs/InP
interface for application of quantum information process- avalanche photodiode (APD) for photon detection at the
ing and communications. Especially, such a conversion telecommunication band. Therefore the observed fidelity
aiming at near-infrared photons in telecommunication of the state after the wavelength conversion is degraded.
bands are essential for transmitting quantum information In this Letter, we demonstrate almost noiseless wave-
over long-distance optical fiber networks with quantum length conversion by suppressing the effects of both
repeaters [11–13]. In the quantum repeaters, the pho- the optical noise from the Raman scattering and the
ton sent to a relay point through an optical fiber needs dark count via newly developed superconducting single-
to be entangled with a quantum memory. At present, photon detectors (SSPDs) for visible and telecommunica-
many of quantum memories and processors based on al- tion wavelengths of the photons [27, 28]. The SSPDs have
kaline atoms, trapped ions and solid states have success- lower dark count rates and smaller timing jitters than
fully created entanglement with photons at around visi- those of typical APDs. Especially, the latter property
ble wavelengths [14–20]. Thus, a quantum interface for enables us to selectively observe well-defined temporal
the wavelength conversion from visible to telecommuni- modes containing the two photons. Because duration of
cation bands with a high fidelity has attracted much in- signal photons in our experiment is of pico-second order
terest for its applications. So far such a quantum in- whereas the optical noise through the wavelength con-
terface has been demonstrated by using four wave mix- version are continuously generated, the use of the SSPDs
ing with a cold atomic cloud [6] or difference frequency will lead to reduction of irrelevant photon detections.
generation (DFG) from a nonlinear optical crystal [7]. The observed fidelity of the two-photon state after the
Among them, nonlinear optical crystals with waveguide wavelength conversion to a maximally entangled state is
structure have practically desirable features. They can 0.93 ± 0.04, which is very close to the initial fidelity of
operate near room temperature and do not require laser 0.97 ± 0.01. We also clearly observe the violation of the
cooling configuration, enabling a compact setup and in- Clauser-Horne-Shimony-Holt-type Bell’s inequality with
tegration into a photonic quantum circuit on a chip us- S = 2.62 ± 0.09.
ing waveguide structures [21]. In addition, they have a
Theoretical treatment of wavelength conversion of a
wider bandwidth, compatible with wide-band quantum
single mode of a pulsed light is as follows [1, 7]. When a
memories [22, 23], resulting in high-clock-rate quantum
pump light at angular frequency ωp is sufficiently strong,
information processing. Such kind of solid-state-based
the Hamiltonian of the wavelength conversion using a
optical quantum interface lead to development of a ma-
second-order
√ nonlinear optical interaction is described by
ture quantum information technology. However, in sev-
Ĥ = i~ ηP (e−iϕ â†c âs − eiϕ â†s âc ). Here âs and âc are
eral demonstrations of the solid-state-based wavelength
annihilation operators of a signal mode at angular fre-
conversion [4, 7, 24–26], they suffered from degradation
quency ωs and a converted mode at angular frequency
of an observed fidelity of a reconstructed quantum state
ωc = ωs − ωp , respectively. P and ϕ are a power and a
after the wavelength conversion due to background noises
phase of the classical pump light, respectively. η is a con-
 2

Lens QWP HWP PBS

tons passing through S1 and L1, respectively. We set
Ti:S laser the time difference between S1 and L1 to about 700
780 nm
L1
L2 ps. The photon in mode B′ goes through a polarization-
SHG maintaining fiber (PMF) and then enters the frequency
B S1 C
B’ Frequency S2 down-converter whose details are shown in the inset of
down-converter
BBO BS PMF BS Fig. 1.
C’ For the DFG of the signal photon at 780 nm, a cw
780 nm 1522 nm pump laser at 1600 nm is used. The linewidth of the
A BG3
pump light is 150 kHz, and its coherence time is much
PPLN
Pump laser longer than the time difference of the photons passing
1600 nm BG2
DM HPF through S1 and L1. The pump light is combined with the
signal photon at a dichroic mirror (DM) after its polar-
Start Stop
TDC ization is set to V by a polarization beamsplitter (PBS)
BG1 Dv Dt and a HWP. Then they are focused on the type-0 quasi-
(sillicon APD or SSPD) (InGaAs/InP APD or SSPD)
phase matched (V → V+V) PPLN waveguide [30] whose
temperature is controlled to be about 50◦ C. The length
FIG. 1: The experimental setup for frequency down-
conversion of one halves of the visible entangled photon pairs. of the PPLN crystal is 20 mm and the acceptable band-
width is calculated to be about 0.3 nm. After passing
through the PPLN waveguide, the strong pump light is
diminished by a high-pass filter (HPF), and the light con-
√ verted to the wavelength of 1522 nm is extracted by BG2
stant and ηP eiϕ represents an effective coupling con-
and BG3 whose bandwidths are 1 nm.
stant. Using this Hamiltonian, an annihilation operator
The photon in mode C from the frequency down-
âc,out of the converted mode coming from the nonlinear
converter is split into a short path (S2) and a long
optical crystal is described by using the time evolution
path (L2) by a BS. The polarization of the photon passing
of âs in the Heisenberg picture as
through L2 is flipped from V to H by a HWP. Time dif-
ference between S2 and L2 is adjusted to be the same as
p p
âc,out = âc (τ ) = e−iϕ sin( ηP τ ) âs + cos( ηP τ ) âc ,(1)
that between S1 and L1. The components of the photon
where τ is the traveling time of the pulses through the from S2 and L2 are recombined by a PBS, and the pho-
crystal. ton in mode C′ is detected by Dt after passing through a
The experimental setup for the DFG-based frequency single-mode fiber. We switch the InGaAs/InP APD and
down-conversion of one half of a polarization-entangled the SSPD for Dt .
photon pair at 780 nm to the wavelength of 1522 nm Each SSPD consists of an 100-nm-thick Ag mirror, a
is shown in Fig. 1. We use a mode-locked Ti:sapphire λ/4 SiO cavity and 4-mm-thick niobium nitride mean-
laser (wavelength: 780 nm; pulse width: 1.2 ps; repeti- der nanowire on a 0.4-mm-thick MgO substrate from the
tion rate: 82 MHz) as a light source. It is frequency dou- top. The nanowire is 80-nm-wide, and it covers an area
bled by a second harmonic generator (SHG), and then the of 15 µm × 15 µm. The respective optical cavity struc-
UV pulse with a power of 250 mW pumps a pair of type- tures of the SSPDs for Dv and Dt are designed for vis-
I phase-matched 1.5-mm-thick β-barium borate (BBO) ible and the telecommunication wavelengths to achieve
crystals to prepare the polarization-entangled photon higher detection efficiencies. The detection efficiencies
pair A and√B which is described as |φ+ iAB ≡ (|HHiAB + are 32 % and 12.5 % for 780 nm and 1522 nm wave-
|VViAB )/ 2 through spontaneous parametric down- lengths, respectively. Each of the SSPD chip is shielded
conversion [29]. Here |Hi and |Vi represent horizon- by a copper block which has a holder of a single-mode
tal (H) and vertical (V) polarization states of a photon. optical fiber followed by a small-gradient index lense for
The spectrum of photon A is narrowed by a Bragg grat- efficient coupling. The blocks are installed in the Gifford-
ing (BG1) with a bandwidth of 0.2 nm, and then the McMahon cryocooler system whose cooling temperature
photon is detected by detector Dv connected to a single- is 2.28 ± 0.02 K.
mode fiber. We switch the silicon APD and the SSPD for To measure the coincidence events of the detections at
Dv . Photon B is split into a short path (S1) and a long Dv and Dt , signals from Dv and Dt are input to a time-
path (L1) according to a polarization of the photon. A to-digital converter (TDC) as a start and stop signals of
half-wave plate (HWP) in S1 flips the polarization from a clock, respectively. By post-selecting the events where
H to V. As a result, a polarization qubit {|Hi, |Vi} in the photon in mode C′ has passed through S1-L2 or L1-
mode B is transformed to a time-bin qubit {|S1i, |L1i} S2, we obtain the polarization-entangled state |φ+ iAC′
in mode B′ , leading to a two-photon state in modes √ in modes A and C′ . In our experiment, we accept such
A and B′ of |ψiAB′ ≡ (|HiA |S1iB′ + |ViA |L1iB′ )/ 2. events in 200-ps time window.
Here |S1i and |L1i represent states of V-polarized pho- For a faithful wavelength conversion, the rate of back-
 3

0.7 70
0.6 60
0.5 50
0.4 40
0.3 30
0.2 20
0.1 10
0.0 0
0 200 400 600 800 1000

FIG. 3: The coincidence counts of the photon pairs recorded

by the TDC with the timing resolution of 100 ps when we
FIG. 2: The dependencies of the conversion efficiency and the used the pair of the silicon APD for Dv and InGaAs/InP
rate of the background noises on the pump power. The former APD for Dt (rhombus), the silicon APD for Dv and the SSPD
has been measured by using a pico-second coherent light in for Dt (square), and the SSPDs for both the detectors (cir-
Ref. [7]. The latter was measured by using the SSPD for Dt cle). Each result of the coincidence counts was fitted by the
without the UV pulse. The first vertical axis is the conversion Gaussian function after the subtraction of the counts from
efficiency. A maximum conversion efficiency is achieved at a the background noises.
pump power of 700 mW. The second vertical axis is detection
rate of the background noises. We fitted the experimental
data by a function of bP + d. b and d are estimated as 80
a b c
Hz/mW and 266 Hz, respectively.
Re Re Re

0.5 0.5 0.5

ground noises must be sufficiently small compared to 0.25 0.25 0.25

that of the converted photons. The background noises 0 0 0

are mainly caused by the Raman scattering of the pump −0.25 −0.25 −0.25

light and dark countings of a photon detector. The for- −0.5 HH

HV
−0.5 HH
HV −0.5
HH
HV
VH VH VH
mer is proportional to the pump power and is written HH HV
VH VV VV HH HV
VH VV VV HH HV
VH VV VV

by bP with a constant b. The latter is described by a

constant d. Taking into account these noise effects and FIG. 4: The real parts of the reconstructed density matri-
using Eq. (1), the signal-to-noise ratio for the observed ces. (a) The initial entangled photon pair ρAB prepared by
converted photons is represented by the BBO crystal. (b) The two-photon state ρAS AC′ after the
√ wavelength conversion when we used the silicon APD for Dv
a sin2 ( ηP τ ) and the SSPD for Dt . (c) The two-photon state ρSSAC′ after
fSNR (P ) ≡ , (2) the wavelength conversion when we used the SSPDs for both
bP + d
detector Dv and Dt .
where a is a constant. When d is comparable to bP
as in the case where both detectors Dv and Dt are
APDs [7], a maximum of fSNR (P ) is achieved near a
pump power Pmax which gives the maximum conver- the time distribution of the coincidence counts was mea-
sion efficiency, and the maximum of fSNR (P ) is close sured to be ≈ 350 ps. When we replaced the InGaAs/InP
to fSNR (Pmax ) = a/(bPmax + d). On the other hand, APD with the SSPD for Dt , the FWHM was measured to
when d is sufficiently small as in the case of using the be ≈ 290 ps. When we used the SSPDs for both detector,
SSPDs, we may improve fSNR (P ) significantly by de- the FWHM became ≈ 150 ps. These results which are
creasing the pump power. Using d/b ≈ 3.3 mW in Fig. 2 shown in Fig. 3 clearly show that the SSPDs gather the
and ητ 2 ≈ 3.6 W−1 [7], we see that fSNR (P ) reaches coincidence counts in the smaller time bins. The use of
its maximum when the pump power is decreased from the SSPDs instead of the APDs is expected to improve
Pmax ≈ 700 mW to 50 mW. In the following experiments, the ratio of the signal photons to the optical noises by a
we chose the pump power to be 160 mW, for which the factor of about 1.8 in our case of 200-ps time window.
conversion efficiency is not severely degraded (about half In the experiment of the frequency down-conversion
the maximum value) and the signal-to-noise ratio is ex- of the visible entangled photon pairs, we first performed
pected to be above 10. quantum state tomography of the initial photon pairs in
Before the wavelength conversion of the entangled pho- modes A and B at 780 nm by rotating a quarter-wave
ton pairs, we measured the variance of jitter in the arrival plate (QWP) and a HWP followed by a PBS in each
time of pico-second photons to see the high timing reso- mode [31]. We used a silicon APD for Dv . The PMF
lution of the SSPDs. When we used the APDs for both before the frequency down-converter was connected to a
detectors, the full width at half maximum (FWHM) of silicon APD for detection of photons in mode B. We ob-
 4

F EOF purity S ing for the faithful wavelength conversion was achieved
ρAB 0.97 ± 0.01 0.97 ± 0.03 0.97 ± 0.02 2.73 ± 0.02 at the cost of the conversion efficiency, it will be attained
ρAS
AC′ 0.87 ± 0.06 0.68 ± 0.15 0.79 ± 0.08 2.35 ± 0.10 without decreasing the conversion efficiency by a faster
ρSS
AC′ 0.93 ± 0.04 0.88 ± 0.10 0.93 ± 0.07 2.62 ± 0.09 time-resolution measurement. Our demonstration shows
ρ′AC′ [7] 0.75 ± 0.06 0.36 ± 0.13 - - the possibility of a noiseless and wide-band solid-state-
based frequency down-conversion. We believe that such a
TABLE I: The observed fidelities, the EOFs the purities, and quantum interface is vital for building quantum networks
the S parameters of the reconstructed operators before and based on repeaters and for performing various quantum
after the wavelength conversion. ρ′AC′ is a reconstructed oper-
communication protocols.
ator of photons in modes A and C′ by using the silicon APD
and the InGaAs/InP APD for Dv and Dt , respectively [7]. This work was supported by the Funding Program for
The attached errors are the standard deviations (1-σ) with World-Leading Innovative R & D on Science and Tech-
the assumption of the Poisson statistics of the counts. nology (FIRST), MEXT Grant-in-Aid for Scientific Re-
search on Innovative Areas 20104003 and 21102008, the
MEXT Global COE Program, and MEXT Grant-in-Aid
for Young scientists(A) 23684035.
served the two-photon state in modes A and B with a de-
tection rate of 444 Hz. Using the iterative maximum like-
lihood method [32], the density operator ρAB was recon-
structed as shown in Fig. 4 (a). From the reconstructed
ρAB , we calculated the fidelity defined by hφ+ |ρAB |φ+ i, REFERENCES
the entanglement of formation (EOF) [33], and the pu-
rity defined by tr(ρ2AB ) as 0.97 ± 0.01, 0.97 ± 0.03 and [1] P. Kumar, Opt. Lett. 15, 1476 (1990).
0.97 ± 0.02, respectively. These results show that the [2] S. Tanzilli et al. Nature 437, 116 (2005).
780-nm photon pair before the wavelength conversion was [3] H. Takesue, Phys. Rev. Lett. 101, 173901 (2008).
highly entangled. Next, we performed quantum state to- [4] M. T. Rakher, L. Ma, O. T. Slattery, X. Tang, and K.
A. Srinivasan, Nature Photonics 4, 786 (2010).
mography of the photon in mode A and the converted
[5] H. J. McGuinness, M. G. Raymer, C. J. McKinstrie, and
photon in mode C′ when we used the silicon APD for Dv S. Radic, Phys. Rev. Lett. 105, 093604 (2010).
and the SSPD for Dt . An observed detection rate of the [6] Y. O. Dudin et al. Phys. Rev. Lett. 105, 260502 (2010).
two photons was 0.324 Hz including background noises [7] R. Ikuta et al., Nature Commun. 2, 537 (2011).
at a rate of 0.039 Hz. The reconstructed density operator [8] S. Ramelow, A. Fedrizzi, A. Poppe, N. K. Langford, and
ρAS A. Zeilinger, Phys. Rev. A 85, 013845 (2012).
AC′ is shown in Fig. 4 (b). The fidelity, EOF and the
purity are shown in Table I. We then switched the silicon [9] S. Zaske et al., e-print arXiv:1204.6253.
[10] S. Ares, I. Agha, A. Badolato, and K. Srinivasan, e-print
APD to the SSPD for Dv , and performed quantum state
arXiv:1205.2218.
tomography. In this case, we detected the two photons [11] H. -J. Briegel, W. Dür, J. I. Cirac, and P. Zoller, Phys.
after the conversion at a rate of 0.280 Hz including noise Rev. Lett. 81, 5932 (1998).
photons at a rate of 0.015 Hz. The reconstructed density [12] L. M. Duan, M. D. Lukin, J. I. Cirac, and P. Zoller,
operator ρSSAC′ is shown in Fig. 4 (c). The observed fi- Nature 414, 413 (2001).
delity, EOF and the purity are shown in Table I. We also [13] N. Gisin, and R. Thew, Nature Photonics 1, 165 (2007).
performed the tests of the Clauser-Horne-Shimony-Holt- [14] J. I. Cirac, and P. Zoller, Nature 404, 579 (2000).
[15] D. N. Matsukevich, and A. Kuzmich, Science 306, 663 -
type Bell’s inequality [34] by observing the S parameter.
666 (2004).
It is known that any local hidden variable theory leads [16] W. Rosenfeld et al., Phys. Rev. Lett. 101, 260403 (2008).
to S ≤ 2. The experimental results of the S parameters [17] C. W. Chou et al., Nature 438, 828.
are shown in Table I. The Bell’s inequality was clearly [18] S. Olmschenk et al., Science 323, 486 (2009).
violated with over 6-σ deviation. From these results, [19] E. Togan et al., Nature 466, 730 (2010).
we see that the high-fidelity frequency down-conversion [20] S. Ritter et al., Nature 484, 195 (2012).
was achieved, which could be more clearly observed with [21] A. Politi, M. J. Cryan, J. G. Rarity, S. Yu, and J. L.
OfBrien, Science 320, 646 (2008).
SSPDs.
[22] K. F. Reim et. al., Nature Photonics 4, 218 (2010).
In conclusion, we have demonstrated the faithful solid- [23] E. Saglamyurek et. al., Nature 469, 512 (2011).
state-based frequency down-conversion of visible photons [24] H. Takesue, Phys. Rev. A 82, 013833 (2010).
at 780 nm to a telecommunication wavelength of 1522 [25] N. Curtz, R. Thew, C. Simon, N. Gisin, and H. Zbinden,
nm. Thanks to the lower dark count rates and the smaller Opt. Express 18, 22099 (2010).
[26] S. Zaske, A. Lenhard, and C. Becher, Opt. Express 19,
timing jitters of the SSPDs, the observed fidelity of the
12825 (2011).
two-photon state after the wavelength conversion was [27] S. Miki, M. Takeda, M. Fujiwara, M. Sasaki, and Z.Wang,
0.93 ± 0.04, which was very close to the value of the fi- Opt. Express 17, 23558 (2009).
delity obtained before the conversion. In this experiment, [28] S. Miki, T. Yamashita, M. Fujiwara, M. Sasaki, and Z.
while the suppression of the effect of the Raman scatter- Wang, Opt. Lett. 35, 2133 (2010).
 5

[29] T. Yamamoto, K. Hayashi, Ş. K. Özdemir, M. Koashi, (2007).

and N. Imoto, Nature Photonics 2, 488 (2008). [33] W. K. Wootters, Phys. Rev. Lett. 80, 2245 (1998).
[30] T. Nishikawa et al., Opt. Express 17, 17792 (2009). [34] J. F. Clauser, M. A. Horne, A. Shimony, and R. A. Holt,
[31] D. F. V. James, P. G. Kwiat, W. J. Munro, and A. G. Phys. Rev. Lett. 23, 880 (1969).
White, Phys. Rev. A 64, 052312 (2001).
[32] J. Řeháček, and Z. Hradil, Phys. Rev. A 75, 042108