
Design of mid-infrared entangled photon sources using lithium niobate
Jin-Long Zhu,1, ∗Wen-Xin Zhu,1, ∗Xiao-Tao Shi,1, ∗Chen-Tao
Zhang,1Xiangying Hao,1, †Zi-Xiang Yang,1, ‡and Rui-Bo Jin1, §
1Hubei Key Laboratory of Optical Information and Pattern Recognition,
Wuhan Institute of Technology, Wuhan 430205, PR China
The mid-infrared (MIR) band entangled photon source is vital for the next generation of quantum
communication, quantum imaging, and quantum sensing. However, the current entangled states are
mainly prepared in visible or near-infrared bands. It is still lack of high-quality entangled photon
sources in the MIR band. In this work, we optimize the poling sequence of lithium niobate to prepare
two kinds of typical entangled states, the Hermit-Gaussian state and the comb-like entangled state
at 3.2 µm. We have also calculated the photon pair rates and estimated the effect of fabrication
resolution in the schemes. Our approach will provide entangled photon sources with excellent
performance for the study of quantum information in the MIR band.
I. INTRODUCTION
The entangled photon source in the mid-infrared
(MIR) band (approximately 2-20 µm) is promising for
the next generation of quantum communication, quan-
tum imaging, and quantum sensing[1,2]. In quan-
tum communication, the entangled photon source in the
wavelength between 3 µm and 5 µm covers the atmo-
spheric transmission window, which has higher trans-
parency than that in the near-infrared band and is ben-
eficial for free-space quantum communications [3]. In
quantum imaging, room temperature objects emit light
at MIR wavelengths, therefore the MIR band entangled
photon sources are compatible with novel applications in
infrared thermal imaging [4,5]. In quantum sensing, the
MIR band entangled photon source has strong absorp-
tion bands of a variety of gases, which leads to essential
applications in gas quantum sensing [6]. With the help
of entanglement, the sensing precision may be improved
from the shot noise limit to the Heisenberg limit [7].
Spontaneous parametric down-conversion (SPDC) is
one of the widely used methods to prepare an entan-
gled photon source. Recently, several works have investi-
gated the generation of entangled photons in MIR range
from an SPDC process in the nonlinear crystal. From
the theoretical side, in 2016, Lee et al reported a scheme
for the generation of polarization-entangled state from
periodically poled potassium niobate (PPKN) covering
3.2 to 4.8 µm [8]; In 2018, McCracken et al numerically
investigated six novel nonlinear crsystals in order to gen-
erate MIR single photons [9]; In 2020, Kundys et al nu-
merically studied the reconfigurable MIR single-photon
sources in PMN-0.38PT crystal at 5.6 µm [10]; In 2021,
Wei et al theoretically investigated the preparation of
MIR spectrally uncorrelated biphotons from an SPDC
process using doped lithium niobate (LN) crystals [11];
∗These authors contributed equally to this work.
†xyhao.321@163.com
‡yangzixiangyzx@foxmail.com
§jin@wit.edu.cn
These schemes for single photon can also be upgraded
to prepare entangled photons. From the experimental
side, in 2020, Prabhakar et al demonstrated an entan-
gled photons source at 2.1 µm generated from type-II
PPLN crystals [12].
PPLN is one of the most promising crystals for MIR
entangled photon source, not only for its large nonlin-
ear coefficient and wide transparency range [13,14], but
also for its group velocity-matched (GVM) wavelengths,
which are in the MIR range [11]. However, from the view-
point of quantum state engineering, the previous entan-
gled source with PPLN is still not optimal [11,12]. For
a standard PPLN crystal, the phase matching function
(PMF) has a “sinc” distribution, which has side lobes
and will harm the spectral purity of heralded single pho-
ton. To overcome the problem of side lobes, one useful
method is to adopt the “customized poling” instead of
“periodical poling”[15–18].
Many previous works have been devoted to the opti-
mization of a poling period in a periodically poled potas-
sium titanyl phosphate (PPKTP) crystal at 1550 nm,
and the optimization can be divided into three categories:
(1) the optimization of poling order: in 2011, Branczyk
et al proposed and experimentally demonstrated the first
optimization design of KTP by arranging the poling order
[19], and this approach was further improved by Kaneda
et al in 2021 [20]. (2) the optimization of duty cycle:
in 2013, Dixon et al proposed to design the duty cycle
of KTP [21], which was verified experimentally in 2017
[22]; In 2019, Cui et al adopted the Adam algorithm in
a machine learning framework to optimize the duty cy-
cle [23]. In 2022, Cai et al optimized the duty cycle of
LN using the particle swarm algorithm [24]; (3) the op-
timization of domain sequence: in 2016, Tambasco et al
optimized the domain sequence in the unit of dual do-
main blocks [25]; In the same year, Dosseva et al pro-
posed to optimize the sequence of single domain blocks
using simulated annealing algorithm [26]; In 2017, Graf-
fitti et al theoretically optimized the domain blocks with
sub-coherence-length [15], and then verified experimen-
tally in 2018 [27,28]. Recently, frequency-bin entangle-
ment generated by domain-engineered down-conversion
arXiv:2210.12466v1 [quant-ph] 22 Oct 2022