Document [original]

FULL PAPER

www.advquantumtech.com

Plug&Play Fiber-Coupled 73 kHz Single-Photon Source

Operating in the Telecom O-Band

Anna Musiał,* Kinga ˙

Zołnacz, Nicole Srocka, Oleh Kravets, Jan Große, Jacek Olszewski,

Krzysztof Poturaj, Grzegorz Wójcik, Paweł Mergo, Kamil Dybka, Mariusz Dyrkacz,

Michał Dłubek, Kristian Lauritsen, Andreas Bülter, Philipp-Immanuel Schneider,

Lin Zschiedrich, Sven Burger, Sven Rodt, Wacław Urba´

nczyk, Grzegorz S˛

ek,

and Stephan Reitzenstein*

A user-friendly, ﬁber-coupled, single-photon source operating at telecom

wavelengths is a key component of photonic quantum networks providing

long-haul, ultra-secure data exchange. To take full advantage of

quantum-mechanical data protection and to maximize the transmission rate

and distance, a true quantum source providing single photons on demand is

highly desirable. This great challenge is tackled by developing a ready-to-use

semiconductor quantum-dot-based device that launches single photons at a

wavelength of 1.3 µm directly into a single-mode optical ﬁber. In the proposed

approach, the quantum dot is deterministically integrated into a

nanophotonic structure to ensure eﬃcient on-chip coupling into a ﬁber. The

whole arrangement is integrated into a 19ʺcompatible housing to enable

stand-alone operation by cooling via a compact Stirling cryocooler. The

realized source delivers single photons with a multiphoton events probability

as low as 0.15 and a single-photon emission rate of up to 73 kHz into a

standard telecom single-mode ﬁber.

Dr. A. Musiał, O. Kravets, Prof. G. S ˛

Laboratory for Optical Spectroscopy of Nanostructures

Department of Experimental Physics

Wrocław University of Science and Technology

Wybrze˙

ze Wyspia´

nskiego 27, Wrocław 50-370, Poland

E-mail: anna.musial@pwr.edu.pl

K. ˙

Zołnacz, J. Olszewski, Prof. W. Urba´

nczyk

Department of Optics and Photonics

Wroclaw University of Science and Technology

Wybrze˙

ze Wyspia´

nskiego 27, Wrocław 50-370, Poland

N. Srocka, J. Große, Dr. S. Rodt, Prof. S. Reitzenstein

Institute of Solid State Physics

Technische Universität Berlin

Hardenbergstraße 36, Berlin 10623, Germany

E-mail: [email protected]

The ORCID identiﬁcation number(s) for the author(s) of this article

can be found under https://doi.org/10.1002/qute.202000018

KGaA, Weinheim. This is an open access article under the terms of the

Creative Commons Attribution License, which permits use, distribution

and reproduction in any medium, provided the original work is properly

cited.

DOI: 10.1002/qute.202000018

1. Introduction

Sources of single photons (SPSs) are

fundamental building blocks for pho-

tonic quantum technology, for example,

secure quantum communication,[1,2]

quantum internet,[3] and linear quantum

computation.[4,5] Recent world-wide activ-

ities on the implementation of quantum

networks,[6–11] including a satellite node,[12]

reﬂect the importance of the ﬁeld. Among

diﬀerent SPS concepts, to date, the purest

single-photon emission is provided by

semiconductor quantum dots (QDs) fea-

turing probabilities of multiphoton events

as low as 10−5for emission wavelengths

below 1 µm[13] (excluding their use for

long-haul transmission),[14–16] and 10−4at

telecom wavelengths under non-resonant

excitation.[17] Thus QDs constitute superb

K. Poturaj, Dr. G. Wójcik, Dr. P. Mergo

Laboratory of Optical Fibers Technology

Institute of Chemical Sciences

Faculty of Chemistry

Maria Curie Sklodowska University

Maria Curie Sklodowska Sq. 3, Lublin 20-031, Poland

K. Dybka, M. Dyrkacz, Dr. M. Dłubek

Fibrain Sp. z o.o.

Zaczernie 190F, Zaczernie 36-062, Poland

K. Lauritsen, A. Bülter

PicoQuant GmbH

Rudower Chaussee 29, Berlin 12489, Germany

Dr. P.-I. Schneider, Dr. L. Zschiedrich, Dr. S. Burger

JCMwave GmbH

Bolivarallee 22, Berlin 14050, Germany

Dr.S.Burger

Zuse Institute Berlin

Takustraße 7, Berlin 14195, Germany

www.advancedsciencenews.com www.advquantumtech.com

quantum emitters in terms of scalability, integration, and com-

patibility with advanced semiconductor technology.[18–25] Agen-

eral drawback hindering real-world applications of In(Ga)As QDs

is the cryogenic operation temperature. In fact, this is the main

reason why commercially available quantum key distribution sys-

tems and experimental quantum networks are almost exclusively

based on sources utilizing spontaneous parametric down conver-

sion or attenuated lasers.[7–12,26–28] This is despite the drawbacks

that the former is probabilistic with low eﬃciency, whereas the

latter does not inherently provide single photons making trans-

mission susceptible to the photon number splitting attack.[29]

In this work, we focus on developing a user-friendly SPS for

quantum communication in the telecom O-band. This spectral

window features a local minimum of loss in silica ﬁbers, zero

dispersion, and is suitable for multiplexing with C-band classical

signals, without the need for expensive dark ﬁbers for the quan-

tum channel, due to good spectral isolation reducing Raman

scattering. Interestingly, ﬁber-based quantum links have already

been implemented using QDs at the telecom O-band by KTH

Stockholm and by Toshiba-Cambridge.[30] However, in all these

reports, the source was operated in complex and bulky experi-

mental setups with QD emission coupled externally to a ﬁber. In

order to take the application of QDs in quantum technologies to

the next level, we developed a user-friendly “plug&play” QD SPS

that is ﬁber-coupled, compact, and portable; includes a cooling

system; and provides a stable train of spectrally ﬁltered single

photons in the O-band via a standard telecommunication single-

mode ﬁber. Importantly, this source is very convenient for the end

user as it does not require any adjustment and is fully operational

after a 15 min cool-down cycle. In contrast, commercially avail-

able QD-based sources[31] utilize a standard bulky and expensive

experimental cryostat and proof-of-principle realizations of com-

pact designs[32] are multimode ﬁber-coupled. Moreover, both ap-

proaches operate at shorter wavelengths below 1 µm and require

external spectral ﬁltering of the single QD emission. A previous

work reporting on a plug&play-like SPS operating in the telecom

O-band[14] is based on the random positioning of a single-mode

ﬁber bundle with respect to a regular non-deterministic micropil-

lar array featuring 30% multiphoton events. Whereas in our

case, the QD-mesa fabrication technology and the positioning of

the ﬁber with respect to the mesa are fully deterministic, which

allow for the utilization of a single-mode ﬁber and the integration

with a highly eﬃcient spectral ﬁltering system. Overall, this is

far beyond what has been reported so far in the ﬁeld.

2. Results

2.1. Approach and Design

Our concept for the realization of a compact ﬁber-coupled, single-

photon source is presented in Figure 1a. The overall scheme

aims at providing the end user with a stable source of internally

or externally triggered 1.3 µm single photons directly in a stan-

dard single-mode ﬁber. The source is easy to handle and oper-

ates at stable photon ﬂux without the need of any alignment or

additional spectral ﬁltering, and can be further used, for exam-

ple, for implementing quantum communication or computation

schemes.

The frame in the scheme marks elements that are placed in

a19ʺcompatible housing shown in the image in Figure 1b.

The desired functionality is realized in the following way: emis-

sion of the pulsed (80 MHz, pulse length <50 ps), non-resonant

(805 nm), ﬁber-coupled, electrically triggered, semiconductor

diode laser is ﬁrst spectrally ﬁltered to transmit only the laser

line itself into the ﬁber arrangement. The laser ﬁlter avoids un-

wanted broadband emission background, for example, sponta-

neous emission from the laser cavity at the ﬁber output of our

quantum device. In fact, while broad background emission (typ-

ically in the range of 600–1600 nm) from the laser is too low

to eﬃciently excite QDs, it is comparable to the single QD sig-

nal. Therefore, it is crucial to ﬁlter this background out directly

after the excitation laser. Once background emission passes the

pumping ﬁlter in the all ﬁber conﬁguration, it would not be ﬁl-

tered out by any other component and it would be spectrally inte-

grated by the single-photon detectors into the photon ﬂux. In that

case, even the low, spectrally broad laser background could result

in a total number of photons exceeding the number of photons

from a single optical QD transition and therefore would make

the whole system completely unusable. The spectrally ﬁltered,

single-mode laser signal is delivered to the sample via a reﬂection

channel (see Figure 1a) of a specially designed three-port ﬁlter for

pumping. The aforementioned ﬁber components are based on

standard telecom ﬁbers and they are spliced to a high numerical

aperture ﬁber (numerical aperture (NA) =0.42), which is pre-

cisely positioned with respect to the single-photon emitter via a

recently developed interferometric method[33] with an alignment

accuracy of 50 nm and ﬁxed to the sample surface by low tempera-

ture compatible epoxy glue (see Experimental Section for details).

It is important for a potential commercialization of our concept

to perform not only the ﬁber alignment in a deterministic way,

but also the device fabrication itself. For that, we apply in situ

electron-beam lithography to preselect suitable QDs via cathodo-

luminescence mapping before integrating them into microstruc-

tures with a process yield >90% and a positioning accuracy of

about 40 nm.[34] This way, cylindrical mesas were deterministi-

cally fabricated around preselected QDs in ≈20 µm ×20 µm writ-

ing ﬁelds (see inset to Figure 1c). In addition, the writing ﬁelds

act as apertures and facilitate the orientation on the sample sur-

face by mapping of its topography via a ﬁber in the alignment

step. An exemplary cathodoluminescence map is shown in Fig-

ure 1c with the corresponding emission intensity color-coded for

a narrow spectral range corresponding to the target wavelength

of 1293–1295 nm at 40 K.[34] The position of the selected QD for

device processing is marked by a black circle in Figure 1c. It is

characterized by emission in the spectral range of interest, high

intensity, and good spatial separation, which indicates that it orig-

inates from a single QD. This preselected QD is deterministically

integrated into a cylindrical mesa with a diameter of 1090 ±50

nm. Emission from this QD-micromesa is collected via the high-

NA ﬁber and enters a transmission channel of the pumping ﬁlter

that at the same time ﬁlters out the excitation signal. Finally, the

target excitonic line of the QD is spectrally selected from emis-

sion of other QD transitions such as the biexciton via a narrow

(0.6 nm) ﬁber-integrated band-pass ﬁlter. The exit port of this ﬁl-

ter is connected to the FC/PC ﬁber connector at the optical output

of our stand-alone single-photon source. The sample design, the

microstructure, and ﬁber geometry, as well as the placement of

www.advancedsciencenews.com www.advquantumtech.com

Figure 1. a) Scheme of the fully ﬁber-coupled single-photon source—the frame marks the device with a standard telecom single-mode FC/PC ﬁber con-

nector as output. b) Image of the actual device with Stirling cryocooler on the left, pulsed excitation laser on the bottom-right, and the ﬁber components

secured on the back wall. The ventilation openings in the housing and extra fans provide air ﬂow for cooling of the Stirling cryocooler, which is operated

in vertical position and mounted to the metal frame. c) Active region and sample patterning: spatial low-temperature (15 K) cathodoluminescence map

in the spectral range corresponding to the target wavelength (1293–1295 nm) with the emission intensity color-coded and the target QD marked by a

black circle. Inset: SEM image of the patterned sample with a mesa etched at the position of the target QD marked by a black circle.

the ﬁber follow the results of numerical optimization of the de-

sign parameters. To this end, the propagation of the light emitted

by the QD was simulated using a ﬁnite-elements method.[35] The

system parameters with optimal ﬁber-coupling eﬃciency were

determined using Bayesian optimization as global optimization

method.[36]

One of many technological challenges in implementing our

device concept is the ﬁltering of all spurious spectral contribu-

tions besides the single photons originating from the target QD

transition from the all-ﬁber coupled system. This is realized by

customized ﬁber components described in detail below. These

optical elements have to provide high spectral isolation of the tar-

get QD transition with minimum insertion loss.

2.2. Specialty Fiber Components

Customized ﬁber components were designed and fabricated to

fulﬁll the requirements of the stand-alone SPS device concept.

Their optical characteristics (see Figure 2) were evaluated using

a supercontinuum light source and an optical spectrum analyzer

(OSA). The laser ﬁlter responsible for cleaning preliminarily the

laser spectrum (see Figure 2a) has high transmission at the laser

line wavelength (805 nm) of –2.5 dB and high attenuation in the

spectral range of the QD emission (–62 dB). For the ﬂexibility of

the optical excitation, this ﬁlter is a broad band-pass to allow for

application of diﬀerent laser wavelengths in the range of 700–

1000 nm. Laser light in this range is spectrally removed in the

transmission channel (Figure 2c) of the ﬁber pumping ﬁlter with

an optical isolation better than –35 dB in the broad range and bet-

ter than –40 dB at the actual laser line position. In the ﬁnal device,

the laser attenuation was increased to >80 dB by combining two

ﬁlters of this type. The results of the measurements are shown

for a single ﬁlter due to limited dynamic range of the OSA. This

channel exhibits high transmission for the optical signal from

the QD with –0.7 dB loss above 1150 nm up to at least 1700 nm

(OSA detection limit) covering both the telecom O-band and C-

band. In the reﬂection channel, the transmission is high in the

range above 750 nm (loss lower than –4.2 dB, see Figure 2b) so

that the optical excitation can be delivered eﬃciently to the sam-

ple. The relatively high loss at the wavelength of the laser line

is not an issue in the case of single QDs as for the investigated

structures the emission is typically saturated at average excita-

tion powers in the single µW range (cf. Figure 3b). The next step

is to isolate a single optical transition from the other emission

signals related to the wetting layer, strain reducing layer, possible

www.advancedsciencenews.com www.advquantumtech.com

Figure 2. Spectral characteristics of the ﬁber components measured with a supercontinuum light source and an OSA: a) Broad band-pass ﬁlter for the

excitation laser. b) Single-mode pumping ﬁlter—reﬂection channel through which the optical excitation is delivered to the sample after cleaning up its

spectrum with the ﬁlter presented in (a). c) Single-mode pumping ﬁlter—transmission channel through which the QD signal is delivered to the detection

system. The laser radiation is blocked by –40 dB. d) Single-mode narrow band-pass ﬂat-top ﬁlter (0.6 nm full width at half maximum) with a central

wavelength of 1294.65 nm; inset: set of ﬁber-based narrow band-pass ﬁlters diﬀering in the adjustable central wavelength covering a 2 nm spectral range

of 1293–1295 nm. All panels: The orange curves show the transmission of the investigated component with respect to the reference signal level and the

blue ones depict the OSA noise level.

defects, or other excitonic complexes conﬁned in the same QD.

This functionality is realized by an ultra-narrow, top-ﬂat, ﬁber-

integrated band-pass ﬁlter (0.6 nm) (see Figure 2d). To provide

some spectral ﬂexibility, a set of seven exchangeable ﬁlters cov-

ering a wavelength range of 1293–1295 nm in the O-band with

isolation better than –45 dB was fabricated.

2.3. Device Performance

The optical properties of the stand-alone telecom single-photon

source are evaluated using a ﬁber-based Hanbury Brown and

Twiss conﬁguration equipped with superconducting nanowire

single-photon photon detectors (SNSPDs). Figure 3a depicts the

corresponding coincidences’ histogram (black curve) obtained

under pulsed excitation at 80 MHz with an average excitation

power of 0.75 µW recorded at the laser input of the customized

ﬁber arrangement (T=40 K), which corresponds to 0.65 µW in-

cident on the sample. The measured histogram was ﬁtted by the

sum of double-sided exponential decays for each maximum[17]

including the background level in between the emission pulses.

The peak height for the non-zero delay peaks determined from

the ﬁtting procedure was further used to normalize the coin-

cidence events histogram to obtain the time delay-dependent

second-order correlation function g(2)(𝜏). The probability of

multiphoton emission events g(2)(0) was determined from the

ﬁtting procedure as the ratio of the height of the central (zero

delay) peak and the peaks at the long time delays and yields

background-corrected g(2)(0) =0.15 ±0.05 proving single-photon

emission from the target optical transition. Here, the level of

uncorrelated background determined in between the emission

pulses is subtracted from the as measured g(2)(0) (see Experimen-

tal Section for details). This uncorrelated background signal is

mainly attributed to non-ideal laser suppression in the full-ﬁber

conﬁguration. This issue can most probably be resolved by fur-

ther increasing the attenuation of the laser blocking ﬁlter in the

future. The associated photoluminescence (PL) spectrum of the

QD at the output of the SPS is shown in the inset of Figure 3a.

The emission is centered at 1294.7 nm, and the linewidth equals

0.43 nm which is a typical value for 1.3 µm QDs,[18,37] where the

quite signiﬁcant inhomogeneous broadening is related mainly to

spectral diﬀusion eﬀects in the case of non-resonant excitation.

At this excitation strength, the total photon ﬂux yields 31 kHz at

the device output which, taking into account the 15% probability

of multiphoton events, corresponds to a true single-photon rate

of 27 kHz (Figure 3b) according to the study by Pelton et al.[38]

These rates were obtained by measuring the demonstrator’s

output using an external detection system and by correcting

it for the detection system’s eﬃciency. Therefore, this is the

actual single-photon rate that the end user will be provided with.

Under these excitation conditions, the probability of the source

to emit a single photon per excitation pulse is in the range of

www.advancedsciencenews.com www.advquantumtech.com

Figure 3. Optical properties of the stand-alone SPS: a) Normalized coincidences histogram measured under pulsed non-resonant excitation (0.75 µW

average power at 80 MHz) in all-ﬁber conﬁguration at T=40 K (black curve) together with a ﬁtting curve (red); inset: corresponding spectrum measured

in all-ﬁber conﬁguration including the ﬁber-based narrow band-pass ﬁlter (its bandwidth is marked with the dashed vertical lines). b) g(2)(0) (left scale,

blue symbols) and single-photon ﬂux rate at the output of the demonstrator (right scale, red symbols), measured under pulsed non-resonant excitation

at T=40 K as a function of average excitation power; the blue dashed line marks the g(2)(0) =0.5 deﬁning the limiting value for single-photon operation;

the error bars were obtained as the sum of the errors of ﬁtting parameters. c) Stability test of the SPS: histogram of the count rates (averaged over

10 min) on one of the SNSPDs measured for 18 h. A statistical analysis yields a mean value of 1151 counts per second (cps) and a standard deviation

of 16 cps that corresponds to a relative standard deviation of 0.014.

1%. This value is limited by a combination of below-saturation

excitation, non-ideal internal quantum eﬃciency of the emitter,

non-ideal coupling eﬃciency into the ﬁber, and losses of the

all-ﬁber conﬁguration as we discuss in Section 3 in more detail.

To investigate the upper limit of the achievable single-photon

ﬂux in the present device, coincidences histograms were mea-

sured as a function of the average excitation power in the range

of 0.75 µW up to 10 µW. The corresponding power dependences

of g(2)(0)—blue symbols and single-photon ﬂux—red symbols,

are presented in Figure 3b. The limit of single-photon emission

(g(2)(0) =0.5) is observed at 10 µW excitation power, suggesting

that at this excitation strength emission from the QD is already

saturated, and that a further increase of the excitation power

results in increased uncorrelated emission background overlap-

ping spectrally with the QD line. The associated maximal true

single-photon ﬂux corresponding to the saturation of QD emis-

sion equals 73 kHz. Its excitation power dependence follows the

emission intensity dependence of the single QD transition.

During collection of the histogram at a given excitation power,

the photon count rate at the SNSPDs was monitored over 18

h to get an insight into the long-term stability of the output of

the source. The SNSPD detector count rates were averaged over

10 min and combined to generate the histogram presented in

Figure 3c. In comparison to the emission rates in Figure 3b,

this rate is decreased by the eﬃciency of the external detection

system, including ﬁber beam splitter, ﬁber connectors, and

quantum eﬃciency of the detectors themselves. The statistical

evaluation of these data yields a mean value of 1151 cps with a

standard deviation of 16 cps, which corresponds to the relative

standard deviation of 0.014. This shows that the long-term

stability of the demonstrator output is better than 1.5%.

3. Discussion and Outlook

For optimized performance, InGaAs QDs need to be cooled

down. To take advantage of the compact and cheap cooling

method provided by Stirling cryocoolers, a direct, rigid, and

thermally as well as mechanically stable ﬁber coupling of the

QD emission to a single-mode ﬁber was developed.[33] This

overcomes the drawback of low-frequency vibrations with the

amplitude in the range of a few micrometers exceeding the size

of the quantum emitter (at most tens of nanometers) inherent to

the operation of the Stirling cryocooler. The proposed solution

with superior performance is obtained by the interplay of vari-

ous developed components, pump laser, growth of high-quality

self-assembled QDs, and the design of a mesa-mirror-ﬁber-

setup based on numerical modeling, deterministic fabrication

of QD-mesas, advanced cooling method, specialty high-NA

ﬁber, high-precision ﬁber positioning, and customized ﬁber

components for spectral ﬁltering. An important challenge of

our device concept is to deliver the excitation eﬃciently and

ﬁlter out the unwanted photons both from the pump laser and

emitted from other parts of the structure. High isolation of the

single transition is needed both in broad range as well as in a

narrow range. Due to relatively small binding energies (in the

range of 1 meV) of various excitonic complexes conﬁned in

the same QD,[37] a very sharp-edge band-pass ﬁlter is required.

Even more important is the loss for the actual single-photon

signal. Commercially available elements can oﬀer arbitrary

good isolation, but the insertion loss, especially if one has to

stack several of such elements with diﬀerent functionalities, is

typically unacceptable with values in the range of 1.5 dB per

element for a signal from a single QD. Therefore, minimizing

the insertion loss of the ﬁber components was crucial for the

realization of the reported source. This can be further optimized

by using low loss splices instead of standard mating sleeves.

The achieved single-photon ﬂux of the source in the range of

27–73 kHz, depending on the excitation conditions and required

single-photon purity (down to 0.15), is far beyond the state-of-the-

art for similar user-friendly sources: ﬁber-coupled and compact,

not requiring any external ﬁltering, but providing the end user

with a train of single photons in a standard single-mode ﬁber

at the output. At shorter wavelengths (about 900 nm), a multi-

mode, ﬁber-coupled, and compact device was demonstrated.[32]

It featured a photon ﬂux of 230 kHz into the multimode optical

ﬁber with g(2)(0) =0.57 under pulsed operation. In that case,

the single-photon character of emission could be demonstrated

under cw excitation with g(2)(0) equal to 0.07. When it comes

to ﬁber-coupled, QD-based, single-photon sources at telecom

www.advancedsciencenews.com www.advquantumtech.com

wavelengths, ﬁrst results were reported by Xu et al.[14] However,

this device was not fabricated deterministically and the reported

results were obtained at 4.2 K while dipping the device in a

liquid-helium storage dewar. For that source, the best obtained

g(2)(0) was equal to 0.2 under cw excitation and the source

eﬃciency was 3.75 ×10−5. Therefore, when compared to the

implementations above, the source reported here constitutes

a substantial improvement over state-of-the-art of any known

compact solutions.

The brightness of our single-photon source can certainly be

further optimized as the probability to have a single photon per

pulse is in the range of 1% currently. The achieved single-photon

ﬂux is modest yet, when compared to QD-based but non-compact

sources at shorter wavelengths. However, the source characteris-

tics of the latter were measured at temperatures below 10 K, that

is, requiring complex liquid-helium-based experimental setups.

In addition, external spectral ﬁltering had to be used to select a

single-photon transition, both making such a solution proof-of-

principle like rather than an application-relevant device.

Photon ﬂuxes achieved for QDs at shorter wavelengths (100

MHz[31] or 143 MHz[39]) for 0.5 GHz optical and electrical excita-

tion show the high potential of QD-based single-photon sources.

In principle, they outperform other (laser-based) approaches

also at highly application-relevant telecom wavelengths. A low

source eﬃciency in our case is mostly related to the non-ideal

photon-extraction eﬃciency of the simple QD-micromesa and

internal quantum eﬃciency of the emitter. Even for an ideal

QD and the fully optimized mesa design, the maximum theo-

retically achievable outcoupling eﬃciency into the single-mode

ﬁber is in the range of 20%.[35] Therefore, more sophisticated

engineering of the photonic environment for maximizing the

extraction eﬃciency, in particular, bulls-eye cavity designs for

which eﬃciency exceeding 90% into low-NA collection optics is

predicted,[40] might be used to improve it in the next development

stages.

Applying a tunable narrow band-pass ﬁlter might also fur-

ther increase the ﬂexibility and functionality of the entire device

concept. Also, for a good isolation of the single-photon emis-

sion from the scattered laser, the excitation wavelength has to be

signiﬁcantly diﬀerent than the QD emission range (by at least

100 nm for our device). Therefore, the strictly resonant or even

quasi-resonant excitation was at this stage not possible. It would

also be less practical as a very speciﬁc diﬀerent wavelength of the

pump would be required for each QD and only one laser source

could be used for each device, which would decrease the univer-

sal character of this design, and would also increase substantially

the cost of the whole system, which is now mostly dictated by the

Stirling cryocooler and the excitation source. One has to keep in

mind that the coherence properties in the case of the investigated

structures will be in any case limited by the operation tempera-

ture (40 K) and not by the non-resonant excitation scheme. On

the other hand, it is straightforward to use our concept, which

can easily be transferred to optical above-bandgap and wetting-

layer excitation as well as electrically triggered structures where

the pump rejection is not required at all, so it depends only on

the availability of suitable QD emitters. The purity of the single

photons is determined by the QD material itself. Thus, it is a mat-

ter of choosing a properly isolated transition with low emission

background, which is not a limitation of the implemented con-

cept itself, but relies more on the development of high-quality

QD material at telecom wavelengths.[25]

4. Conclusion

We demonstrated a user-friendly, fully ﬁber-coupled triggered

source of single photons in the telecom O-band suitable for ap-

plications in long-range quantum communication schemes. The

single photons are emitted by a semiconductor QD, determinis-

tically integrated into a micromesa and on-chip coupled to a high

NA customized single-mode ﬁber. The QD sample is cooled by

a compact Stirling cryocooler at 40 K. The main ingredient of

the proposed solution are specially designed ﬁber components,

the deterministic in situ fabrication of mesa structures follow-

ing the numerically obtained structure design around the target

QD and an ultra-precise interferometric method for ﬁber align-

ment (accuracy below 50 nm) with respect to the mesa center.

Combining these developments resulted in a device performance

with a probability of multiphoton events as low as 15% and the

maximal single-photon generation rate at the single-mode ﬁber

output of 73 kHz. The obtained rate of generation of single pho-

tons is by two orders of magnitude larger than the ≈0.7 kHz

reported by Schlehahn et al.[32] for a multimode, ﬁber-coupled,

stand-alone SPS based on a standard InGaAs QD emitting in

the 900–950 nm wavelength range. This highlights the signiﬁ-

cant advances achieved in the present work, which not only pro-

vides a fully ﬁber-based solution, but also demonstrates single-

mode operation in the O-band—all of which are crucial prerequi-

sites for real-world device applications, for example, in the ﬁeld

of quantum communication. The long-term stability of the op-

tical output of the stand-alone SPS is better than 1.5% (stan-

dard deviation). Our user-friendly device concept does not re-

quire a supply of cryogenic liquids; is robust; and provides a hard-

ware solution being compact, mechanically stable, and portable.

It does not require any additional adjustments or post selection

of the single photons by the user as the ﬁltering ﬁber systems

are already integrated. Therefore, the end user can operate the

source in a plug&play fashion, as at the output, it has a standard

telecommunication single-mode ﬁber that delivers the train of

triggered single photons at 1294.7 nm for the used QD. Notewor-

thy, our approach is independent of the material used and can be

adapted to diﬀerent spectral ranges. The limiting factors are the

structural quality of the QD material and the spatial QD density.

Additionally, the possibility of tuning the QD-based SPSs with

external strain and static electric ﬁeld[41,42] as well as electrical

excitation[43] could be easily integrated in our source, rendering

its application potential even larger. Therefore, these results pave

the way to real-word application of QD-based ﬁber quantum net-

works.

5. Experimental Section

Sample Growth:The QDs were formed by self-organization during

metalorganic chemical vapor deposition in the Stranski–Krastanov growth

mode. Starting with a GaAs wafer, ﬁrst, the bottom distributed Bragg re-

ﬂector with 15 pairs of GaAs/Al0.9Ga0.1As layers with 98.3/113.8 nm thick-

ness on a 300 nm GaAs buﬀer was grown at 700 °C followed by a 505.4 nm

thick additional GaAs layer. For the growth of the QD layer, the temperature

www.advancedsciencenews.com www.advquantumtech.com

was decreased down to 500 °C. The active region constitutes of InGaAs

QDs (formed from 2.5 monolayers (ML) of InGaAs with 66% In content

and ﬂushed with 1 ML of GaAs) and is followed by a 5.5 nm thick InGaAs

strain-reducing layer with an In gradient from 30% at the bottom down to

10% at the top. After initial capping with 2 nm of GaAs, the ﬁnal capping

layer consisting of 612.7 nm of GaAs is grown at 615 °C.

In Situ Electron-Beam Lithography:The applied ﬁber positioning

method[33,44] requires that a single QD is located with high accuracy in the

center of a micrometer-sized nanophotonic structure. For that purpose, in

situ electron-beam lithography (EBL)[45] optimized for 1.3 µm emission

wavelength and with an overall positioning accuracy below 50 nm[34] was

utilized. In this procedure 310 ±2 nm of chemical semi ampliﬁed EBL

resist (measured by ellipsometry) diluted to a solid content of 6.5% was

spin coated on the sample surface. Next, the sample was mounted in the

in situ EBL system and cooled down to cryogenic temperatures (15 K). At

ﬁrst, the cathodoluminescence mapping was performed with 40 ms inte-

gration time for each pixel to identify the most suitable QDs for further

processing. The criteria at this step are the spatial isolation of the QD, the

emission intensity, and the spectral range of emission. In this context, it

is important to note that QD emission needs to ﬁt within the bandwidth

of the ﬁber band-pass ﬁlters which span in the range of 1293–1295 nm.

The mapping dose must be below the onset dose for inverting the resist

and in this particular processing, 8 mC cm−2was used. After identifying a

suitable QD, the lithography step was performed still at cryogenic temper-

atures within the same in situ EBL system. The resist was exposed using

25 mC cm−2electron dose and single cylindrical mesa structures were pat-

terned in each writing ﬁeld. The nominal diameters of the mesas on this

sample are 1050, 1075, and 1090 nm, and the mesa height (corresponding

to etching depth) equals to 620 ±5 nm. Further processing was performed

at room temperature in the cleanroom. It included resist development and

dry etching (reactive ion etching) of the patterned structures. Afterwards,

scanning electron microscopy (SEM) in top-view conﬁguration (no tilt an-

gle) was performed to determine the actual mesa diameters and the etch-

ing depth was veriﬁed via proﬁlometer measurements.

Positioning of the Fiber with Respect to the Mesa:For positioning of the

optimized high-NA single-mode ﬁber with respect to the mesa center prior

to gluing a zirconia ferrule with the ﬁber to the sample surface, an interfer-

ometric method detailed by ˙

Zołnacz et al.[33,44] was used. The alignment

procedure was performed at room temperature which, importantly, does

not rely on the actual QD signal (which is not detectable at room tempera-

ture), but takes advantage of the deterministic character of the mesa struc-

ture fabrication with a single QD in the center. The position of the ﬁber

was adjusted for the center of the mesa based on measurements of the

topography of the sample utilizing the interference between the spectrally

broad signal from a supercontinuum light source reﬂected from the ﬁber

facet and the surface of the sample dependent on the distance between

the ﬁber and the sample surface. Both the ﬁber and the sample surface

were ﬁrst positioned horizontally using piezo actuators and the ﬁber was

further moved across the sample surface at constant distance. The analy-

sis of the spectral interference fringes allowed us to position the ﬁber with

respect to the mesa center with 50 nm accuracy (for micromesas with di-

ameters smaller than 2 µm), that is, with a deviation much smaller than

the diameter (2.5 µm) of the single-mode ﬁber core. After having aligned

the ﬁber with respect to the mesa center leaving an air gap between mesa

and ﬁber end facet of about 0.5 µm, the ﬁber was set in physical contact

with the sample surface which is crucial for long-term mechanical and ther-

mal stability. Then the ﬁber ferrule was glued to the sample surface with

ceramic UV-cured glue exhibiting a low thermal expansion coeﬃcient of

only 14 ppm °C−1(compared to the GaAs coeﬃcient of 5.73 ppm °C−1). It

is important to note that this glue is not transparent in the spectral range

of interest, so it was only applied outside the ferrule leaving the ﬁber end

facet and the mesa top surface free of any glue.

Fiber Components:The ﬁber components specially designed for the

presented single-photon source include a single-mode ﬁber with high

(0.42) numerical aperture, broad band-pass ﬁlter to remove/suppress un-

wanted spontaneous emission background of the pulsed laser source, a

ﬁber ﬁlter responsible for delivering the optical excitation to the sample (re-

ﬂection channel) as well as spectrally suppressing it out from the detection

Table 1. Parameters of the specialty ﬁber ﬁlters.

Parameter Pumping ﬁlter Narrow

band-pass ﬁlter

Transmission

channel

Reﬂection channel

Maximal

insertion

loss [dB]

0.70 4.20 (single-mode)

1.80 (multimode)

0.50

Bandwidth

[nm]

1150 ÷ 1600 785 ÷ 1000 Central wave-

length ±0.3

Laser

attenuation

[dB]

>40 (per chip) >2.50

path to provide high transmission for the actual QD signal (transmission

channel), and ﬁnally a narrow band-pass ﬁlter (0.6 nm) for selecting the

target emission line of the ﬁber-coupled QD. Additionally, the customized

ﬁber coupled to the QD has to be spliced with a standard telecom ﬁber to

provide an easy-to-integrate output of the device. The transmission of each

component was evaluated with respect to the reference measurement in

which a tested component was exchanged with a simple patch cord to ac-

count for spectral characteristics of the source and the detection system.

Additionally, the noise level of the OSA itself was measured each time for

illustrating the dynamic range of the detector to verify whether it is large

enough or the result of the measurements constitutes only a lower limit

of the isolation provided by the respective ﬁber element.

The customized ﬁber was optimized in terms of numerical aperture and

residual thermal stress for safe operation at cryogenic temperatures. The

high numerical aperture (NA =0.42) was achieved by using a highly Ge-

doped core with 40 mol% of GeO2. Such a high level of doping results in

a huge diﬀerence of thermal expansion coeﬃcients between the core and

the cladding of the ﬁber that might lead to ﬁber breaking either during fab-

rication of the ﬁber or cooling it down for low-temperature measurements

of the QD signal. To reduce the stress level, a ﬁber with a three-step doping

proﬁle (40, 13, and 5 mol%), resulting in the similar refractive index pro-

ﬁle (1:2:3 diameter ratios), was fabricated following the design obtained

via the ﬁnite-element method simulations. This approach resulted in a nu-

merical aperture of 0.42 with a cut-oﬀ wavelength for single-mode opera-

tion of 1050 nm for a ﬁber with 2.5 µm core diameter. The customized ﬁber

was terminated with a zirconia ferrule polished into the spherical end-face.

It was glued directly to the QD-micromesa and further sealed by an epoxy

glue in a specially designed vacuum feedthrough to a portable Stirling cry-

ocooler. The customized ﬁber has to be further spliced to a standard tele-

com single-mode ﬁber (core diameter of 9 µm) and the main challenge

here is to overcome the 3.6 factor between the ﬁber core diameters. A low-

loss splice (0.2 dB in both directions) was achieved using glass processing

station with CO2laser splicer via the thermal core expansion technique

that relies on equalizing the ﬁber core diameters by controlled heating of

the splice area causing the GeO2dopant from the core to diﬀuse to the

cladding, eventually creating a gradual low loss splice.

All ﬁlters were fabricated in the thin ﬁlm ﬁlter (TFF) technology. The

band-pass ﬁlters have a bandwidth of 0.6 nm with a ﬂat-top characteristic

that is important to maximize the emission line transmission through the

ﬁlter. It is also possible to fabricate ﬁlters with a diﬀerent bandwidth down

to 0.3 nm (at the expense of higher loss of 1.2 dB). For their fabrication,

a commercially available TFF chip was used and the central wavelength

of the actual ﬁlter can be tuned by 2 nm via the angle of incidence on the

chip that can be tuned on the fabrication stage of the ﬁlter. The tuning

range of the central wavelength is limited by the change in the shape of

the bandwidth and polarization selectivity appearing for high angles of

incidence to 2 nm at maximum and 1 nm in practice. The parameters of

the ﬁber ﬁlters are summed up in Table 1.

www.advancedsciencenews.com www.advquantumtech.com

The characteristics of the ﬁber ﬁlters (see Figure 2) were measured

using ﬁber-coupled supercontinuum light source (NKT Photonics SuperK

Versa) and an OSA (Yokogawa AQ6370B) covering the spectral range

of 600–1700 nm. The transmission of the ﬁlters was determined as

a diﬀerence between the transmission of setup with the ﬁlter and a

reference measurement in which the ﬁlter was replaced by a ﬁber patch

cord. Additionally, the noise level of the OSA was measured showing in

which cases the measurement results (in particular signal attenuation by

the ﬁlter) are limited by the sensitivity of the detection system itself. In

this case, only the lower limit for the attenuation can be determined due

to limited dynamic range of the detector.

Experimental Setup—Spectroscopy Measurements:In general, two ex-

perimental conﬁgurations were used for the spectroscopy study of the SPS.

The common part of the two conﬁgurations was the ﬁber-glued sample

mounted in the Stirling cryocooler (base temperature 38 K) with optical

excitation (laser output ﬁltered with the broad band-pass ﬁlter) delivered

via the customized ﬁber pumping ﬁlter. The non-resonant excitation of

the investigated QD structures was realized using an electrically triggered,

ﬁber-coupled, semiconductor diode laser that was custom-designed by Pi-

coQuant. The laser was built around a preselected laser diode emitting

at 805 nm. Special driving electronics permit freely selectable repetition

rates up to 100 MHz using internal or even external triggering. The laser

emits pulses with a pulse width of <50 ps (full width at half maximum)

with average power of a few milliwatts, which can be further reduced by

a built-in computer-controlled attenuator allowing to adjust the excitation

power to the requirements of the QD structures. The diﬀerence in the two

experimental conﬁgurations appears in the way the signal from the sam-

ple was ﬁltered spectrally. For the pre-characterization of emission from

the sample, the optical signal was out-coupled from the output port of

the ﬁber ﬁlter for pumping to free space and ﬁltered spectrally via a 0.32

m focal length spectrometer with 600 grooves mm–1 grating blazed at

1000 nm, providing 0.4 nm bandwidth at its output. The PL signal was fur-

ther coupled to the single-mode ﬁber connected to a single-photon count-

ing module (superconducting NbN nanowire detector with 20% quantum

eﬃciency—SNSPD). This conﬁguration was used to measure PL spectra of

the QD in a broad spectral range to identify proper excitation conditions

for autocorrelation measurements and, in particular, to select a proper

band-pass ﬁlter for the second, all-ﬁber conﬁguration. For the measure-

ments on the actual fully ﬁber-coupled device, the output of the pumping

ﬁlter was connected to the narrow band-pass ﬁlter and further via the out-

put connector of the stand-alone SPS to a 50:50 beam splitter based on

single-mode ﬁbers. Each output of the beam splitter was then connected

to a SNSPD for autocorrelation measurements that were carried out us-

ing a multichannel picosecond event timer (PicoHarp 300) with 256 ps

time-bin width. The measured histograms were ﬁtted with the following

function[17]

g(2)

ﬁt (𝜏)=gbg +gautof(|𝜏|)+∑

n≠0

f(|

|𝜏−nT0|

|)(1)

where gbg corresponds to the background counts, gauto indicates ﬁgure of

merit—g(2)(0), and f(𝜏) is the normalized biexponential function with T0

being the distance between the consecutive pulses (corresponding to the

repetition rate of the excitation laser) that describe the autocorrelation be-

tween the photons emitted within diﬀerent pulses. The peak height for the

non-zero delay peaks determined from the ﬁtting procedure was further

used to normalize the coincidence events histogram to obtain the time

delay-dependent second-order correlation function g(2)(𝜏).

Acknowledgements

This work was funded by the FI-SEQUR project jointly ﬁnanced by the

European Regional Development Fund (EFRE) of the European Union

in the framework of the programme to promote research, innovation,

and technologies (Pro FIT) in Germany, and the National Centre for Re-

search and Development in Poland within the 2nd Poland-Berlin Photon-

ics Programme, grant number 2/POLBER-2/2016 (project value 2 089 498

PLN). Support from the German Science Foundation via CRC 787, from

the Federal Ministry of Education and Research, BMBF, grant number

05M20ZBM, and from the Polish National Agency for Academic Exchange

is also acknowledged.

Conﬂict of Interest

The authors declare no conﬂict of interest.

Keywords

III-V semiconductor epitaxial quantum dots, ﬁber elements design and

fabrication, photon statistics, quantum communication, quantum-dot-

based devices, quantum optics, single-photon sources

Received: February 10, 2020

Revised: March 26, 2020

Published online: May 11, 2020

[1] D. Bouwmeester, A. Ekert, A. Zeilinger, The Physics of Quantum Infor-

mation, Springer, Berlin 2000.

[2] N. Sangouard, C. Simon, J. Mináˇ

r, H. Zbinden, H. de Riedmatten, N.

Gisin, Phys.Rev.A2007,76, 050301.

[3] H. J. Kimble, Nature 2008,453, 1023.

[4] J. L. O’Brien, Science 2007,318, 1567.

[5] D.J.Brod,E.F.Galvão,A.Crespi,R.Osellame,N.Spagnolo,F.Scia-

rrino, Adv. Photonics 2019,1, 034001.

[6] T. Heindel, C. A. Kessler, M. Rau, C. Schneider, M. Fürst, F. Hargart,

W.-M. Schulz, M. Eichfelder, R. Roßbach, S. Nauerth, M. Lermer, H.

Weier, M. Jetter, M. Kamp, S. Reitzenstein, S. Höﬂing, P. Michler, H.

Weinfurter, A. Forchel, New J. Phys. 2012,14, 083001.

[7] M. Lucamarini, Z. L. Yuan, J. F. Dynes, A. J. Shields, Nature 2018,557,

400.

[8] M. Sasaki, M. Fujiwara, H. Ishizuka, W. Klaus, K. Wakui, M. Takeoka,

S. Miki, T. Yamashita, Z. Wang, A. Tanaka, K. Yoshino, Y. Nambu, S.

Takahashi, A. Tajima, A. Tomita, T. Domeki, T. Hasegawa, Y. Sakai, H.

Kobayashi, T. Asai, K. Shimizu, T. Tokura, T. Tsurumaru, M. Matsui,

T. Honjo, K. Tamaki, H. Takesue, Y. Tokura, J. F. Dynes, A. R. Dixon

et al., Opt. Express 2011,19, 10387.

[9] H.-L. Yin, T.-Y. Chen, Z.-W. Yu, H. Liu, L.-X. You, Y.-H. Zhou, S.-J. Chen,

Y. Mao, M.-Q. Huang, W.-J. Zhang, H. Chen, M. J. Li, D. Nolan, F.

Zhou, X. Jiang, Z. Wang, Q. Zhang, X.-B. Wang, J.-W. Pan, Phys. Rev.

Lett. 2016,117, 190501.

[10] R. Dixon, Z. L. Yuan, J. F. Dynes, A. W. Sharpe, A. J. Shields, Opt.

Express 2008,16, 18790.

[11] X.-S. Ma, T. Herbst, T. Scheidl, D. Wang, S. Kropatschek, W. Naylor, B.

Wittmann, A. Mech, J. Koﬂer, E. Anisimova, V. Makarov, T. Jennewein,

R. Ursin, A. Zeilinger, Nature 2012,489, 269.

[12] S.-K.Liao,W.-Q.Cai,J.Handsteiner,B.Liu,J.Yin,L.Zhang,D.Rauch,

M. Fink, J.-G. Ren, W.-Y. Liu, Y. Li, Q. Shen, Y. Cao, F.-Z. Li, J.-F. Wang,

Y.-M.Huang,L.Deng,T.Xi,L.Ma,T.Hu,L.Li,.N.-L.Liu,F.Koidl,P.

Wang, Y.-A. Chen, X.-B. Wang, M. Steindorfer, G. Kirchner, C.-Y. Lu,

R. Shu, R. Ursin et al., Phys. Rev. Lett. 2018,120, 030501.

[13] L. Schweickert, K. D. Jöns, K. D. Zeuner, S. F. Covre da Silva, H.

Huang, T. Lettner, M. Reindl, J. Zichi, R. Trotta, A. Rastelli, V. Zwiller,

Appl. Phys. Lett. 2018,112, 093106.

[14] X. Xu, F. Brossard, K. Hammura, D. A. Williams, B. Alloing, L. H. Li,

A. Fiore, Appl. Phys. Lett. 2008,93, 021124.

www.advancedsciencenews.com www.advquantumtech.com

[15] M. B. Ward, O. Z. Karimov, D. C. Unitt, Z. L. Yuan, P. See, D. G.

Gevaux, A. J. Shields, P. Atkinson, D. A. Ritchie, Appl. Phys. Lett. 2005,

86, 201111.

[16] C. Zinoni, B. Alloing, C. Monat, V. Zwiller, L. H. Li, A. Fiore, L. Lunghi,

A. Gerardino, H. de Riedmatten, H. Zbinden, N. Gisin, Appl. Phys.

Lett. 2006,88, 131102.

[17] T. Miyazawa, K. Takemoto, Y. Nambu, S. Miki, T. Yamashita, H. Terai,

M. Fujiwara, M. Sasaki, Y. Sakuma, M. Takatsu, T. Yamamoto, Y.

Arakawa, Appl. Phys. Lett. 2016,109, 132106.

[18] M. Paul, J. Kettler, K. Zeuner, C. Clausen, M. Jetter, P. Michler, Appl.

Phys. Lett. 2015,106, 122105.

[19] Ł. Dusanowski, P. Holewa, A. Mary´

nski, A. Musiał, T. Heuser, N.

Srocka, D. Quandt, A. Strittmatter, S. Rodt, J. Misiewicz, S. Reitzen-

stein, G. S˛

ek, Opt. Express 2017,25, 31122.

[20] M. Benyoucef, M. Yacob, J. P. Reithmaier, J. Kettler, P. Michler, Appl.

Phys. Lett. 2013,103, 162101.

[21] A. Musiał, P. Holewa, P. Wyborski, M. Syperek, A. Kors, J. P. Re-

ithmaier, G. S ˛

ek, M. Benyoucef, Adv. Quantum Technol. 2020,3,

1900082.

[22] S. L. Portalupi, M. Jetter, P. Michler, Semicond. Sci. Technol. 2019,34,

053001.

[23] P. Michler, Quantum Dots for Quantum Information Technologies,

Springer, Cham, Switzerland 2017.

[24] P. Lodahl, Quantum Sci. Technol. 2018,3, 013001.

[25] X. Cao, M. Zopf, F. Ding, J. Semicond. 2019,40, 071901.

[26] A. Muller, H. Zbinden, N. Gisin, Europhys. Lett. 1996,33, 335.

[27] S. Wengerowsky, S. K. Joshi, F. Steinlechner, J. R. Zichi, S. M. Dobro-

volskiy, R. van der Molen, J. W. N. Los, V. Zwiller, M. A. M. Versteegh,

A. Mura, D. Calonico, M. Inguscio, H. Hübel, L. Bo, T. Scheidl, A.

Zeilinger, A. Xuereb, R. Ursin, Proc. Natl. Acad. Sci. USA 2019,116,

6684.

[28] A. Boaron, G. Boso, D. Rusca, C. Paul Vulliez, C. Autebert, M. Caloz,

M. Perrenoud, G. Gras, F. Bussières, M.-J. Li, D. Nolan, A. Martin, H.

Zbinden, Phys. Rev. Lett. 2018,121, 190502.

[29] G. Brassard, N. Lütkenhaus, T. Mor, B. C. Sanders, Phys. Rev. Lett.

2000,85, 1330.

[30] Z. Xiang, J. Huwer, R. M. Stevenson, J. S. Szymanska, M. B. Ward, I.

Farrer, D. A. Ritchie, A. J. Shields, presented at Conf. on Lasers and

Electro-Optics, OSA Technical Digest, CLEO: QELS_Fundamental

Science 2019, San Jose, California United States, 5–10 May 2019, pa-

per FM4C.6.

[31] Quandela homepage, http://quandela.com/ (accessed: December

2019).

[32] A. Schlehahn, S. Fischbach, R. Schmidt, A. Kaganskiy, A. Strittmatter,

S. Rodt, T. Heindel, S. Reitzenstein, Sci. Rep. 2018,8, 1340.

[33] K. ˙

Zołnacz, A. Musiał, N. Srocka, J. Große, M. J. Schlösinger, P.-I.

Schneider, O. Kravets, M. Mikulicz, J. Olszewski, K. Poturaj, G. Wój-

cik, P. Mergo, K. Dybka, M. Dyrkacz, M. Dłubek, S. Rodt, S. Burger, L.

Zschiedrich, G. S˛

ek, S. Reitzenstein, W. Urba´

nczyk, Opt. Express 2019,

27, 26772.

[34] N. Srocka, A. Musiał, P.-I. Schneider, P. Mrowi´

nski, P. Holewa, S.

Burger, D. Quandt, A. Strittmatter, S. Rodt, S. Reitzenstein, G. S ˛

ek,

AIP Adv. 2018,8, 085205.

[35] P.-I. Schneider, N. Srocka, S. Rodt, L. Zschiedrich, S. Reitzenstein, S.

Burger, Opt. Express 2018,26, 8479.

[36] P.-I. Schneider, X. G. Santiago, V. Soltwisch, M. Hammerschmidt, S.

Burger, C. Rockstuhl, ACS Photonics 2019,6, 2726.

[37] P. Mrowi´

nski, A. Musiał, K. Gawarecki, Ł. Dusanowski, T. Heuser, N.

Srocka, D. Quandt, A. Strittmatter, S. Rodt, S. Reitzenstein, G. S˛

ek,

Phys. Rev. B 2019,100, 115310.

[38] M. Pelton, C. Santori, J. Vuc˘

kovi´

c, B. Zhang, G. S. Solomon, J. Plant,

Y. Yamamoto, Phys.Rev.Lett.2002,89, 233602.

[39] A. Schlehahn, M. Gaafar, M. Vaupel, M. Gschrey, P. Schnauber, J.-H.

Schulze, S. Rodt, A. Strittmatter, W. Stolz, A. Rahimi-Iman, T. Heindel,

M. Koch, S. Reitzenstein, Appl. Phys. Lett. 2015,107, 041105.

[40] L. Rickert, T. Kupko, S. Rodt, S. Reitzenstein, T. Heindel, Opt. Express

2019,27, 36824.

[41] B. Höfer, F. Olbrich, J. Kettler, M. Paul, J. Höschele, M. Jetter, S. L. Por-

talupi, F. Ding, P. Michler, O. G. Schmidt, AIP Adv. 2019,9, 085112.

[42] Z.-H.Xiang,J.Huwer,J.Skiba-Szymanska,R.M.Stevenson,D.J.P.El-

lis, I. Farrer, M. B. Ward, D. A. Ritchie, A. J. Shields, arXiv:1909.12222,

2019.

[43] M. B. Ward, T. Farrow, P. See, Z. L. Yuan, O. Z. Karimov, A. J. Bennett,

A. J. Shields, P. Atkinson, K. Cooper, D. A. Ritchie, Appl. Phys. Lett.

2007,90, 063512.

[44] K. ˙

Zołnacz, W. Urba´

nczyk, N. Srocka, T. Heuser, D. Quandt, A.

Strittmatter, S. Rodt, S. Reitzenstein, A. Musiał, P. Mrowi´

nski, G. S˛

ek,

K. Poturaj, G. Wójcik, P. Mergo, K. Dybka, M. Dyrkacz, M. Dłubek,

Proc. SPIE 2018,10674, 106741R.

[45] M. Gschrey, A. Thoma, P. Schnauber, M. Seifried, R. Schmidt, B.

Wohlfeil, L. Krüger, J.-H. Schulze, T. Heindel, S. Burger, F. Schmidt,

A. Strittmatter, S. Rodt, S. Reitzenstein, Nat. Commun. 2015,6, 7662.