Abstract
An ondemand singlephoton source is a key element in a series of prospective quantum technologies and applications. Here we demonstrate the operation of a tuneable ondemand microwave photon source based on a fully controllable superconducting artificial atom strongly coupled to an openended transmission line. The atom emits a photon upon excitation by a short microwave πpulse applied through a control line. The intrinsically limited device efficiency is estimated to be in the range 65–80% in a wide frequency range from 7.75 to 10.5 GHz continuously tuned by an external magnetic field. The actual demonstrated efficiency is also affected by the excited state preparation, which is about 90% in our experiments. The singlephoton generation from the singlephoton source is additionally confirmed by antibunching in the secondorder correlation function. The source may have important applications in quantum communication, quantum information processing and sensing.
Introduction
Control and manipulation with light at the singlephoton level^{1,2,3,4} is interesting from fundamental and practical viewpoints. In particular, ondemand singlephoton sources are of high interest because of their promising applications in quantum communication, quantum informatics, sensing and other fields. In spite of several realizations in optics^{5,6,7,8,9}, practical implementation of photon sources imposes a number of requirements, such as high photon generation, collection efficiencies and frequency tunability. Recently developed superconducting quantum systems provide a novel basis for the realization of microwave (MW) photon sources with photons confined in resonator modes^{10,11,12,13,14,15}, which is essentially different from the threedimensional (3D) case of optical photon sources^{16}. However, all the circuits consist of two elements (the resonator and the quantum emitter system) and the generated photon frequency is fixed to the resonator frequency.
In this work, we propose and realize a different approach: a singlephoton source based on a tuneable artificial atom coupled asymmetrically to two openended transmission lines (onedimensional (1D) halfspaces) and a similar scheme was proposed by Lindkvist et al.^{17}. The atom is excited from a weakly coupled control line (c) and emits a photon to a strongly coupled emission line (e). The photon freely propagates in the emission line and can be further processed using, for example, nonlinear circuit elements. Among the advantages of our circuit is its simplicity: it consists of a single element.
Results
Operation principles and device description
An optical analogue of the proposed singlephoton source consists of a twolevel atom situated near a tiny hole (much smaller than the wavelength) in a nontransparent screen (Fig. 1a). The atom is slightly shifted towards the righthandside space, defining an asymmetric coupling to the halfspaces. By applying powerful light from the left side, the atom can be excited by evanescent waves, which cannot propagate in the righthandside space because of their rapid decay. On the other hand, the excited atom emits photons into the righthandside space (Fig. 1b). In practice, the presented layout is difficult to build using natural atoms and, even if one succeeds, another problem must be solved: the low collection efficiency of emitted photons in the 3D space. These problems can be easily avoided by using onchip superconducting quantum circuits coupled to 1D transmission lines^{18,19,20}. Figure 1c shows a circuit with an artificial atom coupled asymmetrically to a pair of openended coplanar transmission lines (1Dhalf spaces), each with Z=50 Ω impedance. The coupling capacitances C_{c} and C_{e} are between the artificial atom and the control and emission lines, respectively (shown on the equivalent circuit in Fig. 1d). The capacitances can be approximated as pointlike objects because their sizes are much smaller than the wavelength of the radiation (∼1 cm). Note also that the transmission lines in the centre of our device (about 80 μm for each line) slightly differ from 50 Ω because of the shifted down ground plane; however we can ignore it because they are also much shorter than the wavelength. A MW pulse is applied from the control line, exciting the atom, and then the atom emits a photon mainly to the emission line because of asymmetric coupling: C_{e}/C_{c}≈30. The following are intrinsic features of the device: the two lines are well isolated from each other so that the excitation pulse does not leak from the control line to the emission line; because of the strong asymmetry, the excited atom emits a photon with up to 1−(C_{c}/C_{e})^{2} output probability; the photon is confined in the 1D transmission line and can be easily delivered to other circuit elements through the line.
The artificial atom schematically shown in Fig. 1d is a controllable twolevel system based on a tuneable gap flux qubit^{21,22,23,24}, that is coupled to two Nb coplanar lines. The atom is fabricated by Al/AlOx/Al shadow evaporation techniques. It contains two identical junctions in series implemented in the loop together with a dcSQUID (called an αloop), shown in the bottom part of the device in Fig. 1d. Here α≈0.7 specifies the nominal ratio between the two critical currents in the dcSQUID and the other two Josephson junctions in the loop. The magnetic fluxes are quantized in the loop: an integer number, N, of the magnetic flux quanta, Φ_{0}, can be trapped. At the magnetic fields where the induced magnetic flux in the loop is equal to Φ=Φ_{0}(N+1/2), two adjacent flux states 0〉 and 1〉 with N and N+1 flux quanta, which is corresponding to oppositely circulating persistent currents, are degenerated. The degeneracy is lifted because of the finite flux tunnelling energy Δ_{N}, determined by the effective dcSQUID Josephson energy and varies between different degeneracy points (depends on N). The energy splitting of the atom is controlled by fine adjustment of the magnetic field δΦ in the vicinity of the degeneracy points, where δΦ=Φ−(N+1/2)Φ_{0} and I_{p} is the persistent current in the main loop. (We neglect the weak dependence of Δ_{N} on δΦ.)
The capacitances of the circuit are estimated to be C_{c}≈0.3 fF and C_{e}≈9 fF. The effective impedance between the two lines because of the capacitive coupling is Z_{C}=1/iω(C_{c}+C_{e}), which is about 2 kΩ for ω/2π=10 GHz and the transmitted part of the power is as low as 2Z/Z_{C}^{2}≈2.5 × 10^{−3}. This enables nearly perfect line decoupling.
Device characterisation
Our experiment is carried out in a dilution refrigerator at a base temperature of around 30 mK. We first characterize our device by measuring the transmission coefficient t_{ce} from the control line to the emission line using a vector network analyser (VNA) and the reflection coefficient r_{e} from the emission line. Figure 2a shows a twodimensional (2D) plot of the normalized transmission amplitude t_{ce}/t_{0} in the frequency range 7.75–10.5 GHz with the magnetic flux bias δΦ from −30 to 30 mΦ_{0} around the energy minimum, where t_{0} is the maximal transmission amplitude. The transmission is suppressed everywhere except in the narrow line that corresponds to the expected atomic resonance at ω_{10} and is a result of the photon emission from the continuously driven atom. The spectroscopic curve is slightly asymmetric with respect to δΦ=0 because of the weak dependence of Δ_{N} on δΦ. From the spectroscopy line we deduce the parameters of the twolevel system: the tunnelling energy is Δ=min(ħω_{10})=h × 7.750 GHz at δΦ=0 and the persistent current in the loop is I_{p}≈45 nA.
To evaluate the coupling of our atom to the emission line, we also measure the reflection at δΦ=0 with different probing powers from −147 to −121 dBm. Next, Fig. 2b shows the reflection coefficient r_{e} mapped in the complex plain, measured in the case of atom excitation from the emission line (opposite to the case of source operation). The curve changes its form from circular to oval, which reflects the transition from the linear weakdriving regime up to the nonlinear strongdriving regime of the twolevel system^{18}.
We next derive the dynamics of the point like atom (the loop size ∼10 μm is much smaller than the wavelength λ∼1 cm) located at x=0 and coupled to the 1D open space via an electrical dipole. We also take into account the fact that . The atom is driven by the oscillating voltage at the frequency ω of the incident wave V_{0}(x, t)=V_{0}e^{−iωt+ikx} in the control line and the resulting driving amplitude of 2V_{0} cos ωt=Re[V_{0}e^{−iωt+ikx}+V_{0}e^{−iωt−ikx}]_{x=0} is the sum of the incident and reflected waves. The Hamiltonian of the atom in the rotatingwave approximation is H=−(ħδωσ_{z}+ħΩσ_{x})/2, where δω=ω−ω_{10} and ħΩ=−2V_{0}C_{c}ν_{a} with the electric dipole moment of the atom ν_{a} (between C_{c} and C_{e}). The atomic voltage creation/annihilation operator is , where . The driven atom generates voltage amplitudes of V_{c,e}(t)/2=iωZC_{c,e}ν_{a}〈σ^{−}〉e^{−iωt} in the control (x<0) and emission (x>0) lines. Substituting the relaxation rates due to voltage quantum noise (S_{V}(ω)=2ħωZ) from the line impedance Z in each line, we obtain
In the ideal case of suppressed pure dephasing (γ=0) and in the absence of nonradiative decay , the power ratio between the control and emission lines generated by the atom under resonance is , which means that up to 99.9% of the power generated by the atom can be emitted into the emission line. This allows us to measure the spectroscopy curve shown in Fig. 2a. To find 〈σ^{−}〉 under continuous driving, we solve the master equation by considering the total relaxation rate , where is the nonradiative relaxation rate (for a photon absorbed by the environment). Here, the dephasing rate is Γ_{2}=Γ_{1}/2+γ, where γ is the pure dephasing rate. The solution is . The reflection in the control line and the transmission coefficient from the control line to the emission line are r_{c}=1+V_{c}(0, t)/V_{0}(0, t) and t_{ce}=V_{e}(0, t)/V_{0}(0, t), respectively. At the weakdriving limit ,
are circular plots in the complex plane. Similar to equation (3), we can write down the following expression for the reflection in the emission line
with the substitution by . We will further use this expression to characterize the coupling strength and efficiency of our device.
Furthermore, the excited atom emits an instantaneous power proportional to the atomic population I_{1}(t) (ref. 19) that can be straightforwardly expressed as
where I_{1}=(1−〈σ_{z}〉)/2. If the atom is prepared in the excited state 1〉 at t=0, the probability decays according to I_{1}(t)=exp(−Γ_{1}t). In addition, the efficiency of the photon emission to the right line is , which ideally can be as high as . The plot in Fig. 2b gives us a measure of the coupling strength of the atom to the emission line. Using equation (5), we estimate the highest possible efficiency of photon generation to be =0.79. Note that in real experiments it is additionally affected by the excited state control efficiency because of the competition between excitation and relaxation processes. However, the manipulation efficiency is not fundamentally limited and can be nearly one, if the available equipment allows to make πpulse lengths much shorter than .
Device operation
Our photon source based on the conversion of an atomic excitation into a MW photon requires efficient control of the quantum states. Figure 3a shows measured quantum oscillations. We monitor the coherent emission from the atom into the emission line by VNA when a train of identical excitation MW pulses, each of length Δt with period T=80 ns, is applied from the control line. The amplitude of the emission oscillates with Δt. The maxima and minima of the oscillations correspond to 〈σ^{±}〉≈±1, when the atom is in the maximally superposed states with 50% population. For the singlephoton source operation, we tune the pulse length to obtain the maximum incoherent emission (defined as a πpulse and its length is Δt_{π}=3.5 ns), emitting a single photon from the atom excited state in every pulse period. The traces are then amplified and digitized with sampling time 4 ns. The traces V(t) of repeated measurements are then squared and accumulated. The typical photon shape P(t) obtained after 2 × 10^{9} times averaging is shown in Fig. 3b. The inset shows the averaged emission power peak excited by the πpulses with repetition time and measured by a spectrum analyser. Using the Lorentzian fit, we obtain a FWHM Δω/2π≈20 MHz, which is equal to the relaxation rate Γ_{1}.
Measurements of correlation functions
The twolevel system operation together with the Rabi oscillations prove that the source generates a single photon at a time. Nevertheless, we provide additional evidence of the singlephoton generation by measuring the secondorder correlation function with linear detectors (microwave amplifiers)^{25}. Such a demonstration is straightforward in optics due to existence of photon counters but extremely demanding in the microwave range, where the microwave amplifiers have typical signaltonoise ratio in the singlephoton regime is less than 10^{−2} in power and, therefore, long accumulation of statistics is required.
The circuit schematically shown in Fig. 4a is implemented to perform Hanbury–Brown–Twiss measurements^{26} using linear detectors^{12,13,14,25,27,28}. We accumulate traces each consisting of a train of 40 pulses with a period of T=160 ns. The emitted photons are then transmitted through an isolator to a 90° hybrid coupler operating as a microwave beam splitter. The idle input port terminated by 50 Ω is a source of vacuum noise. The two signals coming from the splitter output ports of the hybrid coupler are amplified by amplifiers at 4.2 K and at room temperature. We assume that the noise added by the amplifiers are uncorrelated in each channel. Next, the signals, downconverted to close to zerofrequency with mixers, are passed through 30 MHz bandwidth lowpass filters and voltage amplitudes V_{1}(t) and V_{2}(t) are recorded by two digitisers. Finally, the traces are read out and treated by a PC to extract the twopoint correlations functions.
First, we demonstrate the firstorder correlation function of the two signals. The normalized function with subtracted background is exemplified in an inset of Fig. 4b. The central peak (G^{(1)}(0)) corresponds to the total power emitted by the atom and the side peaks (G^{(1)}(nT)) correspond to the coherent emission. The dynamics of the peaks as a function of the excitation pulse length is shown in Fig. 4b. The solid lines demonstrate simulations with the previously extracted device parameters.
Next, we calculate the second order correlated function of the emitted radiation defined as
The result of measurements after averaging of 1.5 × 10^{10} traces is shown in Fig. 4c, where the function is obtained from by subtracting the background and normalization. The traces are additionally smoothed to reduce fluctuations. The red curves show expectations calculated with the measured photon shapes. We observe a series of side peaks spaced by T with a suppressed peak at zero delay time =0. The observed antibunching of the emission demonstrates the singlephoton generation.
Singlephoton source efficiency at different frequencies
We evaluate the devicelimited efficiency of our source over a wide frequency range by tuning the emission frequency ω_{10} controlled by δΦ. First, we would like to point out that in our fluxqubitbased atom, pure dephasing is expected to be strongly suppressed because of the low persistent current I_{p} being one order lower than in conventional designs^{22,23}. Therefore, the pure dephasing should not affect the efficiency too much, even when we detune the energy from the minimal ones. We characterize the coupling strength using equation (5) by measuring the circle radius r_{e} at different frequencies in the complex plane, similar to Fig. 2b. Figure 5 shows the derived device intrinsicallylimited efficiency as a function of frequency. We obtained more than 65% efficiency almost everywhere over the range of frequencies from 7.75 to 10.5 GHz. The efficiency can be affected by nonnegligible pure dephasing γ and/or nonradiative relaxation .
We would like to point out that the efficiency of the excited state preparation in our experiments can be estimated as (ref. 19) reducing the total efficiency at the degeneracy point down to 0.79 × 0.87=0.69. This value is consistent with the one obtained after calibration of our setup, which is found to be 0.67.
Discussion
In conclusion, we demonstrated an onchip tuneable ondemand singlemicrowavephoton source operating with high efficiency over a wide range. The source is expected to be useful for applications including quantum communication, quantum information processing, and sensing.
Data availability
The data that support the findings of this study are available from the corresponding author on request.
Additional information
How to cite this article: Peng, Z. H. et al. Tuneable ondemand singlephoton source in the microwave range. Nat. Commun. 7:12588 doi: 10.1038/ncomms12588 (2016).
References
 1.
Kimble, H. J. The quantum internet. Nature 453, 1023–1030 (2008).
 2.
Duan, L. M. & Monroe, C. Quantum networks with trapped ions. Rev. Mod. Phys. 82, 1209–1224 (2010).
 3.
Sangouard, N. et al. Quantum repeaters based on atomic ensembles and linear optics. Rev. Mod. Phys. 83, 33–80 (2011).
 4.
Northup, T. E. & Blatt, R. Quantum information transfer using photons. Nat. Photon. 8, 356–363 (2014).
 5.
Kim, J. et al. A singlephoton turnstile device. Nature 397, 500–503 (1999).
 6.
Lounis, B. & Moerner, W. E. Single photons on demand from a single molecule at room temperature. Nature 407, 491–493 (2000).
 7.
Kurtsiefer, C. et al. Stable solidstate source of single photons. Phys. Rev. Lett. 85, 290 (2000).
 8.
Keller, M. et al. Continuous generation of single photons with controlled waveform in an iontrap cavity system. Nature 431, 1075–1078 (2004).
 9.
He, Y. M. et al. Ondemand semiconductor singlephoton source with nearunity indistinguishability. Nat. Nanotechnol. 8, 213–217 (2013).
 10.
Houck, A. A. et al. Generating single microwave photons in a circuit. Nature 449, 328–331 (2006).
 11.
Hofheinz, M. et al. Generation of Fock states in a superconducting quantum circuit. Nature 454, 310–314 (2008).
 12.
Bozyigit, D. et al. Antibunching of microwavefrequency photons observed in correlation measurements using linear detectors. Nat. Phys. 7, 154–158 (2011).
 13.
Lang, C. et al. Observation of resonant photon blockade at microwave frequencies using correlation function measurements. Phys. Rev. Lett. 106, 243601 (2011).
 14.
Lang, C. et al. Correlations, indistinguishability and entanglement in HongOuMandel experiments at microwave frequencies. Nat. Phys. 9, 345–348 (2013).
 15.
Pechal, M. et al. Microwavecontrolled generation of shaped single photons in circuit quantum electrodynamics. Phys. Rev. X 4, 041010 (2014).
 16.
Lodahl, P. et al. Interfacing single photons and single quantum dots with photonic nanostructures. Rev. Mod. Phys. 87, 347–400 (2015).
 17.
Lindkvist, J. & Johansson, G. Scattering of coherent pulses on a twolevel system—singlephoton generation. New J. Phys. 16, 055018 (2014).
 18.
Astafiev, O. et al. Resonance fluorescence of a single artificial atom. Science 327, 840–843 (2010).
 19.
Abdumalikov, A. A. et al. Dynamics of coherent and incoherent emission from an artificial atom in a 1D space. Phys. Rev. Lett. 107, 043604 (2011).
 20.
Hoi, I. C. et al. Generation of nonclassical microwave states using an artificial atom in 1D open space. Phys. Rev. Lett. 108, 263601 (2012).
 21.
Mooij, J. E. et al. Josephson persistentcurrent qubit. Science 285, 1036–1039 (1999).
 22.
van der Wal, C. H. et al. Quantum superposition of macroscopic persistentcurrent states. Science 290, 773–777 (2000).
 23.
Paauw, F. G. et al. Tuning the gap of a superconducting flux qubit. Phys. Rev. Lett. 102, 090501 (2009).
 24.
Zhu, X. B. et al. Coherent operation of a gaptunable flux qubit. Appl. Phys. Lett. 97, 102503 (2010).
 25.
da Silva, M. P. et al. Schemes for the observation of photon correlation functions in circuit QED with linear detectors. Phys. Rev. A 82, 043804 (2010).
 26.
Brown, R. H. & Twiss, R. Q. Correlation between photons in two coherent beams of light. Nature 177, 27–29 (1956).
 27.
Gabelli, J. et al. Hanbury BrownCTwiss correlations to probe the population statistics of GHz photons emitted by conductors. Phys. Rev. Lett. 93, 056801 (2004).
 28.
Menzel, E. P. et al. Dualpath state reconstruction scheme for propagating quantum microwaves and detector noise tomography. Phys. Rev. Lett. 105, 100401 (2010).
Acknowledgements
Peng would like to thank Z.R. Lin for help and Y. Kitagawa for preparing Nb wafers. We thank T. Lindstrom for helping in correlation measurements and useful discussions. This work was carried out within the project EXL03 MICROPHOTON of the European Metrology Research Programme (EMRP). EMRP is jointly funded by the EMRP participating countries within EURAMET and the European Union. This work was supported by ImPACT Program of Council for Science, Technology and Innovation (Cabinet Office, Government of Japan). The work has been supported by Russian Science Foundation, grant no. 161200070.
Author information
Affiliations
Physics Department, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK
 Z. H. Peng
 & O. V. Astafiev
Center for Emergent Matter Science, RIKEN, Wako, Saitama 3510198, Japan
 Z. H. Peng
 & J. S. Tsai
National Physical Laboratory, Teddington TW11 0LW, UK
 S. E. de Graaf
 & O. V. Astafiev
Department of Physics, Tokyo University of Science, Kagurazaka, Tokyo 1628601, Japan
 J. S. Tsai
Moscow Institute of Physics and Technology, Dolgoprudny 141700, Russia
 O. V. Astafiev
Authors
Search for Z. H. Peng in:
Search for S. E. de Graaf in:
Search for J. S. Tsai in:
Search for O. V. Astafiev in:
Contributions
Z.H.P. and O.V.A. designed the device, carried out the experiment and analysed data. Z.H.P. and O.V.A. wrote the manuscript. S.E.G. has participated in the correlation function measurements. All the authors participated in discussions and contributed to editing the manuscript.
Competing interests
The authors declare no competing financial interests.
Corresponding authors
Correspondence to Z. H. Peng or O. V. Astafiev.
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Further reading

Observation of microwave absorption and emission from incoherent electron tunneling through a normalmetal–insulator–superconductor junction
Scientific Reports (2018)

Switching Dynamics of an Underdamped Josephson Junction Coupled to a Microwave Cavity
Journal of Low Temperature Physics (2018)

Multiplexing Superconducting Qubit Circuit for Single Microwave Photon Generation
Journal of Low Temperature Physics (2017)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.