Arxiv:2101.05398V2 [Quant-Ph] 12 Apr 2021 Lcdibtent Oma Tmcrsntr Hscon- Atoms This Additional Resonator

Home , Atomic mirror

arXiv:2101.05398v2 [quant-ph] 12 Apr 2021 lcdibtent oma tmcrsntr hscon- atoms This additional resonator. atomic with an mirrors form to Bragg inbetween placed atomic two bine [ n ttsta a nbeepnnilipoeetin improvement [ exponential fidelities enable storage photon may that states ant a of realization [ the gate and single-photon optics non-linear quantum plore nanofiber- [ thousands atoms several trapped with observed was flection [ physics bandgap exploring of 11 possibility unique the offer would one- that with [ waveguides including photonic-crystal demonstrated dimensional studied, also actively was read are photon platforms be single Novel guided can a as that out excitation atomic waveguide-coupled [ effects non-linear exper- to atoms and nanofiber-trapped proposals [ with theoretical of demonstrations variety imental a to has electrodynamics, led quantum waveguide as known field, ligteBagrsnnecniin h nebeof [ ensemble mirror Bragg the a condition, as behave ful- resonance can atoms atoms Bragg of chain the ordered filling periodically a with interacts grto aeas asdalreitrs,wt o only not with [ interest, states large subradiant a extremely raised also have figuration eyatv rao eerh pnn e opportu- new opening [ presently interactions research, is light-matter of photons for area guided nities active with very interact a neu- where atoms platforms tral hybrid harnessing and Developing ∗ † 4 2 3 , , unu ehooisCne,MV oooo ocwSaeU State Moscow Lomonosov M.V. Center, Technologies Quantum ..Pivovarov, V.A. [email protected] [email protected] ie hsscpblte,antrletnini ocom- to is extension natural a capabilities, theses Given nsc rpe tmcary,we h uddmode guided the when arrays, atomic trapped such In ]. 3 5 2 ,rnigfo h elzto fotclmemories optical of realization the from ranging ], oteosraino ueaineadcollective and superadiance of observation the to ] etrfrAvne tde,PtrteGetSt-Petersburg Great the Peter Studies, Advanced for Center igecletv xiaino naoi ra rpe alo trapped array atomic an of excitation collective Single td fcoeaieeiso o ieetaoi hi co chain atomic different for emission cooperative of study a 1 hsc eatet t-eesugAaei University, Academic St.-Petersburg Department, Physics ih,t ietoal eueo nac t n nsm cas some in system. and the it, bounding enhance mirrors or atomic absorb system reduce the or atomic directionally reflecting one-dimensional to either a Addi light, role, such waveguide. passive geometry, the a the o play into varying and light sides array emits both linear and or a mode as arranged superradiant excit are a atomic q collective atoms single and The w a information configurations. of nanoscale dynamics quantum of decay to the vicinity study applications the in with trapped interfaces, arrays atomic Ordered 13 .INTRODUCTION I. 1,2 , 6 15 14 , N-nvri´ S,Cl`g eFac,4paeJsiu 7 Jussieu, place 4 France, Collège de ENS-Université PSL, ..Gerasimov, L.V. 7 .Ti ytmcnb sdt ex- to used be can system This ]. .Coeaieeet nti con- this in effects Cooperative ]. .Rcnl,tegnrto fa of generation the Recently, ]. 17 ]. 16 4 u loslcieyradi- selectively also but ] aoaor ate rse,Sron nvri´,CNRS, Université, Sorbonne Brossel, Kastler Laboratoire 1 12 ,2 3, .Ti emerging This ]. .Sc rg re- Bragg Such ]. .Berroir, J. Dtd pi 4 2021) 14, April (Dated: 4 .Ray, T. 8 9 ]. – hr itne ewe tmcBagmroswieSec. while mirrors Bragg atomic between distances short iuain ftedcydnmc,frdffrn configu- different for Section dynamics, rations. decay the theoret- of general Section simulations the and framework. study ical under system the review conventional to lifetime. analogy [ photon enhanced in complete with detailed a re- cavity-QED was mirror resonators ones the long long to for to respect propa- time, with where negligible sponse regime a the is from delay in transition gation i.e., The resonator, atomic mirrors. electrodynam- short classical quantum with cavity states (QED) to ics of akin achieved, density coupling be strong the can that demonstrating of splitting hence enhancement Rabi predicted, vacuum was the A from anti-nodes. res- its atomic arising of short one very at a onator inside placed strategically [ atom in explored theoretically was figuration ioedpl opigbtenteaosi free-space. in atoms the and between emission coupling spontaneous nanostruc- dipole-dipole mi- includes a intrinsically a of and presence on the ture relies in treatment theory ab-initio Our timescales croscopic two dynamics. in resulting the of mirror, segment governing Bragg a in atomic in prepared larger is a system mode a is superradiant mirror is the atomic configuration configura- which one explored novel where in Another examine case disorder removed. one-sided also of as We such effect in- tions, the We mirrors. as atomic dynamics. such the system’s aspects, the novel ex- to troduce photon respect virtual re- with through non-negligible change interactions non- the with in the associated tardation where is fol- definitions regime We used Markovian two. conventionally the the between low differences [ key in regime, the non-Markovian As that confirming and Markovian mode the waveguide. and both atomic the consider resonator into superradiant the emits a inside cooperatively in placed prepared atoms initially of ensemble an 4 oyehi nvriy 921 t-eesug Russia St.-Petersburg, 195251, University, Polytechnic iest,Lnnky oy13,199,Mso,Russia Moscow, 119991, 1-35, Gory Leninskiye niversity, nti ae,a kthdi Fig. in sketched as paper, this In h ae sognzda olw.I Sec. In follows. as organized is paper The .Laurat, J. hoia8 901S.Ptrbr,Russia St.-Petersburg, 194021 8, Khlopina to ope oawvgiei different in waveguide a to coupled ation n hseiso.W hwta when that show We emission. this ing st oaiei nacvt omdby formed cavity a in it localize to es atmnnlna pis ee we Here, optics. non-linear uantum ol eal ordrc h emitted the redirect to able be could vgie ffroiia light-matter original offer aveguides l emn fte sectdto excited is them of segment a nly inlaoi hispae none on placed chains atomic tional 4 .Urvoy, A. III 05Prs France Paris, 5005 osdr h akva eiewith regime Markovian the considers 4, III ∗ gawaveguide: a ng n ..Kupriyanov D.V. and and nfigurations IV 1 rvd theoretical provide enwconsider now we , 18 ,wt single a with ], 19 II ,showing ], efirst we , 19 ,2, 3, ,we ], † 2

but more importantly, this scheme allows us to model a rather realistic example of an open system in which an emitted photon can experience either Rayleigh or Ra- man scattering. As we will show, even with such an open system, the cooperative processes indeed mainly develop via the Rayleigh scattering channel, whose main role is to transport a polaritonic excitation along the waveguide. The quantization axis is taken along the axis of the waveguide, following the approach of Ref. [20], which is a practical configuration for a tripod scheme, and offers the benefits of a cylindrical symmetry. The atoms are all prepared in the F0, +1 state (spin oriented along the waveguide).1 Subsequently,| i a weak and short coherent FIG. 1. Schematic representation of the system.An atomic external light pulse excites the active atoms and creates array trapped near a sub-wavelength dielectric waveguide is a single collective excitation in a superradiant mode in split in three segments (left, center, right) separated by a dis- the atomic array. wg tance d0. Atoms within each segment are spaced by λ /2 As a typical experimental example, in this study we (unless specified otherwise). Disorder will also be considered consider the F0 = 0 F = 1 transition of the for the left and right atomic mirrors. All atoms have a tripod 87Rb D2 line,| with ai vacuum→ | transitioni wavelength level scheme, as shown in the inset, and are initially spin po- λ0 = 780 nm, coupled to a silica nanofiber of radius larized in the ˆz direction, albeit for the central segment that a 200 nm. The atoms are separated from the surface contains one excitation with the proper phase factors for co- by∼ half the fiber radius a/2 100 nm. This configuration operative emission into the waveguide. The atoms can emit leads to a single-atom coupling∼ ratio of β = γ /γ 0.1, light in free space and in the waveguide in all three polariza- 1D ∼ tions (dashed pink arrows). The red arrows specifically high- where γ1D and γ are the radiative decay rates into the light the coupling of an arbitrary atomic pair on σ− atomic guided mode and in free space (undisturbed), respec- transition via the guided field, which is the only one that can tively [13, 21, 22]. cooperatively couple to the spin-polarized ground state atoms via the Rayleigh channel and thus undergo Bragg reflection from the segments. The aim of this paper is to study the B. Microscopic description of the excitation decay dynamics of the decay process in various configurations. We first consider an atomic chain, with a single collective excitation shared among all NC central atoms at a IV focuses on the non-Markovian one, which maps very given initial time t = 0 (central atomic chain in Fig. 1). well to the conventional cavity-QED setting even for an In the subsequent quantum description, such a collective excitation stored in an ensemble of atoms. atomic state, which we denote as ψ(0) , should be expanded in the complete basis set by| includingi the vacuum environment of the field modes. Since this state is not II. FUNDAMENTALS: SYSTEM AND MODEL an eigenstate of the Hamiltonian of the system, it further experiences a joint dynamics with the field modes, such In this section, we first present the considered scheme that up to arbitrary time t it evolves to for coupling an atomic array to a nanoscale dielectric waveguide and we introduce the parameters that are ex- i perimentally relevant. Then we focus on the theoretical ψ(t) = exp Htˆ ψ(0) U(t, 0) ψ(0) (2.1) | i −~ | i≡ | i model describing the atomic excitation temporal evolution and the photonic modes emerging from this evolu- where ψ(t) is a polariton-type entangled state that su- tion. We conclude this section by presenting the decay perposes| ai single atomic excitation and a single-photon dynamics for a chain of atoms sharing an initial single occupation of the field modes. The decay of the exci- collective excitation, the base case where left and right tation can be traced by the time-dependent probability atomic Bragg mirrors are absent. p = p(t) given by p(t) = TrPˆ ρˆ(t) withρ ˆ(t)= ψ(t) ψ(t) . (2.2) | ih | A. An example: a nanofiber-trapped atomic array

1 In our study, the atoms are modeled by a tripod energy This choice is different than in most experimental implementa- tions, where the quantization axis is taken to be normal to the configuration, with the following spin angular momenta waveguide and typically in the plane of the atoms and waveguide. and projections: F0 = 1, M0 = 0, 1 (ground state m ) If instead of circular we used linear polarizations along either x and F,M =0, 0 (excited state n ).± Such an energy| con-i or y axes, and superposed our atoms between m = ±1 states, we figuration can be found in the hyperfine| i manifold of 87Rb, would obtain similar results to those discussed below. 3

The operator Pˆ defined as a compound state considered at the conventional initial moment t = 0 [26]. In the interaction representation, the N corresponding asymptotic transformation from the initial Pˆ = m1,...,ma−1,n,ma+1,...mN | i state ψ(0) (excited atoms) to the outgoing state ψ out Xa=1 X Xn {mj },j6=a (emitted| photon),i is given by | i m ,...,m ,n,m ,...m 0 0 , (2.3) h 1 a−1 a+1 N | × | ih |Field ψ = Ωˆ (−)† ψ(0) projects the system density matrixρ ˆ(t) onto the Hilbert | iout | i subspace of the atomic excitation, where the photonic (±) i i Ωˆ = lim exp Hˆ τ exp Hˆ0τ field is vacuum and the atomic state is any superposition τ→∞ ∓2~ ±2~ of singly excited states with one specific a’th atom (with = lim U( τ/2, 0)U0( τ/2, 0) (2.5) a running from 1 to N) occupying a Zeeman sublevel n τ→∞ ± ∓ of the excited state, and all N 1 other atoms occupying| i where H is the system Hamiltonian and H0 is the channel a Zeeman sublevel m (j = a)− of the ground state. Note | j i 6 Hamiltonian corresponding to the separated subsystems. that the initial state is one of such states. The probability In our case the output channel consists of the emitted to obtain the system in its initial state can be found by photon and the de-excited atomic chain. Since the latter projecting the final stateρ ˆ(t) onto its initial state with exists in its ground state, then in the rotating wave ap- ˆ ˆ P P0 = ψ(0) ψ(0) in (2.2). In this case, we denote proximation (RWA), we can ignore in H any coupling → | ih | 0 the observation probability as p0(t). with the quantized field. As Pˆ = Pˆ2 and Pˆ ψ(0) = ψ(0) , the excitation prob- The Møller operators are nontrivial mathematical ob- | i | i ability can be expressed in the following form jects which map the states belonging to the continuous spectrum of the compound system onto the separated ˆ ˆ ˆ † ˆ p(t) = Tr PU(t, 0)P ρˆ(0) PU (t, 0)P. (2.4) states for the particular scattering channel. Although the operators are constructed as a product of two uni- This basic equation describes the microscopic dynam- tary operators U(...) and U0(...) describing the inter- ics of the decay process in a multiatomic system, where acting and non-interacting dynamics respectively, if the the optical excitation can be virtually re-emitted many compound system had a set of stable states belonging to times. We considered here only the transformation of its discrete spectrum, the product considered at infinite pure states, but the derivation can be straightforwardly τ generally would not tend to a unitary operator. generalized towards transformation of an arbitrary initial In→ our ∞ system there are no stable and localized coupled mixed state ρ(0). The evolution operator U(t, 0) can be excited states of the field and the atoms and the Møller found in previous work [20, 22–25] and is summarized in operators are unitary. We detail the operator Ωˆ (−)†, con- Appendix A. tributing to the transformation (2.5), in Appendix B. Let us point out that the projection of the evolution- The asymptotic state ψ can be decomposed into ary operator on a single excitation subspace physically out the infinite set of the states| i f G, s . Here G = means that no doubly excited states play a role in the m ,...,m represents any| ofi ≡ the | collectivei atomic| i propagation dynamics through the waveguide. However 1 N |ground eigenstates,i and the quantum number s speci- including doubly excited states is necessary in our model fies the freely-propagating field modes modified by the to account for the near field virtual interaction giving rise presence of the waveguide. One particular mode is the to a static dipole-dipole coupling for any atom with its guided mode s = σ, k parameterized by an azimuthal proximal neighbors. number σ = 1 and a longitudinal wavenumber k. Some details regarding± the description of the field and of the C. Light emitted into the waveguide atom-field interaction Hamiltonian near a nanostructure are discussed in Appendix C. The transition amplitude to a particular f-th final state is eventually given by After describing the general theoretical framework that (−)† gave us access to the evolution of the atomic populations, Mf0 = f Ωˆ ψ(0) let us now extend it to also extract the temporal profile h | | i N ∗ of the light emitted into the waveguide. (s) = i 2π~ωs d D (ra) From the point of view of scattering theory we deal · nma p aX=1 Xn here with an example of a process, where the wave-packet ˆ˜ of an initial state of the compound system evolves into ...ma−1,n,ma+1 ... R(Ef + i0) ψ(0) , × h | | i the final state of two uncoupled subsystems belonging (2.6) to a particular scattering channel. In accordance with general principles of the scattering theory, the dynam- where ωs is the frequency of the photon outgoing to the s- ics of such an incomplete scattering process can be de- th mode. The mode, associated with the photon injected scribed with the formalism of the Møller operators: the into the waveguide, contributes through the spatial pro- system’s states are transformed either from infinite past file of its displacement field D(s)(r) subsequently taken (+) (−) (operator Ωˆ ) or from infinite future (operator Ωˆ ) to at the position r = ra of each atom in the chain. 4

The spatial profile α = α(z) for a pulse outgoing the here from the polarization sensitivity of the cooperative 3 system within any azimuthal guided mode, and when the emission on the σ− transition, see Fig. 1. atoms occupy any ground state, can be constructed as We investigate how the emission by such a chain would change in different conditions. Specifically, this base case ′ 2 ∗ i(k−k )z α(z) = M M ′ e (2.7) of a single excited chain will be compared with two other | | f0 f 0 Xf,f ′ geometries, where we extend the atomic array by additional segments of atom arrays all in the ground state where f = f ′ excepting the longitudinal wavenumbers (see Fig. 1), to the left and/or to the right. In the follow- k and k′. In accordance with the general concept of ing, we will refer to the additional segments as atomic the scattering theory, the transformation (2.5) concerns mirrors, and to the single excitation collective state in only spatial profile of the asymptotic wave emerging the the central segment as a collective emitter. In order to system. Physically we can associate it with free dy- follow the decay process, we will provide three proba- namics of the outgoing light wavepacket by substituting bilities. The first one is p(t), the excitation probability z z v τ, where v is the group velocity.2 shared in the whole atomic array the sample as defined → − g g by (2.2) and (2.3). The second one is p0(t) that cap- tures the decay of the original excited state, as defined in D. The base case: decay dynamics for a chain with Section IIB by swapping Pˆ for Pˆ0. Finally we will also a single atomic excitation compute the probability pa(t):

p (t) = Tr Pˆ ρˆ(t), (2.9) We consider here the case of a single segment of NC a a atoms, with the initial atomic state given by where Pˆa is deﬁned as in (2.3), with the only exception

NC that it projects only onto the central segment subspace. 1 iϑa ψ(0) = e g,g,...,ea,...,g . (2.8) In the considered one-dimensional open system, we can | i √N | i C Xa=1 expect the recurrent incoherent scattering to play only a negligible role. This translates to the balance p0(t) = For the sake of simplicity, g F0, +1 denotes the p (t) being fulfilled, which we will verify for any atomic | i ≡ | i a atomic ground state and e F, 0 the excited state. configuration, within our calculation precision. The ket-vector in the right-hand| i ≡ | sidei of (2.8) can incor- porate the occupied atomic states from all the segments – central, left and right – but the excitation is shared III. RESULTS IN THE MARKOVIAN REGIME only by the atoms of the central segment, as clarified in Fig. 1. In order for the atoms’ emission to constructively We now turn to studying the various configurations of interfere into the waveguide, each phase shift ϑa must be the system with atomic Bragg mirrors added on the left matching the retardation phase inside the waveguide at and/or the right side. The response time of an N-atom the position of the a-th atom. If the atoms are separated −1 Bragg mirror is ΓM , where ΓM Nγ1D/2 is similar by a multiple of a half-wavelength of the guided mode to the cooperative decay rate introduced∼ above. In this wg λ /2, the phases take the values ϑa = π (odd multi- section we assume that the retardation over the length plier) and ϑa = 0 (even multiplier) and the radiation of the full array L is limited compared to the response is symmetrically emitted in both the waveguide direc- −1 time of the atomic mirror, i.e., L/vg ΓM where vg tions. ψ(0) is thus a Dicke state and the superradiant is the group velocity inside the waveguide.≪ This is the | i mode of the central segment, emitting cooperatively into so-called Markovian regime [19, 27, 28], that is achieved the waveguide. Experimentally, the state (2.8) can be when the distance d0 between the segments is small. We prepared by a short external pulse excitation addition- first discuss (case 1 ), where the full chain is in the atomic wg ally phase modulated along the waveguide, or through Bragg mirror configuration, i.e., d0 = λ /2. Then in wg heralding a spin excitation [8]. (case 2 ), d0 = λ /4 and we show that the breakdown of For a separation between the atoms given by a multiple translational invariance for the full chain leads to massive of a half-wavelength of the guided mode – the well-known changes in the response. Finally we comment on the atomic Bragg mirror case – the system is open from both temporal profiles and directionality of the light emitted sides and exhibits a cooperative exponential decay, with into the waveguide in these different cases. a rate given by ΓC NC γ1D/2. The scaling rate γ1D coincides with the evaluation≃ of the single particle self- energy part in [22]. Let us note that the 1/2 factor results 3 Here one should not confuse the σ− transition with the azimuthal mode σ = −1. In the emission process the atoms are coupled with both the modes σ = ±1 and each of the modes overlaps all three transitions σ−,π,σ+. Nevertheless the emission of a single 2 Even with negligible dispersion the group velocity is an impor- atom into the waveguide via the π-transition is typically much tant parameter of the theory, see [20, 22], and typically it is about weaker than via the σ±-transitions that justifies the 1/2 factor, vg ∼ 0.7 c for a nanofiber. see [22] ( Eq. (3.5)) and related comments there. 5

FIG. 2. Translationally-invariant configuration with d0 = wg λ /2. The number of atoms are NL = NC = NR = 100. The asymmetric case with NL = 0 and NR = 200 would yield the same results, as discussed in the text. (a) Decay wg of the probabilities p0(t)= pa(t) (magenta dash-dotted) and FIG. 3. Same as in Fig. 2 (d0 ∼ λ /2, Ntot = 300 to- p(t) (magenta). Also shown are the cooperative emission of tal atoms) but with disordered atomic mirrors, and consid- the unperturbed collective emitter exp(−ΓC t) (gray) and the ering the cases of one-sided and symmetric arrangement of single-atom natural spontaneous decay in free space exp(−γt) the atomic mirrors. (a) The decay probabilities p0(t)= pa(t) (gray dashed). The inset in linear scale clarifies the early stage (dash-dotted purple) coincide with the cooperative emission of the decay process. (b) Schematic of the arrangement of the of the collective emitter alone. The excitation probability for full chain and spatial profiles of the light pulses emitted into the whole sample p(t) decays more initially if the collective the waveguide (purple shaded profiles), see (2.7). The grey emitter is placed at one end of the full array [orange dotted, shaded profiles represent the emission in the absence of atomic configuration shown in (b)] as opposed to between the two dis- mirrors. ordered atomic mirrors [purple, configuration shown in (c)]. (b) and (c) display the one-sided arrangement and the symmetric one, and provide the associated emission profiles. A. Case 1: Bragg-mirror chain

mode of the complete atomic system. This projection In our first example we consider the configuration where wg has amplitude NC/Ntot, hence the initial fast decay in d0 = λ /2. Here, the collective emitter consists in a Fig. 2 lasts onlyp until p(t) has decreased by NC /Ntot. segment of N atoms from a longer array of N = C tot The asymptotes of p (t) and p(t) deviate from the nat- N + N + N atoms, that essentially forms an atomic 0 L C R ural spontaneous decay exp( γt). These later time dy- Bragg mirror with N . Figure 2 shows numerical results tot namics result from the projection− onto the (N 1) other for the decay probabilities, calculated with N = N = tot L C collective modes, which are subradiant into the− waveg- N = 100 atoms, i.e., for 300 atoms in total. R uide and hence decay by spontaneous emission into the We first note that, in this situation, the decay dynam- external modes with a decay rate γext. The presence of ics is insensitive to the actual position of the collective the waveguide restricts spontaneous emission to a smaller emitter within the full array. For a small enough number than 4π solid angle, resulting in γext < γ. 4 This is of atoms such that the Markovian approximation also ap- a specific property of this one-dimensional and transla- plies for propagation along the full array from end to end, tionally invariant atomic configuration, arising from the the retardation for the internal virtual processes is negli- ability to cooperatively emit the photon along the array gible. The initially prepared superradiant mode can thus direction via the Rayleigh channel with only a negligi- be instantly re-expanded in the basis set of the collective ble probability to scatter via the Raman channel. We eigenmodes of the complete resolvent operator and then obey the wavepacket dynamics driven by its complete self-energy part (effective Hamiltonian). The beginning of the decay reveals a stage of fast co- 4 If we combine the single-atom emission rate into the waveguide ext ext operative emission into the waveguide, which arises from γ1D with emission into the external modes γ we get γ + ext the projection of the initial state onto the superradiant γ1D > γ (Purcell effect). Simultaneously we have γ < γ. 6 found that p0(t)= pa(t) within our numerical precision, Fig. 3(c). showing that there is no signature of recurrent scattering in the emission process and that the atomic excitation in the central segment only occupies the superradiant mode B. Case 2: Broken invariance state (2.8). The perfect ordering of atoms is a crucial requirement Let us now consider a different configuration, where the for the data presented in Fig. 2. Disorder in the left translational symmetry along the full array is broken, but and right atomic mirrors, such as completely random- only at the boundary between the segments by shifting ized positions, would break their ability for giving rise to wg wg the separation d0 from its reference value λ /2 to λ /4. coherent Bragg-type interference with the atoms in the At the same time we preserve the periodic distance of central segment. In this case the decay dynamics is mod- λwg/2 between the atoms in each segment and their abil- ified as shown in Fig. 3, where the atoms in the mirrors ity for the Bragg-type interference with the guided light. are disordered, following a uniform probability distribu- With identical left and right atomic mirrors, this con- tion along the length of the mirror with a density of 1 figuation is then the extension of Refs. [18, 19] from a wg atom per λ /2. The initially excited atoms remain fully single atom to a collective emitter in the central segment ordered. Somewhat surprisingly, the beginning stage of of the chain. the decay process in Fig. 3 appears similar to the case In Refs. [18, 19] this configuration has been called a of the ordered system shown in Fig. 2. Even for the dis- quantum cavity, given that the central atoms sit at the ordered configuration there is meaningful probability to anti-node of the photonic mode and therefore maximally project onto a cooperatively emitting collective state of couple to it. A mapping to the Jaynes-Cummings model the full atomic array involving all the atoms and with was made [18], exemplified by the appearance of a vac- specific choice of the phase shifts ϑa in (2.8). That col- uum Rabi splitting, which shows that the density of pho- lective state exhibits fast superradiant emission from the tonic states is indeed enhanced. However we emphasize full array into the waveguide observed at short times in the crucial difference with cavity-QED, already pointed Fig. 3. out in [19], that in this process the left and right segments The most striking difference with the ordered configu- neither operate as classical light reflectors nor as a cavity ration concerns the population of the original state. In trapping the light. This means that the photon lifetime in the tail of the process the dynamics changes and the dis- the system does not extend beyond the natural lifetime ordered mirrors act as absorbers that incoherently re- with the presence of the “atomic Bragg mirrors”. scatter the photon into any mode. As a consequence we In this geometry we eliminate the contribution of the observe an exponential depopulation of the initial state global superradiant mode, which still exists but only has nearly identical to the original undisturbed cooperative negligible overlap with the initially prepared state. But emission: the excitation leaves the collective emitter as if now each atomic segment, considered independently, pos- the atomic mirrors were absent. Just as above, no other sesses a superradiant mode (2.8) as one of the eigenstates collective state than the superradiant mode is populated of its own partial resolvent, and the complete resolvent within the active atoms, i.e. p0(t)= pa(t). operator can be decomposed into the basis of these states The emission at long times, given by the asymptote of and reproduced by a separated matrix block in a part of p(t), is also different from the ordered case. In partic- the atomic Hilbert subspace. As a consequence the initial ular, a bump in the emission appears at long times in state, being a superposition of the eigenstates of this ma- Fig. 3(b), that has its counterpart in p(t) in Fig. 3(a), trix block, would further evolve as a wave-packet, with which originates mainly from the atomic mirrors as the periodic swapping of the excitation between the atomic excitation has depleted from the collective emitter (pa(t) segments. During this swapping process each segment is orders of magnitude smaller than p(t) then). The has the ability for cooperative emission into the waveg- long-term asymptote for p(t) is an exponential decay uide. The probability p0(t) should then reveal an oscillat- ext exp( (γ + γ1D)t), slightly faster than exp( γt), ing behaviour, a signature of partial revival to the initial ∼ − − which originates from the Purcell enhancement of the state. Physically it corresponds to a coupled dynam- single-atom emission by the presence of the waveguide. ics between the atomic segments. During an excitation Interestingly, if the collective emitter is placed in the cen- swapping cycle, a significant part of the optical excita- ter of the full array rather than just on one end, the emis- tion leaks into the waveguide. Since the emission into the sion is delayed. Indeed if one end of the array is open for waveguide has a superradiant nature from all segments, the emission from the initial state, the emitted light in the total excitation probability p(t) depletes faster than that direction is directly lost and cannot be re-absorbed. in the preceding examples, shown in Fig. 2 and Fig. 3. We presented our simulations for a particular ran- Figures 4 and 5 present the result of our numerical sim- dom configuration of the atomic mirror, without ensem- ulations, supporting the above arguments for the same ble averaging, so the emission of the sample indicates a other calculation parameters as in Fig. 2. The strong slight signature of its configurational dependence. That difference with the results of Fig. 2 is evident, with the is more visible for the approximately symmetric configu- above mentioned oscillating behavior being observable ration showing slight beatings in its decay dynamics, see in Fig. 4. These oscillations are the direct counterpart 7

wg FIG. 4. Same as in Fig. 2 but for d0 = λ /4. The atomic mirrors are added symmetrically on both sides. FIG. 5. Same as in Fig. 4 but with one single atomic mirror with NR = 200 atoms. Noticeably, both p(t) and p0(t) decay to the behavior observed in Refs. [18, 19] with a col- faster than the bare collective emitter (grey solid line). lective emitter instead of a single atom, here with full inclusion of dipole-dipole coupling and decay through free space. Noticeably the decay process is slowed down is strongly correlated with the radiation at σ− atomic and its dynamics indicates the reviving oscillations for transition, see Fig. 1. Note that p(t) amounts to these left and right emission profiles summed with the emis- p0(t)= pa(t) as well as certain deviations from exponential decay of p(t). sion into free-space. For all configurations the addition When the atomic mirror is added to one side only, as of atomic mirrors atoms can either reduce or direction- shown in Fig. 5, the excitation exits the system faster ally enhance this emission. The differences in the out- than from the bare collective emitter. In the beginning going pulse shape and in their energy balance between stage, approximately one half of the excitation is emitted two propagating directions are quite sensitive to the lo- directly into the waveguide towards the empty direction. cation of the atomic mirrors with respect to the central The other part of the excitation transfers to the atomic segment. For instance in Fig. 3(b), there is a strong im- mirror and is then cooperatively emitted into the waveg- balance of the cooperative emission into the waveguide uide at a higher rate. In such a geometry the reviving towards the empty left side. This arises from the disordered chain acting as a light absorber. Figure 5 shows a oscillations of p0(t) = pa(t) only appear in a long-term asymptote beyond the scale of the graph reproduced in similar directed emission, but its origin is fundamentally the plot. different. Here the excitation migrates between the segments and the directional dependence is associated with the cooperative emission into the waveguide from the two different segments independently, whereas in the case of C. Emission into the waveguide Fig. 3(b) the cooperative emission comes only from the collective emitter. The only exception to the dependence Let us now briefly discuss the spatial profiles of the on the detailed configuration is the fully ordered, trans- light emitted into the waveguide. For all of the con- lationally symmetric atomic chain, shown in Fig. 2, as sidered configurations a significant portion of the emit- discussed above. ted light is injected into the waveguide, since the ini- In some cases, like with the geometry of Fig. 5, the tial state emits (at least partly) cooperatively inside the emission is found to decay faster than when the atomic waveguide. In Figs 2-5, the purple shaded profiles show mirrors are absent. This is a consequence of the collec- the light fractions propagating within the guided modes, tive effects from the full chain, which makes the coop- while the grey-shaded profiles represent the emission into erative emission into the waveguide generally stronger. the waveguide by the collective emitter without atomic The emission profiles thus further highlight the strong mirrors. Emission into both the azimuthal modes is in- dependence of the system dynamics on the specifics of cluded, but in fact the main part of the light wave-packet the atomic arrangement. Such directionality could for in- is transported in the σ = 1 azimuthal mode, which stance be useful for the implementation of efficient atom- − 8 based quantum optical protocols, as an alternative to in- trinsic chiral systems [29, 30].

IV. RESULTS IN THE NON-MARKOVIAN REGIME

In the previous section, the behavior of the system could qualitatively be understood as the interaction of a collective emitter with atomic Bragg mirrors. In particu- wg lar the symmetric situation of Fig. 4, with d0 = λ /4, shares similarities with a classical cavity-QED configuration, with oscillations arising from a “vacuum Rabi splitting”. Yet, the enhancement of photon lifetime, or photon trapping, a salient feature of optical resonators, does not arise for short atomic resonators. The system can be transformed into a conventional optical resonator if the left and right atomic mirrors are separated by a sufficiently long distance L, as shown on Fig. 6. As commented in Appendix D, such a cavity configuration requires that the time L/vg needed by a light pulse with a group velocity vg to cross the cavity is longer than the response time of the Bragg reflection from −1 the atomic mirrors. The latter being ΓM , this condition translates to L/v > Γ−1. This regime of waveguide- FIG. 7. Excitation probabilities in the case of a long res- g M ∼ QED, where the retardation cannot be neglected, is com- onator with d0 L/2 = 20 cm and for NC = 100 atoms monly referred to as non-Markovian [27, 28, 31, 32]. and NL = NR = 500 atoms. Color and line type conven- tions are the same as in Figs. 2-5. (a) and (b) correspond to Simultaneously, the coherence length of a photon emit- wg wg wg d0 = nλ /2 and d0 = nλ /2+ λ /4 (where n is integer ted into the waveguide should be longer than the cavity number) respectively. length L for the wave-packet to be trapped in the cavity volume and its field amplitude to be enhanced with respect to the open waveguide. With the coherence length −1 −1 We will consider the two distinct cases for d0, as we did given by vg/ΓC with ΓC the duration of the emitted in Section III in the case of a short cavity. The first pulse in the open waveguide, this condition can be writ- wg −1 case is d0 = nλ /2 (with n being large positive inte- ten as L/vg < ΓC . These two inequalities ensure the ger number), where the atoms in the collective emitter closest physical conditions to the textbook model dis- are placed at nodes of a mode of the cavity formed by cussed in Appendix D and they are simultaneously ful- the atomic Bragg mirrors. By construction, this cavity filled if Γ Γ , i.e., if the number of atoms is much M ≫ C mode is resonant with the atomic transition. The collec- larger in the atomic mirrors than in the collective emit- tive emitter is then partly decoupled from the waveguide ter. because its constituent atoms are placed at nodes of the In the following, we extend the study presented in electric field of this resonant cavity mode. For the sec- Ref. [19] by considering a collective emitter placed in- wg wg ond case d0 = nλ /2+λ /4, the atoms in the collective side of a long cavity length L made of atomic mirrors. emitter are placed at the anti-nodes of the electric field and therefore maximally couple to this cavity mode. Figure 7 provides the results of our numerical simulations in these two cases, for NC = 100 atoms, and NL = NR = 500 atoms for each atomic Bragg mirror. The atomic mirrors and collective emitter are separated by d0 L/2 = 20 cm, approximately fulfilling the above requirements.∼ In the weak coupling case shown wg in Fig. 7(a) (d0 = nλ /2), both p(t) and p0(t) initially exhibit a fast decay at a rate of order ΓC , which indicates the initially superradiant light emission through the distant chain segments. But the process is tailed FIG. 6. An atomic Bragg cavity system of length L. In con- asymptotically by a much slower and slightly oscillating trast to the previous studied cases where there was no photon decay. The observed weak oscillations appear as a re- lifetime enhancement, the left and right atomic mirrors are sult of ΓC L/vg being finite, which imperfectly fulfills the now separated by a long distance L, as discussed in the text. condition on the emitted wave-packet, with the spectrum 9 distributed within ΓC . As a result, for sufficiently long the emission dynamics to the atomic arrangement. We cavity the atoms can still couple to the guided modes have obtained that the emission process is expected to proximal to the resonant mode but without coupling to critically depend not only on the original localization of the resonant mode itself. In the strong coupling case the optical excitation but also on the entire distribution wg wg shown in Fig. 7(b) (d0 = nλ /2+ λ /4), as expected of the atoms structuring the complete chain. The cal- in cavity-QED as well, the model predicts much stronger culation results have predicted an oscillation behavior of intra-cavity quantum Rabi oscillations. the occupation probability and directional dependence of The salient features of Rabi oscillations or decay at the pulse emitted into the waveguide. Our results with two time scales are therefore qualitatively similar for long a collective emitter are consistent with a mapping to a and short cavities. There are however crucial differences Jaynes-Cummings Hamiltonian, as previously described with the behavior observed in Figs. 2, 4 and 5, related in the literature for a single emitter [18, 19]. to the above-mentioned photon trapping, which justify In the specific long cavity setting, the system truly the full cavity-QED behavior for long resonators. First behaves as a cavity trapping the emitted light, where of all, in Fig. 7 the probabilities p0(t) and p(t) are nearly the edge atoms play the rule of classical Bragg mirrors. identical p (t) p(t). This indicates that the optical 0 ∼ This cavity regime requires specific settings for the sys- excitation is mainly in a polaritonic superposition of the tem parameters. We have pointed out the main criterion superradiant atomic mode and intra-cavity field mode, of it, namely the importance of low inertia in the reflec- and only a small part of the excitation is transferred into tion response of the Bragg-mirrors in comparison with the atomic mirrors. As a consequence the incoherent the retardation associated with the virtual dynamics of losses from spontaneous emission into the external modes the internal interactions. In this case we have observed arise mostly from the collective emitter. In contrast with quantum Rabi oscillations as well as slowing down of the Section III there is no light leakage through the atomic external emission, which are convincing indicators that mirrors into the open waveguide (i.e., emission from the the light is trapped inside the cavity. Compared to previ- atomic mirrors). It is also worth pointing out that the ous works along these lines, these observations confirmed trapping of light within such an atomic Bragg cavity de- that these ideas extend well from fully 1D configurations lays the emission into outer space [19], as demonstrated with only one atom to a collective emitter and in the pres- by the asymptotic exponential decay trends in Fig. 7(b) ence of dipole-dipole coupling and decay via free space. being slower than in the previous examples of short resonators, as well as slower than single atom decay in ex- Such systems have the potential for implementing tool ternal modes γext. of light trapping in one-dimensional cavity-type atomic We have therefore shown that the regime of a classical structures, which would be of interest in the context of cavity, exhibiting extended photon lifetimes, can also be quantum information processing and interfacing. For ex- achieved with a collective emitter provided that the cav- ample the coupling between atomic qubits stored in the ity is long enough. This was shown both in the regime spin degrees of freedom of atomic segments loaded inside of enhanced coupling through the cavity as well as in the such an atomic cavity would be significantly enhanced. regime of suppressing the emission into a selected guided mode.

ACKNOWLEDGMENTS V. CONCLUSION

In this paper we have theoretically studied the pro- We are deeply grateful to Mark Havey for the long cess of cooperative emission of a collective emitter into and fruitful collaborative work during many years. We a nanoscale dielectric waveguide, in the presence of thank him for many inspiring discussions of the problems atomic Bragg mirrors coupled to the same waveguide. considered in this paper. The atomic ensemble can be prepared and controlled This work was supported by the Russian Foundation with state-of-the-art experimental techniques. The ap- for Basic Research under Grants No. 18-02-00265-A and proach is based on a three-dimensional quantum scatter- No. 19-52-15001-CNRS-a, the Russian Science Founda- ing theory adjusted to the specific geometry of a one- tion under Grant No. 18-72-10039, the Foundation for dimensional atomic array positioned along a waveguide, the Advancement of Theoretical Physics and Mathemat- and primarily interacting with the evanescent field of ics BASIS under Grant No. 18-1-1-48-1, the CNRS PRC the guided modes. The relatively simple atomic energy France-Russia project Quantum1D, the French National structure and the analytically-constructed, near-surface Research Agency NanoStrong Project (ANR-18-CE47- asymptote of the electric field Green’s function allowed us 0008), and the Région Ile-de-France (DIM SIRTEQ). to apply the rigorous three-dimensional analysis tracking This project has also received funding from the Euro- all the main features of the internal coupling and config- pean Union’s Horizon 2020 research and innovation pro- uration dependence of the emission process. gramme under grant agreement No. 899275 (DAALI The numerical simulations show a strong sensitivity of project). 10

Appendix A: Computation of the dynamics be expressed by the following time integral

τ/2 The evolution operator U(t, 0) used in (2.1) is given by ˆ (−)† i i ˆ ˆ i ˆ ˆ Ω = ~ dt exp +~H0 t V exp ~Ht + I the following inverse Fourier transform − Z0 − (B1) ∞ dE 1 i with τ + , and where Vˆ = Hˆ Hˆ0 is the inter- → ∞ − U(t, 0) = exp ~Et (A1) action Hamiltonian in the Schrödinger time independent − Z−∞ 2πi E Hˆ + i0 − − representation, and Iˆ is a unit operator. where the integrand is the resolvent operator of the sys- The required matrix element taken between an ar- tem Hamiltonian bitrary final state f , consisting of the outgoing photon, and the initial| statei ψ(0) , consisting of the excited 1 atoms, is given by | i Rˆ(E)= (A2) ˆ E H f Ωˆ (−)† ψ(0) − h | | i acting in the full atoms-field Hilbert space of infinite di- i τ/2 i = dt f Vˆ Pˆ exp E Hˆ t Pˆ ψ(0) mension. The energy argument E can be an arbitrary ~ ~ f − Z0 h | − | i complex number parameter. It is however crucially help- (B2) ful for the entire derivation that the resolvent operator actually contributes in (2.4) only by being projected onto where we have assumed orthogonality of the states the finite subspace of the atomic subsystem f ψ(0) = 0. Here, E is the energy of the final state h | i f assumed as an eigenstate of Hˆ0. We have also multi- ˜ Rˆ(E)= Pˆ Rˆ(E) P.ˆ (A3) plied and substituted Vˆ Vˆ Pˆ, as made possible by the well-fulfilled assumption→ of the RWA approach. In order to evaluate the excitation probability, one thus The presence of the projector Pˆ in the integrand guar- needs to trace the operator product projected onto the antees its depletion with time and provides rapid con- subspace of large but finite dimension. version of the integral itself. In turn, that lets us make further transfer up to the infinite upper limit τ/2 + The resolvent operator (A3) can be evaluated with → ∞ the invariant perturbation theory technique and with as required by the definition of the Møller operator. We the Feynman diagram method. We refer the reader to arrive to the following expression for the matrix element the supplementary materials in [20, 22–25] for a detailed ˆ (−)† ˆ ˆ˜ presentation of the calculation approach. Here we only f Ω ψ(0) = f V R(Ef + i0) ψ(0) (B3) h | | i h | | i outline the most important calculation steps. The resolvent (A3) can be expressed by a many- where the contributing resolvent operator is defined by particle collective Green’s function transporting the op- (A3) and (A2). tical excitation through the atomic chain, see [23]. This many-particle Green’s function fulfills a generalized Appendix C: Interaction Hamiltonian Dyson-type equation presented in an integral form, whose kernel is expressed by the self-energy part, accumulating the complete internal field coupling between any atoms of The atoms-field interaction near a nanostructure is based the chain. In turn, each coupling diagram is given by the on a general paradigm of statistical physics [33], which single-photon propagator, which near a nanoscale struc- states that the photon propagator can be expressed by ture can be approximately built up in analytical form the retarded-type fundamental solution of the macro- [22]. Then, being considered for a collection of atoms scopic Maxwell equation at rrest, the many particle integral Dyson equation can ∂2 be converted to the set of algebraic equations introduced D(R)(r, r′; ω) D(R)(r, r′; ω) △ µν − ∂x ∂x αν directly for the reduced resolvent (A3). Finally the con- µ α ω2 structed system of algebraic equations can be numerically + [1 + 4πχ(r)] D(R)(r, r′; ω)=4π~ δ δ(r r′) . solved for a chain consisting of an arbitrary number of c2 µν µν − atoms. (C1) Here χ(r) is the dielectric susceptibility of the medium (extended by χ(r) = 0 in free space). We use the Carte- ˆ (−)† Appendix B: Møller operator Ω sian coordinates µ,ν,α and assume the sum over the re- peated indices. In the dipole gauge, the quantum de- Let us focus now on the operator Ωˆ (−)†, which is the scription of the interaction process implies a virtual ex- most important for our consideration since it directly change by the excitation between atoms provided by the contributes to the probability amplitude of the process. causal-type Green’s function of the electric field. If this In accordance with its definition (2.5), this operator can function is known, the resolvent operator (A3) can be 11 compiled by a perturbation theory expansion based on (B3) at the point of the dipole position. The second term the second-quantization formalism, see [22–25]. is the dipoles’ self-energy part, which is mainly important For a nanoscale waveguide as considered here, the for renormalization of the self-action divergencies [24, 34, causal-type Green’s function for the evanescent part of 35]. In the dipole gauge, the operator (C3) represents the the electric field (distributed outside the nanofiber) is microscopic displacement field, which coincides with the assumed to be linked with the photon propagator by the electric field at a certain distance from the dipole. The following transformation difference, as well as the contribution of the second term ∞ in (C5), would be important for evaluation of the single (E) ′ iωτ ′ ′ r r r r ′ particle self-energy associated with the interaction of the Dµν ( , ; ω)= i dτ e T Eµ( ,t) Eν ( ,t ) τ=t−t − Z−∞ h i| dipole with its own field near a nanostructure. ω2 = D(R)(r, r′; ω ) (C2) c2 µν | | where we have assumed a lossless transparent dielectric Appendix D: Simplified model of the decay process medium with a dielectric permittivity ǫ(r)=1+4πχ(r). As shown in [22], the above equations can be fulfilled Here we compare our results with an analytically solvable once we re-expand the electric field operator, originally model of a single atom interacting with a cavity mode introduced in the Schrödinger representation, in the basis [36]. The atom, storing a single excitation, will model of the guided and external modes the collective emitter discussed in the main text. The 1/2 atom is described by the following set of Pauli dyad-type Eˆ(r)= (2π~ω ) i b E(s)(r) d†D(s)∗(r) + ... s s − s operators Xs h i (C3) σ = e e g g Here we have selected only the contribution of the guided z | ih | − | ih | modes, specified by the mode index s and frequency ωs, σ+ = e g for the electric field E(s)(r) and the displacement field | ih | σ− = g e (D1) D(s)(r)= ǫ(r) E(s)(r), where ǫ(r) is the spatially depen- | ih | dent dielectric permittivity of the entire medium (free acting in a two-dimensional energy subspace of the space and dielectric waveguide). The ellipsis in Eq. (C3) atomic ground g and excited e states. Let the atom denote the contribution of the external modes. The mode be coupled with| ai single cavity| modei expressed by the functions are normalized as photon annihilation and creation operators c and c† re- ′ d3rǫ(r) E(s )∗(r) E(s)(r) spectively. In turn, the light in the cavity mode can leak Z · through the cavity mirrors and transform into the incom- ′ 3 (s )∗ (s) ing and outgoing field waves associated with the mode d r D (r) E (r)= δs′s (C4) ≡ Z · propagating in free space and expressed by operators a and a†. The displacement field has a transverse profile verifying Then, as shown in textbooks [36, 37], in the Heisen- D(s) r div ( ) = 0, which is an important requirement for berg picture the system obeys the following set of the the transformation (C3). The mode operators bs and † Heisenberg-Langevin equations ds approximately obey the standard bosonic commuta- † tion rules [bs, d ′ ] = δs,s′ [22]. This furthermore implies † s σ˙ z = 2ga c σ− + σ+c the standard approach of the invariant perturbation the- − ory with applicability of the Wick theorem and diagram σ˙ − =+ga σz c † expansion. The supporting arguments that the micro- σ˙ + =+ga c σz κ scopic quantum description stays applicable in such a c˙ = c + g σ √κa (t) slightly modified form follow from our basic assumptions. − 2 a − − 0 Physically, it is a consequence of the fact that the sub- † κ † † c˙ = c + gaσ+ √κa (t) (D2) wavelength nanoscale object can only weakly affect the − 2 − 0 original vacuum structure of the field, such that the field in any waveguide mode mostly exists in its outer evanes- where ga is a coupling constant between the atom and cent part. cavity mode and κ is a leakage rate through the cavity The interaction of the quantized field with the chain is mirrors. All the operators are time dependent. For the taken in the long wavelength dipole approximation Langevin noise terms in the last two equations, we emphasize that the field operators outside the cavity are N considered in the interaction representation by specifi- Vˆ = dˆ(a) Eˆ(r )+ Hˆ . (C5) − · a self cally indicating the time-dependence, as well as adding a Xa=1 0-subscript. The first term accumulates partial interactions for each The equations (D2) can be solved for the first statis- of an a-th atomic dipole d(a) with the electric field Eˆ(r) tical moments and lead to the following dynamics of the 12 upper state population probability p = p(t) array forming the mirrors. If the collective emitter were loaded between such mirrors separated by a short dis- 1+ σz(t) −1 p(t)= h i tance L < vg ΓM , then the virtual interaction processes 2 would expand so fast that the entire system, consisting 2 2exp( κt/2) κ κ 1/2 of the full chain, would dynamically transform to the col- = − 4g2 + 2g2 + κ2 16g2 κ2 16g2 − a 4 − a 4 − a lective quantum state before it would lose its excitation a − 2 and the photon would be emitted into the outer space. 1/2 t κ κ 1/2 exp κ2 16g2 + 2g2 κ2 16g2 Such a configuration cannot be relevantly described by × − a 2 4 − a − 4 − a the above model and can be only numerically simulated 1/2 t exp κ2 16g2 . (D3) the way it is discussed in the main text. × − − a 2 If we assume the alternative configuration where the separation between the mirrors is sufficiently large such In the two opposite limits of either low or high qual- that L > v Γ−1, then in accordance with (D4) we get ity cavities, this respectively results to either enhanced g M spontaneous decay, or in Rabi-type oscillations initiated κ< (1 R)ΓM . (D6) by the excitation swapping between the cavity field and The cooperative decay of the− collective emitter into the atom. The introduced model can be generalized if instead open cavity ΓC is faster than the benchmark given by of a single atom, we assume a collective atomic emitter (D6), following (D5), at least within a representative consisting of N atoms with the renormalized coupling spectral domain of ω ω0 ΓC . Thus we can expect the constant g √Ng = g in solution (D3). − ∼ a → a C behavior of the compound system to be approximated Let us make some useful estimates based on the scaling by the solution (D3) taken above the critical point with of this model with the parameters of the atomic chain and gC > κ/4. That predicts a signature of the intra-cavity of the subwavelength waveguide, which we have discussed Rabi-oscillations, as we have observed in our numerical in the main text. We suppose the cavity as constructed simulations. by such a waveguide and two atomic Bragg mirrors sep- The important consequence of the above result is that arated by a distance L. Then, we can estimate the rate the trend of the dynamics will not be affected by the of the cavity losses as absolute value of the coupling constant gC as long as the inequality holds. This means that the observation of Rabi κ (1 R)v /L (D4) ∼ − g oscillations will not depend on the external parameters such as the distance of the atoms from the surface, or the where vg is the group velocity of the guided wave, and R is the reflection coefficient of the mirrors, which is ex- number of atoms in the active chain. However, as soon pected to be close to 100%. For the considered atomic as the reverse inequality gC < κ/4 is fulfilled, it critically Bragg mirrors the strong reflection can be actually pro- change the structure of the solution (D3), and the decay vided only in a quite narrow spectral domain near the rate has the same dependence on gC as the open system. For the parameters used in our paper, which we try to point of atomic resonance ω0 where we approximately get keep close to experimental conditions, the critical point g κ/4 would be difficult to reach since (D3) and (D6) C ∼ i 2 leads to cavity losses comparable with or even lower than ΓM the natural decay for a single atomic emitter. In this −2 R(ω) (D5) limit, reducing the number of active atoms down to a ∼ i (ω ω0)+ ΓM few atoms would lead to the loss of the main advantage − 2 of cooperative enhancement, which is crucially required where ΓM is the cooperative decay rate for the atomic to provide the preferable coupling with the guided mode.

[1] D. E. Chang, J. S. Douglas, A. González-Tudela, C.-L. and J. Laurat, Demonstration of a memory Hung, and H. J. Kimble, Colloquium: Quantum mat- for tightly guided light in an optical nanofiber, ter built from nanoscopic lattices of atoms and photons, Phys. Rev. Lett. 114, 180503 (2015). Rev. Mod. Phys. 90, 031002 (2018). [5] C. Sayrin, C. Clausen, B. Albrecht, P. Schneeweiss, [2] T. Nieddu, V. Gokhroo, and S. Nic Chormaic, Optical and A. Rauschenbeutel, Storage of fiber-guided light nanofibres and neutral atoms, J. Opt. 18, 053001 (2016). in a nanofiber-trapped ensemble of cold atoms, [3] P. Solano, J. A. Grover, J. E. Hoffman, S. Ravets, Optica 2, 353 (2015). F. K. Fatemi, L. A. Orozco, and S. L. Rolston, Opti- [6] P. Solano, P. Barberis-Blostein, F. K. Fatemi, L. A. cal Nanofibers: A New Platform for Quantum Optics, Orozco, and S. L. Rolston, Super-radiance reveals in Advances in Atomic, Molecular and Optical Physics, infinite-range dipole interactions through a nanofiber, Vol. 66 (Academic Press Inc., 2017) pp. 439–505. Nature Communications 8, 1857 (2017). [4] B. Gouraud, D. Maxein, A. Nicolas, O. Morin, [7] A. S. Prasad, J. Hinney, S. Mahmoodian, K. Hammerer, 13

S. Rind, P. Schneeweiss, A. S. Sørensen, J. Volz, and nanofiber, Phys. Rev. A 90, 23805 (2014). A. Rauschenbeutel, Correlating photons using the collec- [22] V. A. Pivovarov, A. S. Sheremet, L. V. Gerasimov, V. M. tive nonlinear response of atoms weakly coupled to an Porozova, N. V. Corzo, J. Laurat, and D. V. Kupriyanov, optical mode, Nature Photonics 14, 719 (2020). Light scattering from an atomic array trapped near a [8] N. V. Corzo, J. Raskop, A. Chandra, A. S. one-dimensional nanoscale waveguide: A microscopic ap- Sheremet, B. Gouraud, and J. Laurat, Waveguide- proach, Phys. Rev. A 97, 023827 (2018). coupled single collective excitation of atomic arrays, [23] A. S. Sheremet, A. D. Manukhova, N. V. Larionov, and Nature 566, 359 (2019). D. V. Kupriyanov, Cooperative light scattering on an [9] A. Goban, C. L. Hung, S. P. Yu, J. D. Hood, J. A. atomic system with degenerate structure of the ground Muniz, J. H. Lee, M. J. Martin, A. C. McClung, state, Phys. Rev. A 86, 043414 (2012). K. S. Choi, D. E. Chang, O. Painter, and H. J. [24] D. V. Kupriyanov, I. M. Sokolov, and M. D. Kimble, Atom–light interactions in photonic crystals, Havey, Mesoscopic coherence in light scattering from Nature Communications 5, 3808 (2014). cold, optically dense and disordered atomic systems, [10] J. S. Douglas, T. Caneva, and D. E. Chang, Photon Physics Reports 671, 1 (2017). molecules in atomic gases trapped near photonic crystal [25] A. S. Sheremet, D. F. Kornovan, L. V. Gerasimov, waveguides, Phys. Rev. X 6, 031017 (2016). B. Gouraud, J. Laurat, and D. V. Kupriyanov, Coher- [11] J.-B. Béguin, A. P. Burgers, X. Luan, Z. Qin, S. P. Yu, ent control of light transport in a dense and disordered and H. J. Kimble, Advanced apparatus for the integration atomic ensemble, Phys. Rev. A 91, 053813 (2015). of nanophotonics and cold atoms, Optica 7, 1 (2020). [26] M. Goldberger and K. Watson, Collision Theory, Wiley, [12] I. H. Deutsch, R. J. Spreeuw, S. L. Rolston, and New York-London-Sydney, 1964. 919 p. (1964). W. D. Phillips, Photonic band gaps in optical lattices, [27] H. Zheng and H. U. Baranger, Persistent Phys. Rev. A 52, 1394 (1995). quantum beats and long-distance entangle- [13] N. V. Corzo, B. Gouraud, A. Chandra, A. Goban, ment from waveguide-mediated interactions, A. S. Sheremet, D. V. Kupriyanov, and J. Lau- Phys. Rev. Lett. 110, 113601 (2013). rat, Large bragg reflection from one-dimensional [28] T. Shi, D. E. Chang, and J. I. Cirac, Multiphoton- chains of trapped atoms near a nanoscale waveguide, scattering theory and generalized master equations, Phys. Rev. Lett. 117, 133603 (2016). Phys. Rev. A 92, 053834 (2015). [14] H. L. Sørensen, J.-B. Béguin, K. W. Kluge, I. Iakoupov, [29] P. Lodahl, S. Mahmoodian, S. Stobbe, A. Rauschenbeu- A. S. Sørensen, J. H. Müller, E. S. Polzik, and J. Ap- tel, P. Schneeweiss, J. Volz, H. Pichler, and P. Zoller, pel, Coherent backscattering of light off one-dimensional Chiral quantum optics, Nature 541, 473 (2017). atomic strings, Phys. Rev. Lett. 117, 133604 (2016). [30] R. Jones, G. Buonaiuto, B. Lang, I. Lesanovsky, [15] A. Albrecht, T. Caneva, and D. E. Chang, Chang- and B. Olmos, Collectively Enhanced Chiral Photon ing optical band structure with single photons, Emission from an Atomic Array near a Nanofiber, New Journal of Physics 19, 115002 (2017). Physical Review Letters 124, 093601 (2020). [16] D. F. Kornovan, N. V. Corzo, J. Laurat, and [31] H. Pichler and P. Zoller, Photonic Circuits A. S. Sheremet, Extremely subradiant states with Time Delays and Quantum Feedback, in a periodic one-dimensional atomic array, Physical Review Letters 116, 093601 (2016). Phys. Rev. A 100, 063832 (2019). [32] K. Sinha, P. Meystre, E. A. Goldschmidt, F. K. Fatemi, [17] A. Asenjo-Garcia, M. Moreno-Cardoner, A. Albrecht, S. L. Rolston, and P. Solano, Non-Markovian Collec- H. J. Kimble, and D. E. Chang, Exponential im- tive Emission from Macroscopically Separated Emitters, provement in photon storage fidelities using subra- Physical Review Letters 124, 043603 (2020). diance and “selective radiance” in atomic arrays, [33] L. Landau, E. Lifshitz, and L. Pitaevskii, Phys. Rev. X 7, 031024 (2017). Course of Theoretical Physics: Statistical Physics, Part 2 : by E.M. Lifshitz and L.P. Pitaevskii, [18] D. E. Chang, L. Jiang, A. V. Gorshkov, and v. 9 (1980). H. J. Kimble, Cavity QED with atomic mirrors, [34] C. Cohen-Tannoudji, J. Dupont-Roc, and G. Grynberg, New Journal of Physics 14, 063003 (2012). Atom-Photon Interactions, Wiley & Sons, Inc., New [19] P.-O. Guimond, A. Roulet, H. N. Le, and York-Chichester-Brisbane-Toronto-Singapore, 1993. 656 V. Scarani, Rabi oscillation in a quantum cav- p. (1993). ity: Markovian and non-markovian dynamics, [35] I. M. Sokolov, M. D. Kupriyanova, D. V. Kupriyanov, Phys. Rev. A 93, 023808 (2016). and M. D. Havey, Light scattering from a dense and ul- [20] V. A. Pivovarov, A. S. Sheremet, L. V. Gerasimov, tracold atomic gas, Phys. Rev. A 79, 053405 (2009). J. Laurat, and D. V. Kupriyanov, Quantum interface [36] M. Scully and M. Zubairy, Quantum Optics, Cambrdge between light and a one-dimensional atomic system, University Press, 1997. 630 p. (1997). Phys. Rev. A 101, 053858 (2020). [37] C. Gardiner, Quantum Noise, Springer-Verlag, 1991. 364 [21] F. Le Kien and A. Rauschenbeutel, Anisotropy in scat- p. (1991). tering of light from an atom into the guided modes of a