Giant Atom: Distributed Quantum Emitter

• Giant atom is an artificial quantum emitter coupling to a bosonic bath at multiple spatial points, leading to phase-dependent interference effects.
• Its distributed coupling induces frequency-dependent decay rates, Lamb shifts, and non-Markovian dynamics that allow precise spectral engineering.
• Giant atoms enable decoherence-free interactions and versatile quantum routing in structured environments, promising robust quantum transduction.

A giant atom is an artificial quantum emitter whose coupling to a propagating bosonic bath is distributed over several spatially separated points rather than concentrated at one location. The defining regime is that the coupling-point separations are comparable to, or larger than, the wavelength of the relevant field mode, so the usual point-like or electric-dipole approximation ceases to be adequate. In this regime, the field accumulates nontrivial phases between coupling points, and light–matter or sound–matter interaction becomes intrinsically interferometric: decay rates, Lamb shifts, scattering amplitudes, retardation effects, bound states, dark states, and coherent exchange interactions all become geometry- and frequency-dependent in ways unavailable to ordinary small atoms (Kockum, 2019, He et al., 11 May 2026).

1. Definition and conceptual scope

In ordinary waveguide QED, a small atom couples to the field at a single effective point, so the interaction is local and typically treated as frequency independent over the relevant band. A giant atom instead couples at multiple discrete points $x_k$, and the relevant wavelength is

$\lambda=\frac{2\pi v}{\omega_{\rm a}},$

with $v$ the propagation velocity and $\omega_{\rm a}$ the atomic transition frequency. Once coupling-point separations are a fraction or multiple of $\lambda$, the emitted and absorbed waves acquire relative phases and interfere (Kockum, 2019).

A standard description introduces an effective coupling amplitude as a coherent sum over coupling points,

$A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$

or, in a two-point setting, an interaction factor of the form $g(1+e^{ikd})$. This is the elementary mathematical signature of a giant atom: the bath is sampled nonlocally, and the interaction strength becomes a discrete Fourier transform of the coupling geometry rather than a local constant (Kockum, 2019, Chen et al., 2023).

The term does not imply a literally large atom. In superconducting-circuit realizations, the hardware element may remain a transmon-scale artificial atom, but its effective light–matter interaction geometry is giant because it is connected to the same waveguide at multiple well-separated locations. The 2025 hybrid superconducting–phononic experiment makes this explicit: the “giant atom” is a frequency-tunable transmon qubit coupled to a thin-film lithium niobate phononic waveguide through two interdigital transducers separated by $L=470\,\mu\mathrm{m}$, corresponding to roughly $600$ acoustic wavelengths at $5\,\mathrm{GHz}$ (Xiao et al., 18 Dec 2025, Kannan et al., 2019).

2. Interference, spectral engineering, and the breakdown of locality

The central physical effect of giant-atom coupling is interference between emission and absorption pathways originating at different coupling points. Under the Markov approximation, a symmetric giant atom with $\lambda=\frac{2\pi v}{\omega_{\rm a}},$0 equally spaced, equally strong coupling points has a decay rate

$\lambda=\frac{2\pi v}{\omega_{\rm a}},$1

and a Lamb shift

$\lambda=\frac{2\pi v}{\omega_{\rm a}},$2

where $\lambda=\frac{2\pi v}{\omega_{\rm a}},$3 is the phase accumulated between neighboring coupling points. These expressions show directly that both dissipation and level renormalization become frequency dependent through geometry-controlled interference (Kockum, 2019).

The 2025 phononic giant atom provides a concrete realization of this principle. There, the effective decay rate is modeled as

$\lambda=\frac{2\pi v}{\omega_{\rm a}},$4

with $\lambda=\frac{2\pi v}{\omega_{\rm a}},$5 the intrinsic decay rate, $\lambda=\frac{2\pi v}{\omega_{\rm a}},$6 the IDT-induced radiative coupling, $\lambda=\frac{2\pi v}{\omega_{\rm a}},$7 the amplitude transmittance, and $\lambda=\frac{2\pi v}{\omega_{\rm a}},$8. The delay produces an oscillation period of about $\lambda=\frac{2\pi v}{\omega_{\rm a}},$9 in qubit frequency; over only a $v$0 tuning range, the effective decay rate changes by about four-fold. The extracted fit parameters are $v$1 and $v$2, and the comparison between strongly and weakly coupled operating points yields a Purcell factor exceeding $v$3 (Xiao et al., 18 Dec 2025).

This spectral interference can be used as a design resource. Giant atoms can be nearly decoupled from a waveguide at one frequency, strongly coupled at another, or engineered so that different internal transitions experience sharply different linewidths. In multilevel systems this supports selective relaxation engineering, population inversion, electromagnetically induced transparency, and dissipation-based state preparation (Kockum, 2019).

The same nonlocality also marks the breakdown of the electric-dipole approximation. A modified input-output formulation for giant-atom scattering introduces an additional quasi-direct channel between left- and right-moving modes, represented phenomenologically by a low-$v$4 background path. In this description the scattering amplitude becomes a resonant term plus a background term, naturally producing Fano-type rather than purely Lorentzian spectra. This formalism is intended precisely for the regime in which a giant atom cannot be reduced to a conventional point scatterer (He et al., 11 May 2026).

3. Delay, non-Markovianity, and bound states

When propagation time between coupling points is not negligible compared with the radiative lifetime, giant-atom dynamics become non-Markovian. The canonical crossover is

$v$5

with $v$6 the travel time between coupling points. In this regime the emitter can reabsorb its own earlier emission, the present dynamics depend on the past, decay ceases to be purely exponential, and revivals or algebraically slow long-time behavior can emerge (Kockum, 2019).

The 2025 phononic device is a direct delayed-feedback realization. For $v$7, the transmon relaxes approximately exponentially because phonons emitted at one IDT have not yet reached the other. For $v$8, earlier emitted phonons return and interfere with ongoing emission. The excited-state probability is described as

$v$9

which can be interpreted as a sum over successive backflow events. The contribution from the $\omega_{\rm a}$0-th backflow is suppressed by approximately $\omega_{\rm a}$1, and in the reported device $\omega_{\rm a}$2, so the delayed terms are not negligible (Xiao et al., 18 Dec 2025).

More generally, giant atoms support bound or quasi-bound states generated by destructive interference. The survey literature notes that with three or more coupling points the atom can support oscillating bound states in which energy remains trapped between the outer coupling points. In surface-acoustic-wave models with an extended IDT, sufficiently large spatial extent can suppress relaxation and generate effective vacuum Rabi oscillation with a localized phononic wave packet even in the absence of a cavity; the corresponding toy models exhibit BIC-like purely imaginary poles in the single-excitation response (Kockum, 2019, Guo et al., 2019).

External control can reshape these memory effects. For a two-point giant atom with dynamically modulated transition frequency, the spontaneous-emission amplitude obeys a delay-differential equation in which the retarded term carries a time-dependent dynamical phase $\omega_{\rm a}$3. Fast modulation can effectively average out the delayed feedback, causing the giant atom to behave like a small one, while an additional phase difference between the two coupling paths can generate chiral and tunable temporal profiles of the output fields (Du et al., 2022).

4. Multi-atom geometries, decoherence-free interaction, and dark states

For more than one giant atom, geometry becomes a control parameter for separating coherent exchange from dissipation. The standard two-atom topologies are separate, nested, and braided. In a general Markovian master equation, the coherent exchange $\omega_{\rm a}$4, individual decay rates $\omega_{\rm a}$5, and collective decay $\omega_{\rm a}$6 are all sums of phase-dependent interference terms; cosine terms govern dissipative channels, while sine terms govern coherent exchange (Kockum et al., 2017, Kannan et al., 2019).

The central result is the decoherence-free interaction (DFI) of braided giant atoms. In that configuration one can have

$\omega_{\rm a}$7

This is not a decoherence-free subspace in the usual sense, but geometry-induced cancellation of the waveguide decay channel for the entire multi-atom Hilbert space. The same interference logic extends to larger architectures, including protected nearest-neighbor chains and networks with all-to-all connectivity (Kockum et al., 2017).

Experimentally, superconducting coplanar-waveguide giant atoms have confirmed this separation of loss and interaction. In the two-point braided device, the radiative decay rate is

$\omega_{\rm a}$8

while the exchange interaction is

$\omega_{\rm a}$9

At the decoherence-free point $\lambda$0, $\lambda$1 but $\lambda$2. The experiment reports $\lambda$3, $\lambda$4, and an exchange rate $\lambda$5. A three-point geometry yields two decoherence-free frequencies, $\lambda$6 and $\lambda$7, and supports preparation of $\lambda$8 with a state-preparation fidelity of $\lambda$9 (Kannan et al., 2019).

Chiral waveguides do not eliminate these phenomena; they modify them. In chiral architectures, braided giant atoms still support DFI, while nested giant atoms can exhibit dark states even without coherent driving and regardless of chirality. This contrasts sharply with small atoms, for which nontrivial undriven dark states generally require bidirectional symmetry (Soro et al., 2021).

Structured environments add a caveat. In a one-dimensional tight-binding bath with finite band and band gap, braided giant atoms can realize DFI inside the band, but it is degraded by time delay, non-Markovian decay, and band-edge effects. Outside the band, by contrast, interaction proceeds through overlap of bound states and is fully decoherence-free in the ideal model, with a purely real pole structure (Soro et al., 2022).

5. Structured baths, synthetic dimensions, and higher-dimensional generalizations

A major generalization of giant-atom physics replaces the featureless waveguide by a structured photonic bath. In the most general formulation, a giant atom coupled to several bath sites $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$0 with strengths $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$1 can be represented as a normal atom coupled to a single fictitious site state

$A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$2

This yields a compact Green’s-function theory of atom–bath interaction and a general criterion for decoherence-free Hamiltonians: such a Hamiltonian arises if and only if each giant atom seeds a weak-coupling bound state. The criterion applies inside and outside the continuum and does not depend on the dimensionality or microscopic structure of the bath (Leonforte et al., 2024).

Several structured settings illustrate how multipoint coupling reorganizes the bath itself:

Setting	Interference control	Representative result
1D finite-band structured waveguide	Detuning and topology	DFI inside the band; bound-state-mediated interaction outside the band (Soro et al., 2022)
Cross-stitch ladder lattice	Relative phase $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$3 between two coupling points	$A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$4 selects the dispersive band; $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$5 selects the flat band (Xia et al., 13 Jan 2025)
Synthetic frequency dimension	External drive phase $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$6	Chiral interactions, cascaded transfer, and directional excitation flow in a frequency lattice (Du et al., 2021)
2D structured lattices	Coupling-point geometry	Interfering BICs for a single giant atom and oscillating BICs between many giant atoms (Ingelsten et al., 2024)
Sawtooth lattice with synthetic flux	Broken time-reversal symmetry plus two-point coupling	Chiral spontaneous emission and nonreciprocal delayed light unattainable with a small atom (Du et al., 2022)

The cross-stitch ladder is a particularly transparent example. A small atom coupled locally to one site necessarily addresses both the flat and dispersive bands. A two-point giant atom with relative phase $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$7 has an effective interaction

$A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$8

so $A_m(\omega_j)=\sum_k g_{jkm} e^{i\omega_j x_k/v},$9 cancels flat-band coupling and $g(1+e^{ikd})$0 cancels dispersive-band coupling. In the bandgap regime, this selectivity suppresses leakage and yields cleaner long-range dipole–dipole exchange through bound-state overlap than a comparable small-atom configuration (Xia et al., 13 Jan 2025).

Synthetic dimensions show that literal spatial extent is not essential. In a frequency-lattice implementation based on a dynamically modulated superconducting resonator and a $g(1+e^{ikd})$1-type three-level artificial atom, the effective emitter couples to two distant lattice sites $g(1+e^{ikd})$2 and $g(1+e^{ikd})$3. The external phase $g(1+e^{ikd})$4 then acts as a synthetic gauge flux, breaks momentum-space symmetry, and enables chiral coupling, cascaded interaction, and directional excitation transfer in the frequency domain (Du et al., 2021).

Two-dimensional environments further broaden the concept. Giant atoms in square lattices, photonic graphene, and related baths can support bound states in the continuum and decoherence-free Hamiltonians unavailable to local emitters, while numerical studies of 2D coupled-cavity lattices show that braided geometries are especially favorable for robust decoherence protection (Leonforte et al., 2024, Ingelsten et al., 2024).

6. Experimental platforms, scattering formalisms, applications, and limitations

The first experimental giant atoms were implemented in surface-acoustic-wave platforms, where a superconducting transmon’s interdigitated structure itself provides many wavelength-scale coupling points. Later SAW experiments accessed distances exceeding $g(1+e^{ikd})$5 wavelengths with $g(1+e^{ikd})$6, clearly in the non-Markovian regime, and also demonstrated EIT for a propagating mechanical mode. Microwave transmission-line implementations subsequently established frequency-dependent decay and Lamb shifts, interference-controlled suppression and enhancement of decay, and coherent multi-qubit exchange in braided geometries (Kockum, 2019, Kannan et al., 2019).

The hybrid superconducting–phononic integrated-circuit realization extends this program to guided acoustic phonons in lithium niobate. Its central achievement is simultaneous access to a large physical delay ($g(1+e^{ikd})$7), pronounced frequency-dependent dissipation, non-Markovian phonon backflow, and dissipation-assisted state engineering. Under coherent drive, the steady-state excited-state probability is written as

$g(1+e^{ikd})$8

so the same interference-controlled linewidth that governs spontaneous emission can be used as a bath-engineering resource. In the reported device, the dressed states can be made to experience different decay rates, enabling steady-state purity near $g(1+e^{ikd})$9 even for strong drive in some regimes, with purity oscillations versus $L=470\,\mu\mathrm{m}$0 of period $L=470\,\mu\mathrm{m}$1 (Xiao et al., 18 Dec 2025).

Scattering theory has likewise evolved beyond the multiple-coupling-point picture alone. The modified input-output approach for giant-atom scattering beyond the dipole approximation introduces a quasi-direct background channel, or equivalently an effective low-$L=470\,\mu\mathrm{m}$2 cavity channel, and derives a central reflection amplitude

$L=470\,\mu\mathrm{m}$3

This form captures observed non-Lorentzian Fano spectra and permits extraction of intrinsic dissipation and coupling parameters more accurately than a pure Lorentzian or standard multiple-coupling-point fit (He et al., 11 May 2026).

Applications span both fundamental and device-oriented directions. Giant $L=470\,\mu\mathrm{m}$4-type atoms support phase-dependent elastic and inelastic single-photon scattering, with an optimal frequency-conversion condition shifted by interference phases; with added Sagnac interferometers, unit-efficiency single-photon frequency conversion becomes possible (Du et al., 2021). Dual-rail giant-atom nodes have been proposed as four-port quantum routers capable of targeted routing, non-reciprocal scattering, path-encoded gates, teleportation protocols, and circulator behavior (Gong et al., 2024). Extended input-output calculations for weak coherent pulses show that the phase between coupling points can switch output photon statistics among bunching, antibunching, and coherent regimes, each over a finite phase bandwidth (Yue et al., 29 Mar 2026).

A recurrent misconception is that giant atoms are simply superior versions of small atoms. The structured-bath literature does not support that conclusion in general. In one-dimensional structured environments, DFI inside the band is degraded by time delay and band-edge effects, and small atoms can be nearly comparable in some regimes if allowed larger single-point coupling. Likewise, beyond-dipole scattering theory indicates that the standard multiple-coupling-point model may be incomplete when quasi-direct channels contribute appreciably. The practical behavior of a giant atom therefore depends on geometry, phase, delay, band structure, and experimental coupling constraints rather than on size alone (Soro et al., 2022, He et al., 11 May 2026).

Taken together, these results define giant atoms as a distinct regime of quantum optics and quantum acoustodynamics in which geometry becomes an active dynamical variable. Multipoint coupling turns phase accumulation, retardation, and bath structure into design knobs for dissipation control, bound-state formation, non-Markovian dynamics, decoherence-free interaction, and quantum transduction across microwave, phononic, photonic, synthetic, and higher-dimensional platforms.