betaFeedback

WorksPhysics and Astronomy

Finite-temperature Quantum Quasinormal Mode Theory for Spatiotemporally-resolved Dissipative Cavity Quantum Magnonics

Abstract

Cavity quantum magnonics has ushered in an unprecedented era of hybrid light-matter interfaces in which microwave photons, collective spin excitations, and superconducting or spin qubits can be coupled within a single device architecture, facilitating a versatile platform for quantum information processing. Realizing their full potential, however, requires a theory that can treat non-reciprocal magnetic media, material dissipation and open-cavity radiation losses, as well as finite-temperature fluctuations rigorously and without significant approximation. The dissipative magnonic resonances of such systems are described as quasinormal modes (QNMs), yet a consistent quantization procedure for lossy gyrotropic resonators has remained elusive. In this work, we present a finite-temperature quantum QNM theory, based on macroscopic quantum electrodynamics for non-reciprocal media and complex coordinate transformations, that achieves QNM quantization for a wide class of spatially inhomogeneous, dissipative and dispersive, cavity magnonic systems. We demonstrate the predictive capabilities of our theory via applications to hybrid dielectric-magnonic resonators interacting with both spin and superconducting qubits, evidencing modal interference effects, a continuous transition from coherent to dissipative coupling, spatiotemporal quantum field dynamics, and magnon-mediated qubit synchronization, at both zero and finite temperatures. By addressing the intricacies of modal loss and non-reciprocity in a fully quantum theory of cavity magnonics, our approach enables the quantitative design of hybrid quantum devices that operate under realistic open-system and thermal conditions, without recourse to ad hoc quantization schemes.

Introduction

Magnonics has emerged as a promising platform for the coherent manipulation and transport of information by means of spin waves, the collective excitations of ordered magnetic systems. Owing to their wave nature, magnons enable low-dissipation information processing and afford access to a rich variety of non-reciprocal[1, 2], non-linear[3], non-hermitian[4], and topological[5, 6] phenomena. Advances in materials engineering and nanofabrication have substantially extended the capabilities of magnonic devices, motivating their exploration for both classical and quantum technologies, ranging from compact resonators to bulk and thin-film systems for propagation. Among available material platforms, yttrium iron garnet (Y3Fe5O12\mathrm{Y}_{3}\mathrm{Fe}_{5}\mathrm{O}_{12}, YIG) occupies a position of particular importance by virtue of its exceptionally low magnetic loss, long coherence times, high spin density that permits strong coupling to radio-frequency (RF) and microwave (MW) fields, and readily accessible nonlinear regime[7]. These properties have already established YIG as a standard medium in commercial microwave systems, including filters and phase shifters[8]. At the quantum level, magnons are the quanta of collective spin waves and are described as bosonic quasiparticles. Their low energy consumption, on-chip integrability, nanoscale wavelengths, frequency tunability under an applied bias field, and natural interface with MW circuitry and other classes of quantum excitation render them a promising resource for classical and quantum information processing and for integration into quantum computing architectures[9, 10, 11, 12].

A particularly rich vein of research is cavity magnonics, which addresses the interaction between magnons and confined MW photons. Hybrid cavity-magnonic systems have demonstrated both strong[13] and ultrastrong light-matter coupling[14, 15], thereby enabling the coherent exchange of excitations and the formation of magnon-polaritons. Such systems offer substantial potential for quantum memories[16], optical-to-MW transduction via magneto-optical effects (cavity optomagnonics)[17, 18, 19], magnon-photon-phonon coupling mediated by magnetostrictive forces (cavity magnomechanics)[20, 21], the engineering of synthetic gauge fields[22], and even more exotic applications such as axionic dark matter detectors[23, 24]. At the quantum level, optical-to-MW transducers are expected to play an increasingly important role in quantum networks, since the majority of leading quantum-computing platforms operate at MW frequencies, whereas optical photons propagating in fiber are much better suited to long-distance quantum communication. In addition to the coherent interaction, dissipative magnon-photon[25] and magnon-magnon[26] coupling, arising from the cavity Lenz effect[27] and from coupling to a common radiative reservoir[28] can be observed, leading to level attraction and possibly exceptional points[29, 30] (coalescing eigenmodes) rather than the level repulsion characteristic of coherent coupling; the relative contributions of coherent and dissipative coupling can even be tuned[27, 31]. An instrumental route towards quantum applications is the interfacing of cavity magnonic systems with quantum bits (qubits) required for quantum information processing and computing. The arguably most promising qubit platforms, namely spin qubits (SQs) and ensembles thereof[32] (e.g., nitrogen-vacancy centers in diamond[33], rare-earth ion-doped crystals[34], defects in silicon carbide[35]) and superconducting (SC) qubits[36] (e.g., flux qubits[37] and transmon qubits[38]), have already been interfaced with (cavity) magnonic systems in a variety of configurations. Direct coupling of SQs to magnonic modes has been investigated both theoretically and experimentally, demonstrating coherent strong coupling for ferromagnets and ferrimagnets[39, 40], as well as for van der Waals materials[41], long-distance coupling between SQs mediated by a ferromagnetic coupler[42], simultaneous magnon-photon lasing[43], and single-spin magnetometry at room temperature[44].

Experimentally, SC qubits have so far been coupled to magnons only indirectly, via a MW cavity mode[45, 46]. This arrangement has enabled strong coherent magnon–qubit coupling, the generation of non-classical magnon states[47], dispersive coupling and single-quantum magnon sensing[48, 49, 50], as well as both Kerr and cross-Kerr non-linearities[50, 51, 52]. Direct coupling of SC qubits to magnonic modes through the stray field has not yet been realized. Recent theoretical studies have nevertheless examined the direct coupling of transmon qubits to YIG microresonators, establishing the feasibility of generating non-classical magnonic states[53, 54, 56] [55] and of implementing quantum gates[57]. To the best of our knowledge, no experiment has demonstrated the coupling of a flux qubit (FQ) to a magnonic mode. Theoretical work has, however, explored coherent transfer between a FQ and a SQ via a YIG nanoresonator to which both are directly coupled[58] (with the FQ placed outside the magnetic bias field), as well as the interaction of a FQ with a small, in-plane-magnetized YIG microsphere situated inside the FQ loop[59, 60].

Recent experimental advances further motivate the investigation of directly coupled magnon–qubit systems. FQs that operate at the flux-insensitive sweet spot with zero external bias flux have been demonstrated[61], thereby simplifying on-chip integration and the interfacing with resonators by eliminating the need for an additional flux-bias line. In parallel, a fluxonium qubit capable of operation in the presence of in-plane magnetic fields in the tesla range has been realized[62], an important step toward the experimental feasibility of direct FQ–magnon coupling. It is also possible to interface SC qubits and SQs (or SQ ensembles) directly so as to enable coherent exchange, as has been shown for both ensembles[63, 64, 65] and single spins[66]; such interfaces offer a potential quantum bus for coupling SQs to other systems, including magnonic resonators. Hybrid SQ–SC-qubit architectures could therefore become valuable components of quantum processors, combining the long coherence times of SQs for storage with the stronger couplings characteristic of SC qubits for computation.

Given the diverse phenomenology and underlying interaction mechanisms in cavity quantum magnonics and hybrid architectures based upon it, an urgent demand for a rigorous and unified methodology has emerged, which would allow for a quantitative, spatiotemporally-resolved description of the quantum electromagnetic field dynamics and its interplay with various qubit realizations. Classically, the magnonic modes of ferro- and ferrimagnetic resonators were first described via a magnetostatic approach, namely the coupled Maxwell and Landau–Lifshitz (LL) equations under the approximation of negligible exchange interaction, Gilbert damping, propagation and radiation losses, giving rise to the well-known Walker modes with real frequencies[67, 68, 69, 70]. The simplest such mode, in which all involved spins precess uniformly, is the Kittel mode and is typically observed in ferromagnetic resonance experiments[71], whereas higher-order non-uniform modes can be observed in inhomogeneous bias fields[14, 72]. Still, this approach does not take into account the influence of a magnetic body’s dielectric environment, as needed for the description of magnon-photon hybrid systems. A more general classical approach was used in Ref. [73], in which the electromagnetic scattering on cavity magnonic system was studied, based on generalized Mie theory[74, 75]. In this case, the Landau–Lifshitz–Gilbert (LLG) equation was solved, in which the exchange interaction was still neglected (valid for resonator sizes much larger than the exchange length) but where Gilbert damping was taken into account. Since the full Maxwell’s equations were treated, propagation effects were also included. However, studying scattering alone may not be sufficient to gain a comprehensive understanding of cavity magnonic systems, since, for example, spectral splitting is only a necessary but not a sufficient condition for strong coherent coupling. A similar approach was used in Refs.[76, 77, 78], where the complex quasinormal modes (QNMs) for a bare and a shielded YIG sphere were calculated using an effective scalar permeability. In Ref. [79], the magnon-photon coupling of a YIG sphere within a closed MW resonator was investigated using a perturbative approach to calculate the modified frequencies, assuming that the YIG sphere was significantly smaller than the photonic resonator. The quantum mechanical treatment usually starts from the full Hamiltonian of an interacting spin system, which can be reduced to a Hamiltonian of bosonic modes to describe collective excitations via the Holstein–Primakoff transformation[80]. In principle however, this transformation can only be used for infinite spin systems to derive a Hamiltonian and dispersion for the continuous-wave magnon modes. For compact resonators, the discrete modal frequencies are usually obtained from different approaches, like the magnetostatic theory, and are then used to construct a Hamiltonian with discrete bosonic modes[46, 81], with the losses added phenomenologically. The coupling to a photonic resonator mode is quantified via either an overlap integral of the magnetization arising from the magnonic mode and the magnetic field of the photonic mode[82], or an experimentally determined parameter[81]. In such cases though, a systematic, a priori prediction of the ratio between coherent and dissipative coupling is not possible.

A natural framework in which to describe magnonic modes and their coupling with the modes of photonic structures and qubits from first principles is the QNM theory[83, 84, 85, 86, 87, 88, 89, 90, 91]. Therein, the electromagnetic resonances of an open system correspond to the time-harmonic solutions of Maxwell’s equations subject to suitable outgoing-wave conditions, being characterized by complex frequencies. The non-conservative nature of lossy cavity systems has raised a variety of non-trivial questions regarding a proper, mathematically rigorous formulation of normalization and orthogonality, alongside the issue of whether or not the QNMs truly form a complete set, and thus allow mode expansions for arbitrary electromagnetic field distributions. In spite of the challenges and ambiguities, classical QNM methodology has undergone significant advances in the last decade, including improved understanding of the formal aspects of the theory[91], techniques for computing the QNMs of arbitrary resonators[89, 90, 92, 93, 94, 95] and the release of general-purpose software[96], as well as successful applications to the calculation of generalized effective mode volumes[88, 93, 97], electromagnetic near- and far-field characteristics[98, 99, 100], Green’s functions[87, 95], enhanced spontaneous emission rates[93, 95, 101, 102], the semi-classical treatment of resonator-QE systems in the weak coupling regime[103] and elastic Purcell factors in optomechanical systems[104]. The purely classical electromagnetic QNM theory has also been transcended by the emergence of quantization protocols[105, 106, 107, 108], although to date, numerical studies have largely been confined to dissipative cavity quantum electrodynamics in nano-optical dielectric and plasmonic platforms.

Recently, we introduced a unified theoretical framework, based on macroscopic quantum electrodynamics (MQED)[109, 110, 111, 112, 113, 114, 115, 116] and complex coordinate transformations, that achieves QNM quantization for a wide class of three-dimensional (3D), spatially inhomogeneous, dissipative and dispersive, linear, magnetodielectric resonators[108]. The complex coordinate transformations equivalently convert the radiative losses into non-radiative material dissipation, and via a suitable transformation that reflects all the losses of the resonator, we define creation and annihilation operators that allow the construction of modal Fock states for the joint excitations of electric- and magnetic-field-dressed matter, providing the foundations for a quantum dynamic theory via quantum Langevin and master equations[105]. Whilst our numerical demonstrations concerned Purcell-enhanced spontaneous emission rates and single-emitter strong coupling in open photonic resonators, our formalism bears the notable and, heretofore, unprecedented advantage among quantum QNM theories of accommodating both arbitrary permittivity and permeability distributions. As such, our approach addresses the pressing need for a quantized treatment of the lossy resonances of magnonic cavities and their interaction with spin and SC qubits without recourse to ad hoc quantization schemes. Moreover, our theory admits an important generalization to finite temperatures, offering an account of thermal corrections that are expected to be particularly significant in the RF and MW frequency ranges characteristic of experiments in cavity quantum magnonics.

In this article, we present a finite-temperature quantum QNM theory for arbitrary, 3D, dissipative and dispersive, cavity magnonic systems. Commencing from the classical QNM theory and MQED for non-reciprocal materials[117], we generalize our PML-based QNM quantization approach of Ref.[108] for lossy magnetic resonators in general dielectric environments, enabling the prediction of all magnon-photon couplings and hybridizations from first principles, revealing the quantitative limitations of purely phenomenological quantization protocols, and offering unprecedented insight into the quantum dynamics of magnon-qubit interactions in hybrid quantum architectures. Our methodology affords a truly 4D (spatially and temporally resolved) characterization of quantum magnonic systems and their key figures of merit for device design, optimization and application in the Quantum 2.0 era.

Classical QNM Theory for Non-reciprocal Magnetic Materials

In this section, we discuss the classical electromagnetic QNM theory for non-reciprocal media based on PML regularization. The construction follows Refs.[89, 93, 117, 118] and is reviewed in Ref.[90]. In particular, we focus on magnetic materials like YIG. Whilst a numerical solver for such problems is publicly available[91, 117, 119], utilizing an auxiliary-field formulation to treat gyrotropic resonators with dispersion, not all important derivations are. We hope to fill this gap here, by reviewing the formalism, generalizing it to include anisotropy and exchange interactions, and collecting the important formulae needed for the QNM quantization scheme.

Linearized Magnetization Dynamics

The magnetization dynamics of ferro- and ferrimagnets is commonly described using the LLG equation of classical micromagnetism:

∂tM(r,t)=−γM×Heff+αMsM×∂tM,(1)\partial_t \mathbf{M}(\mathbf{r},t) = -\gamma\mathbf{M} \times\mathbf{H}_{\mathrm{eff}} + \frac{\alpha}{M_s}\mathbf{M} \times\partial_t \mathbf{M}, \tag*{(1)}

in which γ\gamma is the gyromagnetic ratio, α\alpha is the Gilbert damping parameter and MsM_s the saturation magnetization. The effective magnetic field strength Heff\mathbf{H}_{\mathrm{eff}} typically comprises the static external bias field H0\mathbf{H}_0, the applied oscillating RF/MW field h\mathbf{h}, the magnetocrystalline anisotropy field Hani\mathbf{H}_{\mathrm{ani}}, the static demagnetizing field of the saturated body, and if retained, the exchange field Hex=J∇2MH_{\mathrm{ex}} = J\nabla^2 M. In general, the magnetocrystalline anisotropy field depends on the symmetry properties of the crystal and on the orientation of the magnetization relative to the crystal axes; in this work, we shall assume that the external bias field is directed along a principal axis (namely, the easy/hard axis or orthogonal to it for uniaxial anisotropy, or a cube edge for cubic anisotropy), so that the contribution to the linearized magnetization dynamics below assumes the form Hani=−D⋅M\boldsymbol{H}_{\mathrm{ani}}=-\boldsymbol{D}\cdot\boldsymbol{M}, with D=diag⁡(Dx,Dy,Dz)\boldsymbol{D}=\operatorname{diag}(D_x,D_y,D_z) and where the constants DiD_i are determined by the functional derivatives of the anisotropy energy density[120, 121]. For an ellipsoidal body, the static demagnetizing field is uniform, Hd=−N⋅Ms\boldsymbol{H}_{d}=-\boldsymbol{N}\cdot\boldsymbol{M}_s, with N=diag⁡(Nx,Ny,Nz)\boldsymbol{N}=\operatorname{diag}(N_x,N_y,N_z) satisfying Nx+Ny+Nz=1N_x+N_y+N_z=1[122]. It contributes only to a constant shift of the bias field that enters the linearized dynamics. We fix the direction of the saturation magnetization along ez\boldsymbol{e}_z and assume a weak oscillating field h\boldsymbol{h} that produces a small deviation from saturation, M=Msez+m\boldsymbol{M}=M_s\boldsymbol{e}_z+\boldsymbol{m}, with ∣m∣≪Ms\lvert\boldsymbol{m}\rvert\ll M_s. Adding the static demagnetizing field Hd=−NzMsez\boldsymbol{H}_{d}=-N_zM_s\boldsymbol{e}_z, linearizing, and performing a Fourier transform with respect to both the space and time variables then yields

iωm=ez×[γMsh−(γH0−iωα−γMs(Dz+Nz)+γMsD+γMsJk2)m].(2)i\omega\boldsymbol{m}=\boldsymbol{e}_z\times\left[\gamma M_s\boldsymbol{h}-\left(\gamma H_0-i\omega\alpha-\gamma M_s(D_z+N_z)+\gamma M_s\boldsymbol{D}+\gamma M_sJk^2\right)\boldsymbol{m}\right]. \tag*{(2)}

Note that the dynamic demagnetizing field, −N⋅m-\boldsymbol{N}\cdot\boldsymbol{m}, is not added to Heff\boldsymbol{H}_{\mathrm{eff}}, since it is generated automatically when the effective permeability derived here is introduced into the full Maxwell’s equations. Indeed, from the above equation, we can express the magnetization in the form m=χmh\boldsymbol{m}=\chi_m\boldsymbol{h}, and thus identify the effective permeability tensor μ=1+χm\boldsymbol{\mu}=\boldsymbol{1}+\chi_m as

μ(ω,k)=(1+χxx(ω,k)−iκr(ω,k)0iκr(ω,k)1+χyy(ω,k)0001).(3)\boldsymbol{\mu}(\omega,k)=\begin{pmatrix}1+\chi_{xx}(\omega,k)&-i\kappa_r(\omega,k)&0\\i\kappa_r(\omega,k)&1+\chi_{yy}(\omega,k)&0\\0&0&1\end{pmatrix}. \tag*{(3)}

where

κr=ωωM(ωH+AxωM−iαω)(ωH+AyωM−iαω)−ω2.(4)\kappa_r=\frac{\omega\omega_M}{(\omega_H+A_x\omega_M-i\alpha\omega)(\omega_H+A_y\omega_M-i\alpha\omega)-\omega^2}. \tag*{(4)}
χxx=ωM(ωH+AyωM−iαω)(ωH+AxωM−iαω)(ωH+AyωM−iαω)−ω2.(5)\chi_{xx}=\frac{\omega_M(\omega_H+A_y\omega_M-i\alpha\omega)}{(\omega_H+A_x\omega_M-i\alpha\omega)(\omega_H+A_y\omega_M-i\alpha\omega)-\omega^2}. \tag*{(5)}
χyy=ωM(ωH+AxωM−iαω)(ωH+AxωM−iαω)(ωH+AyωM−iαω)−ω2.(6)\chi_{yy}=\frac{\omega_M(\omega_H+A_x\omega_M-i\alpha\omega)}{(\omega_H+A_x\omega_M-i\alpha\omega)(\omega_H+A_y\omega_M-i\alpha\omega)-\omega^2}. \tag*{(6)}

Here, ωM=γMs\omega_M=\gamma M_s, ωH=γ(H0−NzMs)\omega_H=\gamma(H_0-N_zM_s) and Ai=ΔDi+Jk2A_i=\Delta D_i+Jk^2, with ΔDi=Di−Dz\Delta D_i=D_i-D_z. The kk-dependence caused by the exchange interaction implies spatial dispersion (i.e., non-locality). This means that the constitutive relation for the magnetic field is given by a spatial convolution rather than a simple proportionality:

B(r,ω)=[μ(ω)⋆H(ω)](r)=∫d3r′ μ(r,r′,ω)H(r′,ω).(7)\boldsymbol{B}(\boldsymbol{r},\omega)=[\boldsymbol{\mu}(\omega)\star\boldsymbol{H}(\omega)](\boldsymbol{r})=\int\mathrm{d}^3r'\,\boldsymbol{\mu}(\boldsymbol{r},\boldsymbol{r}',\omega)\boldsymbol{H}(\boldsymbol{r}',\omega). \tag*{(7)}

with the spatial inverse Fourier transform

μ(r,r′,ω)=1(2π)3∫d3k μ(k,ω)eik⋅(r−r′).(8)\boldsymbol{\mu}(\boldsymbol{r},\boldsymbol{r}',\omega)=\frac{1}{(2\pi)^3}\int\mathrm{d}^3k\,\boldsymbol{\mu}(\boldsymbol{k},\omega)e^{i\boldsymbol{k}\cdot(\boldsymbol{r}-\boldsymbol{r}')}. \tag*{(8)}

While this expression is valid for infinite magnetic media only, the real-space form of the non-local permeability for finite resonators might be difficult to determine analytically. The linearization of the QNM eigenvalue problem discussed in Sect. II B also provides a method to calculate it numerically. It is noteworthy that the QNM formalism has already been used to describe spatial non-locality in plasmonic systems[123] by means of the hydrodynamic Drude model, in which the metal response is treated via a wavevector-dependent permittivity[124]. There, an auxiliary field was introduced to capture the non-local response; it is essentially analogous to the auxiliary field used to linearize the QNM eigenproblem in the case of spatially local, but temporally non-local systems in standard QNM solvers[89], with the well-known β2k2\beta^2 k^2 pressure contribution restored as a spatial derivative after the inverse Fourier transform. As we now introduce similar auxiliary fields to linearize the present eigenproblem for a frequency-dependent permeability, the exchange interaction can be included in the same way, where the Jk2Jk^2 contribution in μ(ω,k)\mu(\omega,k) is the magnetic counterpart of β2k2\beta^2 k^2.

Left and Right QNMs and Auxiliary-field Formulation

The QNMs are generally defined as time-harmonic solutions of the source-free Maxwell’s equation with outgoing-wave conditions imposed. In the presence of non-reciprocal media, the frequency-domain eigenvalue problem lacks self-adjointness, so that one must distinguish right eigenmodes (Hn(R),En(R))(\boldsymbol{H}_n^{(\mathrm{R})},\boldsymbol{E}_n^{(\mathrm{R})}) of the material parameters (μ,ϵ)(\boldsymbol{\mu},\boldsymbol{\epsilon}) from left eigenmodes (Hn(L),En(L))(\boldsymbol{H}_n^{(\mathrm{L})},\boldsymbol{E}_n^{(\mathrm{L})}) of the corresponding transposed ones (μT,ϵT)(\boldsymbol{\mu}^{\mathrm{T}},\boldsymbol{\epsilon}^{\mathrm{T}}):

\begin{pmatrix} 0 & -i(\boldsymbol{\mu}^{-1})^{\mathrm{T}}\nabla\times\\ i(\boldsymbol{\epsilon}^{-1})^{\mathrm{T}}\nabla\times& 0 \end{pmatrix} \begin{pmatrix} \boldsymbol{H}_n^{(\mathrm{L/R})}\\ \boldsymbol{E}_n^{(\mathrm{L/R})} \end{pmatrix} =\widetilde{\omega}_n \begin{pmatrix} \boldsymbol{H}_n^{(\mathrm{L/R})}\\ \boldsymbol{E}_n^{(\mathrm{L/R})} \end{pmatrix}. \tag*{(9)}\end{pmatrix}

The left and right QNMs share the same eigenvalues but not the same eigenvectors in the case of non-reciprocal materials. For proper normalization, both the left and right QNMs are required[91, 117, 119]. Reflecting the non-hermitian nature of the system, stemming from both radiation and material losses, the eigenvalues ω~n=ωn−iγn\tilde{\omega}_n = \omega_n - i\gamma_n are complex, which causes the QNMs to exponentially diverge in space for ∣r∣→∞|\mathbf{r}| \to\infty. By introducing perfectly matched layers (PMLs), described by transformed material parameters (μ~,ε~)(\tilde{\boldsymbol{\mu}},\tilde{\boldsymbol{\varepsilon}}), the QNMs become exponentially damped inside the PML region and the infinite domain can thus be truncated to a finite computational one ΩPML\Omega_{\mathrm{PML}} with perfect electrical/magnetic conductor boundary conditions[89, 91, 108]. Due to the truncation, the continuum contribution is mapped to a discrete set of PML modes that complement the QNMs. In what follows, all of these modes will be treated on equal footing, and we will simply refer to them as QNMs or modes.

The frequency-dependence of the permeability renders the eigenvalue problem in Eq. (25) non-linear. Nevertheless, the introduction of suitably defined auxiliary fields facilitates its linearization, enabling both practical numerical computations and analytical derivations. For general non-reciprocal media, we define auxiliary magnetization and current fields on the resonator domain Ωres\Omega_{\mathrm{res}} in the form

M(R)=1iω(μ−1−μ∞−1)∇×E(R),J(R)=−iωM(R),M(L)=1iω(μ−T−μ∞−1)∇×E(L),J(L)=−iωM(L).(10)\begin{aligned} \mathbf{M}^{(\mathrm{R})} &= \frac{1}{i\omega}\left(\boldsymbol{\mu}^{-1}-\mu_{\infty}^{-1}\right)\nabla\times\mathbf{E}^{(\mathrm{R})}, \qquad\mathbf{J}^{(\mathrm{R})}=-i\omega\mathbf{M}^{(\mathrm{R})},\\ \mathbf{M}^{(\mathrm{L})} &= \frac{1}{i\omega}\left(\boldsymbol{\mu}^{-\mathrm{T}}-\mu_{\infty}^{-1}\right)\nabla\times\mathbf{E}^{(\mathrm{L})}, \qquad\mathbf{J}^{(\mathrm{L})}=-i\omega\mathbf{M}^{(\mathrm{L})}. \tag*{(10)} \end{aligned}

For the right QNMs, we can write Eq. (10) as i(μ−1−μ∞−1)−1M(R)=∇×E(R)i(\boldsymbol{\mu}^{-1}-\mu_{\infty}^{-1})^{-1}\mathbf{M}^{(\mathrm{R})}=\nabla\times\mathbf{E}^{(\mathrm{R})} and derive the following expression for the xyxy-subblock,

(μ−1−μ∞−1)−1=μ∞ωM((ωH+ωM)I+ωMA)+iωμ∞ωMα,(11)\left(\boldsymbol{\mu}^{-1}-\mu_{\infty}^{-1}\right)^{-1} =\frac{\mu_{\infty}}{\omega_{\mathrm{M}}}\left((\omega_{\mathrm{H}}+\omega_{\mathrm{M}})\mathbf{I}+\omega_{\mathrm{M}}\mathbf{A}\right) +i\omega\frac{\mu_{\infty}}{\omega_{\mathrm{M}}}\boldsymbol{\alpha}, \tag*{(11)}

in which I\mathbf{I} is the 2×22\times2 identity matrix, A=diag⁡(Ax,Ay)\mathbf{A}=\operatorname{diag}(A_x,A_y) and α=(α−11α)\boldsymbol{\alpha}=\begin{pmatrix}\alpha&-1\\1&\alpha\end{pmatrix}. This allows us to rewrite the equation as

iωMμ∞α−1∇×E(R)+1iα−1((ωH+ωM)I+ωMA)J(R)=ωJ(R),(12)\frac{i\omega_{\mathrm{M}}}{\mu_{\infty}}\boldsymbol{\alpha}^{-1}\nabla\times\mathbf{E}^{(\mathrm{R})} +\frac{1}{i}\boldsymbol{\alpha}^{-1}\left((\omega_{\mathrm{H}}+\omega_{\mathrm{M}})\mathbf{I}+\omega_{\mathrm{M}}\mathbf{A}\right)\mathbf{J}^{(\mathrm{R})} =\omega\mathbf{J}^{(\mathrm{R})}, \tag*{(12)}

which is now linear in the eigenvalue. By using the definition of Ax/yA_{x/y} and the inverse Fourier transform, we can write Eq. (12) in the form

iωMμ∞α−1∇×E(R)+1iα−1((ωH+ωM)I+ωMΔD)J(R)−1iα−1ωMJ∇2J(R)=ωJ(R),(13)\frac{i\omega_{\mathrm{M}}}{\mu_{\infty}}\boldsymbol{\alpha}^{-1}\nabla\times\mathbf{E}^{(\mathrm{R})} +\frac{1}{i}\boldsymbol{\alpha}^{-1}\left((\omega_{\mathrm{H}}+\omega_{\mathrm{M}})\mathbf{I}+\omega_{\mathrm{M}}\Delta\mathbf{D}\right)\mathbf{J}^{(\mathrm{R})} -\frac{1}{i}\boldsymbol{\alpha}^{-1}\omega_{\mathrm{M}}J\nabla^2\mathbf{J}^{(\mathrm{R})} =\omega\mathbf{J}^{(\mathrm{R})}, \tag*{(13)}

which now also contains differential operators acting on J(R)\mathbf{J}^{(\mathrm{R})} stemming from the exchange interaction, similar to the additional terms in the hydrodynamic Drude model[123].

Inner Product, (Bi-)Orthogonality and Completeness

Consequently, the linearized eigenvalue problem can be expressed as H^(R)ψ=ωψ\hat{H}^{(\mathrm{R})}\boldsymbol{\psi}=\omega\boldsymbol{\psi}, with the augmented vector (H(R),E(R),M(R),J(R))T(\mathbf{H}^{(\mathrm{R})},\mathbf{E}^{(\mathrm{R})},\mathbf{M}^{(\mathrm{R})},\mathbf{J}^{(\mathrm{R})})^{\mathrm{T}} and the effective frequency-independent Hamiltonian

H^(R)=(01iμ∞∇×0iiε∇×000000iiωMμ∞α−1∇×00H^J,J(R))(14)\hat{H}^{(\mathrm{R})}= \begin{pmatrix} 0 & \frac{1}{i\mu_{\infty}}\nabla\times& 0 & i\\ \frac{i}{\varepsilon}\nabla\times& 0 & 0 & 0\\ 0 & 0 & 0 & i\\ \frac{i\omega_{\mathrm{M}}}{\mu_{\infty}}\boldsymbol{\alpha}^{-1}\nabla\times& 0 & 0 & \hat{H}_{J,J}^{(\mathrm{R})} \end{pmatrix} \tag*{(14)}

with H^J,J(R)=1iα−1((ωH+ωM−ωMJ∇2)I+ωMΔD)\hat{H}_{J,J}^{(\mathrm{R})}=\frac{1}{i}\boldsymbol{\alpha}^{-1}\left((\omega_{\mathrm{H}}+\omega_{\mathrm{M}}-\omega_{\mathrm{M}}J\nabla^2)\mathbf{I}+\omega_{\mathrm{M}}\Delta\mathbf{D}\right). The same construction for the left QNMs with transposed matrix αT\boldsymbol{\alpha}^{\mathrm{T}} defines H^(L)\hat{H}^{(\mathrm{L})}. In the absence of the exchange interaction (J=0J=0), the equations for M(L/R)/J(L/R)\mathbf{M}^{(\mathrm{L/R})}/\mathbf{J}^{(\mathrm{L/R})} are of algebraic nature without any restrictions for their boundary values on ∂Ωres\partial\Omega_{\mathrm{res}}. For J≠0J\ne0, they become partial differential equations, which have to be solved with the additional exchange boundary condition ∂nM(L/R)=0\partial_n\mathbf{M}^{(\mathrm{L/R})}=0. In the presence of additional surface torques (for instance, Néel surface anisotropy), the more general Rado-Weertman boundary condition has to be imposed[125, 126].

This linearization permits an efficient computational solution of the QNM and scattering problems without explicit knowledge of the non-local permeability in real space. However, for some applications it might be required, as for the QNM quantization in § III. The frequency-dependent non-local part of the permeability can be reconstructed by solving the eigenvalue problem H^J,J(L/R)ul(L/R)=λlul(L/R)\hat{H}_{J,J}^{(\mathrm{L/R})}\mathbf{u}_l^{(\mathrm{L/R})}=\lambda_l\mathbf{u}_l^{(\mathrm{L/R})} on the resonator domain Ωres\Omega_{\mathrm{res}}, similar to Ref. 127. Eq. (14) then leads to the representation:

(μ−1(ω)−μ∞−1)(r,r′,ω)=iωMμ∞∑lul(R)(r)⊗ul(L)(r′)λl−ωα−1,(15)\left(\boldsymbol{\mu}^{-1}(\omega)-\mu_{\infty}^{-1}\right)(\mathbf{r},\mathbf{r}',\omega) =i\frac{\omega_{\mathrm{M}}}{\mu_{\infty}}\sum_l\frac{\mathbf{u}_l^{(\mathrm{R})}(\mathbf{r})\otimes\mathbf{u}_l^{(\mathrm{L})}(\mathbf{r}')}{\lambda_l-\omega}\boldsymbol{\alpha}^{-1}, \tag*{(15)}

with the normalized eigenfunctions ∫Ωresd3r ul(L)(r)⋅ul(R)(r)=1\int_{\Omega_{\mathrm{res}}}\mathrm{d}^3r\,\mathbf{u}_l^{(\mathrm{L})}(\mathbf{r})\cdot\mathbf{u}_l^{(\mathrm{R})}(\mathbf{r})=1.

Inner Product, (Bi-)Orthogonality and Completeness

To derive an inner product and the orthogonality relation for the QNMs, we define the matrix

A^=(−μ∞0000ε0000−μ∞(ωHωM−J∇2)I+ΔD00000).(16)\hat{A}= \begin{pmatrix} -\mu_{\infty} & 0 & 0 & 0\\ 0 & \varepsilon& 0 & 0\\ 0 & 0 & -\mu_{\infty}\left(\frac{\omega_{\mathrm{H}}}{\omega_{\mathrm{M}}}-J\nabla^2\right)\mathbf{I}+\Delta\mathbf{D} & 0\\ 0 & 0 & 0 & 0 \end{pmatrix}. \tag*{(16)}

Utilizing the divergence theorem, we can ascertain the relation

∫Ωd3r ψn(L)A^ψn′(R)=ω~n∫Ωd3r ψn(L)A^ψn′(R)+1i∫∂Ωds⋅[En(L)×Hn′(R)−En′(R)×Hn(L)].(17)\int_{\Omega}\mathrm{d}^3r\,\boldsymbol{\psi}_n^{(\mathrm{L})}\hat{A}\boldsymbol{\psi}_{n'}^{(\mathrm{R})} =\tilde{\omega}_n\int_{\Omega}\mathrm{d}^3r\,\boldsymbol{\psi}_n^{(\mathrm{L})}\hat{A}\boldsymbol{\psi}_{n'}^{(\mathrm{R})} +\frac{1}{i}\int_{\partial\Omega}\mathrm{d}\mathbf{s}\cdot\left[\mathbf{E}_n^{(\mathrm{L})}\times\mathbf{H}_{n'}^{(\mathrm{R})}-\mathbf{E}_{n'}^{(\mathrm{R})}\times\mathbf{H}_n^{(\mathrm{L})}\right]. \tag*{(17)}

for an arbitrary domain Ω\Omega. Using ∫Ωd3r ψn(L)A^H^(R)ψn′(R)=ω~n′∫Ωd3r ψn(L)A^ψn′(R)\int_{\Omega} \mathrm{d}^{3}r\,\psi_{n}^{(L)}\hat{A}\hat{H}^{(R)}\psi_{n'}^{(R)}=\widetilde{\omega}_{n'}\int_{\Omega}\mathrm{d}^{3}r\,\psi_{n}^{(L)}\hat{A}\psi_{n'}^{(R)} leads to the unconjugated (source-free) form of the Lorentz reciprocity theorem[89, 117]. When integrating over the entire PML-truncated domain ΩPML\Omega_{\mathrm{PML}}, the boundary integrals in Eq. (17) vanish due to the imposed boundary conditions, resulting in (ω~n−ω~n′)∫ΩPMLd3r ψn(L)A^ψn′(R)=0(\widetilde{\omega}_{n}-\widetilde{\omega}_{n'})\int_{\Omega_{\mathrm{PML}}}\mathrm{d}^{3}r\,\psi_{n}^{(L)}\hat{A}\psi_{n'}^{(R)}=0. Consequently, the orthonormality condition for the QNMs is given by

⟨ψn(L)∣ψn′(R)⟩A=∫ΩPMLd3r ψn(L)A^ψn′(R)=δnn′.(18)\langle\psi_{n}^{(L)}|\psi_{n'}^{(R)}\rangle_{A}=\int_{\Omega_{\mathrm{PML}}}\mathrm{d}^{3}r\,\psi_{n}^{(L)}\hat{A}\psi_{n'}^{(R)}=\delta_{nn'} . \tag*{(18)}

Since it can be written as ⟨ψn(L)∣ψn′(R)⟩A=⟨A^ψn(L)∗∣ψn′(R)⟩L2\langle\psi_{n}^{(L)}|\psi_{n'}^{(R)}\rangle_{A}=\langle\hat{A}\psi_{n}^{(L)*}|\psi_{n'}^{(R)}\rangle_{L^{2}}, the modes {ψn(R)}\{\psi_{n}^{(R)}\} and their dual counterparts {A^ψn(L)∗}\{\hat{A}\psi_{n}^{(L)*}\} form a biorthogonal basis with respect to the standard (complex) L2L^{2}-inner product. The induced QNM norm QNn2=⟨ψnL∣ψnR⟩A\mathrm{QN}_{n}^{2}=\langle\psi_{n}^{L}|\psi_{n}^{R}\rangle_{A} admits two equivalent expressions, namely

QNn2=∫ΩPMLd3r[En(L)εEn(R)−Hn(L)μ∞Hn(R)−Mn(L)μ∞((ωHωM−J∇2)I+ΔD)Mn(R)].(19)\mathrm{QN}_{n}^{2}=\int_{\Omega_{\mathrm{PML}}}\mathrm{d}^{3}r\left[\mathbf{E}_{n}^{(L)}\varepsilon\mathbf{E}_{n}^{(R)}-\mathbf{H}_{n}^{(L)}\mu_{\infty}\mathbf{H}_{n}^{(R)}-\mathbf{M}_{n}^{(L)}\mu_{\infty}\left(\left(\frac{\omega_{\mathrm{H}}}{\omega_{\mathrm{M}}}-J\nabla^{2}\right)I+\Delta D\right)\mathbf{M}_{n}^{(R)}\right]. \tag*{(19)}

and

QNn2=∫ΩPMLd3r[En(L)εEn(R)−Hn(L)∂ω(ωμ(ω))Hn(R)].\mathrm{QN}_{n}^{2}=\int_{\Omega_{\mathrm{PML}}}\mathrm{d}^{3}r\left[\mathbf{E}_{n}^{(L)}\varepsilon\mathbf{E}_{n}^{(R)}-\mathbf{H}_{n}^{(L)}\partial_{\omega}\left(\omega\mu(\omega)\right)\mathbf{H}_{n}^{(R)}\right].

Henceforth, the QNMs are understood to be normalized so that QNn=1\mathrm{QN}_{n}=1, unless otherwise stated. In the presence of the exchange interaction, the last equation would involve the calculation of a convolution integral requiring the analytical form of the non-local permeability, whereas this is not the case for the linearized equation. Completeness of the QNMs alone within the resonator (for trivial background geometries) and of the set comprising both QNMs and PML modes everywhere inside the unmapped domain enclosed by the PMLs is assumed, as in the established dielectric and plasmonic contexts[89]; whilst no rigorous proof has been formulated for gyrotropic and possibly spatially dispersive media, it is here justified a posteriori by the agreement of truncated QNM expansions with the analytic Green’s function in § IV.

Modal Expansions and Green's Function

Assuming completeness of the QNMs and PML modes, one can solve the generalized scattering/radiation problem of the form H^(R)ψ(R)(r)=ωψ(R)(r)+S(ω)\hat{H}^{(R)}\psi^{(R)}(\mathbf{r})=\omega\psi^{(R)}(\mathbf{r})+S(\omega), where S(ω)S(\omega) is a source term. The solution can be expanded in the right QNMs as ψ(R)(r,ω)=∑nαn(ω)ψn(R)(r)\psi^{(R)}(\mathbf{r},\omega)=\sum_{n}\alpha_{n}(\omega)\psi_{n}^{(R)}(\mathbf{r}). Inserting this expansion into the previous equation and projecting both sides onto a dual QNM using the inner product in Eq. (18), we obtain:

αn(ω)=−1ω−ω~n⟨ψn(L)∣S(ω)⟩A.(20)\alpha_{n}(\omega)=-\frac{1}{\omega-\widetilde{\omega}_{n}}\langle\psi_{n}^{(L)}|S(\omega)\rangle_{A}. \tag*{(20)}

For a general electric current density j(r,ω)\mathbf{j}(\mathbf{r},\omega) located inside the unmapped domain Ω∖ΩPML\Omega\setminus\Omega_{\mathrm{PML}}, so that S(ω)=(0,iε−1j(r,ω),0,0)TS(\omega)=(0,i\varepsilon^{-1}\mathbf{j}(\mathbf{r},\omega),0,0)^{T}, this leads to the expression

αn(ω)=−iω−ω~n∫ΩPMLd3r En(L)(r)⋅j(r,ω),(21)\alpha_{n}(\omega)=-\frac{i}{\omega-\widetilde{\omega}_{n}}\int_{\Omega_{\mathrm{PML}}}\mathrm{d}^{3}r\,\mathbf{E}_{n}^{(L)}(\mathbf{r})\cdot\mathbf{j}(\mathbf{r},\omega), \tag*{(21)}

while for a point electric/magnetic dipole emitter at location re\mathbf{r}_{e} with j(r,ω)=−iωdδ(r−re)+∇δ(r−re)×m\mathbf{j}(\mathbf{r},\omega)=-i\omega\mathbf{d}\delta(\mathbf{r}-\mathbf{r}_{e})+\nabla\delta(\mathbf{r}-\mathbf{r}_{e})\times\mathbf{m}, we obtain

αn(ω)=−ωω−ω~nEn(L)(re)⋅d+ωω−ω~nBn(L)(re)⋅m.(22)\alpha_{n}(\omega)=-\frac{\omega}{\omega-\widetilde{\omega}_{n}}\mathbf{E}_{n}^{(L)}(\mathbf{r}_{e})\cdot\mathbf{d}+\frac{\omega}{\omega-\widetilde{\omega}_{n}}\mathbf{B}_{n}^{(L)}(\mathbf{r}_{e})\cdot\mathbf{m}. \tag*{(22)}

Classically, the spontaneous decay rate Γ=∑nΓn=∑nΓne+Γnm\Gamma=\sum_{n}\Gamma_{n}=\sum_{n}\Gamma_{n}^{e}+\Gamma_{n}^{m} can be calculated from the field generated by the emitter via Γn(ω)=1ℏIm⁡(αn(ω)d∗⋅En(R)(re))\Gamma_{n}(\omega)=\frac{1}{\hbar}\operatorname{Im}(\alpha_{n}(\omega)\mathbf{d}^{*}\cdot\mathbf{E}_{n}^{(R)}(\mathbf{r}_{e})) and Γnm(ω)=1ℏIm⁡(αn(ω)m∗⋅Bn(R)(re))\Gamma_{n}^{m}(\omega)=\frac{1}{\hbar}\operatorname{Im}(\alpha_{n}(\omega)\mathbf{m}^{*}\cdot\mathbf{B}_{n}^{(R)}(\mathbf{r}_{e})). Inserting Eq. (22) leads to

Γn=2ℏRe⁡(g~n2)γnω(ω−ωn)2+γn2(1−Im⁡(g~2)ω−ωnRe⁡(g~2)1γn)≈2ℏRe⁡(g~n2)γnωn(ω−ωn)2+γn2(1−Im⁡(g~2)ω−ωnRe⁡(g~2)1γn)=2ℏRe⁡(gn2)γn(ω−ωn)2+γn2(1−Im⁡(g2)ω−ωnRe⁡(g2)1γn),(23)\begin{aligned} \Gamma_{n}&=\frac{2}{\hbar}\operatorname{Re}(\widetilde{g}_{n}^{2})\frac{\gamma_{n}\omega}{(\omega-\omega_{n})^{2}+\gamma_{n}^{2}}\left(1-\frac{\operatorname{Im}(\widetilde{g}^{2})\omega-\omega_{n}}{\operatorname{Re}(\widetilde{g}^{2})}\frac{1}{\gamma_{n}}\right)\\ &\approx\frac{2}{\hbar}\operatorname{Re}(\widetilde{g}_{n}^{2})\frac{\gamma_{n}\omega_{n}}{(\omega-\omega_{n})^{2}+\gamma_{n}^{2}}\left(1-\frac{\operatorname{Im}(\widetilde{g}^{2})\omega-\omega_{n}}{\operatorname{Re}(\widetilde{g}^{2})}\frac{1}{\gamma_{n}}\right)\\ &=\frac{2}{\hbar}\operatorname{Re}(g_{n}^{2})\frac{\gamma_{n}}{(\omega-\omega_{n})^{2}+\gamma_{n}^{2}}\left(1-\frac{\operatorname{Im}(g^{2})\omega-\omega_{n}}{\operatorname{Re}(g^{2})}\frac{1}{\gamma_{n}}\right), \tag*{(23)} \end{aligned}

with g2=g~2ωn=ωn(En(L)(re)⋅d−Bn(L)(r)⋅m)(En(R)(re)⋅d∗+Bn(R)(r)⋅m∗)g^{2}=\widetilde{g}^{2}\omega_{n}=\omega_{n}(\mathbf{E}_{n}^{(L)}(\mathbf{r}_{e})\cdot\mathbf{d}-\mathbf{B}_{n}^{(L)}(\mathbf{r})\cdot\mathbf{m})(\mathbf{E}_{n}^{(R)}(\mathbf{r}_{e})\cdot\mathbf{d}^{*}+\mathbf{B}_{n}^{(R)}(\mathbf{r})\cdot\mathbf{m}^{*}). In the second line, we approximated ω≈ωn\omega\approx\omega_{n} in the numerator near the resonance. The first term in Eq. (23) is the usual Lorentzian contribution, similar to the classical result obtained by Purcell; the second term encodes non-Lorentzian lineshapes stemming from overlapping and interfering QNMs as well as from non-Markovian effects. In the single-mode case, gg can be used as the coupling constant of a quantum emitter with transition dipole moment d/m\mathbf{d}/m in the interaction Hamiltonian H^int=ga^σ++h.c.\hat{H}_{\mathrm{int}}=g\hat{a}\sigma^{+}+\mathrm{h.c.}

From Eq. (22), we also obtain an expansion for the dyadic Green’s function of the electric field, defined by ∇×μ−1∇×G(r,r′,ω)−(ω2/c2)εG(r,r′,ω)=δ(r−r′)\nabla\times\mu^{-1}\nabla\times G(\mathbf{r},\mathbf{r}',\omega)-(\omega^{2}/c^{2})\varepsilon G(\mathbf{r},\mathbf{r}',\omega)=\delta(\mathbf{r}-\mathbf{r}'):

G(r,r′,ω)=ε0c2∑n1ω(ω−ω~n)En(R)(r)⊗En(L)(r′),(24)G(\mathbf{r},\mathbf{r}',\omega)=\varepsilon_{0}c^{2}\sum_{n}\frac{1}{\omega(\omega-\widetilde{\omega}_{n})}\mathbf{E}_{n}^{(R)}(\mathbf{r})\otimes\mathbf{E}_{n}^{(L)}(\mathbf{r}'), \tag*{(24)}

from which Purcell factors and other significant quantities can be obtained. Another important case is the excitation of the modes by an incoming wave field (Einc(r,ω),Hinc(r,ω))T(\mathbf{E}_{\mathrm{inc}}(\mathbf{r},\omega),\mathbf{H}_{\mathrm{inc}}(\mathbf{r},\omega))^{T}, satisfying Maxwell’s equations

(0−iμb−1∇×iεb−1∇×0)(HincEinc)=ω(HincEinc)(25)\begin{pmatrix} 0 & -i\mu_{b}^{-1}\nabla\times\\ i\varepsilon_{b}^{-1}\nabla\times& 0 \end{pmatrix} \begin{pmatrix} \mathbf{H}_{\mathrm{inc}}\\ \mathbf{E}_{\mathrm{inc}} \end{pmatrix} = \omega \begin{pmatrix} \mathbf{H}_{\mathrm{inc}}\\ \mathbf{E}_{\mathrm{inc}} \tag*{(25)} \end{pmatrix}

in the background medium (εb\varepsilon_b, μb\mu_b). After separating the scattered field (E(R),H(R))=(Einc,Hinc)+(Esca(R),Hsca(R))(\mathbf{E}^{(\mathrm{R})},\mathbf{H}^{(\mathrm{R})}) = (\mathbf{E}_{\mathrm{inc}},\mathbf{H}_{\mathrm{inc}}) + (\mathbf{E}_{\mathrm{sca}}^{(\mathrm{R})},\mathbf{H}_{\mathrm{sca}}^{(\mathrm{R})}), we can expand (Esca(R),Hsca(R))(\mathbf{E}_{\mathrm{sca}}^{(\mathrm{R})},\mathbf{H}_{\mathrm{sca}}^{(\mathrm{R})}) in the right QNMs. By using Eq. (21) with j=−iω(ε−εb)Einc−1iω∇×(μ−1(ω)−μb−1)∇×Einc\mathbf{j}=-\mathrm{i}\omega(\varepsilon-\varepsilon_b)\mathbf{E}_{\mathrm{inc}}-\frac{1}{\mathrm{i}\omega}\nabla\times(\mu^{-1}(\omega)-\mu_b^{-1})\nabla\times\mathbf{E}_{\mathrm{inc}}, we can ascertain the formula for the excitation coefficients of the scattered field:

αn(ω)=−1ω(ω−ω~n)∫Ωresd3r[ω2En(L)(ε−εb)Einc−(∇×En(L))(μ−1(ω)−μb−1)(∇×Einc)](26)\alpha_n(\omega)=-\frac{1}{\omega(\omega-\widetilde{\omega}_n)}\int_{\Omega_{\mathrm{res}}}\mathrm{d}^3r\left[\omega^2\mathbf{E}_n^{(\mathrm{L})}(\varepsilon-\varepsilon_b)\mathbf{E}_{\mathrm{inc}}-(\nabla\times\mathbf{E}_n^{(\mathrm{L})})(\mu^{-1}(\omega)-\mu_b^{-1})(\nabla\times\mathbf{E}_{\mathrm{inc}})\right] \tag*{(26)}

In the presence of the exchange interaction, the second term cannot be reduced to the action of simple derivative operators as in Eq. (19). Its evaluation is therefore non-trivial, and requires the spatial convolution in Eq. (7) with an analytic form or the constructed spatial representation of μ(r,r′,ω)\mu(\mathbf{r},\mathbf{r}',\omega) in Eq. (15). How good that approximation is and how the presence of exchange affects the modal spectrum are important questions to investigate, but lie beyond the scope of the present investigation. Nevertheless, the QNMs themselves, their normalization, the Green’s function and the system’s response to local sources can still be ascertained without constructing μ(r,r′,ω)\mu(\mathbf{r},\mathbf{r}',\omega).

It is important to note that the normalization of the left/right QNMs E~n(L/R)\widetilde{\mathbf{E}}_n^{(\mathrm{L/R})} is not unique. Indeed, a constant rescaling E~n(L)→aE~n(L)\widetilde{\mathbf{E}}_n^{(\mathrm{L})}\to a\widetilde{\mathbf{E}}_n^{(\mathrm{L})} in the normalization integral of Eq. (19) would transform the normalized left/right QNM in the manner E~n(L)→aE~n(L)\widetilde{\mathbf{E}}_n^{(\mathrm{L})}\to\sqrt{a}\widetilde{\mathbf{E}}_n^{(\mathrm{L})}, E~n(R)→1aE~n(R)\widetilde{\mathbf{E}}_n^{(\mathrm{R})}\to\frac{1}{\sqrt{a}}\widetilde{\mathbf{E}}_n^{(\mathrm{R})}. However, all of the physically significant predictions, like the expanded scattered field, the Green’s function and the coupling constants of QEs, always involve products of E~n(L/R)\widetilde{\mathbf{E}}_n^{(\mathrm{L/R})}, rendering them invariant under the aforementioned rescaling.

QNM Quantization for Non-reciprocal Magnetic Resonators at Finite Temperatures

In this section, we briefly review the theory of MQED in non-reciprocal media, before introducing a quantization scheme (in the spirit of Refs.[105, 108]) for the regularized QNMs of arbitrary, 3D, linear, dissipative (with possible gain contributions) and dispersive, magnetic resonators. Furthermore, we extend our zero-temperature quantum theory to finite temperatures, which constitutes a particularly relevant generalization for cavity quantum magnonics given its characteristic RF and MW spectral domains.

Macroscopic QED in Non-Reciprocal Media

We now generalize the PML-based QNM quantization scheme from Ref.[108] to non-reciprocal magnetic resonators, which are describable within the classical framework of § II. The quantization is based on MQED[109, 113] and its generalization to non-local and non-reciprocal media[116], treating the quantization of the electromagnetic field in the presence of spatially inhomogeneous, lossy/amplifying, dispersive media utilizing the Green’s function formalism. For a general non-local conductivity tensor Q(r,r′,ω)\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega) the frequency components of the electric field operator E^(r,ω)\hat{\mathbf{E}}(\mathbf{r},\omega) satisfy the equation

[∇×∇×−ω2c2]E^(r,ω)−iμ0ω[Q(ω)⋆E^(ω)](r)=iμ0ωj^F(r,ω)(27)\left[\nabla\times\nabla\times-\frac{\omega^2}{c^2}\right]\hat{\mathbf{E}}(\mathbf{r},\omega)-\mathrm{i}\mu_0\omega[\mathbf{Q}(\omega)\star\hat{\mathbf{E}}(\omega)](\mathbf{r})=\mathrm{i}\mu_0\omega\hat{\mathbf{j}}_{\mathrm{F}}(\mathbf{r},\omega) \tag*{(27)}

with a fluctuation or noise current density operator

j^F(r,ω)=ℏωπ[Re⁡(Q(ω))]1/2⋆f^(ω)(r)(28)\hat{\mathbf{j}}_{\mathrm{F}}(\mathbf{r},\omega)=\sqrt{\frac{\hbar\omega}{\pi}}\left[\operatorname{Re}(\mathbf{Q}(\omega))\right]^{1/2}\star\hat{\mathbf{f}}(\omega)(\mathbf{r}) \tag*{(28)}

in accordance with the fluctuation-dissipation theorem. For simplicity, we omit the superscript (R) on the quantum mechanical operators. The operators f^(†)(r,ω)\hat{\mathbf{f}}^{(\dagger)}(\mathbf{r},\omega) are bosonic annihilation and creation operators of the field-dressed matter excitations. These modes are constructed to diagonalize the coupled field-matter Hamiltonian, such that it can be written in the form H^=∫dω∫d3r ℏωf^†(r,ω)f^(r,ω)\hat{H}=\int\mathrm{d}\omega\int\mathrm{d}^3r\,\hbar\omega\hat{\mathbf{f}}^\dagger(\mathbf{r},\omega)\hat{\mathbf{f}}(\mathbf{r},\omega). For non-reciprocal and non-local tensors, with Q(r,r′,ω)≠QT(r′,r,ω)\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega)\ne\mathbf{Q}^{\mathrm{T}}(\mathbf{r}',\mathbf{r},\omega), the real and imaginary parts are defined as Re⁡(Q(r,r′,ω))=(Q(r,r′,ω)+Q†(r′,r,ω))/2\operatorname{Re}(\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega))=(\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega)+\mathbf{Q}^{\dagger}(\mathbf{r}',\mathbf{r},\omega))/2 and Im⁡(Q(r,r′,ω))=(Q(r,r′,ω)−Q†(r′,r,ω))/(2i)\operatorname{Im}(\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega))=(\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega)-\mathbf{Q}^{\dagger}(\mathbf{r}',\mathbf{r},\omega))/(2\mathrm{i}). The (anti)-hermitian part of the corresponding integral operator thus satisfies

Re⁡/Im⁡(Q(r′,r,ω))ji=Re⁡/Im⁡(Q(r,r′,ω))ij(29)\operatorname{Re}/\operatorname{Im}(\mathbf{Q}(\mathbf{r}',\mathbf{r},\omega))_{ji}=\operatorname{Re}/\operatorname{Im}(\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega))_{ij} \tag*{(29)}

and should not be confused with the componentwise real and imaginary parts. In our case, we can express the conductivity in terms of the permittivity and permeability, and decompose the electric/magnetic contributions, allowing us to write the conductivity and fluctuation current as

Q(r,r′,ω)=1iμ0ω∇×(μ~−1(r,r′,ω)−1)×∇←′−iε0ωε~(r,ω)δ(r−r′)(30)\mathbf{Q}(\mathbf{r},\mathbf{r}',\omega)=\frac{1}{\mathrm{i}\mu_0\omega}\nabla\times\left(\widetilde{\mu}^{-1}(\mathbf{r},\mathbf{r}',\omega)-1\right)\times\overleftarrow{\nabla}'-\mathrm{i}\varepsilon_0\omega\widetilde{\varepsilon}(\mathbf{r},\omega)\delta(\mathbf{r}-\mathbf{r}') \tag*{(30)}

The permeability in Eq. (3) satisfies the Schwarz reflection principle in both ω\omega- and kk-space (i.e., μ∗(ω,k)=μ(−ω∗,−k∗)\mu^*(\omega,k)=\mu(-\omega^*,-k^*)), ensuring real-valued response functions at both time and real space. Furthermore, it is analytic in the upper ω\omega half-plane and admits a finite limit for ∣ω∣→∞|\omega|\to\infty, thus satisfying the Kramers-Kronig relations and ensuring causality.

Using the dyadic Green’s function of Eq. (27), the electric field operator in the Schrödinger picture can be written as

E^(r,ω)=iμ0∫0∞dω [G(ω)⋆j^F(ω)](r)+h.c.=E^(+)(r)+E^(−)(r)(31)\hat{\mathbf{E}}(\mathbf{r},\omega)=\mathrm{i}\mu_0\int_0^\infty\mathrm{d}\omega\,[\mathbf{G}(\omega)\star\hat{\mathbf{j}}_{\mathrm{F}}(\omega)](\mathbf{r})+\mathrm{h.c.} =\hat{\mathbf{E}}^{(+)}(\mathbf{r})+\hat{\mathbf{E}}^{(-)}(\mathbf{r}) \tag*{(31)}

where the decomposition into positive and negative frequency components is facilitated by the Schwarz reflection principle. In Ref.[116], it was shown that an important identity for the imaginary part of the Green’s function still holds for non-reciprocal media:

μ0ω∫d3s∫d3s′ G(r,s,ω)Re⁡(Q(s,s′,ω))G†(r′,s′,ω)=Im⁡(G(r,r′,ω)).(32)\mu_0\omega\int\mathrm{d}^3s \int\mathrm{d}^3s'\,G(\mathbf{r},\mathbf{s},\omega)\operatorname{Re}(Q(\mathbf{s},\mathbf{s}',\omega))G^\dagger(\mathbf{r}',\mathbf{s}',\omega)=\operatorname{Im}(G(\mathbf{r},\mathbf{r}',\omega)). \tag*{(32)}

Employing this identity, the field expansion in Eq. (31) and the fluctuation current density in Eq. (28), the vacuum correlation function of the electric field can be ascertained in the form

⟨E^i(+)(r)E^j(−)(r′)⟩0=ℏμ0π∫0∞dω ω2Im⁡(G(r,r′,ω))ij.(33)\left\langle\hat{E}_i^{(+)}(\mathbf{r})\hat{E}_j^{(-)}(\mathbf{r}')\right\rangle_0=\frac{\hbar\mu_0}{\pi}\int_0^\infty\mathrm{d}\omega\,\omega^2\operatorname{Im}(G(\mathbf{r},\mathbf{r}',\omega))_{ij}. \tag*{(33)}

The vacuum correlation function has the same form as in the reciprocal case, but crucially, the non-reciprocity manifests in the loss of a symmetry property: ⟨E^i(+)(r)E^j(−)(r′)⟩0≠⟨E^j(+)(r′)E^i(−)(r)⟩0\left\langle\hat{E}_i^{(+)}(\mathbf{r})\hat{E}_j^{(-)}(\mathbf{r}')\right\rangle_0\ne\left\langle\hat{E}_j^{(+)}(\mathbf{r}')\hat{E}_i^{(-)}(\mathbf{r})\right\rangle_0, (i.e., directional quantum correlations without invariance under exchange of positions and directions). The effect of this can be seen, for example, in the interaction of two (or more) dipolar emitters, which is essentially governed by the spectral density matrix Kij∝di∗Im⁡(G(ri,rj,ω))djK_{ij}\propto\mathbf{d}_i^*\operatorname{Im}(G(\mathbf{r}_i,\mathbf{r}_j,\omega))\mathbf{d}_j. In the non-reciprocal case, it is possible that ∣K12∣≠∣K21∣\lvert K_{12}\rvert\ne\lvert K_{21}\rvert (i.e., the transfer of excitation between the emitters depends on the direction). For circularly polarized emitters, this implies ∣K++∣≠∣K−−∣\lvert K_{++}\rvert\ne\lvert K_{--}\rvert, meaning that the interaction of the emitters depends on handedness, since the bias field selects a sense of spin precession in the magnetic medium.

From the hermiticity of the imaginary part of the Green’s function in Eq. (29), it only follows that ⟨E^i(+)(r)E^j(−)(r′)⟩0=⟨E^j(+)(r′)E^i(−)(r)⟩0∗\left\langle\hat{E}_i^{(+)}(\mathbf{r})\hat{E}_j^{(-)}(\mathbf{r}')\right\rangle_0=\left\langle\hat{E}_j^{(+)}(\mathbf{r}')\hat{E}_i^{(-)}(\mathbf{r})\right\rangle_0^*. Using the right-hand side of Eq. (32) and the QNM expansion of the Green’s function in Eq. (24), the vacuum correlation function can be expressed as

⟨E^i(+)(r)E^j(−)(r′)⟩0=⟨E^i(+)(r),E^j(−)(r′)⟩=∑nn′Pnn′E~n,i(R)(r)E~n′,j(R)∗(r′),(34)\left\langle\hat{E}_i^{(+)}(\mathbf{r})\hat{E}_j^{(-)}(\mathbf{r}')\right\rangle_0=\left\langle\hat{E}_i^{(+)}(\mathbf{r}),\hat{E}_j^{(-)}(\mathbf{r}')\right\rangle=\sum_{n n'}P_{n n'}\widetilde{E}_{n,i}^{(R)}(\mathbf{r})\widetilde{E}_{n',j}^{(R)*}(\mathbf{r}'), \tag*{(34)}

with

Pnn′=ℏμ0π∫0∞dω ω2∫d3s ω2c2E~n(L)(s)Im⁡(ε~(s,ω))E~n′(L)∗(s)−∇×E~n(L)(s)[Im⁡(μ~(ω))⋆(∇×E~n′(L)∗)](s).(35)P_{n n'}=\frac{\hbar\mu_0}{\pi}\int_0^\infty\mathrm{d}\omega\,\omega^2\int\mathrm{d}^3s\,\frac{\omega^2}{c^2}\widetilde{\mathbf{E}}_n^{(L)}(\mathbf{s})\operatorname{Im}(\widetilde{\boldsymbol{\varepsilon}}(\mathbf{s},\omega))\widetilde{\mathbf{E}}_{n'}^{(L)*}(\mathbf{s})-\nabla\times\widetilde{\mathbf{E}}_n^{(L)}(\mathbf{s})\left[\operatorname{Im}(\widetilde{\boldsymbol{\mu}}(\omega))\star\left(\nabla\times\widetilde{\mathbf{E}}_{n'}^{(L)*}\right)\right](\mathbf{s}). \tag*{(35)}

At this point, it is evident that both left and right QNMs are required for the quantization scheme, although the PP-matrix in Eq. (35), needed for the mode symmetrization itself, can be computed solely from the left QNMs. Just as for reciprocal media, we use PMLs to truncate the open space to a finite domain ΩPML\Omega_{\mathrm{PML}}, and the spatial integral in Eq. (35) is then performed over all of ΩPML\Omega_{\mathrm{PML}}. As discussed in Ref.[108], the PMLs also contain gain components (i.e., components of the effective material parameters (μ~,ε~)(\widetilde{\boldsymbol{\mu}},\widetilde{\boldsymbol{\varepsilon}}) with negative imaginary parts). If the resonator itself is comprised of gain media, the noise current would be modified and the MQED creation/annihilation operators f^(†)(r,ω)\hat{f}^{(\dagger)}(\mathbf{r},\omega) would interchange roles in the amplifying regions[108, 114, 129]. However, the gain components of the PMLs are not explicitly taken into account, but simply treated as an effective reduced loss in the integral of Eq. (35), so that the QNM expansion of the vacuum correlation function in Eq. (34) reproduces the correct MQED result in Eq. (33).

QNM Quantization

Analogous to the reciprocal case, we now introduce a QNM expansion for the electric field operator E^(+)(r)=∑nE~n(R)(r)ξ^n\hat{\mathbf{E}}^{(+)}(\mathbf{r})=\sum_n\widetilde{\mathbf{E}}_n^{(R)}(\mathbf{r})\hat{\xi}_n[105, 108, 129]. Inserting this expansion into the commutator and comparing it to Eq. (34) shows that

[ξ^n,ξ^n′†]=Pnn′\left[\hat{\xi}_n,\hat{\xi}_{n'}^\dagger\right]=P_{n n'}

(i.e., non-bosonic commutation relations arising from the non-hermitian nature of the system). The symmetrization transformation a^n=Pnn′−1/2ξ^n′\hat{a}_n=P_{n n'}^{-1/2}\hat{\xi}_{n'} yields bosonic modal operators a^n\hat{a}_n satisfying [a^n,a^n′†]=δnn′\left[\hat{a}_n,\hat{a}_{n'}^\dagger\right]=\delta_{n n'}, and the expansion E^(+)(r)=∑nE~n(R,sym)(r)a^n\hat{\mathbf{E}}^{(+)}(\mathbf{r})=\sum_n\widetilde{\mathbf{E}}_n^{(R,\mathrm{sym})}(\mathbf{r})\hat{a}_n with the symmetrized QNMs E~n(R,sym)(r)=Pnn′1/2E~n′(R)(r)\widetilde{\mathbf{E}}_n^{(R,\mathrm{sym})}(\mathbf{r})=P_{n n'}^{1/2}\widetilde{\mathbf{E}}_{n'}^{(R)}(\mathbf{r}). The modal operators themselves can be written in the form

a^n=iμ0Pnn′−1/2∫0∞dω ωAn(ω)∫d3r[ωℏε0πIm⁡(ε(r,ω))E~n′(L)(r)⋅f^e(r,ω)+ℏμ0π[Im⁡(μ−1)1/2⋆(∇×E~n′(L))](r)⋅f^m(r,ω)].(36)\hat{a}_n=i\mu_0P_{n n'}^{-1/2}\int_0^\infty\mathrm{d}\omega\,\omega A_n(\omega)\int\mathrm{d}^3r\left[ \omega\sqrt{\frac{\hbar\varepsilon_0}{\pi}\operatorname{Im}(\varepsilon(\mathbf{r},\omega))}\widetilde{\mathbf{E}}_{n'}^{(L)}(\mathbf{r})\cdot\hat{\mathbf{f}}_{\mathrm{e}}(\mathbf{r},\omega) +\sqrt{\frac{\hbar}{\mu_0\pi}}\left[\operatorname{Im}(\boldsymbol{\mu}^{-1})^{1/2}\star(\nabla\times\widetilde{\mathbf{E}}_{n'}^{(L)})\right](\mathbf{r})\cdot\hat{\mathbf{f}}_{\mathrm{m}}(\mathbf{r},\omega) \right]. \tag*{(36)}

Namely, as superpositions of the continuum MQED modes constituting the (symmetrized) QNMs. They satisfy the Heisenberg-Langevin equation

da^ndt=iℏ[H^eff,a^n]−χnn′−a^n′+F^n(t),(37)\frac{\mathrm{d}\hat{a}_n}{\mathrm{d}t}=\frac{i}{\hbar}\left[\hat{H}_{\mathrm{eff}},\hat{a}_n\right]-\chi_{n n'}^{-}\hat{a}_{n'}+\hat{F}_n(t), \tag*{(37)}

with the effective Hamiltonian Heff=ℏχnn′+a^n′†a^n′H_{\mathrm{eff}}=\hbar\chi_{n n'}^{+}\hat{a}_{n'}^\dagger\hat{a}_{n'}, the hermitian/antihermitian parts χnn′±=i2(χnn′±χnn′†)\chi_{n n'}^{\pm}=\frac{i}{2}(\chi_{n n'}\pm\chi_{n n'}^\dagger) of the matrix χnn′=Pnn′−1/2ωmPmn′1/2\chi_{n n'}=P_{n n'}^{-1/2}\omega_mP_{m n'}^{1/2}, and the Langevin noise term

F^n(t)=iμ0Pnn′−12∫0∞dω ω(ω~n′−ω)An′(ω)∫d3r[ωℏε0πIm⁡(ε(r,ω))E~n′(L)(r)⋅f^e(r,ω)+ℏμ0π[Im⁡(μ−1)12⋆(∇×E~n′(L))](r)⋅f^m(r,ω)].(38)\hat{F}_n(t)=\mathrm{i}\mu_0 P_{nn'}^{-\frac{1}{2}}\int_0^\infty\mathrm{d}\omega\,\omega(\widetilde{\omega}_{n'}-\omega)A_{n'}(\omega)\int\mathrm{d}^3r\left[\omega\sqrt{\frac{\hbar\varepsilon_0}{\pi}\operatorname{Im}(\varepsilon(\mathbf{r},\omega))}\widetilde{\mathbf{E}}_{n'}^{(L)}(\mathbf{r})\cdot\hat{\mathbf{f}}_e(\mathbf{r},\omega)+\sqrt{\frac{\hbar}{\mu_0\pi}}\left[\operatorname{Im}(\mu^{-1})^{\frac{1}{2}}\star(\nabla\times\widetilde{\mathbf{E}}_{n'}^{(L)})\right](\mathbf{r})\cdot\hat{\mathbf{f}}_m(\mathbf{r},\omega)\right]. \tag*{(38)}

Approximating this term as white noise and invoking Itô-Stratonovich calculus leads to an effective Markovian master equation[105, 129, 130],

dρdt=−iℏ[H^eff,ρ]+χnn′Dnn′ρ,(39)\frac{\mathrm{d}\rho}{\mathrm{d}t}=-\frac{\mathrm{i}}{\hbar}\left[\hat{H}_{\mathrm{eff}},\rho\right]+\chi_{nn'}\mathcal{D}_{nn'}\rho, \tag*{(39)}

with the dissipators Dnn′ρ=2a^n′ρa^n†−{ρ,a^n†a^n′}\mathcal{D}_{nn'}\rho=2\hat{a}_{n'}\rho\hat{a}_n^\dagger-\{\rho,\hat{a}_n^\dagger\hat{a}_{n'}\}. The symmetrized QNMs couple both coherently and dissipatively via the effective Hamiltonian and the dissipators, respectively, which facilitates a rigorous treatment of dissipative photon-magnon coupling.

As in the classical case, we can show that the physical predictions of the theory are independent of any rescaling of the left and right QNMs. Rescaling of the QNMs before their normalization as E~(L/R)(r)→aL/RE~(L/R)(r)\widetilde{\mathbf{E}}^{(L/R)}(\mathbf{r})\to a_{L/R}\widetilde{\mathbf{E}}^{(L/R)}(\mathbf{r}) leads to the normalized QNMs E~(L/R)(r)→U(L/R)E~(L/R)(r)\widetilde{\mathbf{E}}^{(L/R)}(\mathbf{r})\to U^{(L/R)}\widetilde{\mathbf{E}}^{(L/R)}(\mathbf{r}), with transformation matrices Unn′(L/R)=an(L/R)/(an(L/R))12δnn′U_{nn'}^{(L/R)}=a_n^{(L/R)}/(a_n^{(L/R)})^{\frac{1}{2}}\delta_{nn'}, being non-unitary by themselves, but together satisfying U(L)U(R)=U(R)†U(L)†=IU^{(L)}U^{(R)}=U^{(R)\dagger}U^{(L)\dagger}=I. Correspondingly, the PP-matrix transforms as P→U(L)PU(L)†P\to U^{(L)}PU^{(L)\dagger}. It can be evidenced that under these transformations, the vacuum correlation function remains invariant:

⟨Ei(+)(r)Ej(−)(r′)⟩0=E~(R)(r)PE~(R)∗(r′)→E~(R)(r)U(R)U(L)PU(L)†U(R)†E~(R)∗(r′)=⟨Ei(+)(r)Ej(−)(r′)⟩0.(40)\begin{aligned} \langle E_i^{(+)}(\mathbf{r})E_j^{(-)}(\mathbf{r}')\rangle_0=\widetilde{\mathbf{E}}^{(R)}(\mathbf{r})P\widetilde{\mathbf{E}}^{(R)*}(\mathbf{r}') \\ &\to\widetilde{\mathbf{E}}^{(R)}(\mathbf{r})U^{(R)}U^{(L)}PU^{(L)\dagger}U^{(R)\dagger}\widetilde{\mathbf{E}}^{(R)*}(\mathbf{r}') \\ &=\langle E_i^{(+)}(\mathbf{r})E_j^{(-)}(\mathbf{r}')\rangle_0 . \tag*{(40)} \end{aligned}

The symmetrized QNMs transform as

E~(R,sym)→(U(L)†PTU(L))12U(R)E~(R)=UsymE~(R,sym),(41)\widetilde{\mathbf{E}}^{(R,\mathrm{sym})}\to\left(U^{(L)\dagger}P^TU^{(L)}\right)^{\frac{1}{2}}U^{(R)}\widetilde{\mathbf{E}}^{(R)}=U^{\mathrm{sym}}\widetilde{\mathbf{E}}^{(R,\mathrm{sym})}, \tag*{(41)}

with Usym=(U(L)†PTU(L))12U(R)(PT)−12U^{\mathrm{sym}}=\left(U^{(L)\dagger}P^TU^{(L)}\right)^{\frac{1}{2}}U^{(R)}(P^T)^{-\frac{1}{2}} being a unitary transformation:

Usym†Usym=(PT)−12U(R)†U(L)†PTU(L)U(R)(PT)−12=(PT)−12PT(PT)−12=I.(42)\begin{aligned} U^{\mathrm{sym}\dagger}U^{\mathrm{sym}}=(P^T)^{-\frac{1}{2}}U^{(R)\dagger}U^{(L)\dagger}P^TU^{(L)}U^{(R)}(P^T)^{-\frac{1}{2}} \\ &=(P^T)^{-\frac{1}{2}}P^T(P^T)^{-\frac{1}{2}}=I . \tag*{(42)} \end{aligned}

Although the scaling generally leads to a non-unitary transformation of the QNMs, the symmetrized QNMs always transform unitarily (i.e., they are uniquely determined up to a unitary transformation). The bosonic modal operators transform as a^n→Usym∗a^n\hat{a}_n\to U^{\mathrm{sym}*}\hat{a}_n, and from its definition, we can show that χ\chi transforms accordingly as χ→Usym∗χ(Usym∗)−1\chi\to U^{\mathrm{sym}*}\chi(U^{\mathrm{sym}*})^{-1}. In conjunction with the transformation of the noise operator, we find that the QNM master equation (39) and thus the quantum dynamics remain invariant under these transformations, as demanded.

Finite Temperature

So far, QNM quantization has been considered only at zero temperature[105, 108]. This has proven adequate for the photonic and plasmonic structures to which the formalism has most commonly been applied, where the bosonic mode frequencies lie in the optical range and thermal occupation is negligible. In quantum nano-optical scenarios, finite-temperature effects are often accounted for simply by an increased pure dephasing rate of the QEs embedded in the system. In contrast, since the excitation frequencies of magnonic modes and spin/SC qubits are often in the MHz to GHz range, thermal excitation of the modes is significantly more pronounced, even at cryogenic temperatures, and must be taken into account.

In the general theory of open quantum systems, thermal effects are typically described by assuming that the mode is coupled to a continuous bath in thermal equilibrium, leading to an incoherent pumping term in the corresponding master equation of the form Dpρ=γnT(2a^†ρa^−{ρ,a^a^†})\mathcal{D}_p\rho=\gamma n_T(2\hat{a}^\dagger\rho\hat{a}-\{\rho,\hat{a}\hat{a}^\dagger\}), with the Bose-Einstein distribution nT=1/(exp⁡(ℏω/kT)−1)n_T=1/(\exp(\hbar\omega/kT)-1), while the incoherent loss term becomes Dlρ=γ(1+nT)(2a^ρa^†−{ρ,a^†a^})\mathcal{D}_l\rho=\gamma(1+n_T)(2\hat{a}\rho\hat{a}^\dagger-\{\rho,\hat{a}^\dagger\hat{a}\}), including both spontaneous and thermally stimulated emission. For the expectation value of the bosonic number operator, this leads to the equation of d⟨n^(t)⟩/dt=−2γ(⟨n^(t)⟩−nT)\mathrm{d}\langle\hat{n}(t)\rangle/\mathrm{d}t=-2\gamma(\langle\hat{n}(t)\rangle-n_T), with solution ⟨n^(t)⟩=⟨n^(0)⟩e−2γt+nT(1−e−2γt)\langle\hat{n}(t)\rangle=\langle\hat{n}(0)\rangle e^{-2\gamma t}+n_T(1-e^{-2\gamma t}), so that the mode relaxes towards the thermal equilibrium boson number nTn_T with the spontaneous decay rate γ\gamma.

Building on the QNM quantization discussed earlier in this section, we now derive an effective thermal QNM master equation, in which interference effects arising from the shared bath are rigorously taken into account, spatially dependent temperature profiles T(r)T(\mathbf{r}) can be described, and all relevant quantities can be derived and calculated from traditional MQED. Assuming that the MQED continuum modes are in thermal equilibrium, namely ⟨f^λ,i(r,ω)f^λ′,j†(r′,ω′)⟩=(nT(r)(ω)+1)δ(ω−ω′)δijδλλ′\langle\hat{f}_{\lambda,i}(\mathbf{r},\omega)\hat{f}_{\lambda',j}^{\dagger}(\mathbf{r}',\omega')\rangle=(n_T(\mathbf{r})(\omega)+1)\delta(\omega-\omega')\delta_{ij}\delta_{\lambda\lambda'}, the thermal correlation function becomes

⟨E^i(+)(r)E^j(−)(r′)⟩T=ℏμ0π∫0∞dω (nT(ω)+1)ω2∫d3s∫d3s′[G(r,s,ω)Re⁡(Q(s,s′,ω))G†(r′,s′,ω)]\langle\hat{E}_i^{(+)}(\mathbf{r})\hat{E}_j^{(-)}(\mathbf{r}')\rangle_T =\frac{\hbar\mu_0}{\pi}\int_0^\infty\mathrm{d}\omega\,(n_T(\omega)+1)\omega^2\int\mathrm{d}^3s\int\mathrm{d}^3s'\left[G(\mathbf{r},\mathbf{s},\omega)\operatorname{Re}(Q(\mathbf{s},\mathbf{s}',\omega))G^\dagger(\mathbf{r}',\mathbf{s}',\omega)\right]
T(r)≡Tℏμ0π∫0∞dω (nT(ω)+1)ω2Im⁡(G(r,r′,ω))ij=(Pnn′+PT,nn′)E~n,i(r)E~n′,j∗(r′).(43)\begin{aligned} T(\mathbf{r})\equiv T\frac{\hbar\mu_0}{\pi}\int_0^\infty\mathrm{d}\omega\,(n_T(\omega)+1)\omega^2\operatorname{Im}(G(\mathbf{r},\mathbf{r}',\omega))_{ij} \\ &=(P_{nn'}+P_{T,nn'})\widetilde{E}_{n,i}(\mathbf{r})\widetilde{E}_{n',j}^{*}(\mathbf{r}') . \tag*{(43)} \end{aligned}

where the penultimate equality holds under the assumption of a spatially uniform temperature distribution; note that the common representation through the imaginary part of the Green’s function can no longer be used in the case of a spatially varying T(r)T(\mathbf{r}). For the last equality, we have simply used Eq. (24) and the (hermitian) thermal PP-matrix Pnn′TP^{T}_{nn'} is defined as in Eq. (35), with the additional Bose-Einstein factor nT(ω)n_T(\omega) in the frequency integral. Using the expansion of the electric field operator in terms of the symmetrized QNMs, this expectation value can also be expressed as

⟨E^i(+)(r)E^j(−)(r′)⟩T=E~n,i(R,sym)(r)E~n′,j(R,sym)∗(r′)⟨a^na^n′†⟩T=Pmn−12Pn′m′−12E~m,i(R)(r)E~m′,j(R)∗(r′)⟨a^na^n′†⟩T=!(Pmm′+Pmm′T)E~m,i(r)E~m′,j∗(r′).(44)\left\langle\hat{E}^{(+)}_i(\mathbf{r})\hat{E}^{(-)}_j(\mathbf{r}') \right\rangle_T = \widetilde{E}^{(R,\mathrm{sym})}_{n,i}(\mathbf{r})\widetilde{E}^{(R,\mathrm{sym})*}_{n',j}(\mathbf{r}')\left\langle\hat{a}_n\hat{a}^{\dagger}_{n'}\right\rangle_T = P^{-\frac{1}{2}}_{mn}P^{-\frac{1}{2}}_{n'm'}\widetilde{E}^{(R)}_{m,i}(\mathbf{r})\widetilde{E}^{(R)*}_{m',j}(\mathbf{r}')\left\langle\hat{a}_n\hat{a}^{\dagger}_{n'}\right\rangle_T \overset{!}{=}\left(P_{mm'}+P^T_{mm'}\right)\widetilde{E}_{m,i}(\mathbf{r})\widetilde{E}^{*}_{m',j}(\mathbf{r}') . \tag*{(44)}

The modal thermal expectation values are thus given by ⟨a^n†a^n′⟩T=Pnm−12PT,mm′∗Pm′n′−12\left\langle\hat{a}^{\dagger}_n\hat{a}_{n'}\right\rangle_T=P^{-\frac{1}{2}}_{nm}P^*_{T,mm'}P^{-\frac{1}{2}}_{m'n'}. To derive the thermal QNM master equation, we now make the ansatz

dρdt=−iℏ[H^eff,ρ]+(χnn′−+χnn′G(T))Dnn′Lρ+χnn′G(T)Dnn′Gρ,(45)\frac{d\rho}{dt}=-\frac{i}{\hbar}\left[\hat{H}_{\mathrm{eff}},\rho\right]+\left(\chi^-_{nn'}+\chi^G_{nn'}(T)\right)D^L_{nn'}\rho+\chi^G_{nn'}(T)D^G_{nn'}\rho, \tag*{(45)}

with a modified loss term and an incoherent pumping dissipator Dnn′Gρ=2a^n†ρa^n′−{ρ,a^n′a^n†}D^G_{nn'}\rho=2\hat{a}^{\dagger}_n\rho\hat{a}_{n'}-\{\rho,\hat{a}_{n'}\hat{a}^{\dagger}_n\} that includes cross-ing terms arising from the shared bath. From Eq. (45), we can derive equations of motion for the expectation values ⟨a^n†a^n′⟩\left\langle\hat{a}^{\dagger}_n\hat{a}_{n'}\right\rangle:

ddt⟨a^n†a^n′⟩=−(iχn′m′++χn′m′−)⟨a^n†a^m′⟩+(iχmn+−χmn−)⟨a^m†a^n′⟩+2χn′nG(T).(46)\frac{d}{dt}\left\langle\hat{a}^{\dagger}_n\hat{a}_{n'}\right\rangle =-\left(i\chi^+_{n'm'}+\chi^-_{n'm'}\right)\left\langle\hat{a}^{\dagger}_n\hat{a}_{m'}\right\rangle +\left(i\chi^+_{mn}-\chi^-_{mn}\right)\left\langle\hat{a}^{\dagger}_m\hat{a}_{n'}\right\rangle +2\chi^G_{n'n}(T) . \tag*{(46)}

for which the thermal steady-state condition ddt⟨a^n†a^n′⟩=!0\frac{d}{dt}\left\langle\hat{a}^{\dagger}_n\hat{a}_{n'}\right\rangle\overset{!}{=}0 leads to

χn′nG(T)=i2(χnm′⟨a^n′†a^m′⟩T−χmn′⟨a^m†a^n⟩T).(47)\chi^G_{n'n}(T)=\frac{i}{2}\left(\chi_{nm'}\left\langle\hat{a}^{\dagger}_{n'}\hat{a}_{m'}\right\rangle_T-\chi_{mn'}\left\langle\hat{a}^{\dagger}_m\hat{a}_n\right\rangle_T\right). \tag*{(47)}

Choosing the incoherent gain matrix χn′nG(T)\chi^G_{n'n}(T) such that it reproduces the modal thermal steady states ⟨a^n†a^n′⟩T\left\langle\hat{a}^{\dagger}_n\hat{a}_{n'}\right\rangle_T calculated from MQED leads to

χG(T)=i2(χP−12PT∗P−12−P−12PT∗P−12χ†)=−Im⁡(χP−12PTP−12)=−Im⁡(P−12PTω~P−12),(48)\chi^G(T)=\frac{i}{2}\left(\chi P^{-\frac{1}{2}}P_T^*P^{-\frac{1}{2}}-P^{-\frac{1}{2}}P_T^*P^{-\frac{1}{2}}\chi^\dagger\right) =-\operatorname{Im}\left(\chi P^{-\frac{1}{2}}P_TP^{-\frac{1}{2}}\right) =-\operatorname{Im}\left(P^{-\frac{1}{2}}P_T\widetilde{\omega}P^{-\frac{1}{2}}\right), \tag*{(48)}

with ω~=diag⁡(ω~n)\widetilde{\omega}=\operatorname{diag}(\widetilde{\omega}_n) and Im⁡(A)=1/(2i)(A−A†)\operatorname{Im}(A)=1/(2i)(A-A^\dagger) being the anti-hermitian part of the matrix.

Numerical Examples

Complementary to the general formalism introduced in the preceding sections, we now illustrate the predictive capabilities of our quantum QNM methodology by studying, in a spatiotemporally resolved manner, both photon-magnon and magnon-qubit interactions. In § IV A, we begin by considering a 1D, half-open magnonic slab, capturing both the fundamental FMR and higher-order modes, as well as their complex interplay. Second, we consider a similar magnonic slab embedded in a photonic cavity, illustrating the ability of our method to predict both coherent and dissipative magnon-photon coupling regimes and their interpolation, together with a fully spatial and temporal characterization of the quantum field dynamics. We extended our approach to a 3D case in § IV B, exploring a spherically symmetric magnon-photon hybrid resonator and its interaction with SQs and FQs.

For the purpose of our numerical demonstrations, we adopt the canonical magnonic material YIG[131, 132], with Gilbert damping parameter α=3⋅10−4\alpha=3\cdot10^{-4}, saturation magnetization μ0Ms=175 mT\mu_0M_s=175\,\mathrm{mT} and relative permittivity ϵYIG=15\epsilon_{\mathrm{YIG}}=15. The gyromagnetic ratio is chosen for the spin-only moment of Fe3+\mathrm{Fe}^{3+}, so that γ/(2π)=28 GHz/T\gamma/(2\pi)=28\,\mathrm{GHz/T}. Throughout this section, we shall neglect magnetocrystalline anisotropy and the exchange interaction; whilst the former is negligible in the case of YIG magnonics, the influence of the latter could become significant for nanoscale systems and shall be studied elsewhere.

1D Systems

A. 1D Systems

a. Magnonic Slab Resonator. We consider a YIG slab of length LYIGL_{\mathrm{YIG}} in the zz-direction and of infinite extent in the xx- and yy-directions, subject to an out-of-plane bias field H0=H0ez\mathbf{H}_0=H_0\mathbf{e}_z with H0≡H0−NzMs\mathbf{H}_0\equiv H_0-N_zM_s, Nz=1N_z=1. We impose a perfect electrical conductor boundary condition at z=0z=0, such that the slab radiates into a half-open space. The permeability in Eq. (3) has μxx=μyy\mu_{xx}=\mu_{yy} and eigenvalues {μxx±κ,1}\{\mu_{xx}\pm\kappa,1\}, with corresponding eigenvectors {(1,±i,0)T,(0,0,1)T}\{(1,\pm i,0)^{\mathrm{T}},(0,0,1)^{\mathrm{T}}\}. We consider waves propagating along the zz-direction and treat the system as 1D. As apparent from Eq. (35), the left and right circularly polarized eigenmodes do not couple and can be treated separately in the quantization; here, we retain only the left circularly polarized modes, since the FMR occurs for this polarization. The right QNMs have the form E~n(R)(z)=En(R)(z)(1,i,0)T/2\widetilde{\mathbf{E}}^{(R)}_n(z)=E^{(R)}_n(z)(1,i,0)^{\mathrm{T}}/\sqrt{2} and can be calculated from the 1D wave equation ∂z(μeff−1∂zEn(R))+(ω2/c2)ϵEn(R)=0\partial_z(\mu_{\mathrm{eff}}^{-1}\partial_zE^{(R)}_n)+(\omega^2/c^2)\epsilon E^{(R)}_n=0, with the effective relative permeability μeff(z,ω)=μxx(ω)+κ(ω)\mu_{\mathrm{eff}}(z,\omega)=\mu_{xx}(\omega)+\kappa(\omega) (z<LYIGz<L_{\mathrm{YIG}}) and μeff(z,ω)=1\mu_{\mathrm{eff}}(z,\omega)=1 (z≥LYIGz\geq L_{\mathrm{YIG}}), together with the imposed boundary condition En(R)(z=0)=0E^{(R)}_n(z=0)=0. The corresponding left QNMs are given by E~n(L)(z)=En(R)(z)(1,−i,0)T/2\widetilde{\mathbf{E}}^{(L)}_n(z)=E^{(R)}_n(z)(1,-i,0)^{\mathrm{T}}/\sqrt{2} (for transposed permeability), and the PP-matrix elements in Eq. (35) reduce to

Pnn′=ℏμ0πA∫0∞dω ω2An(ω)An′∗(ω)∫ΩPMLds[ω2c2Im⁡(ε~(s,ω))En(R)(s)En′(R)∗(s)−Im⁡(μ~eff−1(s,ω))∂sEn(R)(s)∂sEn′(R)∗(s)],(49)P_{nn'}=\frac{\hbar\mu_0}{\pi A}\int_0^\infty d\omega\,\omega^2 A_n(\omega)A_{n'}^*(\omega)\int_{\Omega_{\mathrm{PML}}}ds\left[\frac{\omega^2}{c^2}\operatorname{Im}(\tilde{\varepsilon}(s,\omega))E_n^{(\mathrm{R})}(s)E_{n'}^{(\mathrm{R})*}(s)-\operatorname{Im}(\tilde{\mu}_{\mathrm{eff}}^{-1}(s,\omega))\partial_sE_n^{(\mathrm{R})}(s)\partial_sE_{n'}^{(\mathrm{R})*}(s)\right], \tag*{(49)}

with the normalization area AA in the xyxy-plane. Considering a two-level magnetic QE with frequency ωQE\omega_{\mathrm{QE}} coupled to the resonator, the QE decay rate can be obtained from the QNM master equation in the bad-cavity limit[105, 108, 133]:

ΓQE=1ℏ2Re⁡(gnPnn′121i(ωQE−ω~n′∗)Pn′n′′12gn′′∗).(50)\Gamma_{\mathrm{QE}}=\frac{1}{\hbar^2}\operatorname{Re}\left(g_nP_{nn'}^{\frac{1}{2}}\frac{1}{i(\omega_{\mathrm{QE}}-\tilde{\omega}_{n'}^*)}P_{n'n''}^{\frac{1}{2}}g_{n''}^*\right). \tag*{(50)}

The QE decay rate can also be calculated exactly from the 1D Green’s function of the system via ΓQE=(μ0∣mQE∣2/ℏA)lim⁡z′→z∂z∂z′Im⁡(G(z,z′,ω))\Gamma_{\mathrm{QE}}=(\mu_0|m_{\mathrm{QE}}|^2/\hbar A)\lim_{z'\to z}\partial_z\partial_{z'}\operatorname{Im}(G(z,z',\omega)), which is known analytically for this system. Normalization with respect to the 1D free space decay rate ΓQE0=μ0ωQE∣mQE∣2/(2ℏcA)\Gamma_{\mathrm{QE}}^0=\mu_0\omega_{\mathrm{QE}}|m_{\mathrm{QE}}|^2/(2\hbar cA) yields an mQEm_{\mathrm{QE}}- and AA-independent magnetic Purcell factor.

Figure 1(a) presents the six lowest complex QNM frequencies ω~n\tilde{\omega}_n, along with the corresponding magnetic flux density profiles ∣Bn(z)∣|B_n(z)|, for a YIG slab of length LYIG=500 μmL_{\mathrm{YIG}}=500\,\mu\mathrm{m} magnetized along the positive zz-direction and subject to a bias field of H0=1.5 MsH_0=1.5\,M_s. At the pole of the effective permeability, ω~pole=(ωH−ωM)/(1+iα)\tilde{\omega}_{\mathrm{pole}}=(\omega_{\mathrm{H}}-\omega_{\mathrm{M}})/(1+i\alpha), there is an accumulation point of QNMs (i.e., an infinite set of QNMs with ω~n→ω~pole\tilde{\omega}_n\to\tilde{\omega}_{\mathrm{pole}}), as also appear in plasmonic systems[91]. The lowest mode is the nearly uniform Kittel mode, which appears to have the lowest QQ-factor, whereas the higher-order modes are spatially oscillatory. The accumulation point arises from the divergence of the refractive index, so that the YIG slab of fixed length can accommodate an arbitrarily large number of oscillations as the wavelength in the material tends to zero at ω~pole\tilde{\omega}_{\mathrm{pole}}.

Numerical investigation of a 1D YIG slab in 1D half-open space

Figure 1. Numerical investigation of a 1D YIG slab in 1D half-open space, magnetized in the positive zz-direction with H0=1.5MsH_0 = 1.5M_{\mathrm{s}}. (a) Magnetic flux density profiles ∣B~n(z)∣|\tilde{\mathbf{B}}_n(z)| and complex frequencies ω~n\tilde{\omega}_n of the lowest left circularly polarized QNMs for a slab of length LYIG=500 μmL_{\mathrm{YIG}} = 500\,\mu\mathrm{m}. (b) Frequency-dependent magnetic Purcell factor for a QE positioned at the open end of the slab (z=LYIGz = L_{\mathrm{YIG}}), obtained from the full quantum QNM theory, the phenomenological QNM quantization scheme and the analytical treatment. (c) Corresponding magnetic Purcell factor for a slab of length LYIG=10 μmL_{\mathrm{YIG}} = 10\,\mu\mathrm{m}, calculated from a single-mode (Kittel) expansion in the quantum QNM theory, compared with the analytical result. (d) Real and imaginary parts of the complex QNM frequency of the Kittel mode as a function of LYIGL_{\mathrm{YIG}}.

In Figure 1(b), the corresponding Purcell factor for a magnetic QE positioned at the open end of the slab resonator (z=LYIGz=L_{\mathrm{YIG}}) is shown, calculated from both the quantum QNM methodology, retaining the six modes of Figure 1(a), and the analytical Green’s function. The Purcell factors are found to be in excellent quantitative agreement. Notably, the higher-order modes lead to dips rather than additional peaks – stemming from interference effects in the dissipative coupling of these modes – which the quantum QNM theory reproduces very well. Our calculated Purcell factors imply that the dynamics of the Kittel mode, and of QEs interacting with it, are directly impacted by the presence of the higher-order modes. In addition, we plot the Purcell factor calculated from the phenomenological quantization of the QNMs, using the QE-QNM coupling constants from Eq. (23) and uncoupled QNMs (i.e., H^eff=∑nℏωna^n†a^n\hat{H}_{\mathrm{eff}}=\sum_n\hbar\omega_n\hat{a}_n^\dagger\hat{a}_n). Evidently, such a phenomenological approach is not capable of reproducing the dips, illustrating the need for a rigorous QNM quantization to capture the interplay between the Kittel and the higher-order modes.

In Figure 1(c), we present the magnetic Purcell factor for a YIG slab of length LYIG=50 μmL_{\mathrm{YIG}}=50\,\mu\mathrm{m}, comparing the analytical result with a single-mode (Kittel) expansion in the quantum QNM treatment. For this thinner slab, a single-mode expansion already yields very good agreement, indicating that the influence of the higher-order modes becomes negligible for decreasing size of the YIG slab; moreover, although not shown in this plot, we find that the phenomenological quantization protocol suffices to reproduce the analytical result as well. To provide additional insight into the role of the cavity length, we also present the LYIGL_{\mathrm{YIG}}-dependence of Re⁡(ω~n)\operatorname{Re}(\tilde{\omega}_n) and Im⁡(ω~n)\operatorname{Im}(\tilde{\omega}_n) for the Kittel mode in Figure 1(d). We observe that ω~n\tilde{\omega}_n approaches ω~pole\tilde{\omega}_{\mathrm{pole}} as LYIG→0L_{\mathrm{YIG}}\to0, and since the effective wavenumber tends to a finite value in this limit, the Kittel mode profile becomes homogeneous inside the YIG slab. It is worth recalling that we have neglected the exchange interaction, which must be taken into account if the size of the magnetic structure approaches the exchange length, typically within the nanometer range[134, 135], especially for the higher-order modes with non-uniform magnetization profile. The present simple system would provide a useful platform to systematically study the effects of the exchange interaction on the QNMs and their coupling with magnetic QEs at small slab dimensions LYIGL_{\mathrm{YIG}}. However, such an investigation shall be reported elsewhere.

b. Hybrid Magnonic-photonic Slab Resonator An even more enriching and stringent case for the application of our quantum QNM theory concerns the problem of magnon-photon coupling in a hybrid configuration, where the YIG slab is embedded in a half-open photonic cavity comprising a high-index dielectric slab (ε=20\varepsilon=20). The permittivity distribution is defined by

ε(z,ω)={1:z<LairεYIG:z∈[Lair,Lair+LYIG]1:z∈[Lair+LYIG,Lcav]20:z∈[Lcav,Lcav+Lε](51)\varepsilon(z,\omega)= \begin{cases} 1 & : z<L_{\mathrm{air}}\\ \varepsilon_{\mathrm{YIG}} & : z\in[L_{\mathrm{air}},L_{\mathrm{air}}+L_{\mathrm{YIG}}]\\ 1 & : z\in[L_{\mathrm{air}}+L_{\mathrm{YIG}},L_{\mathrm{cav}}]\\ 20 & : z\in[L_{\mathrm{cav}},L_{\mathrm{cav}}+L_{\varepsilon}] \end{cases} \tag*{(51)}

where LairL_{\mathrm{air}} prescribes the placement of the YIG slab within the cavity, LYIG=50 μmL_{\mathrm{YIG}}=50\,\mu\mathrm{m} is the length of the YIG slab, Lcav=40.860 mmL_{\mathrm{cav}}=40.860\,\mathrm{mm} is the length of the cavity, and Lε=2 mmL_{\varepsilon}=2\,\mathrm{mm} is the thickness of the dielectric slab making up the right boundary of the cavity. In the left (right) column of Figure 2, we consider a YIG slab placed at Lair=10.753 mmL_{\mathrm{air}}=10.753\,\mathrm{mm} (Lair=11.674 mmL_{\mathrm{air}}=11.674\,\mathrm{mm}). Our choice of the YIG slab length LYIGL_{\mathrm{YIG}} ensures that higher-order magnonic modes are negligible, while the cavity length LcavL_{\mathrm{cav}} is chosen so that the second-order cavity mode is tuned to the Kittel mode (i.e., two QNMs are considered in our calculations).

Numerical investigation of a 1D YIG slab embedded in a half-open cavity

Figure 2. Numerical investigation of a 1D YIG slab of length LYIG=50 μmL_{\mathrm{YIG}} = 50\,\mu\mathrm{m} embedded in a half-open cavity of length Lcav=40.860 mmL_{\mathrm{cav}} = 40.860\,\mathrm{mm}, magnetized in the positive zz-direction. The permittivity of the cavity material is defined by Eq. (51). The second-order mode of the cavity is tuned to the Kittel mode of the YIG slab, and those two QNMs are retained. Left column: Lair=10.753 mmL_{\mathrm{air}} = 10.753\,\mathrm{mm}. Right column: Lair=11.674 mmL_{\mathrm{air}} = 11.674\,\mathrm{mm}. (a),(f) Real parts of the QNM frequencies, Re⁡(ω~n)\operatorname{Re}(\tilde{\omega}_n), as a function of bias field. (b),(g) Imaginary parts of the QNM frequencies, Im⁡(ω~n)\operatorname{Im}(\tilde{\omega}_n), as a function of bias field. (c),(h) QNM magnetic flux density profiles ∣B~n(R)(z)∣|\tilde{\mathbf{B}}_n^{(R)}(z)| at the bias fields H0H_0 marked by the dashed lines in the panels above. (d),(i) Frequency-dependent magnetic Purcell factor at the QE position marked in (c),(h), obtained from the full quantum QNM theory, the phenomenological QNM quantization scheme and the analytical treatment.

Figure 2(a),(Figure 2f) present the H0H_0-dependence of the QNM spectral positions, Re⁡(ω~n)\operatorname{Re}(\tilde{\omega}_n), evidencing level attraction (repulsion) for the left (right) configuration and signifying a transition from dissipative to coherent magnon-photon coupling. In Figure 2(b),(Figure 2g) we show the H0H_0-dependence of the QNM decay rates Im⁡(ω~n)\operatorname{Im}(\widetilde{\omega}_n). The right configuration exhibits damping attraction, a typical signature of coherent coupling. For the left configuration, there is still a damping attraction, however, the rates do not cross. This is in contrast to the damping repulsion observed in the phenomenological model of dissipative magnon-photon coupling employed in [25, 27], indicating an intermediate regime which the QNM theory is capable of describing from first principles. By further tuning the parameters to the transition point between these regimes of dissipative and coherent coupling, an exceptional point can be located, although we do not discuss it here. Next, Figure 2(c),(Figure 2h) present the QNM magnetic flux density profiles ∣B~n(z)∣|\widetilde{\mathbf{B}}_n(z)| for bias fields marked by the gray dashed lines in the panels above, namely H0=1.4844 MsH_0=1.4844\,M_s (H0=1.48725 MsH_0=1.48725\,M_s) for the left (right) configuration (i.e., where the damping rates cross). The two QNMs have very similar mode profiles in both cases, particularly in the case of the right configuration where coherent coupling leads to the hybridization of the modes. Note that the YIG slab resides at almost the same location relative to the cavity standing-wave mode in the two configurations, close to a node, which explains the sensitivity of the coupling mechanism to the slab position.

Figure 2(d),(Figure 2g) exhibit the magnetic Purcell factor for a QE position indicated by the grey dashed line in Figure 2(c),(Figure 2h), namely zQE=Lcav/2=20.430 mmz_{\mathrm{QE}}=L_{\mathrm{cav}}/2=20.430\ \mathrm{mm}. For both configurations, the quantum QNM theory yields excellent agreement with the semi-analytical result obtained via the Green's function expansion in Eq. (24). In each case, we observe an apparent spectral splitting of the modes. For the left configuration, this splitting arises from the dissipative coupling of the modes, where the second mode contributes with opposite sign. This becomes especially clear upon a comparison with the predictions of the phenomenological QNM quantization scheme, which overestimates the Purcell factor and fails to reproduce the split-like lineshape. This is because the system operates in proximity to an exceptional point in its parameter space, where the two modal contributions separately diverge, while their sum remains finite[136]. Anomalous spontaneous emission spectra at exceptional points with a double peak structure have previously been reported for whispering-gallery-mode cavities[137]. For illustration, we plot the two QNM contributions to the Green's function in Eq. (24) individually, showing that one of them is completely negative. For the right configuration, the spectral splitting arises from the strong coherent coupling; in this case, the phenomenological QNM quantization correctly reproduces a spectral splitting, but not of the correct magnitude due to the additional dissipative coupling present in the system.

Since our QNM quantization methodology provides a modal expansion of the electromagnetic field operators, it allows us to study the dynamics of the magnetic field in a spatially- and temporally-resolved manner, providing a powerful complement to the more common spatiocspectral analyses in the literature. As a non-trivial example, Figure 3(a),(Figure 3b) present the spatiotemporal evolution of the magnetic flux-density amplitude ⟨B~−(z,t)B~+(z,t)⟩12\langle\widetilde{\mathbf{B}}^{-}(z,t)\widetilde{\mathbf{B}}^{+}(z,t)\rangle^{\frac{1}{2}} for the decay of the initial two-QNM Bell state ∣Ψ−⟩=(∣0⟩∣1⟩−∣1⟩∣0⟩)/2|\Psi^{-}\rangle=(|0\rangle|1\rangle-|1\rangle|0\rangle)/\sqrt{2}. For the left configuration, both the magnonic and photonic parts of the modes decay exponentially, whereas for the right configuration, coherent Rabi oscillations between the magnonic and photonic parts can be observed, enabled by the strong coherent magnon-photon coupling. Note that the QNMs are already the hybridized modes (i.e., superpositions of the magnonic and photonic parts), so an initial excitation of a single symmetrized mode would decay with much less pronounced Rabi oscillations; the Bell state is chosen precisely because it is a superposition in the two-mode QNM Fock basis. For the same reason, the Rabi oscillations are not simply a consequence of the off-diagonal coupling terms χnn′±\chi_{nn'}^{\pm}, but would also appear without them.

Spatiotemporally-resolved magnetic flux-density amplitude for the decay of the two-QNM Bell state

Figure 3. Spatiotemporally-resolved magnetic flux-density amplitude for the decay of the two-QNM Bell state ∣Ψ−⟩=(∣0⟩∣1⟩−∣1⟩∣0⟩)/2|\Psi^{-}\rangle= (|0\rangle|1\rangle- |1\rangle|0\rangle)/\sqrt{2} in the same cavity configuration of Figure 2 with (a) Lair=10.753L_{\mathrm{air}} = 10.753 mm and (b) Lair=11.674L_{\mathrm{air}} = 11.674 mm.

These 1D examples demonstrate the versatility of our quantum QNM methodology, capturing both coherent and dissipative regimes of magnon-photon coupling from first principles, and revealing the limitations of purely phenomenological quantization approaches in describing cavity-QE interactions. The elucidation of the quantum field dynamics in both real space and real time constitutes a particularly valuable capability, and offers a truly four-dimensional characterization of quantum magnonic devices.

3D Spherical Systems

a. Geometry and Qubit Couplings In order to demonstrate the predictive capabilities of our quantum QNM theory in a 3D context, we explore a spherically symmetric, hybrid magnonic-photonic resonator and its coupling to FQss and SQss. The device is schematized in Figure 4. Specifically, we consider a spherical YIG core of radius rYIG=550 μmr_{\mathrm{YIG}}=550\,\mu\mathrm{m} surrounded by a yttrium orthosilicate (Y2SiO5\mathrm{Y}_2\mathrm{SiO}_5, YSO) shell of outer radius R=4615 μmR=4615\,\mu\mathrm{m}, with a 10 μm10\,\mu\mathrm{m} air gap between them. Two FQss (characterized by the persistent current Ip=800 nAI_{\mathrm{p}}=800\,\mathrm{nA} and loop radius RFQ=10 μmR_{\mathrm{FQ}}=10\,\mu\mathrm{m}) reside on opposite sides of the YSO-air interface, with their loops oriented parallel to the direction of the bias field H0=H0ez\mathbf{H}_0=H_0\mathbf{e}_z (H0≈H0≡NzMsH_0\approx H_0\equiv N_zM_s, Nz=1/3N_z=1/3) with nFQ=ey\mathbf{n}_{\mathrm{FQ}}=\mathbf{e}_y. YSO can host rare-earth ions that function as SQss, and has previously been used as a substrate to couple SQss to a superconducting resonator[138]. For our numerical example, we choose an ensemble of Ce3+ ⁣:Y2SiO5\mathrm{Ce}^{3+}\!:\mathrm{Y}_2\mathrm{SiO}_5 with N=3⋅107N=3\cdot10^7 (corresponding to a doping of ≈0.32%\approx0.32\% of Ce1 sites in 1 μm31\,\mu\mathrm{m}^3, assuming the Ce2 concentration to be negligible[139]), described as spin-1/2 systems with no zero field splitting and an effective gg-tensor of g=diag⁡(0.545,1.609,2.319)g=\operatorname{diag}(0.545,1.609,2.319) [140]. The ensembles are assumed to be located inside the FQ loops, supporting a collective N\sqrt{N} scaling of the coupling constants. An important feature of our configuration is that, since the SQ ensembles reside inside the FQ loops, their direct interaction with the resonator modes is negligible and the dominant coupling is with the local FQ; the latter thus acts as a quantum interface to the resonator itself.

Schematic representation of the investigated spherically symmetric, core-shell, hybrid photonic-magnonic resonator device

Figure 4. Schematic representation of the investigated spherically symmetric, core-shell, hybrid photonic-magnonic resonator device, comprising a YIG core and concentric YSO shell, together with embedded FQs and SQ ensembles. Relevant geometrical and physical parameters are indicated.

To describe the quantum dynamics arising from the QNM-FQ and FQ-SQ interactions, we now formulate the Hamiltonian of the system. Beginning with the FQ, its Hamiltonian in the persistent current basis is given by H^FQ=−(ϵτx+Δτz)/2\hat{H}_{\mathrm{FQ}} = -(\epsilon\tau_x + \Delta\tau_z)/2, where we denote the FQ Pauli matrices by τi\tau_i and ϵ=2Ip(Φext−Φ0/2)\epsilon= 2I_{\mathrm{p}}(\Phi_{\mathrm{ext}}-\Phi_0/2), where Φext\Phi_{\mathrm{ext}} is the external magnetic flux through the FQ and Φ0=h/(2e)\Phi_0 = h/(2e) is the magnetic flux quantum[37, 65]. Decomposing the external magnetic flux into a static contribution, which maintains the qubit at the sweet spot (ϵ=0\epsilon= 0), and the resonator flux[53]

Φ^res=∫AFQdA B^(r)⋅nFQ≈AFQ(B^(rFQ)⋅nFQ),(52)\hat{\Phi}_{\mathrm{res}} = \int_{A_{\mathrm{FQ}}} \mathrm{d}A\,\hat{\mathbf{B}}(\mathbf{r}) \cdot\mathbf{n}_{\mathrm{FQ}} \approx A_{\mathrm{FQ}}\bigl(\hat{\mathbf{B}}(\mathbf{r}_{\mathrm{FQ}}) \cdot\mathbf{n}_{\mathrm{FQ}}\bigr), \tag*{(52)}

leads to a Hamiltonian for the FQ coupled to the QNMs of the form H^FQ=−Δ2τx+∑ngnFQa^nτz+gnFQ∗a^n†τz\hat{H}_{\mathrm{FQ}} = -\frac{\Delta}{2}\tau_x + \sum_n g_n^{\mathrm{FQ}}\hat{a}_n\tau_z + g_n^{\mathrm{FQ}*}\hat{a}_n^\dagger\tau_z, where gnFQ=−IpAFQ(B~(rFQ)⋅nFQ)g_n^{\mathrm{FQ}} = -I_{\mathrm{p}}A_{\mathrm{FQ}}(\tilde{\mathbf{B}}(\mathbf{r}_{\mathrm{FQ}}) \cdot\mathbf{n}_{\mathrm{FQ}}). In principle, the integration in Eq. (52) should be performed over the FQ

surface AFQA_{\mathrm{FQ}}, with nFQ\mathbf{n}_{\mathrm{FQ}} a unit normal vector to that surface. The simple approximation used here is justified only if the symmetrized resonator field is nearly uniform over AFQA_{\mathrm{FQ}}, which holds here as the magnetic resonator is much larger than the FQ and is placed in close proximity to it.

We model the interaction of a spin-1/2 SQ (neglecting zero-field splitting) with the resonator magnetic field via the Hamiltonian H^SQ=μBB0(rSQ)gzσz/2+μBB^(rSQ)gσ/2\hat{H}_{\mathrm{SQ}} = \mu_{\mathrm{B}}B_{0}(\mathbf{r}_{\mathrm{SQ}})g_{z}\sigma_{z}/2 + \mu_{\mathrm{B}}\hat{\mathbf{B}}(\mathbf{r}_{\mathrm{SQ}})\mathbf{g}\boldsymbol{\sigma}/2, with the gg-tensor g=diag⁡(gx,gy,gz)\mathbf{g} = \operatorname{diag}(g_{x},g_{y},g_{z}) (diagonal in a system of principal axes) and the vector of Pauli matrices σ=(σx,σy,σz)T\boldsymbol{\sigma} = (\sigma_{x},\sigma_{y},\sigma_{z})^{\mathrm{T}}. The first term describes the Zeeman splitting in the static bias field B0B_{0}, and the second term the coupling to the resonator. Approximating the FQ as a circular loop of radius RFQR_{\mathrm{FQ}} with the SQ at its center, the magnetic flux density generated by the persistent current is B^FQ(rSQ)=μ0I^PnFQ/2RFQ=μ0IPnFQτz/2RFQ\hat{\mathbf{B}}_{\mathrm{FQ}}(\mathbf{r}_{\mathrm{SQ}}) = \mu_{0}\hat{I}_{\mathrm{P}}\mathbf{n}_{\mathrm{FQ}}/2R_{\mathrm{FQ}} = \mu_{0}I_{\mathrm{P}}\mathbf{n}_{\mathrm{FQ}}\tau_{z}/2R_{\mathrm{FQ}}, and the FQ-SQ coupling Hamiltonian is given by H^FQ+SQ=μBB^FQ(rSQ)gσ/2\hat{H}_{\mathrm{FQ+SQ}} = \mu_{\mathrm{B}}\hat{\mathbf{B}}_{\mathrm{FQ}}(\mathbf{r}_{\mathrm{SQ}})\mathbf{g}\boldsymbol{\sigma}/2. Collecting these various contributions, applying the unitary transformation U^=exp⁡(−iπτy/4)\hat{U} = \exp(-i\pi\tau_{y}/4) to transform the Hamiltonian into the FQ eigenbasis, and using the rotating wave approximation (i.e., keeping only excitation-conserving terms)[66] leads to an effective Hamiltonian for the system comprising a FQ-SQ pair and the QNMs:

H^=+Δ2τz+μ0B0gz2σz+ℏχnn′a^n†a^n′+gnFQa^nτ++gnFQ∗a^n†τ−+gnSQa^nσ++gnSQ∗a^n†σ−+gnSQ,za^nσz+gnSQ,z∗a^n†σz+gFQ−SQτ+σ−+gFQ−SQ∗τ−σ++gFQ−SQzτ+σz+gFQ−SQz∗τ−σz,(53)\hat{H} = +\frac{\Delta}{2}\tau_{z} + \frac{\mu_{0}B_{0}g_{z}}{2}\sigma_{z} + \hbar\chi_{nn'}\hat{a}_{n}^{\dagger}\hat{a}_{n'} + g_{n}^{\mathrm{FQ}}\hat{a}_{n}\tau^{+} + g_{n}^{\mathrm{FQ}*}\hat{a}_{n}^{\dagger}\tau^{-} + g_{n}^{\mathrm{SQ}}\hat{a}_{n}\sigma^{+} + g_{n}^{\mathrm{SQ}*}\hat{a}_{n}^{\dagger}\sigma^{-} + g_{n}^{\mathrm{SQ},z}\hat{a}_{n}\sigma_{z} + g_{n}^{\mathrm{SQ},z*}\hat{a}_{n}^{\dagger}\sigma_{z} + g_{\mathrm{FQ-SQ}}\tau^{+}\sigma^{-} + g_{\mathrm{FQ-SQ}}^{*}\tau^{-}\sigma^{+} + g_{\mathrm{FQ-SQ}}^{z}\tau^{+}\sigma_{z} + g_{\mathrm{FQ-SQ}}^{z*}\tau^{-}\sigma_{z}, \tag*{(53)}

with the transverse and longitudinal SQ-QNM coupling constants gnSQ=μB(gxB~n,xsym(rSQ)−igyB~n,ysym(rSQ))/2g_{n}^{\mathrm{SQ}} = \mu_{\mathrm{B}}(g_{x}\widetilde{B}_{n,x}^{\mathrm{sym}}(\mathbf{r}_{\mathrm{SQ}}) - ig_{y}\widetilde{B}_{n,y}^{\mathrm{sym}}(\mathbf{r}_{\mathrm{SQ}}))/2, gnSQ,z=μBgzB~n,zsym/2g_{n}^{\mathrm{SQ},z} = \mu_{\mathrm{B}}g_{z}\widetilde{B}_{n,z}^{\mathrm{sym}}/2, and the FQ-SQ coupling constants gFQ−SQ=μBμ0Ip(gxnFQ,x+igynFQ,y)/4RFQg_{\mathrm{FQ-SQ}} = \mu_{\mathrm{B}}\mu_{0}I_{\mathrm{p}}(g_{x}n_{\mathrm{FQ},x} + ig_{y}n_{\mathrm{FQ},y})/4R_{\mathrm{FQ}}, gFQ−SQz=μBμ0IpgznFQ,z/4RFQg_{\mathrm{FQ-SQ}}^{z} = \mu_{\mathrm{B}}\mu_{0}I_{\mathrm{p}}g_{z}n_{\mathrm{FQ},z}/4R_{\mathrm{FQ}}. To account for qubit decay and pure dephasing effects, we supplement the quantum master equation with Lindblad dissipators of the form D[c^]=c^ρ^c^†−1/2{c^†c^,ρ^}\mathcal{D}[\hat{c}] = \hat{c}\hat{\rho}\hat{c}^{\dagger} - 1/2\{\hat{c}^{\dagger}\hat{c},\hat{\rho}\}, where c^=γσ−\hat{c} = \sqrt{\gamma}\sigma^{-} and γϕσz\sqrt{\gamma_{\phi}}\sigma_{z} respectively, with γ=1/T1\gamma= 1/T_{1} and γϕ=1/T2−1/(2T1)\gamma_{\phi} = 1/T_{2} - 1/(2T_{1}), in which T1T_{1} and T2T_{2} are the corresponding relaxation times. The FQs are treated analogously.

b. Spatiospectral Characterization To calculate the QNMs and scattering cross-sections of the hybrid resonator described above, we employ the generalized Mie theory for anisotropic magnetic spheres[74], in conjunction with a semi-analytical transfer matrix approach for multilayer systems[73]. Note that in the basis of vector spherical harmonics, the anisotropy of the permeability results in a mixing of transverse-electric (TElm\mathrm{TE}_{lm}) and transverse-magnetic (TMlm\mathrm{TM}_{lm}) modes with different ll; only the azimuthal index mm remains a good quantum number.

Figure 5(a) shows the H0H_{0}- and ω\omega-dependent scattering efficiency QscaQ_{\mathrm{sca}}, assuming an incident electromagnetic plane wave propagating along the zz-direction. We observe a marked anticrossing of two modes; the H0H_{0}-independent branch corresponds to the lowest-order TE1m\mathrm{TE}_{1m} photonic mode of the YSO shell, while the H0H_{0}-dependent one corresponds to the magnonic Kittel mode of the YIG sphere, predominantly TE11\mathrm{TE}_{11} in character. Additionally, we plotted the H0H_{0}-dependent Zeeman transition frequency of the SQs, determined by the gg-tensor component gz=2.319,g_{z} = 2.319, and marked the bias field H0=1.614H_0 = 1.614 chosen for the preceding calculations where the SQs are tuned to the lower hybridized mode. Apart from the Kittel and photonic modes, three additional H0H_0-dependent features can be observed in the scattering efficiency map, manifesting higher-order magnonic QNMs. The middle one progressively weakens with increasing bias field and becomes negligible for H0=1.614H_0 = 1.614; the other two QNMs, indicated by black arrows and corresponding spatial field profiles ∣H~n(R)(r)∣|\widetilde{H}_n^{(\mathrm{R})}(\mathbf{r})| in Figure 5(a), have a core-localized nature with predominantly (TE11+TM21+TE31)(\mathrm{TE}_{11} + \mathrm{TM}_{21} + \mathrm{TE}_{31}) character (i.e., mixed spherical multipoles), barely coupling to the photonic mode due to their inhomogeneous radial profile in the TE11\mathrm{TE}_{11}-component. Figure 5(b) exhibits the magnetic field strength profiles pertaining to the two hybridized magnon-photon modes, with complex frequencies ω~n/(2π)=(9.1711−0.1603i)\widetilde{\omega}_n/(2\pi) = (9.1711 - 0.1603\mathrm{i}) GHz (upper plot) and (9.7941−0.3476i)(9.7941 - 0.3476\mathrm{i}) GHz (lower plot) for H0=1.614H_0 = 1.614. The lower-frequency mode exhibits a stronger magnonic character and is mostly homogeneous within the YIG sphere, reflecting its origin in the Kittel mode. Near the FMR, one of the eigenvalues of the permeability becomes negative (Re⁡(μxx(ω)+κ(ω))<0\operatorname{Re}(\mu_{xx}(\omega) + \kappa(\omega)) < 0), giving rise to a surface-localized field profile analogous to a plasmonic resonance, yet not a true surface plasmon of the kind observed in metallic nanostructures. The higher-frequency mode possesses more photonic character, but also shows a hotspot inside the YIG sphere rather than surface field localization.

Spatiopsectral characterization of the hybrid photonic-magnonic resonator

Figure 5. Spatiopspectral characterization of the hybrid photonic-magnonic resonator schematized in Fig. 4. (a) Scattering efficiency QscaQ_{\mathrm{sca}} as a function of bias field H0H_0 and frequency ω/(2π)\omega/(2\pi) for an incident, linearly polarized plane wave propagating along the positive zz-direction. Additionally, the H0H_0-dependent SQ transition frequency and the magnetic field strength profiles ∣H~n(R)(r)∣|\widetilde{H}_{n}^{(R)}(\mathbf{r})| of two higher-order magnonic modes with complex frequencies ω~n/(2π)=(8.5159−0.0026i) GHz\widetilde{\omega}_{n}/(2\pi)=(8.5159-0.0026i)\,\mathrm{GHz} (lower left) and (9.8244−0.0035i) GHz(9.8244-0.0035i)\,\mathrm{GHz} (upper right) for the bias field H0=1.614 mTH_0=1.614\,\mathrm{mT} (marked by the white dashed line) are shown. (b) Spatial magnetic field strength profiles ∣H~n(R)∣|\widetilde{H}_{n}^{(R)}| of the two hybridized magnon-photon modes with complex frequencies ω~n/(2π)=(9.1711−0.1603i) GHz\widetilde{\omega}_{n}/(2\pi)=(9.1711-0.1603i)\,\mathrm{GHz} (upper panel) and (9.7941−0.3476i) GHz(9.7941-0.3476i)\,\mathrm{GHz} (lower panel). (c) Frequency-dependent magnetic Purcell factors of the FQs calculated from the quantum QNM theory incorporating the four QNMs from (a) and (b), compared with the semi-analytical result obtained from the QNM expansion of the Green’s function. Additionally, the semi-analytical result for the same hybrid resonator device, but with a YSO shell of radius R=1000 μmR=1000\,\mu\mathrm{m}, is shown.

Similar to § IV A, we further characterize our hybrid resonator via a calculation of the magnetic Purcell factor, plotted for the FQs in Figure 5(c). Our results, obtained from a four-mode QNM expansion in the full quantum treatment, stand in good agreement with the semi-analytical prediction via the Green’s function. Notably, the hybridized magnon-photon modes are much broader than the higher-order magnonic modes and have much lower Purcell factors (∼0.88×106\sim0.88 \times10^6 and 0.21×1060.21 \times10^6 for the higher-order modes), attributable to the hybridization with the photonic mode. For comparison, the same (semi-analytical) calculation with a thinner YSO shell (R=1000 μmR = 1000\ \mu\mathrm{m}), which supports no photonic mode in the considered frequency range, shows that the unhybridized Kittel mode has a much larger Purcell factor. Interestingly however, the lower-frequency, higher-order magnon mode offers a slightly larger Purcell factor and a higher quality factor, so that if there are relevant figures of merit in given device application, the higher-order magnonic modes might actually be more suitable.

Magnon-mediated Qubit Synchronization and Entanglement Dynamics

Turning to the temporal evolution of the system, Figure 6(a) exhibits the qubit dynamics assuming one SQ initially excited and the FQ transition tuned to the SQs. Each SQ undergoes strong coupling with its local FQ and Rabi oscillations can be observed. Through their interaction with a common resonator, excitation is transferred to the opposite FQ-SQ pair, where similar Rabi oscillations develop. After about four periods, both the SQs and the FQs are synchronized in their oscillations. This phenomenon is in accordance with noise-induced quantum synchronization in qubit chains [141, 142], where the application of noise to the transition frequency of the middle qubit in the chain creates a decoherence-free subspace (i.e., a decoupling from the environment) that sustains the qubit oscillations. In our case, the middle qubit is replaced by the resonator, the noise is provided by the lossy mode, and we utilize different types of qubits (FQs and SQs). Even though the mode exhibits loss rather than just noise and the qubits couple to it, the amount of excitation present in the system does not further decrease once the qubits are synchronized (i.e., once the pair is synchronized, the lossy mode does not drain excitation).

Qubit dynamics in the hybrid photonic-magnonic resonator

Figure 6. Qubit dynamics in the hybrid photonic-magnonic resonator schematized in Figure 4. (a),(b) SQ and FQ dynamics for one initially excited SQ in the case of no qubit damping or dephasing (a) and for relaxation times T1=1.47 μsT_1 = 1.47\,\mu\mathrm{s} and T2=1.45 μsT_2 = 1.45\,\mu\mathrm{s} (b). (c),(d) Time-dependent concurrences for two-qubit subsystems C(ρAB)C(\rho_{AB}) in the case of no qubit damping or dephasing (c), and for damped qubits (d). (e) Real part of the two-SQ reduced density-matrix elements Re⁡((ρSQ1,SQ2)ij)\operatorname{Re}((\rho_{\mathrm{SQ1,SQ2}})_{ij}) in the case of no qubit damping or dephasing, at times of maximum concurrence C(ρSQ1,SQ2)C(\rho_{\mathrm{SQ1,SQ2}}).

Such behavior is not expected in the presence of qubit relaxation effects however. In Figure 6(b), we present the qubit dynamics in which both the SQs and FQs possess the relatively short relaxation times T1=1.47 μsT_1 = 1.47\ \mu\mathrm{s} and T2=1.45 μsT_2 = 1.45\ \mu\mathrm{s}, taken from recently reported experimental results for zero-flux-biased FQs [61]. Note that rare-earth SQs usually have better T1,2T_{1,2} values, but since the decoherence of the FQs is the limiting factor here, we simply choose the same values for the SQs. The synchronization and the transient time to reach the quasi-steady state are not affected by the presence of qubit damping or dephasing, the excitation present in the system only decays over time according to the timescale T1T_1; such predictions also remain valid for other combinations of SQ/FQ T1,2T_{1,2} values.

Deeper insight into the quantum dynamics of the multiqubit-resonator system can be acquired by exploring the time-dependent entanglement of the various two-qubit subsystems. To this end, we quantify the degree of entanglement between qubits AA and BB via the concurrence C(ρAB)\mathcal{C}(\rho_{AB}) [143], shown in Figure 6(c) for the idealized case in which there is no qubit damping or dephasing. For the SQ-SQ and FQ-FQ subsystems, the concurrences follow the excitation oscillations with maxima of max⁡(C(ρSQ1,SQ2(t)))≈max⁡(C(ρFQ1,FQ2(t)))≈0.5\max(\mathcal{C}(\rho_{\mathrm{SQ1,SQ2}}(t))) \approx\max(\mathcal{C}(\rho_{\mathrm{FQ1,FQ2}}(t))) \approx0.5, implying that the two SQs or FQs possess the maximum entanglement allowed by their reduced excitation probabilities in Figure 6(a). The FQ-SQ concurrences oscillate at double that frequency, with a phase shift, and have max⁡(C(ρSQ1,FQ1/2(t)))≈max⁡(C(ρSQ1/2,FQ1(t)))≈0.25\max(\mathcal{C}(\rho_{\mathrm{SQ1,FQ1/2}}(t))) \approx\max(\mathcal{C}(\rho_{\mathrm{SQ1/2,FQ1}}(t))) \approx0.25 at the times where FQs and SQs have equal excitation probability. Each SQ is thus entangled with the FQ on both the same side and the opposite side. The real parts of the two-SQ reduced density-matrix elements Re⁡((ρSQ1,SQ2)ij)\operatorname{Re}((\rho_{\mathrm{SQ1,SQ2}})_{ij}) are shown in Figure 6(e) for a temporal instant of maximum concurrence, resembling the state ρ=p1∣Ψ−⟩⟨Ψ−∣+p2∣00⟩⟨00∣+p3∣Ψ+⟩⟨Ψ+∣+p4∣00⟩⟨00∣\rho= p_1|\Psi^-\rangle\langle\Psi^-| + p_2|00\rangle\langle00| + p_3|\Psi^+\rangle\langle\Psi^+| + p_4|00\rangle\langle00|, in which p2=p3=1/2p_2 = p_3 = 1/2 and p1=p4=0p_1 = p_4 = 0, with a fidelity of F≈99.94%F \approx99.94\% (adopting the convention used in QuTiP [144]). The same result is found for ρFQ1,FQ2\rho_{\mathrm{FQ1,FQ2}}. Notably, this is a different state from that of the five-qubit chain investigated in [142], where p1=1/2p_1 = 1/2, p2=1/3p_2 = 1/3, p3=1/6p_3 = 1/6 and p4=0p_4 = 0, though in both cases, they are maximally entangled mixed states[145].

At times where the FQs and SQs have equal excitation probability, the FQ-SQ subsystems are in the states ρFQ1,SQ1=ρFQ2,SQ2=0.25∣Ψ−⟩⟨Ψ−∣+0.75∣00⟩⟨00∣\rho_{\mathrm{FQ1,SQ1}}=\rho_{\mathrm{FQ2,SQ2}}=0.25|\Psi^{-}\rangle\langle\Psi^{-}|+0.75|00\rangle\langle00| and ρFQ1,SQ2=ρFQ2,SQ1=0.25∣Ψ+⟩⟨Ψ+∣+0.75∣00⟩⟨00∣\rho_{\mathrm{FQ1,SQ2}}=\rho_{\mathrm{FQ2,SQ1}}=0.25|\Psi^{+}\rangle\langle\Psi^{+}|+0.75|00\rangle\langle00|, with ∣Ψ±⟩=(∣01⟩±i∣10⟩)/2|\Psi^{\pm}\rangle=(|01\rangle\pm i|10\rangle)/\sqrt{2} (i.e., also maximally entangled mixed states).

The impact of decay and pure dephasing are evidenced in Figure 6(d), which presents the same concurrences in the case of qubits with relaxation times of T1=1.47 μsT_1=1.47\,\mu\mathrm{s} and T2=1.45 μsT_2=1.45\,\mu\mathrm{s}. The concurrences follow the same qualitative behavior, but with maxima that follow a decaying envelope as they are bounded by the reduced excitation probabilities. Additionally, we plot the combined SQ excitation probabilities pSQs=⟨σSQ1+σSQ1−+σSQ2+σSQ2−⟩p_{\mathrm{SQs}}=\langle\sigma_{\mathrm{SQ1}}^{+}\sigma_{\mathrm{SQ1}}^{-}+\sigma_{\mathrm{SQ2}}^{+}\sigma_{\mathrm{SQ2}}^{-}\rangle, showing that the concurrence C(ρSQ1,SQ2)C(\rho_{\mathrm{SQ1,SQ2}}) does not fully reach its upper bound at the times of its local maxima (i.e., the fidelity of the state is reduced). However, for t>400 nst>400\,\mathrm{ns}, fidelity values of >99.3%>99.3\% are still achieved for the SQ-SQ subsystem and ρ(t)=pSQs(t)∣Ψ+⟩⟨Ψ+∣+(1−pSQs(t))∣00⟩⟨00∣\rho(t)=p_{\mathrm{SQs}}(t)|\Psi^{+}\rangle\langle\Psi^{+}|+(1-p_{\mathrm{SQs}}(t))|00\rangle\langle00|.

Whilst not discussed in further detail here, two general conclusions can be made from our simulated quantum dynamics. First, the loss rate of the resonator mode affects the transition time required for the system to attain a quasistationary synchronized state and an optimum appears to exist (i.e., for too large or too small loss rates, the transient time increases). However, the amount of excitation present in the system in its quasistationary state does not depend on the loss rate of the mode. The same is also true for the FQ-resonator coupling. Second, detuning the FQs from the SQs leaves the maximally entangled mixed state of the SQ subsystem intact at the times of maximum concurrence, but reduces the fraction of excitation that is transferred to the FQs. This is visible for the undamped case with ωFQ−ωSQ=1.479gFQ−SQ\omega_{\mathrm{FQ}}-\omega_{\mathrm{SQ}}=1.479g_{\mathrm{FQ-SQ}} in Figure 7(a); the FQs reach a maximally entangled mixed state of lower excitation, and the transient time for synchronization is lengthened. Increasing the detuning further suppresses the transfer until the SQs retain nearly all of the excitation, so that the system settles into a long-lived, weakly oscillating state without continuous driving. This regime is illustrated in Figure 7(b) for damped qubits (T1=40 μsT_1=40\,\mu\mathrm{s} and T2=85 μsT_2=85\,\mu\mathrm{s} for the FQs[146] and T1=T2=3 msT_1=T_2=3\,\mathrm{ms} for the SQs) at ωFQ−ωSQ=7.393gFQ−SQ\omega_{\mathrm{FQ}}-\omega_{\mathrm{SQ}}=7.393g_{\mathrm{FQ-SQ}}; the longer coherence times were chosen so that the extended transient remains observable, given that the transient time also grows with the detuning. The SQ subsystem reaches the maximally entangled mixed state with a fidelity of approximately 99.85%99.85\%. Note that for rare-earth ions in YSO, the superhyperfine interaction with neighboring nuclear spins could limit the achievable coherence times in practice, but nevertheless, optimized host crystals have demonstrated T2T_2 values up to approximately 23 ms23\,\mathrm{ms}[147].

Numerical results for detuned qubit dynamics

Figure 7. Numerical results for detuned qubit dynamics, assuming one initially excited SQ. (a) FQs are detuned to ωFQ=1.0005ωSQ\omega_{\mathrm{FQ}} = 1.0005\omega_{\mathrm{SQ}} (ωFQ−ωSQ=1.479gFQ−SQ\omega_{\mathrm{FQ}}-\omega_{\mathrm{SQ}} = 1.479g_{\mathrm{FQ-SQ}}) in the absence of decay and pure dephasing. (b) FQs are detuned to ωFQ=1.0025ωSQ\omega_{\mathrm{FQ}} = 1.0025\omega_{\mathrm{SQ}} (ωFQ−ωSQ=7.393gFQ−SQ\omega_{\mathrm{FQ}}-\omega_{\mathrm{SQ}} = 7.393g_{\mathrm{FQ-SQ}}), in the presence of decay and pure dephasing with T1=40 μsT_1 = 40\,\mu\mathrm{s} and T2=85 μsT_2 = 85\,\mu\mathrm{s} (FQs), and T1=T2=3 msT_1 = T_2 = 3\,\mathrm{ms} (SQs).

d. Finite-temperature Effects Finally, we demonstrate the capability of our quantum QNM approach to capture finite-temperature effects in the hybrid

magnonic-photonic resonator device of Figure 4, following the formalism in § III C. In Figure 8(a), we present the radial dependence of the magnetic-field thermal correlation function ⟨B^(+)(r)B^(−)(r)⟩T\langle\hat{B}^{(+)}(\mathbf{r})\hat{B}^{(-)}(\mathbf{r})\rangle_T (see Eq. (43)) in the xyxy plane, for T=450 mKT=450\,\mathrm{mK}, where the frequency integrals were computed for ω/(2π)∈[0 GHz,10.5 GHz]\omega/(2\pi)\in[0\,\mathrm{GHz},10.5\,\mathrm{GHz}]. We compare the result from the QNM quantization with the semi-analytical one obtained from the imaginary part of the Green’s function, evidencing excellent agreement. Figure 8(b) shows the qubit dynamics (T1=1.47 μsT_1=1.47\,\mu\mathrm{s} and T2=1.45 μsT_2=1.45\,\mu\mathrm{s}) in an environment with T=200 mKT=200\,\mathrm{mK}. The synchronization itself is not affected, however there is additional dephasing and the system relaxes towards a non-oscillating thermal steady state. With increasing temperature, the system relaxes more rapidly towards its steady state, as shown in Figure 8(c) for T=700 mKT = 700\,\mathrm{mK}. As only barely half a synchronized oscillation can be observed, the critical temperature of this device (i.e., the temperature above which the desired functionality can no longer be observed) is approximately Tcrit=700 mKT_{\mathrm{crit}} = 700\,\mathrm{mK}.

Numerical results for the hybrid photonic-magnonic resonator at finite temperatures

Figure 8. Numerical results for the hybrid photonic-magnonic resonator schematized in Figure 4 and characterized in Figure 5 at finite temperatures. (a) Comparison of spatially resolved magnetic-field thermal correlation functions at T=450 mKT = 450\,\mathrm{mK}, calculated from the quantum QNM theory and semi-analytical approach. (b),(c) Qubit dynamics of one initially excited SQ at T=200 mKT = 200\,\mathrm{mK} (b) and T=700 mKT = 700\,\mathrm{mK} (c), assuming T1=1.47 μsT_1 = 1.47\,\mu\mathrm{s} and T2=1.45 μsT_2 = 1.45\,\mu\mathrm{s} for all qubits.

Conclusion and Outlook

In this work, we have introduced a finite-temperature quantum QNM theory for 3D, dissipative and dispersive, linear, cavity magnonic systems and their interaction with spin and superconducting qubits. At the level of the classical theory, we have generalized the auxiliary-field eigenvalue methodology for the computation of the classical QNMs of gyrotropic magnetic resonators to include both the exchange interaction and anisotropy effects, and discussed QNM expansions for the Green’s function and the calculation of other important quantities. Invoking the theory of MQED for non-reciprocal media, we have then generalized our PML-based QNM quantization approach of Ref. 108 for lossy magnonic resonators in arbitrary (linear) dielectric environments, enabling the prediction of all magnon-photon couplings and hybridizations from first principles, revealing the quantitative limitations of purely phenomenological quantization protocols, and offering unprecedented insight into the quantum dynamics of magnon-qubit interactions in hybrid quantum architectures. Moreover, we have taken into account finite-temperature effects, which play a more significant role in cavity magnonic systems, relative to dielectric or plasmonic ones, by virtue of their lower (RF and MW) excitation frequencies.

The predictive capabilities of our methodology have been demonstrated via application to a 1D, YIG slab resonator, capturing the accumulation of higher-order magnonic modes at the pole of the effective permeability and their interference with the Kittel mode. When the same slab is placed inside a photonic cavity, our method has enabled an essentially exact prediction of the continuous tuning, via sub-millimeter geometrical changes, from dissipative to coherent magnon–photon coupling, with the associated Purcell spectra and spatiotemporally-resolved field dynamics ascertained. Indeed, even in the proximity of an exceptional point, where two modal contributions to the Green’s function themselves diverge and a phenomenological quantization scheme fails completely, the quantum QNM method is able to correctly describe the nature of the system. Motivated by recent progress in directly interfacing SQs or FQs and magnonic resonators[61, 62, 63, 64, 65, 66], we have furthermore demonstrated the application of our quantum QNM theory to a 3D YIG-YSO hybrid resonator coupled to such qubits, where the photonic component (YSO shell) is similar in size to the YIG sphere, and consequently perturbative treatments of the magnon-photon coupling may break down[79]. Our calculations not only yield the hybrid modes and qubit couplings, but show how the tuning of losses enabled by the hybrid magnonic-photonic nature of the construct can enable functionalities like magnon-mediated synchronization and entanglement of spatially separate qubits, as well as attaining almost-steady states in the absence of continuous pumping, including at finite temperature and in the presence of realistic qubit decoherence effects. Ultimately, our numerical examples exhibit how the quantum QNM approach can be utilized as a quantitative basis for the design, optimization and analysis of quantum cavity magnonic devices as a functional elements of hybrid quantum processors operating under open-system and thermal conditions.

Our work admits a variety of natural extensions and seeds important further developments in the theoretical treatment of quantum cavity magnonics. For example, incorporating the exchange interaction in numerical QNM solvers, such as the MAN programs[96, 117, 119], and quantifying its impact on the behavior and performance of quantum magnonic devices will be crucial as such technologies are driven down to ever smaller length scales. Furthermore, it is of interest to explore nonlinear effects that can, for example, lead to the suppression of magnon-photon hybridization[148], while another promising route is the inclusion of time-periodic fields that opens the route to Floquet cavity electromagnetics, with a plethora of intriguing phenomena like the emergence of sidebands, magnonic Autler-Townes splittings, Floquet ultrastrong coupling, and the coherent control of magnon-photon coupling[149, 150, 151]. Recent progress in classical QNM theory[152, 153] and MQED[154] for Floquet systems have already established important foundations, so that such developments may lie within reach.

Acknowledgments

This publication has emanated from research conducted with the financial support of Research Ireland, under Grants Number 18/RP/6236 and 23/RC/12197 at Rinn Quantum.

Conflict of Interest

The authors have no conflicts to disclose.

Data Availability

The data that support the findings of this study are available from the corresponding authors upon reasonable request.

References

  1. [1]Y.-P. Wang, J. Rao, Y. Yang, P.-C. Xu, Y. Gui, B. Yao, J. You, and C.-M. Hu, “Nonreciprocity and unidirectional invisibility in cavity magnonics,” Physical review letters 123, 127202 (2019).DOI
  2. [2]M. Kim, A. Tabesh, T. Zegray, S. Barzanjeh, and C.-M. Hu, “Nonreciprocity in cavity magnonics at milikelvin temperature,” arXiv preprint arXiv:2303.04358 (2023).arxiv.org/abs/2303.04358
  3. [3]S. Zheng, Z. Wang, Y. Wang, F. Sun, Q. He, P. Yan, and H. Yuan, “Tutorial: nonlinear magnonics,” Journal of Applied Physics 134 (2023).DOI
  4. [4]J. Kim, B. Kim, and S.-K. Kim, “Non-hermitian cavity magnonics for microwave applications: Pt/anti-pt symmetry, complex photon–magnon coupling, and spectral singularities,” Journal of Physics D: Applied Physics 59, 243002 (2026).DOI
  5. [5]X. Wang and X. Wang, “Topological magnonics,” Journal of Applied Physics 129 (2021).DOI
  6. [6]T. Yu, J. Zou, B. Zeng, J. Rao, and K. Xia, “Non-hermitian topological magnonics,” Physics Reports 1062, 1–86 (2024), non-Hermitian topological magnonics.
  7. [7]E. Mallmann, A. Sombra, J. Goes, and P. Fechine, “Yttrium iron garnet: properties and applications review,” Solid State Phenomena 202, 65–96 (2013).DOI
  8. [8]K. O. Levchenko, K. Davíðková, J. Mikkelsen, and A. V. Chumak, “Review on spin-wave rf applications,” IEEE Transactions on Magnetics (2026).DOI
  9. [9]A. Khitun, “Magnonic memory and logic devices: From proposal to device,” Applied Physics Letters 127 (2025).DOI
  10. [10]Z. Jiang, J. Lim, Y. Li, W. Pfaff, T.-H. Lo, J. Qian, A. Schleife, J.-M. Zuo, V. Novosad, and A. Hoffmann, “Integrating magnons for quantum information,” Applied Physics Letters 123 (2023).arxiv.org/abs/2305.03164
  11. [11]H. Yuan, Y. Cao, A. Kamra, R. A. Duine, and P. Yan, “Quantum magnonics: When magnon spintronics meets quantum information science,” Physics Reports 965, 1–74 (2022).
  12. [12]R. O. Serha, C. Dubs, and A. V. Chumak, “Magnetic materials for quantum magnonics,” APL Materials 14 (2026).arxiv.org/abs/2510.09331
  13. [13]X. Zhang, C.-L. Zou, L. Jiang, and H. X. Tang, “Strongly coupled magnons and cavity microwave photons,” Phys. Rev. Lett. 113, 156401 (2014).DOI
  14. [14]M. Goryachev, W. G. Farr, D. L. Creedon, Y. Fan, M. Kostylev, and M. E. Tobar, “High-cooperativity cavity qed with magnons at microwave frequencies,” Physical Review Applied 2, 054002 (2014).DOI
  15. [15]J. Bourhill, N. Kostylev, M. Goryachev, D. Creedon, and M. Tobar, “Ultrahigh cooperativity interactions between magnons and resonant photons in a yig sphere,” Physical Review B 93, 144420 (2016).DOI
  16. [16]X. Zhang, C.-L. Zou, N. Zhu, F. Marquardt, L. Jiang, and H. X. Tang, “Magnon dark modes and gradient memory,” Nature communications 6, 8914 (2015).DOI
  17. [17]R. Hisatomi, A. Osada, Y. Tabuchi, T. Ishikawa, A. Noguchi, R. Yamazaki, K. Usami, and Y. Nakamura, “Bidirectional conversion between microwave and light via ferromagnetic magnons,” Physical Review B 93, 174427 (2016).DOI
  18. [18]D. Lachance-Quirion, Y. Tabuchi, A. Gloppe, K. Usami, and Y. Nakamura, “Hybrid quantum systems based on magnons,” Applied Physics Express 12, 070101 (2019).
  19. [19]K. Wang, Y.-P. Gao, R. Jiao, and C. Wang, “Recent progress on optomagnonic coupling and optical manipulation based on cavity-optomagnonics,” Frontiers of Physics 17, 42201 (2022).DOI
  20. [20]Z. Shen, G.-T. Xu, M. Zhang, Y.-L. Zhang, Y. Wang, C.-Z. Chai, C.-L. Zou, G.-C. Guo, and C.-H. Dong, “Coherent coupling between phonons, magnons, and photons,” Physical Review Letters 129, 243601 (2022).DOI
  21. [21]X. Zhang, C.-L. Zou, L. Jiang, and H. X. Tang, “Cavity magnomechanics,” Science advances 2, e1501286 (2016).DOI
  22. [22]A. Gardin, G. Bourcin, J. Bourhill, V. Vlaminck, C. Person, C. Fumeaux, G. C. Tettamanzi, and V. Castel, “Engineering synthetic gauge fields through the coupling phases in cavity magnonics,” Physical Review Applied 21, 064033 (2024).DOI
  23. [23]R. Barbieri, M. Cerdonio, G. Fiorentini, and S. Vitale, “Axion to magnon conversion. a scheme for the detection of galactic axions,” Physics Letters B 226, 357–360 (1989).
  24. [24]R. Barbieri, C. Braggio, G. Carugno, C. S. Gallo, A. Lombardi, A. Ortolan, R. Pengo, G. Ruoso, and C. C. Speake, “Searching for galactic axions through magnetized media: The quax proposal,” Physics of the Dark Universe 15, 135–141 (2017).DOI
  25. [25]Y.-P. Wang and C.-M. Hu, “Dissipative couplings in cavity magnons,” Journal of Applied Physics 127 (2020).DOI
  26. [26]J. M. Nair, D. Mukhopadhyay, and G. S. Agarwal, “Cavity-mediated level attraction and repulsion between magnons,” Physical Review B 105, 214418 (2022).DOI
  27. [27]M. Harder, Y. Yang, B. Yao, C. Y. Yu, J. Rao, Y. Gui, R. Stamps, and C.-M. Hu, “Level attraction due to dissipative magnon-photon coupling,” Physical review letters 121, 137203 (2018).DOI
  28. [28]B. Yao, T. Yu, X. Zhang, W. Lu, Y. Gui, C.-M. Hu, and Y. M. Blanter, “The microscopic origin of magnon-photon level attraction by traveling waves: Theory and experiment,” Physical Review B 100, 214426 (2019).DOI
  29. [29]H. Liu, D. Sun, C. Zhang, M. Groesbeck, R. Mclaughlin, and Z. V. Vardeny, “Observation of exceptional points in magnonic parity-time symmetry devices,” Science advances 5, eaax9144 (2019).DOI
  30. [30]D. Zhang, X.-Q. Luo, Y.-P. Wang, T.-F. Li, and J. You, “Observation of the exceptional point in cavity magnon-polaritons,” Nature communications 8, 1368 (2017).DOI
  31. [31]I. Boventer, M. Kläui, R. Macêdo, and M. Weides, “Steering between level repulsion and attraction: broad tunability of two-port driven cavity magnon-polaritons,” New Journal of Physics 21, 125001 (2019).
  32. [32]G. Burkard, T. D. Ladd, A. Pan, J. M. Nichol, and J. R. Petta, “Semiconductor spin qubits,” Reviews of Modern Physics 95, 025003 (2023).DOI
  33. [33]S. Pezzagna and J. Meijer, “Quantum computer based on color centers in diamond,” Applied Physics Reviews 8 (2021).DOI
  34. [34]S. Bertaina, S. Gambarelli, A. Tkachuk, I. Kurkin, B. Malkin, A. Stepanov, and B. Barbara, “Rare-earth solid-state qubits,” Nature nanotechnology 2, 39–42 (2007).
  35. [35]I. Capan, “Defects in silicon carbide as quantum qubits: Recent advances in defect engineering,” Applied Sciences 15, 5606 (2025).DOI
  36. [36]P. Krantz, M. Kjaergaard, F. Yan, T. P. Orlando, S. Gustavsson, and W. D. Oliver, “A quantum engineer’s guide to superconducting qubits,” Applied physics reviews 6, 021318 (2019).DOI
  37. [37]A. Y. Dmitriev and O. Astafiev, “A perspective on superconducting flux qubits,” Applied Physics Letters 119 (2021).DOI
  38. [38]T. E. Roth, R. Ma, and W. C. Chew, “An introduction to the transmon qubit for electromagnetic engineers,” arXiv preprint arXiv:2106.11352 (2021).DOI
  39. [39]T. Neuman, D. S. Wang, and P. Narang, “Nanomagnonic cavities for strong spin-magnon coupling and magnon-mediated spin-spin interactions,” Phys. Rev. Lett. 125, 247702 (2020).DOI
  40. [40]D. R. Candido, G. D. Fuchs, E. Johnston-Halperin, and M. E. Flatté, “Predicted strong coupling of solid-state spins via a single magnon mode,” Materials for Quantum Technology 1, 011001 (2021).DOI
  41. [41]D. García-Pons, J. Pérez-Bailón, C. Boix-Constant, I. Gómez-Muñoz, X. del Arco, S. Mañas-Valero, E. Coronado, D. Zueco, and M. J. Martínez-Pérez, “Strong spin-magnon coupling in a van der waals magnet with tunable chiral symmetry,” Newton (2026).
  42. [42]L. Trifunovic, F. L. Pedrocchi, and D. Loss, “Long-distance entanglement of spin qubits via ferromagnet,” Physical review X 3, 041023 (2013).DOI
  43. [43]S. Ribeiro and P. Kumar, “Dual-domain coherent emission: Photon-magnon laser in a hybrid quantum system,” Physical Review A 112, 043711 (2025).DOI
  44. [44]L. Trifunovic, F. L. Pedrocchi, S. Hoffman, P. Malentinsky, A. Yacoby, and D. Loss, “High-efficiency resonant amplification of weak magnetic fields for single spin magnetometry,” arXiv preprint arXiv:1409.1497 (2014).DOI
  45. [45]Y. Tabuchi, S. Ishino, A. Noguchi, T. Ishikawa, R. Yamazaki, K. Usami, and Y. Nakamura, “Coherent coupling between a ferromagnetic magnon and a superconducting qubit,” Science 349, 405–408 (2015).DOI
  46. [46]Y. Tabuchi, S. Ishino, A. Noguchi, T. Ishikawa, R. Yamazaki, K. Usami, and Y. Nakamura, “Quantum magnonics: The magnon meets the superconducting qubit,” Comptes Rendus. Physique 17, 729–739 (2016).DOI
  47. [47]Y. Fan, J. Li, and Y. Wu, “Nonclassical magnon pair generation and cauchy-schwarz inequality violation,” Physical Review A 108, 053715 (2023).DOI
  48. [48]S. P. Wolski, D. Lachance-Quirion, Y. Tabuchi, S. Kono, A. Noguchi, K. Usami, and Y. Nakamura, “Dissipation-based quantum sensing of magnons with a superconducting qubit,” Physical Review Letters 125, 117701 (2020).DOI
  49. [49]S. Rani, X. Cao, E. Baptista, A. Hoffmann, and W. Pfaff, “High-dynamic-range quantum sensing of magnons and their dynamics using a superconducting qubit,” Physical Review Applied 23, 064032 (2025).DOI
  50. [50]D. Lachance-Quirion, Y. Tabuchi, S. Ishino, A. Noguchi, T. Ishikawa, R. Yamazaki, and Y. Nakamura, “Resolving quanta of collective spin excitations in a millimeter-sized ferromagnet,” Science advances 3, e1603150 (2017).
  51. [51]Y. Hu, G.-Q. Ge, S. Chen, X.-F. Yang, and Y.-L. Chen, “Cross-kerr-effect induced by coupled josephson qubits in circuit quantum electrodynamics,” Physical Review A—Atomic, Molecular, and Optical Physics 84, 012329 (2011).DOI
  52. [52]E. T. Holland, B. Vlastakis, R. W. Heeres, M. J. Reagor, U. Vool, Z. Leghtas, L. Frunzio, G. Kirchmair, M. H. Devoret, M. Mirrahimi, et al., “Single-photon-resolved cross-kerr interaction for autonomous stabilization of photon-number states,” Physical review letters 115, 180501 (2015).DOI
  53. [53]M. Kounalakis, G. E. Bauer, and Y. M. Blanter, “Analog quantum control of magnonic cat states on a chip by a superconducting qubit,” Physical review letters 129, 037205 (2022).DOI
  54. [54]M. Kounalakis, S. Viola Kusminsky, and Y. M. Blanter, “Engineering entangled coherent states of magnons and phonons via Y. Ren, K. W. Huang, X. Wang, and H. Xiong, “Preparing magnon–photon pairs in a cavity–magnon–qubit system,” Advanced Photonics Research 6, 2500029 (2025).
  55. [56]Z.-y. Jin and J. Jing, “Magnon blockade in magnon-qubit systems,” Physical Review A 108, 053702 (2023).DOI
  56. [57]M. Dols, S. Sharma, L. Bechara, Y. M. Blanter, M. Kounalakis, and S. Viola Kusminsky, “Magnon-mediated quantum gates for superconducting qubits,” Physical Review B 110, 104416 (2024).DOI
  57. [58]Y.-Y. Lai, G.-D. Lin, J. Twamley, and H.-S. Goan, “Single-nitrogen-vacancy-center quantum memory for a superconducting flux qubit mediated by a ferromagnet,” Physical Review A 97, 052303 (2018).DOI
  58. [59]Y.-B. Hou, X.-L. Hei, X.-F. Pan, J.-K. Xie, Y.-L. Ren, S.-L. Ma, F.-L. Li, and P.-B. Li, “Robust generation of a magnonic cat state via a superconducting flux qubit,” Physical Review A 110, 013711 (2024).DOI
  59. [60]G. Liu, J. Liu, F. Qiao, and R.-C. Yang, “Generation of magnonic squeezed state and its superposition in a hybrid qubit-magnon system,” arXiv preprint arXiv:2604.02924 (2026).arxiv.org/abs/2604.02924
  60. [61]S. Kim, L. V. Abdurakhimov, D. Pham, W. Qiu, H. Terai, S. Ashhab, S. Saito, T. Yamashita, and K. Semba, “Superconducting flux qubit with ferromagnetic josephson π-junction operating at zero magnetic field,” Communications Materials 5, 216 (2024).DOI
  61. [62]S. Günzler, J. Beck, D. Rieger, N. Gosling, N. Zapata, M. Field, S. Geisert, A. Bacher, J. K. Hohmann, M. Spieker, et al., “Spin environment of a superconducting qubit in high magnetic fields,” Nature Communications 16, 6564 (2025).
  62. [63]X. Zhu, S. Saito, A. Kemp, K. Kakuyama, S.-i. Karimoto, H. Nakano, W. J. Munro, Y. Tokura, M. S. Everitt, K. Nemoto, et al., “Coherent coupling of a superconducting flux qubit to an electron spin ensemble in diamond,” Nature 478, 221–224 (2011).
  63. [64]J. Twamley and S. D. Barrett, “Superconducting cavity bus for single nitrogen-vacancy defect centers in diamond,” Physical Review B—Condensed Matter and Materials Physics 81, 241202 (2010).DOI
  64. [65]D. Marcos, M. Wubs, J. Taylor, R. Aguado, M. D. Lukin, and A. S. Sørensen, “Coupling nitrogen-vacancy centers in diamond to superconducting flux qubits,” Physical review letters 105, 210501 (2010).DOI
  65. [66]T. Chang, I. Holzman, S. Q. Lim, D. Holmes, B. C. Johnson, D. N. Jamieson, and M. Stern, “Strong coupling of a superconducting flux qubit to single bismuth donors,” Nature Communications 16, 9832 (2025).DOI
  66. [67]L. R. Walker, “Magnetostatic modes in ferromagnetic resonance,” Physical Review 105, 390 (1957).DOI
  67. [68]L. Walker, “Resonant modes of ferromagnetic spheroids,” Journal of Applied Physics 29, 318–323 (1958).DOI
  68. [69]P. Fletcher and R. Bell, “Ferrimagnetic resonance modes in spheres,” Journal of applied physics 30, 687–698 (1959).
  69. [70]I. Solt and P. Fletcher, “Magnetic anisotropy from the magnetostatic modes,” Journal of Applied Physics 31, S100–S102 (1960).DOI
  70. [71]C. Kittel, “On the theory of ferromagnetic resonance absorption,” Physical review 73, 155 (1948).DOI
  71. [72]R. L. White and I. H. Solt Jr., “Multiple ferromagnetic resonance in ferrite spheres,” Physical Review 104, 56 (1956).DOI
  72. [73]B. Zare Rameshti, Y. Cao, and G. E. Bauer, “Magnetic spheres in microwave cavities,” Physical Review B 91, 214430 (2015).DOI
  73. [74]Z. Lin and S. Chui, “Electromagnetic scattering by optically anisotropic magnetic particle,” Physical Review E—Statistical, Nonlinear, and Soft Matter Physics 69, 056614 (2004).pubmed.ncbi.nlm.nih.gov/15244970
  74. [75]Y.-L. Geng, X.-B. Wu, L.-W. Li, and B.-R. Guan, “Mie scattering by a uniaxial anisotropic sphere,” Physical Review E—Statistical, Nonlinear, and Soft Matter Physics 70, 056609 (2004).pubmed.ncbi.nlm.nih.gov/15600781
  75. [76]J. Krupka, B. Salski, P. Kopyt, and W. Gwarek, “Electrodynamic study of yig filters and resonators,” Scientific reports 6, 34739 (2016).DOI
  76. [77]J. Krupka, P. Aleshekvych, B. Salski, P. Kopyt, and A. Pacewicz, “Ferromagnetic resonance revised–electrodynamic approach,” Scientific reports 7, 5750 (2017).DOI
  77. [78]J. Krupka, A. Pacewicz, B. Salski, P. Kopyt, J. Bourhill, M. Goryachev, and M. Tobar, “Electrodynamic improvements to the theory of magnetostatic modes in ferrimagnetic spheres and their applications to saturation magnetization measurements,” Journal of Magnetism and Magnetic Materials 487, 165331 (2019).DOI
  78. [79]R. Macêdo, R. C. Holland, P. G. Baity, L. J. McLellan, K. L. Livesey, R. L. Stamps, M. P. Weides, and D. A. Bozhko, “Electromagnetic approach to cavity spintronics,” Physical Review Applied 15, 024065 (2021).DOI
  79. [80]T. Holstein and H. Primakoff, “Field dependence of the intrinsic domain magnetization of a ferromagnet,” Physical Review 58, 1098 (1940).DOI
  80. [81]V. Bittencourt, C. Potts, Y. Huang, J. Davis, and S. Viola Kusminsky, “Magnomechanical backaction corrections due to coupling to higher-order walker modes and kerr nonlinearities,” Physical Review B 107, 144411 (2023).
  81. [82]Ö. O. Soykal and M. Flatté, “Strong field interactions between a nanomagnet and a photonic cavity,” Physical review letters 104, 077202 (2010).
  82. [83]H. M. Lai, P. T. Leung, K. Young, P. W. Barber, and S. C. Hill, “Time-independent perturbation for leaking electromagnetic modes in open systems with application to resonances in microdroplets,” Phys. Rev. A 41, 5187–5198 (1990).pubmed.ncbi.nlm.nih.gov/9903746
  83. [84]P. T. Leung, S. Y. Liu, and K. Young, “Completeness and orthogonality of quasinormal modes in leaky optical cavities,” Phys. Rev. A 49, 3057–3067 (1994).pubmed.ncbi.nlm.nih.gov/9910588
  84. [85]P. T. Leung, S. Y. Liu, S. S. Tong, and K. Young, “Time-independent perturbation theory for quasinormal modes in leaky optical cavities,” Phys. Rev. A 49, 3068–3073 (1994).pubmed.ncbi.nlm.nih.gov/9910589
  85. [86]P. T. Leung and K. M. Pang, “Completeness and time-independent perturbation of morphology-dependent resonances in dielectric spheres,” J. Opt. Soc. Am. B 13, 805–817 (1996).DOI
  86. [87]K. M. Lee, P. T. Leung, and K. M. Pang, “Dyadic formulation of morphology-dependent resonances. i. completeness relation,” J. Opt. Soc. Am. B 16, 1409–1417 (1999).DOI
  87. [88]P. T. Kristensen and S. Hughes, “Modes and mode volumes of leaky optical cavities and plasmonic nanoresonators,” ACS Photonics 1, 2–10 (2014).DOI
  88. [89]W. Yan, R. Faggiani, and P. Lalanne, “Rigorous modal analysis of plasmonic nanoresonators,” Phys. Rev. B 97, 205422 (2018).DOI
  89. [90]P. Lalanne, W. Yan, K. Vynck, C. Sauvan, and J.-P. Hugonin, “Light interaction with photonic and plasmonic resonances,” Laser & Photonics Reviews 12, 1700113 (2018).
  90. [91]C. Sauvan, T. Wu, R. Zarouf, E. A. Muljarov, and P. Lalanne, “Mquasinormal, orthogonality, and completeness of quasinormal modes of open systems: the case of electromagnetism [invited],” Opt. Express 30, 6846–6885 (2022).DOI
  91. [92]Q. Bai, M. Perrin, C. Sauvan, J.-P. Hugonin, and P. Lalanne, “Efficient and intuitive method for the analysis of light scattering by a resonant nanostructure,” Opt. Express 21, 27371–27382 (2013).pubmed.ncbi.nlm.nih.gov/24216959
  92. [93]C. Sauvan, J. P. Hugonin, I. S. Maksymov, and P. Lalanne, “Theory of the spontaneous optical emission of nanosize photonic and plasmon resonators,” Phys. Rev. Lett. 110, 237401 (2013).pubmed.ncbi.nlm.nih.gov/25167528
  93. [94]R.-C. Ge and S. Hughes, “Design of an efficient single photon source from a metallic nanorod dimer: a quasi-normal mode finite-difference time-domain approach,” Opt. Lett. 39, 4235–4238 (2014).DOI
  94. [95]R.-C. Ge, P. T. Kristensen, J. F. Young, and S. Hughes, “Quasi-normal mode approach to modelling light-emission and propagation in nanoplasmonics,” New Journal of Physics 16, 113048 (2014).
  95. [96]T. Wu, D. Arriault, W. Yan, and P. Lalanne, “Modal analysis of electromagnetic resonators: User guide for the man program,” Computer Physics Communications 284, 108627 (2023).DOI
  96. [97]T. Kristensen, C. V. Vlack, and S. Hughes, “Generalized effective mode volume for leaky optical cavities,” Opt. Lett. 37, 1649–1651 (2012).DOI
  97. [98]N. Kongsuwan, A. Demetriadou, M. Horton, R. Chikkaraddy, J. J. Baumberg, and O. Hess, “Plasmonic nanocavity modes: From near-field to far-field radiation,” ACS Photonics 7, 463–471 (2020).arxiv.org/abs/1910.02273
  98. [99]J. Ren, S. Franke, A. Knorr, M. Richter, and S. Hughes, “Near-field to far-field transformations of optical quasinormal modes and efficient calculation of quantized quasinormal modes for open cavities and plasmonic resonators,” Phys. Rev. B 101, 205402 (2020).DOI
  99. [100]K. Bedingfield, E. Elliott, A. Gisdkadis, N. Kongsuwan, J. J. Baumberg, and A. Demetriadou, “Multi-faceted plasmonic nanocavities,” Nanophotonics 12, 3931–3944 (2023).DOI
  100. [101]M. Kamandar Dezfouli, R. Gordon, and S. Hughes, “Modal theory of modified spontaneous emission of a quantum emitter in a hybrid plasmonic photonic-crystal cavity system,” Phys. Rev. A 95, 013846 (2017).DOI
  101. [102]J. Ren, S. Franke, B. VanDrunen, and S. Hughes, “Classical purcell factors and spontaneous emission decay rates in a linear gain medium,” Phys. Rev. A 109, 013513 (2024).
  102. [103]J. Yang, M. Perrin, and P. Lalanne, “Analytical formalism for the interaction of two-level quantum systems with metal nanoresonators,” Physical Review X 5, 021008 (2015).DOI
  103. [104]A.-W. El-Sayed and S. Hughes, “Quasinormal-mode theory of elastic purcell factors and fano resonances of optomechanical beams,” Phys. Rev. Res. 2, 043290 (2020).
  104. [105]S. Franke, S. Hughes, M. K. Dezfouli, P. T. Kristensen, K. Busch, A. Knorr, and M. Richter, “Quantization of quasinormal modes for open cavities and plasmonic cavity quantum electrodynamics,” Phys. Rev. Lett. 122, 213901 (2019).DOI
  105. [106]S. Franke, J. Ren, S. Hughes, and M. Richter, “Fluctuation-dissipation theorem and fundamental photon commutation relations in lossy nanostructures using quasinormal modes,” Physical Review Research 2, 033332 (2020).DOI
  106. [107]S. Franke, M. Richter, J. Ren, A. Knorr, and S. Hughes, “Quantized quasinormal-mode description of nonlinear cavity-qed effects from coupled resonators with a fano-like resonance,” Phys. Rev. Res. 2, 033456 (2020).DOI
  107. [108]L. Meschede, D. D. Clarke, and O. Hess, “Quantum quasinormal-mode theory for dissipative nano-optics and magnetodielectric cavity quantum electrodynamics,” APL Quantum 3 (2026).arxiv.org/abs/2507.05233
  108. [109]S. Scheel, L. Knöll, and D.-G. Welsch, “Qed commutation relations for inhomogeneous kramers-kronig dielectrics,” Phys. Rev. A 58, 700–706 (1998).DOI
  109. [110]H. T. Dung, L. Knöll, and D.-G. Welsch, “Three-dimensional quantization of the electromagnetic field in dispersive and absorbing inhomogeneous dielectrics,” Phys. Rev. A 57, 3931–3942 (1998).DOI
  110. [111]T. Gruner and D.-G. Welsch, “Green-function approach to the radiation-field quantization for homogeneous and inhomogeneous kramers-kronig dielectrics,” Phys. Rev. A 53, 1818–1829 (1996).pubmed.ncbi.nlm.nih.gov/9913077
  111. [112]C. Raabe, S. Scheel, and D.-G. Welsch, “Unified approach to qed in arbitrary linear media,” Physical Review A—Atomic, Molecular, and Optical Physics 75, 053813 (2007).DOI
  112. [113]S. Scheel and S. Y. Buhmann, “Macroscopic quantum electrodynamics-concepts and applications,” Acta Phys. Slovaca 58, 675–809 (2008).DOI
  113. [114]C. Raabe and D.-G. Welsch, “Qed in arbitrary linear media: Amplifying media,” The European Physical Journal Special Topics 160, 371–381 (2008).DOI
  114. [115]T. G. Philbin, “Canonical quantization of macroscopic electromagnetism,” New Journal of Physics 12, 123008 (2010).
  115. [116]S. Y. Buhmann, D. T. Butcher, and S. Scheel, “Macroscopic quantum electrodynamics in nonlocal and nonreciprocal media,” New Journal of Physics 14, 083034 (2012).DOI
  116. [117]T. Wu and P. Lalanne, “Qnmnonreciprocal-resonators: an openly available toolbox for computing the quasinormal modes of nonreciprocal resonators,” arXiv preprint arXiv:2106.05502 (2021).arxiv.org/abs/2106.05502
  117. [118]B. Vial, F. Zolla, A. Nicolet, and M. Commandré, “Quasimodal expansion of electromagnetic fields in open two-dimensional structures,” Physical Review A 89, 023829 (2014).DOI
  118. [119]T. Wu, M. Giurioli, and P. Lalanne, “Nanoscale light confinement: the q’s and v’s,” ACS Photonics 8, 1522–1538 (2021).DOI
  119. [120]B. Schneider, “Effect of crystalline anisotropy on the magnetostatic spin wave modes in ferromagnetic plates. i. theoretical discussion for infinite plates,” physica status solidi (b) 51, 325–338 (1972).DOI
  120. [121]J. E. Miltat and M. J. Donahue, “Numerical micromagnetics: Finite difference methods,” Handbook of magnetism and advanced magnetic materials (2007).DOI
  121. [122]J. A. Osborn, “Demagnetizing factors of the general ellipsoid,” Physical review 67, 351 (1945).DOI
  122. [123]M. Kamandar Dezfouli, C. Tserkezis, N. A. Mortensen, and S. Hughes, “Nonlocal quasinormal modes for arbitrarily shaped three-dimensional plasmonic resonators,” Optica 4, 1503–1509 (2017).DOI
  123. [124]J. M. McMahon, S. K. Gray, and G. C. Schatz, “Nonlocal optical response of metal nanostructures with arbitrary shape,” Physical review letters 103, 097403 (2009).pubmed.ncbi.nlm.nih.gov/19792829
  124. [125]G. T. Rado and J. R. Weertman, “Spin-wave resonance in a ferromagnetic metal,” Journal of Physics and chemistry of solids 11, 315–333 (1959).DOI
  125. [126]K. Y. Gushlienko and A. Slavin, “Boundary conditions for magnetization in magnetic nanoelements,” Physical Review B—Condensed Matter and Materials Physics 72, 014463 (2005).
  126. [129]S. Franke, J. Ren, and S. Hughes, “Quantized quasinormal-mode theory of coupled lossy and amplifying resonators,” Phys. Rev. A 105, 023702 (2022).DOI
  127. [130]C. W. Gardiner and M. J. Collett, “Input and output in damped quantum systems: Quantum stochastic differential equations and the master equation,” Physical Review A 31, 3761 (1985).pubmed.ncbi.nlm.nih.gov/9895956
  128. [131]V. Cherenpanov, I. Kolokolov, and V. Lvov, “The saga of yig: Spectra, thermodynamics, interaction and relaxation of magnons in a complex magnet,” Physics Reports 229, 81–144 (1993).
  129. [132]A. A. Serga, A. V. Chumak, and B. Hillebrands, “Yig magnon-ics,” Journal of Physics D: Applied Physics 43, 264002 (2010).
  130. [133]J. I. Cirac, “Interaction of a two-level atom with a cavity mode in the bad-cavity limit,” Physical Review A 46, 4354 (1992).pubmed.ncbi.nlm.nih.gov/9908636
  131. [134]G. S. Abo, Y.-K. Hong, J. Park, J. Lee, W. Lee, and B.-C. Choi, “Definition of magnetic exchange length,” IEEE Transactions on Magnetics 49, 4937–4939 (2013).DOI
  132. [135]K. Xue and R. H. Victora, “High data rate spin-wave transmitter,” Scientific reports 14, 23129 (2024).DOI
  133. [136]A. Pick, B. Zhen, O. D. Miller, C. W. Hsu, F. Hernandez, A. W. Rodriguez, M. Soljačić, and S. G. Johnson, “General theory of spontaneous emission near exceptional points,” Optics express 25, 12325–12348 (2017).DOI
  134. [137]Y. Lu, Y. Zhao, R. Li, and J. Liu, “Anomalous spontaneous emission dynamics at chiral exceptional points,” Optics Express 30, 41784–41803 (2022).DOI
  135. [138]G. Dold, C. W. Zollitsch, J. O’Sullivan, S. Welinski, A. Ferrier, P. Goldner, S. De Graaf, T. Lindström, and J. J. Morton, “High-cooperativity coupling of a rare-earth spin ensemble to a superconducting resonator using yttrium orthosilicate as a substrate,” Physical Review Applied 11, 054082 (2019).
  136. [139]V. Laguta, Y. Zorenko, M. Buryi, V. Gorbenko, T. Zorenko, J. Mares, and M. Nikl, “Epr study of ce3+ luminescent centers in the y2sio5 single crystalline films,” Optical Materials 72, 833–837 (2017).DOI
  137. [140]M. Buryi, V. Laguta, D. Savchenko, and M. Nikl, “Electron paramagnetic resonance study of lu2sio5 and y2sio5 scintillators doped by cerium,” Advanced Science, Engineering and Medicine 5, 573–576 (2013).DOI
  138. [141]F. Schmolke and E. Lutz, “Noise-induced quantum synchronization,” Physical Review Letters 129, 250601 (2022).DOI
  139. [142]Z. Tao, F. Schmolke, C.-K. Hu, W. Huang, Y. Zhou, J. Zhang, J. Chu, L. Zhang, X. Sun, Z. Guo, et al., “Noise-induced quantum synchronization with entangled oscillations,” Nature Communications 16, 8457 (2025).DOI
  140. [143]R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki, “Quantum entanglement,” Rev. Mod. Phys. 81, 865–942 (2009).DOI
  141. [144]N. Lambert, E. Giguère, P. Menczel, B. Li, P. Hopf, G. Suárez, M. Gali, J. Lishman, R. Gadhvi, R. Agarwal, et al., “QuTiP 5: The quantum toolbox in python,” arXiv preprint arXiv:2412.04705 (2024).DOI
  142. [145]S. Ishizaka and T. Hiroshima, “Maximally entangled mixed states under nonlocal unitary operations in two qubits,” Physical Review A 62, 022310 (2000).DOI
  143. [146]F. Yan, S. Gustavsson, A. Kamal, J. Birenbaum, A. P. Sears, D. Hover, T. J. Gudmundsen, D. Rosenberg, G. Samach, S. Weber, et al., “The flux qubit revisited to enhance coherence and reproducibility,” Nature communications 7, 12964 (2016).DOI
  144. [147]M. Le Dantec, M. Rancic, S. Lin, E. Billaud, V. Ranajan, D. Flanigan, S. Bertaina, T. Chanelliere, P. Goldner, A. Erb, et al., “Twenty-three-millisecond electron spin coherence of erbium ions in a natural-abundance crystal,” Science advances 7, eabj9786 (2021).arxiv.org/abs/2106.14974
  145. [148]O. Lee, K. Yamamoto, M. Umeda, C. W. Zollitsch, M. Elyasi, T. Kikkawa, E. Saitoh, G. E. Bauer, and H. Kurebayashi, “Non-linear magnon polaritons,” Physical Review Letters 130, 046703 (2023).
  146. [149]J. Xu, C. Zhong, X. Han, D. Jin, L. Jiang, and X. Zhang, “Floquet cavity electromagnetics,” Physical Review Letters 125, 237201 (2020).
  147. [150]T. Aguiar and M. De Oliveira, “Perturbative input–output theory of floquet cavity magnonics and magnon energy shifts,” New Journal of Physics 28, 043501 (2026).DOI
  148. [151]S. H. Hosseini Biuki, B. Rejaei, and M. Memarian, “Time-varying ferrimagnetic media: Magnetization dynamics, wave propagation, and spectral shaping,” Physical Review B 114, 094429 (2026).
  149. [152]B. Vial and R. V. Craster, “Quasinormal modes of floquet media slabs,” Physical Review B 112, 214201 (2025).arxiv.org/abs/2507.02784
  150. [153]Y. Sun, S. Fan, and G. Hu, “Quasinormal mode basis for open floquet photonic systems,” Physical Review Letters 136, 206903 (2026).DOI
  151. [154]G. Horsley and R. Baker, “Macroscopic qed and noise currents in time-varying media,” Physical Review A 111, 053511 (2025).arxiv.org/abs/2409.11873

Paper details

vlnc:W_ba5hwrs2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 6

downloads · 5

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents