betaFeedback

WorksPhysics and Astronomy

Nonadiabatic Theory of Phonon Magnetic Moments in Insulators and Metals

Abstract

We develop a nonadiabatic theory of phonon magnetic moments applicable to both insulators and metals. By relating the phonon magnetic moment to the force-velocity response of ions in a magnetic field, we derive a gauge-invariant expression using a gauge-covariant Wigner expansion. The formalism naturally separates Fermi-sea and Fermi-surface contributions and captures the full dependence on phonon frequency. In gapped systems, our theory reduces to previous adiabatic expressions in the low-frequency limit. Beyond this limit, it reveals additional contributions arising from resonant interband processes and the Fermi surface. Applying our theory to Pb1−x_{1-x}Snx_xTe, we find that the Fermi-surface contribution substantially enhances the phonon magnetic moment, reproducing the same order of magnitude as the experimental observation. Our results provide a unified framework for describing phonon magnetic moments beyond the adiabatic regime.

Introduction

Introduction.—Phonons are among the most fundamental quasiparticles in condensed matter, playing a central role in thermodynamic, spectral, and transport properties of materials [1]. In the past two decades, phonons carrying finite angular momentum, known as chiral or axial phonons, have attracted growing interest due to their nontrivial modification of these properties [2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18]. They are closely related to various intriguing phenomena such as the phonon Zeeman effect [4, 7, 9, 19, 20, 21], the thermal Hall effect [2, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34], and the Einstein-de Haas effect [3, 35, 36]. Moreover, due to their nature of broken time-reversal symmetry, chiral phonons also carry magnetic moments, allowing the exciting possibility of generating spin and orbital magnetizations dynamically [37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48].

Despite substantial theoretical progress, a quantitative understanding of phonon magnetic moments remains incomplete. Existing quantum theories, such as those based on adiabatic current pumping [14, 49, 50, 51], density functional perturbation theory [30], and density-matrix perturbation approaches [52], are primarily formulated for insulating systems. However, recent experiments in doped semiconductors and metals have reported phonon magnetic moments reaching the scale of μB\mu_B, far exceeding the predictions of these approaches [19, 20, 21]. Phenomenological extensions to Fermi-surface electrons have been proposed in Dirac semimetals [53, 54], but their model dependence limits their applicability. Together, these results indicate that essential physical ingredients are missing in the adiabatic description, particularly in the presence of finite carrier density or when phonon frequencies are comparable to electronic energy scales.

In this Letter, we resolve this issue by developing a nonadiabatic theory of phonon magnetic moments that treats insulators and metals on equal footing. Our approach expresses the phonon magnetic moment as a linear response of ionic forces to velocities in a magnetic field and derives a gauge-invariant formula using a gauge-covariant Wigner expansion. The resulting expression naturally separates Fermi-sea and Fermi-surface contributions and retains the full frequency dependence of phonons. In the low-frequency limit for gapped systems, it reduces to previous adiabatic theories [14, 49, 51]. Beyond this limit, it captures additional contributions arising from nonadiabatic effects and from the Fermi surface.

To illustrate these effects, we apply our theory to a low-energy model of the topological crystalline insulator Pb1−xSnxTe\mathrm{Pb}_{1-x}\mathrm{Sn}_{x}\mathrm{Te}. We find that nonadiabatic corrections are substantial even in the insulating regime when the electronic gap is narrow. Upon doping into the metallic regime, Fermi-surface electrons contribute significantly and eventually dominate the phonon magnetic moment. Using model parameters consistent with experimental measurements [21], we show that the Fermi-surface contribution plays a central role, leading to a phonon magnetic moment reaching the order of μB\mu_B, comparable to experimental observations. Our framework thus provides a systematic route for studying phonon magnetic moments in general systems, including both insulators and metals.

Chiral phonon splitting

Conventionally, ionic motion is described under the assumption that time-reversal symmetry (TRS) is preserved. Within the harmonic approximation, it is governed by the dynamical matrix D(q)D(\mathbf{q}), whose eigenvalue ωqλ2\omega_{\mathbf{q}\lambda}^{2} is the squared frequency of the λ\lambda-th phonon branch with quasi-wavevector q\mathbf{q}. The corresponding eigenmode eqλ\mathbf{e}_{\mathbf{q}\lambda} describes the vibrating pattern uλ(q,t)∝eqλexp⁡(iq⋅R−iωqλt)\mathbf{u}_{\lambda}(\mathbf{q},t) \propto\mathbf{e}_{\mathbf{q}\lambda}\exp(i\mathbf{q}\cdot\mathbf{R}-i\omega_{\mathbf{q}\lambda}t) of ions [1], where R\mathbf{R} denotes the equilibrium positions of ions and u\mathbf{u} is the displacement.

This description is modified when TRS is broken. An additional Lorentz-like force emerges, which couples ionic displacements with their velocities [2, 30]. The equation of motion then becomes P^˙(q)=−D^(q)u^(q)+G^(q,ω)P^(q)\dot{\hat{\mathbf{P}}}(\mathbf{q})=-\hat{D}(\mathbf{q})\hat{\mathbf{u}}(\mathbf{q})+\hat{G}(\mathbf{q},\omega)\hat{\mathbf{P}}(\mathbf{q}), where P^(q)\hat{\mathbf{P}}(\mathbf{q}) is the ion momentum, and G^(q,ω)\hat{G}(\mathbf{q},\omega) is a frequency-dependent anti-Hermitian matrix representing the Lorentz-like force. This term lifts the

@page 1/7 starts_mid=0 ends_mid=1 the degeneracy of phonon modes. degeneracy of phonon modes and gives rise to chiral phonons. To see this, consider two degenerate linearly polarized modes ex(q)e_x(\boldsymbol{q}) and ey(q)e_y(\boldsymbol{q}). In this subspace, D^(q)=ω02(q)I^\hat{D}(\boldsymbol{q})=\omega_0^2(\boldsymbol{q})\hat{I} and

G(q,ω)=[0g(q,ω)−g∗(q,ω)0].(1)G(\boldsymbol{q},\omega)=\begin{bmatrix}0 & g(\boldsymbol{q},\omega) \\ -g^*(\boldsymbol{q},\omega) & 0\end{bmatrix}. \tag*{(1)}

Solving the equation of motion, the eigenmodes become eR(L)(q)=(ex(q)±iey(q))/2+O(g(q))e_{R(L)}(\boldsymbol{q})=(e_x(\boldsymbol{q})\pm i e_y(\boldsymbol{q}))/\sqrt{2}+O(g(\boldsymbol{q})), corresponding to right- and left-handed circularly polarized vibrations. In the weak-coupling limit ∣g(q,ω0)∣≪ω0(q)|g(\boldsymbol{q},\omega_0)|\ll\omega_0(\boldsymbol{q}), the resulting chiral-phonon splitting is

Δω=∣g(q,ω0(q))∣.(2)\Delta\omega=|g(\boldsymbol{q},\omega_0(\boldsymbol{q}))|. \tag*{(2)}

Under a magnetic field B\boldsymbol{B}, the term proportional to B\boldsymbol{B} directly defines the phonon magnetic moment. In the following, we will consider general cases that the matrices G^\hat{G} and D^\hat{D} contain all ionic degrees of freedom, and focus on the TRS-breaking effect in the matrix G^(q,ω)\hat{G}(\boldsymbol{q},\omega).

In lattice dynamics, the TRS-breaking effects come from two sources (Figure 1 (a)). The first is the direct coupling of ionic charges to an external magnetic field, which produces a splitting via the Lorentz force [4, 29]

Figure 1. (a) Schematic demonstration of how phonons couple to external magnetic fields, i.e., direct Lorentz-coupling of ions to BextB_{\mathrm{ext}} and coupling through EPC to electrons under magnetic fields. (b) Feynman diagrams for calculating chiral phonon splitting. (c) and (d) Feynman diagrams for calculating phonon magnetic moments.

Gκa,κ′b(L)(q)=∑cδκκ′Zκe2MκϵabcBc,(3)G^{(\mathrm{L})}_{\kappa a,\kappa' b}(\boldsymbol{q})=\sum_c\delta_{\kappa\kappa'}\frac{Z_\kappa e}{2M_\kappa}\epsilon_{abc}B_c, \tag*{(3)}

where Zκe=−Zκ∣e∣Z_\kappa e=-Z_\kappa|e| is the bare charge of ion κ\kappa, MκM_\kappa is its mass and B\boldsymbol{B} is the magnetic field. The Levi-Civita tensor ϵabc\epsilon_{abc} denotes antisymmetric permutations of Cartesian directions aa, bb and cc.

The second, and equally important, source arises from electron-phonon coupling (EPC), where time-reversal symmetry (TRS) is broken in the electronic sector, either by an external magnetic field or intrinsic magnetic order. A key step is to formulate this contribution in a linear-response framework [55, 56], by identifying G^\hat{G} as the response function relating the force F^κ(q)=−∑leiq⋅Rl∂H^/∂Rlκ\hat{\boldsymbol{F}}_\kappa(\boldsymbol{q})=-\sum_l e^{i\boldsymbol{q}\cdot\boldsymbol{R}_l}\partial\hat{H}/\partial\boldsymbol{R}_{l\kappa} acting on ion κ\kappa to the velocity u˙κ′(q)\dot{\boldsymbol{u}}_{\kappa'}(\boldsymbol{q}) of another ion κ′\kappa'. Within this formulation, G^\hat{G} naturally takes the form of a Kubo-type response function. By noticing that u˙κ′=−iωuκ′\dot{\boldsymbol{u}}_{\kappa'}=-i\omega\boldsymbol{u}_{\kappa'}, for a non-interacting system, we obtain [57]

Gκa,κ′b(EPC)(q,ω)=−iℏ2ωMκMκ′{1ℏβ∑iTr⁡[F^κa(q)G^(k,ipi)F^κ′b(−q)G^(k+q,ipi+iωj)]}iωj→ω+iη−(κa↔κ′b).(4)G^{(\mathrm{EPC})}_{\kappa a,\kappa' b}(\boldsymbol{q},\omega)=\frac{-i\hbar}{2\omega\sqrt{M_\kappa M_{\kappa'}}}\left\{\frac{1}{\hbar\beta}\sum_i\operatorname{Tr}\left[\hat{F}_{\kappa a}(\boldsymbol{q})\hat{G}(\boldsymbol{k},ip_i)\hat{F}_{\kappa' b}(-\boldsymbol{q})\hat{G}(\boldsymbol{k}+\boldsymbol{q},ip_i+i\omega_j)\right]\right\}_{i\omega_j\to\omega+i\eta}-(\kappa a\leftrightarrow\kappa' b). \tag*{(4)}

Here G^(k,ipi)\hat{G}(\boldsymbol{k},ip_i) stands for electron Green’s functions with quasi-wavevector k\boldsymbol{k} and fermionic Matsubara frequency ipi=i(2i+1)π/ℏβip_i=i(2i+1)\pi/\hbar\beta, with β=1/kBT\beta=1/k_BT being the inverse

temperature. The wavevector q\boldsymbol{q} and frequency transfer iωji\omega_j correspond to those of phonons absorbed or emitted in the process. The notation iωj→ω+iηi\omega_j\to\omega+i\eta denotes an analytical continuation after summation over pip_i, which gives the dependence of G^\hat{G} on phonon frequency ω\omega. The corresponding Feynman diagram is shown in Figure 1 (b).

Formulating G^(EPC)\hat{G}^{(\mathrm{EPC})} as a linear response function provides a unified framework on chiral phonon splitting and phonon magnetic moments. First, it provides a systematic route for incorporating additional perturbations and many-body effects into G(EPC)G^{(\mathrm{EPC})}. In particular, the phonon magnetic moment studied here can be obtained by introducing magnetic-field perturbations. It also allows disorder and Coulomb-interaction effects to be included by replacing the single-particle linear response function with the disorder-averaged or interaction-dressed one. Second, it naturally builds a bridge between several earlier theories. When TRS breaking is intrinsic, Eq. (4) reduces to previously developed nonadiabatic theories in magnetic systems [55, 56, 58]. In the adiabatic limit ω→0\omega\to0, it further reduces to the molecular Berry curvature obtained within the Born-Oppenheimer approximation [2, 8, 13, 59], corresponding to the geometric phase accumulated by electrons under cyclic ionic motion. Furthermore, in the presence of magnetic fields, Eq. (4) naturally connects to the recently developed gauge theory of phonon magnetic moments in Dirac semimetals [53], where ∂H^/∂R∝∂H^/∂k\partial\hat{H}/\partial\boldsymbol{R}\propto\partial\hat{H}/\partial\boldsymbol{k}. In this case, the force operator becomes proportional to the electronic velocity operator, implying that the chiral phonon splitting under magnetic fields scales with the Hall conductivity σxy\sigma_{xy}. In the following, we make no assumptions about the form of the force operator and focus on nonmagnetic materials, where TRS is broken by an external magnetic field.

Phonon magnetic moment.—To evaluate the phonon magnetic moment, we calculate the phonon splitting to linear order in BB. More specifically, we need to calculate γκa,κ′b,c(EPC)≡ℏ∂Gκa,κ′b(EPC)/∂Bc\gamma^{(\mathrm{EPC})}_{\kappa a,\kappa' b,c} \equiv\hbar\partial G^{(\mathrm{EPC})}_{\kappa a,\kappa' b}/\partial B_c, where cc labels the Cartesian component of BB. For the two-mode model introduced earlier, this linear coefficient directly determines the phonon magnetic moment. For more general systems, one substitutes G(EPC)=γ(EPC)⋅BG^{(\mathrm{EPC})}=\boldsymbol{\gamma}^{(\mathrm{EPC})}\cdot B back into the equation of motion and solves for the resulting phonon eigenfrequencies, from which the splitting, and hence the magnetic moment, can be obtained.

To include the effect of magnetic fields in a gauge-invariant way, we transform Eq. (4) to the Wigner representation [60, 61, 62, 63]. In this representation, the magnetic field modifies operator products according to A^1A^2⇒A^1A^2+∑cieBcℏ2ϵμνc(∂pμA^1)(∂pνA^2)\hat{A}_1\hat{A}_2\Rightarrow\hat{A}_1\hat{A}_2+\sum_c\frac{ieB_c\hbar}{2}\epsilon_{\mu\nu c}(\partial_{p_\mu}\hat{A}_1)(\partial_{p_\nu}\hat{A}_2) up to the linear order in BB, where μ,ν=x,y,z\mu,\nu=x,y,z denotes the Cartesian direction of momentum pp. This leads to two types of corrections to G(EPC)G^{(\mathrm{EPC})}. The first type is to the Green’s function, leading to G^B≈G^+∑cieBcℏ2ϵμνcG^v^μG^v^νG^\hat{G}^{B}\approx\hat{G}+\sum_c\frac{ieB_c\hbar}{2}\epsilon_{\mu\nu c}\hat{G}\hat{v}_\mu\hat{G}\hat{v}_\nu\hat{G} for a non-interacting system [60, 62]. The second type of corrections is to the product between Green’s functions and force operators by replacing A^1\hat{A}_1 and A^2\hat{A}_2 above with G^\hat{G} and F^\hat{F}. Details can be found in the Supplemental Materials (SM) [63]. Combining these contributions, we obtain

γκa,κ′b,c(EPC)(ω)=eℏ2ϵμνc4ωMκMκ′{1ℏβ∑i[κa,μ,ν,κ′b]i+j+[(κa,μ),ν,κ′b]i+j−(j→−j)}iωj→ω+iη−(κa↔κ′b),(5)\gamma^{(\mathrm{EPC})}_{\kappa a,\kappa' b,c}(\omega)=\frac{e\hbar^{2}\epsilon_{\mu\nu c}}{4\omega\sqrt{M_\kappa M_{\kappa'}}}\left\{\frac{1}{\hbar\beta}\sum_i[\kappa a,\mu,\nu,\kappa' b]_{i+j}+[(\kappa a,\mu),\nu,\kappa' b]_{i+j}-(j\to-j)\right\}_{i\omega_j\to\omega+i\eta}-(\kappa a\leftrightarrow\kappa' b), \tag*{(5)}

where we define [κa,μ,ν,κ′b]i+j≡Tr⁡[F^κaG^(ipi)v^μG^(ipi)v^νG^(ipi)F^κ′bG^(ipi+iωj)][\kappa a,\mu,\nu,\kappa' b]_{i+j}\equiv\operatorname{Tr}[\hat{F}_{\kappa a}\hat{G}(i p_i)\hat{v}_\mu\hat{G}(i p_i)\hat{v}_\nu\hat{G}(i p_i)\hat{F}_{\kappa' b}\hat{G}(i p_i+i\omega_j)] and [(κa,μ),ν,κ′b]i+j≡Tr⁡[(∂κav^μ)G^(ipi)v^νG^(ipi)F^κ′bG^(ipi+iωj)][(\kappa a,\mu),\nu,\kappa' b]_{i+j}\equiv\operatorname{Tr}[(\partial_{\kappa a}\hat{v}_\mu)\hat{G}(i p_i)\hat{v}_\nu\hat{G}(i p_i)\hat{F}_{\kappa' b}\hat{G}(i p_i+i\omega_j)]. The corresponding Feynman diagrams are shown in Figure 1(c) and (Figure 1d). The second term exists only when the non-kinetic part of the Hamiltonian is non-local, so that ∂κav^μ≡∂v^μ/∂Rκa≠0\partial_{\kappa a}\hat{v}_\mu\equiv\partial\hat{v}_\mu/\partial R_{\kappa a}\ne0. In DFT, this is generally the case because pseudopotentials are non-local, so we need to keep both terms for completeness.

After completing the Matsubara frequency summation and analytical continuation, we obtain the formula for phonon magnetic moments in general systems. The formula contains not only Fermi-sea terms, but also Fermi-surface contributions, i.e.,

γκa,κ′b,c(EPC)=γκa,κ′b,c(EPC,sea)+γκa,κ′b,c(EPC,FS).(6)\gamma^{(\mathrm{EPC})}_{\kappa a,\kappa' b,c}=\gamma^{(\mathrm{EPC,sea})}_{\kappa a,\kappa' b,c}+\gamma^{(\mathrm{EPC,FS})}_{\kappa a,\kappa' b,c}. \tag*{(6)}

As a result, it can be applied to both metals and insulators. In an insulator, the Fermi-surface term vanishes. We find that the frequency dependence of the formula comes entirely from the inter-gap transition process [63]. In the low-frequency insulating limit (ℏω≪Δgap\hbar\omega\ll\Delta_{\mathrm{gap}}), the formula reduces exactly to the adiabatic results of Refs. [14, 49, 51]. The complete formula and proof can be found in the SM [63]. The current theory, however, does not rely on the adiabatic approximation and captures the full frequency-dependence of the phonon magnetic moment, making the theory applicable even when ℏω>Δgap\hbar\omega>\Delta_{\mathrm{gap}}.

The Fermi-surface contribution, absent in previous microscopic theories, is given by

γκa,κ′b,c(EPC,FS)(ω)=eℏ2ϵμνc4ωMκMκ′∑nmlp(Fκa)nm(vμ)ml(vν)lp(Fκ′b)pn×{fm′Δmp(Δmn+ℏω+iΓ)δϵm,ϵlδˉϵl,ϵp+(cyclic of m,l,p)−(ω+iΓ→−ω−iΓ)}+eℏϵμνc4ωMκMκ′∑nmlp(∂κavμ)nm(vν)ml(Fκ′b)ln{fm′Δmn+ℏω+iΓδϵm,ϵl−(ω+iΓ→−ω−iΓ)}−(κa↔κ′b).(7)\begin{aligned} \gamma^{(\mathrm{EPC,FS})}_{\kappa a,\kappa' b,c}(\omega) ={}&\frac{e\hbar^{2}\epsilon_{\mu\nu c}}{4\omega\sqrt{M_\kappa M_{\kappa'}}}\sum_{nmlp}(F_{\kappa a})_{nm}(v_\mu)_{ml}(v_\nu)_{lp}(F_{\kappa' b})_{pn}\\ &\times\left\{\frac{f'_m}{\Delta_{mp}(\Delta_{mn}+\hbar\omega+i\Gamma)}\delta_{\epsilon_m,\epsilon_l}\bar{\delta}_{\epsilon_l,\epsilon_p}+\left(\text{cyclic of }m,l,p\right)-\left(\omega+i\Gamma\to-\omega-i\Gamma\right)\right\}\\ &+\frac{e\hbar\epsilon_{\mu\nu c}}{4\omega\sqrt{M_\kappa M_{\kappa'}}}\sum_{nmlp}(\partial_{\kappa a}v_\mu)_{nm}(v_\nu)_{ml}(F_{\kappa' b})_{ln}\left\{\frac{f'_m}{\Delta_{mn}+\hbar\omega+i\Gamma}\delta_{\epsilon_m,\epsilon_l}-\left(\omega+i\Gamma\to-\omega-i\Gamma\right)\right\}\\ &-(\kappa a\leftrightarrow\kappa' b). \tag*{(7)} \end{aligned}

where Δmp=ϵm−ϵp\Delta_{mp}=\epsilon_m-\epsilon_p is the energy difference between the bands, δϵm,ϵl=1\delta_{\epsilon_m,\epsilon_l}=1 when ϵm=ϵl\epsilon_m=\epsilon_l and otherwise 0, δˉϵm,ϵl=1−δϵm,ϵl\bar{\delta}_{\epsilon_m,\epsilon_l}=1-\delta_{\epsilon_m,\epsilon_l} and we use (cyclic of m,l,pm,l,p) to denote the two additional terms obtained by cyclic permutations (m,l,p)→(l,p,m)→(p,m,l)(m,l,p)\to(l,p,m)\to(p,m,l). One may notice that Fermi-surface contribution for Δmn=0\Delta_{mn}=0 diverges when ω→0\omega\to0. This is regulated by introducing a finite electron lifetime by replacing the infinitesimal η\eta with a finite Γ\Gamma (see SM for details [63]). In practice, the linewidth Γ\Gamma can be induced by EPC, disorder and so on. To this end, we have completed the construction of a nonadiabatic theory of phonon magnetic moment applicable to both insulators and metals.

Importantly, the formulas above suggest two processes that could generate phonon magnetic moments substantially larger than the conventional adiabatic term in insulators. The first is the inter-gap resonant process mediated by phonons in the Fermi-sea contribution. This involves virtual transitions between occupied and unoccupied states whose energy difference is close to ℏω\hbar\omega. The corresponding terms take the form of (Δnm+ℏω+iΓ)−r(\Delta_{nm}+\hbar\omega+i\Gamma)^{-r}, and (near) resonance leads to enhancements proportional to the inverse of the electronic linewidth 1/(Γ)r1/(\Gamma)^r, where r=1r=1 or 22. The second mechanism is the intraband process on the Fermi surface, which contributes terms proportional to 1/(ℏω+iΓ)21/(\hbar\omega+i\Gamma)^2 in the small-ω\omega limit. In the absence of interband resonance, such as in a single-band system (e.g., free electrons) or doped Dirac semimetals where 2EF≫ℏω2E_F\gg\hbar\omega, this intraband Fermi-surface contribution dominates the phonon magnetic moment. In addition, it is worth noting that the behavior is very similar to that of optical Hall conductivity obtained from Drude theory [65].

Application to Doped Pb1−x_{1-x}Snx_xTe

We now apply our theory to a massive anisotropic Dirac k⋅pk\cdot p model described by the Hamiltonian H^=H^0+H^EPC\hat{H}=\hat{H}_0+\hat{H}_{\mathrm{EPC}}, where H^0=\hat{H}_0=

(m+ck32+fk12+fk22)s^0σ^z+vF(k1s^y−k2s^x)σ^x+v3k3s^0σ^y\left(m+ck_3^2+fk_1^2+fk_2^2\right)\hat{s}_0\hat{\sigma}_z+v_F(k_1\hat{s}_y-k_2\hat{s}_x)\hat{\sigma}_x+v_3k_3\hat{s}_0\hat{\sigma}_y

with the sign of the gap 2m2m controlling the topological order, and the coupling to phonons is given by H^EPC=ζ(uxs^0σ^x−uys^zσ^y)\hat{H}_{\mathrm{EPC}}=\zeta(u_x\hat{s}_0\hat{\sigma}_x-u_y\hat{s}_z\hat{\sigma}_y). Both s^\hat{s} and σ^\hat{\sigma} are Pauli matrices, denoting spin and orbital degrees of freedom, respectively. This model can be used to describe Pb1−x_{1-x}Snx_xTe [21, 64]. In the calculation below, the electron linewidth Γ\Gamma is set to 1 meV unless stated explicitly. Using this model, we demonstrate how finite phonon frequency and finite carrier density can amplify and even reverse the EPC-induced phonon magnetic moment.

We begin with the narrow-gap-insulator case to demonstrate the nonadiabatic effect. The gap is set to be Δgap=2m=10\Delta_{\mathrm{gap}}=2m=10 meV and the Fermi level is placed within the gap. In Figure 2(b), the dotted horizontal line shows the frequency-independent adiabatic formula in Ref. [14, 49, 51]. We can see that when ℏω≪Δ\hbar\omega\ll\Delta, our result agrees well with the adiabatic theory. When ℏω\hbar\omega approaches the gap size, the phonon magnetic moment is strongly enhanced because of the near-resonance between inter-gap states. When ℏω\hbar\omega exceeds the gap size, it undergoes an abrupt change and even reverses sign as ℏω\hbar\omega further increases. The results indicate the important role of nonadiabatic effects in determining phonon magnetic moment.

Electronic dispersion, phonon magnetic moment, and reciprocal-space distributions

Figure 2. (a) Electronic dispersion along k3k_3 at k1,k2=0k_1,k_2=0. The gap size is Δgap=2m=10\Delta_{\mathrm{gap}}=2m=10 meV. Other parameters are set as c=0.9c=0.9 eV, f=4.5f=4.5 eV, v3=0.95v_3=0.95 eV, vF=2.4v_F=2.4 eV [21, 64], ζ=0.5\zeta=0.5 eV/Å, Γ=1\Gamma=1 meV. The phonon mass is set to the reduced mass MPbMTe/(MPb+MTe)≈79M_{\mathrm{Pb}}M_{\mathrm{Te}}/(M_{\mathrm{Pb}}+M_{\mathrm{Te}})\approx79 Da. (b) Dependence of phonon magnetic moment on the phonon frequency ℏω\hbar\omega. Results using both Eq. (5) in this work and adiabatic theory in Ref. [14, 49, 51] are shown for comparison. The vertical dashed line denotes the energy where ℏω=Δgap\hbar\omega=\Delta_{\mathrm{gap}}. (c-d) Distribution of phonon magnetic moment in the reciprocal space at kz=0k_z=0. Phonon frequencies are set to the adiabatic limit (ℏω=0.1\hbar\omega=0.1 meV) and values exceeding the gap size (ℏω=12\hbar\omega=12 meV), respectively. The inset in (e) shows the distribution along kxk_x at ky=0k_y=0.

The pronounced frequency dependence originates from these near-resonant processes. In Figure 2(c,Figure 2d), we show the momentum-resolved distribution of the phonon magnetic moment. For ℏω<Δgap\hbar\omega<\Delta_{\mathrm{gap}} [Fig. 2(c)], no exact resonance occurs, but fluctuations become increasingly concentrated near k=0\boldsymbol{k}=0 as the phonon frequency increases and resonance is approached. For example, at ℏω=8\hbar\omega=8 meV, the distribution remains centered near k=0\boldsymbol{k}=0 while its magnitude increases by a factor of four (Fig. S1). When ℏω>Δgap\hbar\omega>\Delta_{\mathrm{gap}} (Fig. 2(d)), resonance sets in, producing sharp features in momentum space and accompanying a sign reversal of the phonon magnetic moment.

When the system is doped into a metallic regime, an additional Fermi-surface contribution emerges. Figure 3(a) shows both Fermi-sea and Fermi-surface contributions to the phonon magnetic moment as functions of the Fermi level EFE_F at different phonon frequencies. At low doping, the response is dominated by the Fermi-sea term due to the small number of Fermi-surface electrons. As EFE_F increases, the growing density of states at the Fermi level enhances the Fermi-surface contribution, which rapidly becomes dominant. This Fermi-surface term exhibits a strong frequency dependence, scaling approximately as (ℏω+iΓ)−2(\hbar\omega+i\Gamma)^{-2} as discussed above. Consequently, in the low-phonon-frequency limit, it can exceed the Fermi-sea contribution by up to two orders of magnitude. At higher phonon frequencies (e.g., ℏω=15\hbar\omega=15 meV), interband resonance becomes active, leading to enhanced Fermi-sea contributions and a sign change in both components when 2EF2E_F crosses ℏω\hbar\omega. These results highlight the central role of Fermi-surface contributions in determining phonon magnetic moments in metals.

Dependence of phonon magnetic moment on Fermi level and Sn concentration

Figure 3. (a) Dependence of Fermi-surface (solid lines) and Fermi-sea (dashed lines) contributions to the phonon magnetic moment on the position of the Fermi level EFE_{\mathrm{F}}. Results for phonon at ℏω=0.1\hbar\omega= 0.1 meV (top), 5 meV (middle) and 15 meV (bottom) are shown for comparison. The EFE_{\mathrm{F}}-axis is split into low-doping and high-doping level region for clarity. The gap size is Δgap=2m=10\Delta_{\mathrm{gap}} = 2m = 10 meV. (b) Dependence of the magnetic moment of a 5-meV phonon on the Sn concentration xx in Pb1−xSnxTe\mathrm{Pb}_{1-x}\mathrm{Sn}_{x}\mathrm{Te}. Gap sizes and positions of the Fermi level are determined according to the hole concentrations measured in experiments [21]. Parameters are set to ζ=0.5\zeta= 0.5 eV/Å and Γ=10\Gamma= 10 meV. The phonon mass is set to the reduced mass MPb1−xSnxMTe/(MPb1−xSnx+MTe)M_{\mathrm{Pb}_{1-x}\mathrm{Sn}_{x}}M_{\mathrm{Te}}/(M_{\mathrm{Pb}_{1-x}\mathrm{Sn}_{x}} + M_{\mathrm{Te}}).

Next, we connect our results to recent experiments in Pb1−xSnxTe\mathrm{Pb}_{1-x}\mathrm{Sn}_{x}\mathrm{Te}, where the phonon magnetic moment was measured as a function of substitution level xx. For each xx, we set the Fermi level according to the experimentally measured hole concentration and use the corresponding gap size mm. The magnetic moment is evaluated for phonons at 5 meV [21]. Figure 3(b) shows the calculated Fermi-sea and Fermi-surface contributions as functions of xx. The total γEPC\gamma^{\mathrm{EPC}} increases markedly with Sn concentration. Using an EPC strength ζ∼0.5\zeta\sim0.5 eV/Å fitted from DFT [63], a scattering rate Γ∼10\Gamma\sim10 meV estimated from mobility measurements [66], and including contributions from all four valleys, we obtain ∣γEPC∣≈0.11 μB|\gamma^{\mathrm{EPC}}| \approx0.11\,\mu_{\mathrm{B}} (x=0x = 0), 0.29 μB0.29\,\mu_{\mathrm{B}} (x=0.24x = 0.24), and 1.47 μB1.47\,\mu_{\mathrm{B}} (x=0.42x = 0.42). These values are in reasonable agreement with the experimental measurements (0.04 μB0.04\,\mu_{\mathrm{B}}, 0.12 μB0.12\,\mu_{\mathrm{B}}, and 1 μB1\,\mu_{\mathrm{B}}, respectively, and both experimentally resolved branches (see below) have comparable magnitudes [21]). At low doping, the Fermi-sea and Fermi-surface contributions are comparable, but the latter quickly becomes dominant as xx increases.

Despite this quantitative agreement, some experimental features are not captured. In particular, the two experimentally resolved phonon branches behave differently across the topological crystalline insulator transition near xc≈0.32x_{\mathrm{c}} \approx0.32: one exhibits a sign change, while the other does not. This branch-dependent behavior is absent in our calculation due to the harmonic approximation. Since the primitive cell contains two atoms, the harmonic theory supports only a single transverse optical (TO) doublet; the two observed branches arise from anharmonic splitting of this mode [21, 67, 68]. Capturing this effect requires going beyond the harmonic approximation.

**Summary.—**We have developed a nonadiabatic framework for phonon magnetic moments that applies to both insulating and metallic systems. The theory identifies two key contributions beyond the adiabatic limit: resonant interband processes and an intraband Fermi-surface term that can become dominant in metals, leading to substantial corrections to previous adiabatic estimates. As a result, the phonon magnetic moment acquires a strong dependence on both frequency and carrier density. Application to Pb1−xSnxTe\mathrm{Pb}_{1-x}\mathrm{Sn}_{x}\mathrm{Te} shows that these effects account for the experimentally observed magnitude and trends, placing phonon magnetic moments on the order of μB\mu_{\mathrm{B}}. Extending the present framework to incorporate anharmonic phonon effects remains an important direction for future work.

**Acknowledgment.—**We gratefully acknowledge Xiaodong Hu, Xiaoyu Liu, Junren Shi, Zhezhu Wei, and Xiao-Wei Zhang for valuable discussions. The theory development is supported by DOE Award No. DE-SC0012509, and the first-principles calculations and materials modeling are supported by NSF Award No. DMR-2339995.

References

  1. [1]F. Giustino, Reviews of Modern Physics 89, 015003 (2017), publisher: American Physical Society.
  2. [2]T. Qin, J. Zhou, and J. Shi, Physical Review B 86, 104305 (2012), publisher: American Physical Society.
  3. [3]L. Zhang and Q. Niu, Physical Review Letters 112, 085503 (2014), publisher: American Physical Society.
  4. [4]D. M. Juraschek, M. Fechner, A. V. Balatsky, and N. A. Spaldin, Physical Review Materials 1, 014401 (2017), publisher: American Physical Society.
  5. [5]H. Zhu, J. Yi, M.-Y. Li, J. Xiao, L. Zhang, C.-W. Yang, R. A. Kaindl, L.-J. Li, Y. Wang, and X. Zhang, Science 359, 579 (2018).
  6. [6]X. Xu, H. Chen, and L. Zhang, Physical Review B 98, 134304 (2018), publisher: American Physical Society.
  7. [7]D. M. Juraschek and N. A. Spaldin, Physical Review Materials 3, 064405 (2019), publisher: American Physical Society.
  8. [8]D. Saparov, B. Xiong, Y. Ren, and Q. Niu, Physical Review B 105, 064303 (2022), publisher: American Physical Society.
  9. [9]D. M. Juraschek, T. Neuman, and P. Narang, Physical Review Research 4, 013129 (2022), publisher: American Physical Society.
  10. [10]H. Ueda, M. García-Fernández, S. Agrestini, C. P. Romao, J. van den Brink, N. A. Spaldin, K.-J. Zhou, and U. Staub, Nature 618, 946 (2023), publisher: Nature Publishing Group.
  11. [11]D. M. Juraschek, R. M. Geilhufe, H. Zhu, M. Basini, P. Baum, A. Baydin, S. Chaudhary, M. Fechner, B. Fle bus, G. Grissonnanche, A. I. Kirilyuk, M. Lemeshko, S. F. Maehrlein, M. Mignolet, M. Murakami, Q. Niu, U. Nowak, C. P. Romao, H. Rostami, T. Satoh, N. A. Spaldin, H. Ueda, and L. Zhang, Nature Physics 21, 1532 (2025).
  12. [12]N. Shabala, F. Tietjen, and R. M. Geilhufe, Axial phono-magnetic effects (2025), arXiv:2511.03329 [cond-mat].arxiv.org/abs/2511.03329
  13. [13]J. Bonini, S. Ren, D. Vanderbilt, M. Stengel, C. E. Dreyer, and S. Coh, Physical Review Letters 130, 086701 (2023), publisher: American Physical Society.
  14. [14]Y. Ren, C. Xiao, D. Saparov, and Q. Niu, Physical Review Letters 127, 186403 (2021), publisher: American Physical Society.
  15. [15]T. Zhang, Y. Liu, H. Miao, and S. Murakami, Advances in phonons: From band topology to phonon chirality (2025), version Number: 2.arxiv.org/abs/2505.06179
  16. [16]R. Yang, Y.-Y. Zhu, M. StegIeIder, Y.-C. Liu, C.-C. Liu, X.-G. Qiu, T. Zhang, and M. Dressel, Physical Review Letters 134, 196905 (2025), publisher: American Physical Society.
  17. [17]M. Che, J. Liang, Y. Cui, H. Li, B. Lu, W. Sang, X. Li, X. Dong, L. Zhao, S. Zhang, T. Sun, W. Jiang, E. Liu, F. Jin, T. Zhang, and L. Yang, Physical Review Letters 134, 196906 (2025), publisher: American Physical Society.
  18. [18]D. Li, G. Yang, T. Qin, J. Zhou, and Y. Yao, Phonon Dichroisms Revealing Unusual Electronic Quantum Geometry (2025), arXiv:2511.16141 [cond-mat].arxiv.org/abs/2511.16141
  19. [19]B. Cheng, T. Schumann, Y. Wang, X. Zhang, D. Barbalas, S. Stemmer, and N. P. Armitage, Nano Letters 20, 5991 (2020), publisher: American Chemical Society.
  20. [20]A. Baydin, F. G. Hernandez, M. Rodriguez-Vega, A. K. Okazaki, F. Tay, G. T. Noe, I. Katayama, J. Takeda, H. Nojiri, P. H. Rappl, E. Abramof, G. A. Fiete, and J. Kono, Physical Review Letters 128, 075901 (2022), publisher: American Physical Society.
  21. [21]F. G. Hernandez, A. Baydin, S. Chaudhary, F. Tay, I. Katayama, J. Takeda, H. Nojiri, A. K. Okazaki, P. H. Rappl, E. Abramof, M. Rodriguez-Vega, G. A. Fiete, and J. Kono, Science Advances 9, eadj4074 (2023), publisher: American Association for the Advancement of Science.
  22. [22]C. Strohm, G. L. J. A. Rikken, and P. Wyder, Physical Review Letters 95, 155901 (2005), publisher: American Physical Society.
  23. [23]T. Ideue, T. Kurumaji, S. Ishiwata, and Y. Tokura, Nature Materials 16, 797 (2017).
  24. [24]T. Saito, K. Misaki, H. Ishizuka, and N. Nagaosa, Physical Review Letters 123, 255901 (2019), publisher: American Physical Society.
  25. [25]X. Zhang, Y. Zhang, S. Okamoto, and D. Xiao, Physical Review Letters 123, 167202 (2019), publisher: American Physical Society.
  26. [26]M.-E. Boulanger, G. Grissonnanche, S. Badoux, A. Allaire, . Lefrançois, A. Legros, A. Gourgout, M. Dion, C. H. Wang, X. H. Chen, R. Liang, W. N. Hardy, D. A. Bonn, and L. Taillefer, Nature Communications 11, 5325 (2020).
  27. [27]X. Li, B. Fauqué, Z. Zhu, and K. Behnia, Physical Review Letters 124, 105901 (2020), publisher: American Physical Society.
  28. [28]J.-Y. Chen, S. A. Kivelson, and X.-Q. Sun, Physical Review Letters 124, 167601 (2020), publisher: American Physical Society.
  29. [29]K. Sun, Z. Gao, and J.-S. Wang, Physical Review B 103, 214301 (2021), publisher: American Physical Society.
  30. [30]A. Zabalo, C. E. Dreyer, and M. Stengel, Physical Review B 105, 094305 (2022), publisher: American Physical Society.
  31. [31]X. Li, Y. Machida, A. Subedi, Z. Zhu, L. Li, and K. Behnia, Nature Communications 14, 1027 (2023), number: 1 Publisher: Nature Publishing Group.
  32. [32]A. Ataei, G. Grissonnanche, M.-E. Boulanger, L. Chen, . Lefrançois, V. Brouet, and L. Taillefer, Nature Physics 20, 585 (2024), publisher: Nature Publishing Group.
  33. [33]T. Oh and N. Nagaosa, Physical Review X 15, 011036 (2025), publisher: American Physical Society.
  34. [34]L. Mangeolle and J. Knolle, Physical Review X 16, 011048 (2026), publisher: American Physical Society.
  35. [35]C. Dornes, Y. Acremann, M. Savoini, M. Kubli, M. J. Neugebauer, E. Abreu, L. Huber, G. Lantz, C. a. F. Vaz, H. Lemke, E. M. Bothschafter, M. Porer, V. Esposito, L. Rettig, M. Buzzi, A. Alberca, Y. W. Windsor, P. Beaud, U. Staub, D. Zhu, S. Song, J. M. Glownia, and S. L. Johnson, Nature 565, 209 (2019), publisher: Nature Publishing Group.
  36. [36]H. Zhang, N. Peschcherenko, F. Yang, T. Z. Ward, P. Raghunvanshi, L. Lindsay, C. Felser, Y. Zhang, J.-Q. Yan, and H. Miao, Nature Physics 21, 1387 (2025), publisher: Nature Publishing Group.
  37. [37]T. F. Nova, A. Cartella, A. Cantalupi, M. Först, D. Bossini, R. V. Mikhaylovskiy, A. V. Kimpel, R. Merlin, and A. Cavalleri, Nature Physics 13, 132 (2017), publisher: Nature Publishing Group.
  38. [38]J. Holanda, D. S. Maior, A. Azevedo, and S. M. Rezende, Nature Physics 14, 500 (2018), publisher: Nature Publishing Group.
  39. [39]A. S. Disa, M. Fechner, T. F. Nova, B. Liu, M. Först, D. Prabhakaran, P. G. Radaelli, and A. Cavalleri, Nature Physics 16, 937 (2020), publisher: Nature Publishing Group.
  40. [40]D. Afanasiev, J. R. Hortensius, B. A. Ivanov, A. Sasani, E. Bousquet, Y. M. Blanter, R. V. Mikhaylovskiy, A. V. Kimel, and A. D. Caviglia, Nature Materials 20, 607 (2021), publisher: Nature Publishing Group.
  41. [41]A. Stupakiewicz, C. S. Davies, K. Szerenos, D. Afanasiev, K. S. Rabinovich, A. V. Boris, A. Caviglia, A. V. Kimel, and A. Kirilyuk, Nature Physics 17, 489 (2021), publisher: Nature Publishing Group.
  42. [42]E. A. Mashkovich, K. A. Grishunin, R. M. Dubrovin, A. K. Zvezdin, R. V. Pisarev, and A. V. Kimel, Science 374, 1608 (2021), publisher: American Association for the Advancement of Science.
  43. [43]R. Sasaki, Y. Nii, and Y. Onose, Nature Communications 12, 2599 (2021), publisher: Nature Publishing Group.
  44. [44]J. Luo, T. Lin, J. Zhang, X. Chen, E. R. Blackert, R. Xu, B. I. Jakobson, and H. Zhu, Science 382, 698 (2023), publisher: American Association for the Advancement of Science.
  45. [45]Y. Ren, M. Rudner, and D. Xiao, Physical Review Letters 132, 096702 (2024), publisher: American Physical Society.
  46. [46]M. Basini, M. Pancaldi, B. Wehinger, M. Udina, V. Unikandannunni, T. Tadano, M. C. Hoffmann, A. V. Balatsky, and S. Bonetti, Nature 628, 534 (2024), publisher: Nature Publishing Group.
  47. [47]N. Shabala and R. M. Geilhufe, Physical Review Letters 133, 266702 (2024), publisher: American Physical Society.
  48. [48]L. Klebl, A. Schobert, M. Eckstein, G. Sangiovanni, A. V. Balatsky, and T. O. Wehling, Physical Review Letters 134, 016705 (2025), publisher: American Physical Society.
  49. [49]X.-W. Zhang, Y. Ren, C. Wang, T. Cao, and D. Xiao, Physical Review Letters 130, 226302 (2023), publisher: American Physical Society.
  50. [50]C. Xiao, Y. Ren, and B. Xiong, Physical Review B 103, 115432 (2021), publisher: American Physical Society.
  51. [51]L. Trifunovic, S. Ono, and H. Watanabe, Physical Review B 100, 054408 (2019), publisher: American Physical Society.
  52. [52]R. Xue, Z. Qiao, Y. Gao, and Q. Niu, Physical Review Letters 135, 106605 (2025), publisher: American Physical Society.
  53. [53]W. Chen, X.-W. Zhang, Y. Su, T. Cao, D. Xiao, and S.-Z. Lin, Physical Review B 111, 035126 (2025), publisher: American Physical Society.
  54. [54]W. Chen, X.-W. Zhang, T. Cao, S.-Z. Lin, and D. Xiao, Geometric Origin of Phonon Magnetic Moment in Dirac Materials (2025), arXiv:2505.09732 [cond-mat].arxiv.org/abs/2505.09732
  55. [55]D. Liu and J. Shi, Physical Review Letters 119, 075301 (2017), publisher: American Physical Society.
  56. [56]F. Wang, X. Liu, H. Sun, H. Wang, S. Murakami, L. Zhang, H. Zhang, and D. Xing, Physical Review Letters 135, 256701 (2025), publisher: American Physical Society.
  57. [57]G. Giuliani and G. Vignale, Quantum Theory of the Electron Liquid, 1st ed. (Cambridge University Press, 2005).
  58. [58]O. Bistritzer, F. Mauri, and M. Calandra, Physical Review Letters 126, 225703 (2021), publisher: American Physical Society.
  59. [59]C. A. Mead, Reviews of Modern Physics 64, 51 (1992), publisher: American Physical Society.
  60. [60]S. Onoda, N. Sugimoto, and N. Nagaosa, Progress of Theoretical Physics 116, 61 (2006).
  61. [61]G. Zhu, S. A. Yang, C. Fang, W. M. Liu, and Y. Yao, Physical Review B 86, 214415 (2012).
  62. [62]A. Shitade, Gradient expansion formalism for generic spin torques (2017), arXiv:1708.03424 [cond-mat].arxiv.org/abs/1708.03424
  63. [63]See Supplemental Material at [{URL}] will be inserted by publisher] for derivations, numerical details, and supplemental numerical results, which includes Refs. [69-74].
  64. [64]A. Lau and C. Ortix, Physical Review Letters 122, 186801 (2019).
  65. [65]S. A. Hartnoll, A. Lucas, and S. Sachdev, Holographic quantum matter (2018), arXiv:1612.07324 [hep-th].arxiv.org/abs/1612.07324
  66. [66]B. A. Assaf, T. Phuphachong, V. V. Volobuev, G. Bauer, G. Springholz, L.-A. De Vaulchier, and Y. Guldner, npj Quantum Materials 2, 26 (2017).
  67. [67]O. Delaire, J. Ma, K. Marty, A. F. May, M. A. McGuire, M.-H. Du, D. J. Singh, A. Podlesnyak, G. Ehlers, M. D. Lumsden, and B. C. Sales, Nature Materials 10, 614 (2011), publisher: Nature Publishing Group.
  68. [68]G. A. S. Ribeiro, L. Paulatto, R. Bianco, I. Errea, F. Mauri, and M. Calandra, Physical Review B 97, 014306 (2018).

Paper details

vlnc:W_szqhqa2m
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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