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 PbSnTe, 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 , 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 . 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 , 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 , whose eigenvalue is the squared frequency of the -th phonon branch with quasi-wavevector . The corresponding eigenmode describes the vibrating pattern of ions [1], where denotes the equilibrium positions of ions and 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 , where is the ion momentum, and 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 and . In this subspace, and
Solving the equation of motion, the eigenmodes become , corresponding to right- and left-handed circularly polarized vibrations. In the weak-coupling limit , the resulting chiral-phonon splitting is
Under a magnetic field , the term proportional to directly defines the phonon magnetic moment. In the following, we will consider general cases that the matrices and contain all ionic degrees of freedom, and focus on the TRS-breaking effect in the matrix .
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 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.
where is the bare charge of ion , is its mass and is the magnetic field. The Levi-Civita tensor denotes antisymmetric permutations of Cartesian directions , and .
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 as the response function relating the force acting on ion to the velocity of another ion . Within this formulation, naturally takes the form of a Kubo-type response function. By noticing that , for a non-interacting system, we obtain [57]
Here stands for electron Green’s functions with quasi-wavevector and fermionic Matsubara frequency , with being the inverse
temperature. The wavevector and frequency transfer correspond to those of phonons absorbed or emitted in the process. The notation denotes an analytical continuation after summation over , which gives the dependence of on phonon frequency . The corresponding Feynman diagram is shown in Figure 1 (b).
Formulating 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 . 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 , 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 . 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 . 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 . More specifically, we need to calculate , where labels the Cartesian component of . For the two-mode model introduced earlier, this linear coefficient directly determines the phonon magnetic moment. For more general systems, one substitutes 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 up to the linear order in , where denotes the Cartesian direction of momentum . This leads to two types of corrections to . The first type is to the Green’s function, leading to 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 and above with and . Details can be found in the Supplemental Materials (SM) [63]. Combining these contributions, we obtain
where we define and . 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 . 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.,
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 (), 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 .
The Fermi-surface contribution, absent in previous microscopic theories, is given by
where is the energy difference between the bands, when and otherwise 0, and we use (cyclic of ) to denote the two additional terms obtained by cyclic permutations . One may notice that Fermi-surface contribution for diverges when . This is regulated by introducing a finite electron lifetime by replacing the infinitesimal with a finite (see SM for details [63]). In practice, the linewidth 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 . The corresponding terms take the form of , and (near) resonance leads to enhancements proportional to the inverse of the electronic linewidth , where or . The second mechanism is the intraband process on the Fermi surface, which contributes terms proportional to in the small- limit. In the absence of interband resonance, such as in a single-band system (e.g., free electrons) or doped Dirac semimetals where , 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 PbSnTe
We now apply our theory to a massive anisotropic Dirac model described by the Hamiltonian , where
with the sign of the gap controlling the topological order, and the coupling to phonons is given by . Both and are Pauli matrices, denoting spin and orbital degrees of freedom, respectively. This model can be used to describe PbSnTe [21, 64]. In the calculation below, the electron linewidth 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 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 , our result agrees well with the adiabatic theory. When approaches the gap size, the phonon magnetic moment is strongly enhanced because of the near-resonance between inter-gap states. When exceeds the gap size, it undergoes an abrupt change and even reverses sign as further increases. The results indicate the important role of nonadiabatic effects in determining phonon magnetic moment.

Figure 2. (a) Electronic dispersion along at . The gap size is meV. Other parameters are set as eV, eV, eV, eV [21, 64], eV/Å, meV. The phonon mass is set to the reduced mass Da. (b) Dependence of phonon magnetic moment on the phonon frequency . 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 . (c-d) Distribution of phonon magnetic moment in the reciprocal space at . Phonon frequencies are set to the adiabatic limit ( meV) and values exceeding the gap size ( meV), respectively. The inset in (e) shows the distribution along at .
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 [Fig. 2(c)], no exact resonance occurs, but fluctuations become increasingly concentrated near as the phonon frequency increases and resonance is approached. For example, at meV, the distribution remains centered near while its magnitude increases by a factor of four (Fig. S1). When (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 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 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 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., meV), interband resonance becomes active, leading to enhanced Fermi-sea contributions and a sign change in both components when crosses . These results highlight the central role of Fermi-surface contributions in determining phonon magnetic moments in metals.

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 . Results for phonon at meV (top), 5 meV (middle) and 15 meV (bottom) are shown for comparison. The -axis is split into low-doping and high-doping level region for clarity. The gap size is meV. (b) Dependence of the magnetic moment of a 5-meV phonon on the Sn concentration in . Gap sizes and positions of the Fermi level are determined according to the hole concentrations measured in experiments [21]. Parameters are set to eV/Å and meV. The phonon mass is set to the reduced mass .
Next, we connect our results to recent experiments in , where the phonon magnetic moment was measured as a function of substitution level . For each , we set the Fermi level according to the experimentally measured hole concentration and use the corresponding gap size . 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 . The total increases markedly with Sn concentration. Using an EPC strength eV/Å fitted from DFT [63], a scattering rate meV estimated from mobility measurements [66], and including contributions from all four valleys, we obtain (), (), and (). These values are in reasonable agreement with the experimental measurements (, , and , 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 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 : 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 shows that these effects account for the experimentally observed magnitude and trends, placing phonon magnetic moments on the order of . 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]F. Giustino, Reviews of Modern Physics 89, 015003 (2017), publisher: American Physical Society.
- [2]T. Qin, J. Zhou, and J. Shi, Physical Review B 86, 104305 (2012), publisher: American Physical Society.
- [3]L. Zhang and Q. Niu, Physical Review Letters 112, 085503 (2014), publisher: American Physical Society.
- [4]D. M. Juraschek, M. Fechner, A. V. Balatsky, and N. A. Spaldin, Physical Review Materials 1, 014401 (2017), publisher: American Physical Society.
- [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]X. Xu, H. Chen, and L. Zhang, Physical Review B 98, 134304 (2018), publisher: American Physical Society.
- [7]D. M. Juraschek and N. A. Spaldin, Physical Review Materials 3, 064405 (2019), publisher: American Physical Society.
- [8]D. Saparov, B. Xiong, Y. Ren, and Q. Niu, Physical Review B 105, 064303 (2022), publisher: American Physical Society.
- [9]D. M. Juraschek, T. Neuman, and P. Narang, Physical Review Research 4, 013129 (2022), publisher: American Physical Society.
- [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]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]N. Shabala, F. Tietjen, and R. M. Geilhufe, Axial phono-magnetic effects (2025), arXiv:2511.03329 [cond-mat].arxiv.org/abs/2511.03329
- [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]Y. Ren, C. Xiao, D. Saparov, and Q. Niu, Physical Review Letters 127, 186403 (2021), publisher: American Physical Society.
- [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]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]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]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]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]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]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]C. Strohm, G. L. J. A. Rikken, and P. Wyder, Physical Review Letters 95, 155901 (2005), publisher: American Physical Society.
- [23]T. Ideue, T. Kurumaji, S. Ishiwata, and Y. Tokura, Nature Materials 16, 797 (2017).
- [24]T. Saito, K. Misaki, H. Ishizuka, and N. Nagaosa, Physical Review Letters 123, 255901 (2019), publisher: American Physical Society.
- [25]X. Zhang, Y. Zhang, S. Okamoto, and D. Xiao, Physical Review Letters 123, 167202 (2019), publisher: American Physical Society.
- [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]X. Li, B. Fauqué, Z. Zhu, and K. Behnia, Physical Review Letters 124, 105901 (2020), publisher: American Physical Society.
- [28]J.-Y. Chen, S. A. Kivelson, and X.-Q. Sun, Physical Review Letters 124, 167601 (2020), publisher: American Physical Society.
- [29]K. Sun, Z. Gao, and J.-S. Wang, Physical Review B 103, 214301 (2021), publisher: American Physical Society.
- [30]A. Zabalo, C. E. Dreyer, and M. Stengel, Physical Review B 105, 094305 (2022), publisher: American Physical Society.
- [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]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]T. Oh and N. Nagaosa, Physical Review X 15, 011036 (2025), publisher: American Physical Society.
- [34]L. Mangeolle and J. Knolle, Physical Review X 16, 011048 (2026), publisher: American Physical Society.
- [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]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]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]J. Holanda, D. S. Maior, A. Azevedo, and S. M. Rezende, Nature Physics 14, 500 (2018), publisher: Nature Publishing Group.
- [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]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]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]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]R. Sasaki, Y. Nii, and Y. Onose, Nature Communications 12, 2599 (2021), publisher: Nature Publishing Group.
- [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]Y. Ren, M. Rudner, and D. Xiao, Physical Review Letters 132, 096702 (2024), publisher: American Physical Society.
- [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]N. Shabala and R. M. Geilhufe, Physical Review Letters 133, 266702 (2024), publisher: American Physical Society.
- [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]X.-W. Zhang, Y. Ren, C. Wang, T. Cao, and D. Xiao, Physical Review Letters 130, 226302 (2023), publisher: American Physical Society.
- [50]C. Xiao, Y. Ren, and B. Xiong, Physical Review B 103, 115432 (2021), publisher: American Physical Society.
- [51]L. Trifunovic, S. Ono, and H. Watanabe, Physical Review B 100, 054408 (2019), publisher: American Physical Society.
- [52]R. Xue, Z. Qiao, Y. Gao, and Q. Niu, Physical Review Letters 135, 106605 (2025), publisher: American Physical Society.
- [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]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]D. Liu and J. Shi, Physical Review Letters 119, 075301 (2017), publisher: American Physical Society.
- [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]G. Giuliani and G. Vignale, Quantum Theory of the Electron Liquid, 1st ed. (Cambridge University Press, 2005).
- [58]O. Bistritzer, F. Mauri, and M. Calandra, Physical Review Letters 126, 225703 (2021), publisher: American Physical Society.
- [59]C. A. Mead, Reviews of Modern Physics 64, 51 (1992), publisher: American Physical Society.
- [60]S. Onoda, N. Sugimoto, and N. Nagaosa, Progress of Theoretical Physics 116, 61 (2006).
- [61]G. Zhu, S. A. Yang, C. Fang, W. M. Liu, and Y. Yao, Physical Review B 86, 214415 (2012).
- [62]A. Shitade, Gradient expansion formalism for generic spin torques (2017), arXiv:1708.03424 [cond-mat].arxiv.org/abs/1708.03424
- [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]A. Lau and C. Ortix, Physical Review Letters 122, 186801 (2019).
- [65]S. A. Hartnoll, A. Lucas, and S. Sachdev, Holographic quantum matter (2018), arXiv:1612.07324 [hep-th].arxiv.org/abs/1612.07324
- [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]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]G. A. S. Ribeiro, L. Paulatto, R. Bianco, I. Errea, F. Mauri, and M. Calandra, Physical Review B 97, 014306 (2018).
