betaFeedback

WorksPhysics and Astronomy

Unified theory of the photovoltaic Hall effect by field- and light-induced Berry curvatures

Abstract

Photovoltaic Hall effect, i.e., generation of a photocurrent perpendicular to the bias electric field, is an interesting platform of Berry curvature engineering by external fields. Floquet engineering aims at generation of light-induced Berry curvature associated with topological phase transition in solids, which may manifest itself as a light-induced anomalous Hall effect. However, recent studies have pointed out a larger contribution by momentum asymmetry of photocarriers, termed a field-induced circular photogalvanic effect. Except for numerical studies, the two mechanisms have been described by different theoretical frameworks, hindering a coherent understanding. Here, we develop a unified theory of the photovoltaic Hall effect capable of describing both mechanisms on an equal footing. We reveal that the bias electric field alters the interband transition dipole moment and transition energy, both contributing to the field-induced circular photogalvanic effect in nonmagnetic materials. These effects are governed by an electric field-induced Berry curvature and the shift vector coupled to bias field, respectively. A resonant enhancement of the transverse photocurrent is found in GaAs owing to the topological character of the valence band. We also clearly distinguish the anomalous Hall effect by light-dressed states within the density matrix calculation using the length gauge. Our theory unifies a number of nonlinear optical processes in a physically transparent way and presents a geometric picture of the third-order nonlinear response under light and bias fields, shedding new light on Berry curvature engineering.

Introduction

Berry curvature associated with a geometric phase is essential in determining the properties of materials with broken time-reversal (TT) or inversion (II) symmetry [1, 2, 3]. In the TT-breaking case, Berry curvature leads to the anomalous Hall effect (AHE), in which an electric current generates a transverse voltage without an external magnetic field [4, 5, 6]. A close relationship between geometry and topology has guided the discovery of a quantized version of the AHE [7, 8, 9]. By contrast, in the II-breaking case, Berry curvature leads to the circular photogalvanic effect (CPGE), where circularly polarized light generates a photocurrent known as an injection current [10, 11, 12, 13]. The possibility of a quantized CPGE has been discussed for topological Weyl semimetals [14, 15, 16]. Even in nonmagnetic and centrosymmetric materials, Berry curvature engineering by external fields can be divided to induce these effects, because circularly polarized light and an electric field can be used to break TT and II, respectively.

Existing studies on the photovoltaic Hall effect, i.e., generation of a photocurrent perpendicular to a bias electric field, have focused on the function of circularly polarized light that breaks TT to induce the AHE. The mechanisms include Floquet engineering, which modifies the band structure to generate a transient Berry curvature [17, 18, 19, 20, 21, 22, 23, 24, 25], and the optical selection rule, which allows interband excitation of electrons carrying a nonvanishing Berry curvature [26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36]. However, recent studies have shown that momentum asymmetry of photoexcited carriers contributes more significantly to the Hall current, when the light and bias fields are applied simultaneously [37, 38, 39, 40, 41]. Although this mechanism has been conceived to be a field-induced CPGE caused by II breaking under a bias electric field [40, 41], its relation with the field-induced Berry curvature has not been addressed yet. Deeper insight into the field-induced CPGE is required to clarify the geometric aspects of light–matter interaction. Furthermore, it is strongly demanded to synthesize the contributions by light-dressed Floquet states and excited carriers into a single theoretical framework, since the existing theories have described them separately using different techniques.

In this study, we develop a general theory of the photovoltaic Hall effect in the third-order nonlinear regime, treating the field-induced CPGE and the light-induced AHE on an equal footing. First, we elucidate the physical processes underlying the field-induced CPGE, which are summarized in Figure 1. We establish the relationship between the field-induced CPGE and field-induced Berry curvature for the first time, as illustrated in Figure 1(a). We also find an additional contribution to the Hall current by a change in the transition energy, which is illustrated in Figure 1(b). This effect involves the shift vector, another geometric quantity that measures the spatial shift of electron clouds accompanying interband transitions [10]. The field-induced Berry curvature and energy shift, combined with the anomalous velocity of photoexcited carriers [Fig. 1(c)], describe the photovoltaic Hall effect in nonmagnetic insulators. We demonstrate our results for a massive Dirac electron system and a typical semiconductor, GaAs. The topological character of the valence band in the latter leads to divergence in all the mechanisms, explaining resonance peaks in the photon energy dependence of the photovoltaic Hall effect observed recently [41]. Next, we calculate the contribution by light-dressed states in metallic systems using the same density matrix method. Our result not only agrees with the prediction by Floquet theory but also reveals a close link between light-induced AHE and optical rectification, both related to a light-induced Berry connection. These findings provide a unified picture of the photovoltaic Hall effect and underline its geometric aspect.

Schematic diagrams of Berry curvature, energy shift, and anomalous velocity

Figure 1. (a) Band-resolved Berry curvature ΔΩnm\Delta\Omega_{nm} generated by a bias electric field E0\mathbf{E}_0, which causes a momentum asymmetry of photoexcited carriers. (b) Field-induced energy shift of interband transitions, arising from the potential gradient corresponding to the shift vector Rnm(e)\mathbf{R}_{nm}(\mathbf{e}). (c) Anomalous velocity of photoexcited carriers generated by the Berry curvature Ωn\Omega_n.

This paper is organized as follows. In Sec. II, we derive the field-induced Berry curvature and prove its role in the field-induced CPGE. Section III reveals another contribution arising from a coupling between the shift vector and electric field. Anomalous velocity of photocarriers is briefly outlined in Sec. IV to obtain a comprehensive picture of the photovoltaic Hall effect in insulators. Photon energy dependence of each mechanism is analyzed in § 4, revealing an enhancement of the transverse current by a degeneracy point in band structure. § 5 completes the theory by presenting the contribution from light-dressed states and shows its relationship with optical rectification. § 6 compares our theory with previous studies. Finally, § 7 summarizes our findings.

Field-induced Berry curvature

We begin with a brief review of the CPGE. In a nonmagnetic insulator, it is described by [10, 14, 42]

jinj=e32ℏ2∑k∑n∑m>nfnm(Γnm⋅Ωnm)(∇kωnm)(1)\mathbf{j}_{\mathrm{inj}}=\frac{e^3}{2\hbar^2}\sum_{\mathbf{k}}\sum_n\sum_{m>n} f_{nm}\left(\mathbf{\Gamma}_{nm}\cdot\mathbf{\Omega}_{nm}\right)\left(\nabla_{\mathbf{k}}\omega_{nm}\right) \tag*{(1)}

where e<0e<0 denotes the electron charge, fnmf_{nm} is the difference in occupancy between the initial (mm) and final (nn) states, and ωnm\omega_{nm} is the transition frequency. k\mathbf{k} dependence of each quantity is omitted for brevity. (1) consists of velocity (∇kωnm\nabla_{\mathbf{k}}\omega_{nm}) and the number (∝fnmΓnm⋅Ωnm\propto f_{nm}\mathbf{\Gamma}_{nm}\cdot\mathbf{\Omega}_{nm}) of photoexcited carriers. The latter depends on two material characteristics: transition energy ℏωnm\hbar\omega_{nm}, which is included in Γnm=2πE1(ω)×\mathbf{\Gamma}_{nm}=2\pi\mathbf{E}_1(\omega)\times

E1∗(ω)δ(ω−ωnm),\mathbf{E}_1^*(\omega)\delta(\omega-\omega_{nm}),

and band-resolved Berry curvature Ωnm=iξnm×ξnm\mathbf{\Omega}_{nm}=i\boldsymbol{\xi}_{nm}\times\boldsymbol{\xi}_{nm}, derived from the Berry connection matrix ξnm=⟨un∣∇k∣um⟩\boldsymbol{\xi}_{nm}=\langle u_n| \nabla_{\mathbf{k}} |u_m\rangle. The well-known Berry curvature Ωn=∇k×ξnn\mathbf{\Omega}_n=\nabla_{\mathbf{k}}\times\boldsymbol{\xi}_{nn} is equal to the sum of band-resolved one, ∑m≠nΩnm\sum_{m\ne n}\mathbf{\Omega}_{nm}.

Appearance of Ωnm\mathbf{\Omega}_{nm} in Eq. (1) can be understood as follows. The transition matrix element for an interband transition m→nm\to n is given by Mnm(e)=(e⋅ξnm)(e∗⋅ξnm)M_{nm}(\mathbf{e})=(\mathbf{e}\cdot\boldsymbol{\xi}_{nm})(\mathbf{e}^*\cdot\boldsymbol{\xi}_{nm}), where e\mathbf{e} denotes the polarization vector of light. This equation enables us to relate the band-resolved Berry curvature to the circular dichroism,

Ωnmz=Mnm(eR)−Mnm(eL),(2)\Omega_{nm}^{z}=M_{nm}(\mathbf{e}_{\mathrm{R}})-M_{nm}(\mathbf{e}_{\mathrm{L}}), \tag*{(2)}

where eL,R=(1/2)(1,±i,0)\mathbf{e}_{\mathrm{L,R}}=(1/\sqrt{2})(1,\pm i,0) is the polarization vector for left- and right-circularly polarized light propagating in the zz direction. The band-resolved Berry curvature thus quantifies the circular dichroism at each k\mathbf{k} point. When this circular dichroism is combined with the inversion symmetry breaking, a photocurrent is generated in a direction dependent on the helicity of incident light, which is nothing but the CPGE. In two-band systems, the band-resolved Berry curvature is equivalent to the total Berry curvature, i.e., Ω1=Ω12\mathbf{\Omega}_1=\mathbf{\Omega}_{12} and Ω2=Ω21\mathbf{\Omega}_2=\mathbf{\Omega}_{21}. In this case, CPGE is described purely by the total Berry curvature, which makes geometrical and topological consideration easy [12, 14]. In systems with more energy bands, the band-resolved Berry curvature deviates from the mathematical definition of curvature unless it is summed over bands, hindering simple link with geometry or topology. Nevertheless, Ωnm\mathbf{\Omega}_{nm} can be considered to quantify how much the mm-th band bends the subspace of nn-th band.

When a static electric field E0\mathbf{E}_0 is applied, the interband Berry connection acquires a correction Δξnm=CnmE0\Delta\boldsymbol{\xi}_{nm}=\mathbf{C}_{nm}\mathbf{E}_0, where

Cnmab=ieℏDa(ξnmbωnm)+eℏ∑l≠n,m(ξnlaξlmbωlm−ξnlbξlmaωnl),(3)C_{nm}^{ab}=\frac{ie}{\hbar}D^a\left(\frac{\xi_{nm}^{b}}{\omega_{nm}}\right)+\frac{e}{\hbar}\sum_{l\ne n,m}\left(\frac{\xi_{nl}^{a}\xi_{lm}^{b}}{\omega_{lm}}-\frac{\xi_{nl}^{b}\xi_{lm}^{a}}{\omega_{nl}}\right), \tag*{(3)}

is the interband Berry connection polarizability, and DaΩnm=(∂a−iξnna+iξmma)ΩnmD^a\Omega_{nm}=(\partial^a-i\xi_{nn}^a+i\xi_{mm}^a)\Omega_{nm} is a generalized [43] or covariant [44] derivative. Among the Berry connection polarizability, intraband components (n=mn=m) were first recognized in the theory of nonlinear transport [45, 46, 47]. (3) extends this concept to the interband components (n≠mn\ne m). The modified Berry connection leads to a correction ΔΩnm=ΠnmE0\Delta\mathbf{\Omega}_{nm}=\mathbf{\Pi}_{nm}\mathbf{E}_0 to the band-resolved Berry curvature, governed by another polarizability

Πnmad=i∑bcϵabcdTnmbcd.(4)\Pi_{nm}^{ad}=i\sum_{bc}\epsilon^{abcd}T_{nm}^{bcd}. \tag*{(4)}

Here, ϵabc\epsilon^{abc} is the Levi-Civita symbol, and Tnmbcd=ξnmbCmncd+CnmbdξmncT_{nm}^{bcd}=\xi_{nm}^{b}C_{mn}^{cd}+C_{nm}^{bd}\xi_{mn}^{c}. The electric field-induced Berry curvature adds a correction to Eq. (1),

jinj,dip=e32ℏ2∑k∑n∑m>nfnm(Γnm⋅ΔΩnm)(∇kωnm),(5)\mathbf{j}_{\mathrm{inj,dip}} = \frac{e^3}{2\hbar^2} \sum_{\mathbf{k}} \sum_n \sum_{m>n} f_{nm} \left(\mathbf{\Gamma}_{nm} \cdot\Delta\mathbf{\Omega}_{nm}\right)\left(\nabla_{\mathbf{k}}\omega_{nm}\right), \tag*{(5)}

which can be nonzero even in centrosymmetric materials. Although we presented equations for a non-degenerate case, systems with degenerate bands can be treated similarly by appropriately choosing the bases (see Appendix C). A more general expression applicable to any choice of bases is given by Eq. (37) in Appendix § 10.

As a simple example, we consider a massive Dirac electron system shown in Figure 2(a). The dispersion relation is given by ϵ1,2=±Λ\epsilon_{1,2} = \pm\Lambda, where Λ=Δ2+ℏ2v2k2\Lambda= \sqrt{\Delta^2 + \hbar^2 v^2 k^2}. We set 2Δ=2.32\Delta= 2.3 eV and v=1×106v = 1 \times10^6 m/s, considering a lead halide perovskite [48]. This group of materials is expected to be useful in spintronics [49]. The field-induced Berry curvature takes the form of ΔΩ12=(e4v4/2Λ5)k×E0\Delta\mathbf{\Omega}_{12} = (e^4v^4/2\Lambda^5)\mathbf{k} \times\mathbf{E}_0 [47]. Figure 2(b) illustrates the zz component of ΔΩ12\Delta\mathbf{\Omega}_{12} on the kz=0k_z = 0 plane, assuming a bias field in the xx direction. A sign change occurs across the horizontal axis, generating a photocurrent in the yy direction following Eq. (5).

Band structure and field-induced Berry curvature for a massive Dirac electron system and GaAs

Figure 2. (a) Band structure of a massive Dirac electron system. (b) Field-induced correction to the zz component of the band-resolved Berry curvature in a massive Dirac electron system. A bias field E0=10E_0 = 10 kV/cm in the xx direction is assumed. (c) Band structure of GaAs described using an eight-band Kane model. (d) Field-induced correction to the zz component of the band-resolved Berry curvature between CB and HH in GaAs. A bias field E0=10E_0 = 10 kV/cm in the xx direction is assumed. kzk_z is fixed at 00 nm−1^{-1}, and kxk_x is varied as indicated in the figure. The curve for kx=0k_x = 0 nm−1^{-1} diverges at ky=0k_y = 0 nm−1^{-1}.

A semiconductor GaAs exhibits a unique behavior despite its prototypical appearance. Figure 2(c) shows the

band structure approximated using a first-order Kane model. The conduction (CB), heavy-hole (HH), light-hole (LH), and split-off bands were labeled 1 to 4, respectively. We calculated the field-induced Berry curvature near the Γ\Gamma point using the k⋅p\mathbf{k} \cdot\mathbf{p} perturbation method. While ΔΩ14∝eE0×k\Delta\mathbf{\Omega}_{14} \propto e\mathbf{E}_0 \times\mathbf{k} exhibits a structure similar to that of a massive Dirac system, ΔΩ12=−ΔΩ13=(3e/2Egk2)E0×k\Delta\mathbf{\Omega}_{12} = -\Delta\mathbf{\Omega}_{13} = (3e/2E_gk^2)\mathbf{E}_0 \times\mathbf{k} diverges at the zone center, as shown in Figure 2(d). This divergence originates from a strong mixing of energetically close HH and LH bands under E0\mathbf{E}_0, enforced by degeneracy at the Γ\Gamma point. Notably, this degeneracy entails monopole charges Q=±3Q = \pm3 in the momentum space, which serves as a source of the Berry curvature through ∇k⋅Ωn=2πδ(k)Q\nabla_{\mathbf{k} \cdot\mathbf{\Omega}_n} = 2\pi\delta(\mathbf{k})Q for n=2,3n = 2,3 [50]. The sign of QQ depends on the helicity of electrons associated with the total angular momentum j=3/2j = 3/2. As such, the valence band in GaAs intrinsically bears a topological character, which is hidden in thermal equilibrium owing to the cancellation of the Berry curvature by equally occupied electronic states. The divergence of ΔΩ12\Delta\mathbf{\Omega}_{12} and ΔΩ13\Delta\mathbf{\Omega}_{13} can be considered as an expression of such a nontrivial topology at the top of the valence band under a bias electric field. We demonstrate later that this divergence can be observed through the resonant enhancement of the field-induced CPGE. A previous study on semiconductors proposed a similar mechanism for the photovoltaic Hall effect [51]. However, the existence of a Fermi surface was required, whereas our jinj,dip\mathbf{j}_{\mathrm{inj,dip}} appears irrespective of carrier doping.

Field-Induced Energy Shift

The second correction to the CPGE is described by

jinj,ene=e32ℏ2∑k∑n∑m>nfnm(∇kωnm)×(Δω~nm∂Γnm∂ωnm⋅Ωnm),(6)\begin{aligned} \mathbf{j}_{\mathrm{inj,ene}} = \frac{e^3}{2\hbar^2} \sum_{\mathbf{k}} \sum_n \sum_{m>n} f_{nm} \left(\nabla_{\mathbf{k}}\omega_{nm}\right) \\ &\times\left(\Delta\widetilde{\omega}_{nm}\frac{\partial\mathbf{\Gamma}_{nm}}{\partial\omega_{nm}} \cdot\mathbf{\Omega}_{nm}\right), \tag*{(6)} \end{aligned}

where

Δω~nm=∣e⋅ξnm∣2Δωnm(e)−(e→e∗)∣e⋅ξnm∣2−(e→e∗),(7)\Delta\widetilde{\omega}_{nm} = \frac{\left|\mathbf{e} \cdot\boldsymbol{\xi}_{nm}\right|^2\Delta\omega_{nm}(\mathbf{e}) - (\mathbf{e} \rightarrow\mathbf{e}^*)}{\left|\mathbf{e} \cdot\boldsymbol{\xi}_{nm}\right|^2 - (\mathbf{e} \rightarrow\mathbf{e}^*)}, \tag*{(7)}

is the weighted average of Δωnm(e)=−eE0⋅Rnm(e)/ℏ\Delta\omega_{nm}(\mathbf{e}) = -e\mathbf{E}_0 \cdot\mathbf{R}_{nm}(\mathbf{e})/\hbar and Δωnm(e∗)\Delta\omega_{nm}(\mathbf{e}^*). Here, Rnm(e)=ξnn−ξmm−∇karg⁡(e⋅ξnm)\mathbf{R}_{nm}(\mathbf{e}) = \boldsymbol{\xi}_{nn} - \boldsymbol{\xi}_{mm} - \nabla_{\mathbf{k}}\arg(\mathbf{e} \cdot\boldsymbol{\xi}_{nm}) is the well-known shift vector for interband transition e⋅ξnm\mathbf{e} \cdot\boldsymbol{\xi}_{nm} [10]. ℏΔωnm(e)\hbar\Delta\omega_{nm}(\mathbf{e}) is equal to the difference in the electrostatic potential between the initial and final states at k\mathbf{k}, and ℏΔωnm(e∗)\hbar\Delta\omega_{nm}(\mathbf{e}^*) is that at −k-\mathbf{k}. A previous study proposed that Rashba-type spin-orbit coupling should modify the band structure under a bias field to generate a Hall current [52], similar to the picture suggested by Eq. (6). However, Rnm(e)\mathbf{R}_{nm}(\mathbf{e}) and ℏΔωnm(e)\hbar\Delta\omega_{nm}(\mathbf{e}) depend on the polarization vector e\mathbf{e} of light, indicating that the band structure under a bias field should appear different when observed using light with different polarizations. A description based on Rashba-type spin-orbit coupling is valid only when the shift vector is independent of light polarization. Materials with degenerate bands can be dealt with by (41) in Appendix B or by single value decomposition in Appendix C.

Shift vectors, transition-energy changes, and GaAs shift vectors

Figure 3. (a), (b) Shift vectors for left- and right-circularly polarized light, respectively, in a massive Dirac electron system. The arrow on the right side corresponds to a length of 0.2 nm. (c), (d) Magnitude of the transition energy shift for paths 1 and 2 caused by the bias field E0=10E_0 = 10 kV/cm along the xx axis. (e) Shift vectors for left-circularly polarized light in GaAs. The first and second quadrants resolve the two transition paths from HH to CB, and the third and fourth quadrants resolve those from LH to CB. (f) The xx component of the shift vectors at k=(0,ky,0)\boldsymbol{k} = (0,k_y,0).

Figures 3(a) and 3(b) illustrate a map of the shift vectors on the kz=0k_z = 0 plane derived from the massive Dirac electron model. Here, we distinguish degenerate states with different angular momenta using pseudospins ∣⇑⟩| \Uparrow\rangle and ∣⇓⟩| \Downarrow\rangle. In a lead halide perovskite, the pseudospin corresponds to the ordinary spin in the ss-like valence band and the total angular momentum jz=±1/2j_z = \pm1/2 in the pp-like conduction band [48]. Panel (a) illustrates the shift vector for ∣⇓⟩→∣⇑⟩| \Downarrow\rangle\to| \Uparrow\rangle driven by left-circularly polarized light, and panel (b) illustrates that for ∣⇑⟩→∣⇓⟩| \Uparrow\rangle\to| \Downarrow\rangle driven by right-circularly polarized light. Figures 3(c) and 3(d) illustrate the corresponding changes in the transition energy under a bias field in the xx direction. The transition energy increases (decreases) at ky>0k_y > 0 (ky<0k_y < 0) for left-circularly polarized light, whereas an inverse behavior is observed for right-circularly polarized light. This contrast results in a photocurrent Jinj,eneJ_{\mathrm{inj,ene}} in the yy direction, with a polarity depending on the helicity of light.

GaAs exhibits a remarkable behavior. Figure 3(e) shows the shift vectors on the kz=0k_z = 0 plane calculated for the left-circularly polarized light. The four quadrants distinguish the four transition paths: two from HH to CB and two from LH to CB, each arising from the two-fold degeneracy of every band. At each kk point, all four shift vectors point in the same direction, and are inverted for right-circularly polarized light (not shown). Notably, all shift vectors diverge as 1/k1/k at k→0k \to0, as illustrated in Fig. 3(f). In addition, they do not depend on material parameters, such as the band gap energy and effective mass. For example, the shift vectors for the transition HH →\to CB are given by R12(eL)=−(−ky,kx,0)/k(k≠kz)\boldsymbol{R}_{12}(\boldsymbol{e}_{\mathrm{L}}) = -(-k_y,k_x,0)/k(k \ne k_z) for eL=(1/2)(1,i,0)\boldsymbol{e}_{\mathrm{L}} = (1/\sqrt{2})(1,i,0). Insensitivity to the details of the band structure and appearance of a singularity reflect the topological character of the valence band. Similar to ΔΩnm\Delta\Omega_{nm}, the divergence in Δωnm(e)\Delta\omega_{nm}(\boldsymbol{e}) causes resonance in the field-induced CPGE as shown later.

Anomalous velocity of photocarriers

We briefly mention the remaining correction to (1),

Jinj,ano=e32ℏ2∑k∑m>n∑αfnm(Γnm⋅Ωnm)×[−eℏE0×(Ωn−Ωm)].(8)\begin{aligned} \boldsymbol{J}_{\mathrm{inj,ano}} = \frac{e^3}{2\hbar^2} \sum_k \sum_{m>n} \sum_{\alpha} f_{nm} \left(\Gamma_{nm} \cdot\boldsymbol{\Omega}_{nm}\right) \\ &\times\left[-\frac{e}{\hbar}\boldsymbol{E}_0 \times\left(\boldsymbol{\Omega}_n-\boldsymbol{\Omega}_m\right)\right]. \tag*{(8)} \end{aligned}

The second line represents the difference between the anomalous velocities of the initial and final states, which is considered as a correction of ∇kωnm\nabla_k\omega_{nm} in Eq. (1). This is the most conventional mechanism of the photovoltaic Hall effect and is referred to differently depending on the degrees of freedom associated with the Berry curvature Ωn\boldsymbol{\Omega}_n. The most familiar example is the inverse spin Hall effect, where optically excited spin-polarized carriers generate a transverse current under a bias field [26, 27, 28, 29, 30, 31, 32, 33]. The valley Hall effect, which arises from the valley degree of freedom in two-dimensional transition metal dichalcogenides, is also well known [35, 36]. Other examples include the orbital Hall effect, associated with the orbital angular momentum [53, 54], and the isospin Hall effect, which is specific for Dirac semimetals [34]. To be precise, Eq. (8) describes only the intrinsic part of these effects; extrinsic processes induced by scatterers, such as side jump and skew scattering [55], are not included. Although they are essential in the low-frequency regime, they can be neglected on ultrafast timescales before the occurrence of scattering events [33]. In GaAs,

Ωn=Qk/2k3\Omega_n = Q\mathbf{k}/2k^3 (n=2,3n = 2, 3) exhibits a singularity at k=0k = 0 because of the aforementioned monopole charges. This causes a resonant enhancement of Jinj,anoJ_{\mathrm{inj,ano}} at the bandgap energy, as discussed next.

Effect of Degeneracy Points

We condense the above results into a tensor KiabK_i^{ab} such that

Jinj,ia=∑bKiabE0b,(i=dip, ene, ano)(9)J_{\mathrm{inj},i}^{a} = \sum_b K_i^{ab} E_0^b,\qquad(i = \mathrm{dip},\ \mathrm{ene},\ \mathrm{ano}) \tag*{(9)}

where aa and bb represent the Cartesian coordinates. Adding the phenomenological damping term −Jinj,ia/τi-J_{\mathrm{inj},i}^{a}/\tau_i to the right-hand side of Eq. (9) yields a DC optical conductivity tensor σiab=τiKiab\sigma_i^{ab} = \tau_i K_i^{ab}. Figure 4(a) illustrates the dependence of KiyxK_i^{yx} on the photon energy of light for the massive Dirac model irradiated by left-circularly polarized light. Figure 4(b) shows the sum Kyx=∑iKiyxK^{yx} = \sum_i K_i^{yx} and absorption spectrum. Our calculations reproduce the sign change in KyxK^{yx} at ℏω=4Δ\hbar\omega= 4\Delta predicted by [56]. The dominant origin shifts from an anomalous velocity to a field-induced Berry curvature with this sign change.

In comparison, Figure 4(c) illustrates KiyxK_i^{yx} for GaAs. All the contributions exhibit singular behavior in the form of (ℏω−Eg)−1/2(\hbar\omega- E_g)^{-1/2} around the band gap energy (Eg=1.426E_g = 1.426 eV). This behavior originates from the above-mentioned topological nature of the valence band. Although the anomalous velocity of photoexcited carriers dominates the simple sum KyxK^{yx} [Figure 4(d)], use of a terahertz pulse E0(t)E_0(t) instead of a static field E0E_0 allows the different mechanisms to be detected separately. ΔΩnm\Delta\Omega_{nm} and Δωnm(e)\Delta\omega_{nm}^{(e)} should follow the slow temporal variation of E0(t)E_0(t), leading to J˙inj,i(t)∼E0(t)I1(t)\dot{J}_{\mathrm{inj},i}(t) \sim E_0(t)I_1(t) for i=dipi = \mathrm{dip} and ene\mathrm{ene}, where I1(t)I_1(t) denotes the time-dependent light intensity. By contrast, Jinj,anoJ_{\mathrm{inj,ano}} is proportional to the anomalous velocity multiplied by the number of carriers; thus, Jinj,ano(t)∼E0(t)∫tdt′I1(t′)J_{\mathrm{inj,ano}}(t) \sim E_0(t)\int^t dt' I_1(t'). Such a qualitative difference in the dynamic response can be used to distinguish the different mechanisms, as demonstrated by our recent experiment [41]. For strong excitations, Jinj,dip+Jinj,eneJ_{\mathrm{inj,dip}} + J_{\mathrm{inj,ene}} even became dominant because of the high suppression of Jinj,anoJ_{\mathrm{inj,ano}} [41]. Strikingly, the resonance peak in Jinj,dip+Jinj,eneJ_{\mathrm{inj,dip}} + J_{\mathrm{inj,ene}} shifted to a slightly higher energy because of the resonance between the terahertz pulse and HH–LH transitions [41]. Although the essence of the field-induced CPGE can be captured by the present theory, such a dynamic effect deserves further investigation.

Three contributions to the field-induced CPGE in a massive Dirac electron system and in GaAs

Figure 4. (a) Three contributions to the field-induced CPGE in a massive Dirac electron system. The electric field amplitude of light is set to 100 kV/cm. (b) The sum of the three contributions in panel (a). The dashed line represents the real part of optical conductivity in equilibrium. (c) Three contributions to the field-induced CPGE in GaAs. KdipyxK_{\mathrm{dip}}^{yx} (orange) and KeneyxK_{\mathrm{ene}}^{yx} (green) almost overlap. The electric field amplitude of light is set to 100 kV/cm. (d) The sum of the three contributions in panel (c). The dashed line represents the real part of optical conductivity in equilibrium.

Dressed State Contribution

Now, we look into the contribution by light-dressed states formed in metallic systems. For a non-degenerate case, the anomalous Hall current carried by the Fermi surface turns out to be

Jdre,ano=e∑k∑nfn(−eℏE0×δΩn),(10)\mathbf{J}_{\mathrm{dre,ano}} = e\sum_{\mathbf{k}}\sum_n f_n\left(-\frac{e}{\hbar}\mathbf{E}_0 \times\delta\mathbf{\Omega}_n\right), \tag*{(10)}

with δΩn=∇k×δξnn\delta\mathbf{\Omega}_n = \nabla_{\mathbf{k}} \times\delta\boldsymbol{\xi}_{nn}. As shown by Eq. (51) in Appendix § 10, δξnn\delta\boldsymbol{\xi}_{nn} is proportional to the intensity of light. Importantly, δΩn\delta\mathbf{\Omega}_n coincides with the light-induced Berry curvature predicted by Floquet formalism in the lowest order. Thus, our theory successfully describes the AHE by Floquet states within the third-order nonlinearity. It can also be shown that the second-order electric polarization responsible for optical rectification,

Prec,0=e∑k∑nfnδξnn,(11)\mathbf{P}_{\mathrm{rec},0} = e\sum_{\mathbf{k}}\sum_n f_n\delta\boldsymbol{\xi}_{nn}, \tag*{(11)}

is governed by δξnn\delta\boldsymbol{\xi}_{nn}. Given the established role of Berry connection in modern theory of electric polarization [1, 2, 3], δξnn\delta\boldsymbol{\xi}_{nn} can be interpreted as a light-induced Berry connection, in accord with the relationship δΩn=∇k×δξnn\delta\mathbf{\Omega}_n = \nabla_{\mathbf{k}} \times\delta\boldsymbol{\xi}_{nn}. This work provides the first analytic formulation of the field-induced CPGE, light-induced AHE by Floquet states, and optical rectification using a single theoretical framework.

Comparison with Previous Studies

Finally, we compare our results with those in previous studies on the photovoltaic Hall effect. A general theory of the bulk photovoltaic effect under a bias field has been presented in the literature [57]. Although its results for the field-induced CPGE were consistent with ours, the physical meaning of each term was not as clear as revealed here. For example, Jinj,dipJ_{\mathrm{inj,dip}} in our theory was split into two terms and was interpreted differently from the field-induced Berry curvature in Ref. [57]. Jinj,eneJ_{\mathrm{inj,ene}} was also given an interpretation different from the field-induced energy shift. Our theory provides an alternative perspective for understanding the photovoltaic Hall effect induced by resonant light. A more recent theory [56] rearranged the result in Ref. [57] in terms of Riemannian geometry. Although this viewpoint reveals an interesting aspect of light–matter interaction, the obtained equations cannot easily be understood from a physical point of view. For example, the contribution of the Hermitian curvature, which was the main focus of Ref. [56], was a mixture of Jinj,anoJ_{\mathrm{inj,ano}} and Jinj,eneJ_{\mathrm{inj,ene}} in our theory. These two components should be distinguished both for physical interpretation and for practical use, because the relaxation times for photocurrent generally differ from each other; Jinj,dip+Jinj,eneJ_{\mathrm{inj,dip}}+J_{\mathrm{inj,ene}} decays through momentum relaxation, whereas Jinj,anoJ_{\mathrm{inj,ano}} decays through the relaxation of the spin or other quantities that break the time-reversal symmetry. To incorporate this qualitative difference, Jinj,dip+Jinj,eneJ_{\mathrm{inj,dip}}+J_{\mathrm{inj,ene}} and Jinj,anoJ_{\mathrm{inj,ano}} should be distinguished.

Conclusion

In summary, we developed a comprehensive theory of the photovoltaic Hall effect in nonmagnetic materials. We revealed three contributions in field-induced CPGE: (i) field-induced Berry curvature, (ii) field-induced energy shift, and (iii) anomalous velocity of photoexited carriers. These contributions were related to either the Berry curvature or shift vector, demonstrating the geometric aspect of the field-induced CPGE. The topological character of the valence band in GaAs generated a resonance peak for all three mechanisms. We also desribed AHE by light-dressed states within the same theoretical framework and revealed its close relationship with optical rectification. Our results provide a solid foundation for interpreting the experimental results on the photovoltaic Hall effect and a possible counterpart to the Floquet [17] and strain [58] engineering of the Berry curvature. Extension to correlated systems is an interesting future direction [59].

Acknowledgments

This study was supported by JSPS KAKENHI (Grants No. JP24K16988 and No. JP24K00550), JST FOREST (Grant No. JPMJFR2240), and JST CREST (Grant No. JPMJCR20R4). Y.M. and R.M. conceived the project. Y.M. developed the theoretical formalism with help of T.F. and performed numerical calculations. All authors discussed the result. Y.M. wrote the manuscript with feedback from all coauthors.

Formulation

In this appendix, we outline the derivation of equations presented in the main text.

Perturbation by a static electric field

We start from a Schrödinger equation expressed in the length gauge [43],

iℏ∂∂t∣Φ(k)⟩=[H(k)−ieE⋅∇k]∣Φ(k)⟩,(12)\mathrm{i}\hbar\frac{\partial}{\partial t}|\Phi(\mathbf{k})\rangle= [H(\mathbf{k})-\mathrm{i}e\mathbf{E}\cdot\nabla_{\mathbf{k}}]|\Phi(\mathbf{k})\rangle, \tag*{(12)}

where E\mathbf{E} is the electric field. Our goal is to calculate the current density,

J=e∑k⟨Φ(k)∣[1ℏ∇kH(k)]∣Φ(k)⟩.(13)\mathbf{J}=e\sum_{\mathbf{k}}\langle\Phi(\mathbf{k})|\left[\frac{1}{\hbar}\nabla_{\mathbf{k}}H(\mathbf{k})\right]|\Phi(\mathbf{k})\rangle. \tag*{(13)}

For brevity, we omit k\mathbf{k} dependence of every quantity hereafter. We diagonalize HH so that each group of degenerate states forms a block:

H=U†HU=diag⁡(ϵ1I1,ϵ2I2,⋯ ),(14)\mathcal{H}=U^{\dagger}HU=\operatorname{diag}(\epsilon_1 I_1,\epsilon_2 I_2,\cdots), \tag*{(14)}

where ϵν\epsilon_\nu denotes the dispersion relation of the ν\nu-th band, and IνI_\nu a unit matrix with a dimension equal to its degeneracy. We label each energy band with a Greek letter (ν\nu, μ\mu, etc.) and each basis in it with a Roman letter (nn, mm, etc.). In this notation, Unm=∣um⟩nU_{nm}=|u_m\rangle_n can be regarded as the nn-th component of the mm-th eigenstate, ∣um⟩|u_m\rangle. Below, the energy eigenvalues are arranged in descending order, i.e., ϵν>ϵν+1\epsilon_\nu>\epsilon_{\nu+1}. Ability to treat degenerate bands is essential because the photovoltaic Hall effect can occur even in nonmagnetic and centrosymmetric materials. Using a new wavefunction ∣ϕ⟩≡U†∣Φ⟩|\phi\rangle\equiv U^\dagger|\Phi\rangle, Eq. (A1) is rewritten as

iℏ∂∂t∣ϕ⟩=[H−eE⋅(ξ+i∇k)]∣ϕ⟩,(15)\mathrm{i}\hbar\frac{\partial}{\partial t}|\phi\rangle=[\mathcal{H}-e\mathbf{E}\cdot(\boldsymbol{\xi}+\mathrm{i}\nabla_{\mathbf{k}})]|\phi\rangle, \tag*{(15)}

where ξ=U†i∇kU\boldsymbol{\xi}=U^\dagger\mathrm{i}\nabla_{\mathbf{k}}U is a Berry connection matrix. It is apparent that ξ+i∇k\boldsymbol{\xi}+\mathrm{i}\nabla_{\mathbf{k}} serves as a position operator acting on ∣ϕ⟩|\phi\rangle [10, 43, 60]. In particular, its off-diagonal component ξνμ\boldsymbol{\xi}_{\nu\mu} (ν≠μ\nu\ne\mu) acts as a transition dipole moment, which determines the transition matrix element Mνμ(e)≡(e⋅ξνμ)(e∗⋅ξμν)\mathcal{M}_{\nu\mu}(e)\equiv(\mathbf{e}\cdot\boldsymbol{\xi}_{\nu\mu})(\mathbf{e}^{*}\cdot\boldsymbol{\xi}_{\mu\nu}) for the transition μ→ν\mu\to\nu. Even in systems with degenerate bands, a relationship between Berry curvature and circular dichroism, similar to Eq. (2), holds:

Ωνμz=Mνμ(eR)−Mνμ(eL),(16)\Omega_{\nu\mu}^{z}=\mathcal{M}_{\nu\mu}(e_{\mathrm{R}})-\mathcal{M}_{\nu\mu}(e_{\mathrm{L}}), \tag*{(16)}

where the band-resolved Berry curvature is defined by Ωνμ≡iξνμ×ξμν\Omega_{\nu\mu} \equiv i\boldsymbol{\xi}_{\nu\mu} \times\boldsymbol{\xi}_{\mu\nu}. Summation of Ωνμ\Omega_{\nu\mu} over μ\mu gives the total Berry curvature for the ν\nu-th band [see Eq. (A15)].

Now we apply a static electric field E0\mathbf{E}_0 to the system. In this case, it is helpful to introduce another wavefunction ∣ϕ~⟩≡U~†∣Φ⟩|\tilde{\phi}\rangle\equiv\tilde{U}^{\dagger}|\Phi\rangle, which transforms Eq. (12) into

iℏ∂∂t∣ϕ~⟩=(H~−ieE0⋅∇k)∣ϕ~⟩(17)i\hbar\frac{\partial}{\partial t}|\tilde{\phi}\rangle= \left(\tilde{H} - ie\mathbf{E}_0 \cdot\nabla_{\mathbf{k}}\right)|\tilde{\phi}\rangle \tag*{(17)}

with a modified Hamiltonian H~≡U~†HU~−eE0⋅ξ~\tilde{H} \equiv\tilde{U}^{\dagger}H\tilde{U} - e\mathbf{E}_0 \cdot\tilde{\boldsymbol{\xi}} and a modified Berry connection matrix ξ~≡U~†i∇kU~\tilde{\boldsymbol{\xi}} \equiv\tilde{U}^{\dagger}i\nabla_{\mathbf{k}}\tilde{U}. We require H~\tilde{H} to be block-diagonal like Eq. (14). Up to the first order in E0\mathbf{E}_0, this can be done by putting U~=U+ΔU\tilde{U}=U+\Delta U with (U†ΔU)νμ=eE0⋅ξνμ/ℏωνμ(U^{\dagger}\Delta U)_{\nu\mu}=e\mathbf{E}_0 \cdot\boldsymbol{\xi}_{\nu\mu}/\hbar\omega_{\nu\mu} (ν≠μ)(\nu\ne\mu), where ωνμ≡(ϵν−ϵμ)/ℏ\omega_{\nu\mu}\equiv(\epsilon_\nu-\epsilon_\mu)/\hbar is the transition frequency. We can safely put (U†ΔU)νν=0(U^{\dagger}\Delta U)_{\nu\nu}=0. In the end, H~\tilde{H} retains only the block-diagonal components, H~νν=ϵνIν−eE0⋅ξνν\tilde{H}_{\nu\nu}=\epsilon_\nu I_\nu-e\mathbf{E}_0\cdot\boldsymbol{\xi}_{\nu\nu}.

Correction to transition dipole moment

The static field E0\mathbf{E}_0 modifies the Berry connection. The correction in ξ~=ξ+Δξ\tilde{\boldsymbol{\xi}}=\boldsymbol{\xi}+\Delta\boldsymbol{\xi} can be expressed as Δξ=CE0\Delta\boldsymbol{\xi}=C\mathbf{E}_0, with the Berry connection polarizability

Cνμab=−eℏ∑μ≠νξνμaξμνb+ξνμbξμνaωνμ(18)C_{\nu\mu}^{ab}=-\frac{e}{\hbar}\sum_{\mu\ne\nu}\frac{\xi_{\nu\mu}^{a}\xi_{\mu\nu}^{b}+\xi_{\nu\mu}^{b}\xi_{\mu\nu}^{a}}{\omega_{\nu\mu}} \tag*{(18)}
Cνμab=ieℏDa(ξνμbωνμ)+eℏ∑λ≠ν,μ(ξνλaξλμbωλμ−ξνλbξλμaωνλ).(19)C_{\nu\mu}^{ab}=\frac{ie}{\hbar}D^{a}\left(\frac{\xi_{\nu\mu}^{b}}{\omega_{\nu\mu}}\right)+\frac{e}{\hbar}\sum_{\lambda\ne\nu,\mu}\left(\frac{\xi_{\nu\lambda}^{a}\xi_{\lambda\mu}^{b}}{\omega_{\lambda\mu}}-\frac{\xi_{\nu\lambda}^{b}\xi_{\lambda\mu}^{a}}{\omega_{\nu\lambda}}\right). \tag*{(19)}

(ν≠μ)(\nu\ne\mu)

Here, Dk=(Dx,Dy,Dz)\mathbf{D}_{\mathbf{k}}=(D^{x},D^{y},D^{z}) represents a generalized [43] or covariant [44] derivative defined by DkO≡−i[ri,O]\mathbf{D}_{\mathbf{k}}O\equiv-i[\mathbf{r}_{i},O] using the intraband position operator ri,νμ≡δνμ(ξνν+i∇k)r_{i,\nu\mu}\equiv\delta_{\nu\mu}(\xi_{\nu\nu}+i\nabla_{\mathbf{k}}). More specifically, DkOνμ=∇kOνμ−i(ξννOνμ−Oνμξμμ)D_{\mathbf{k}}O_{\nu\mu}=\nabla_{\mathbf{k}}O_{\nu\mu}-i(\boldsymbol{\xi}_{\nu\nu}O_{\nu\mu}-O_{\nu\mu}\boldsymbol{\xi}_{\mu\mu}). Berry curvature is also changed into Ω~νμ≡iξ~νμ×ξ~μν\tilde{\Omega}_{\nu\mu}\equiv i\tilde{\boldsymbol{\xi}}_{\nu\mu}\times\tilde{\boldsymbol{\xi}}_{\mu\nu}, which includes a correction ΔΩνμ=ΠνμE0\Delta\Omega_{\nu\mu}=\Pi_{\nu\mu}\mathbf{E}_0. The band-resolved Berry curvature polarizability Πνμ\Pi_{\nu\mu} is given by

Πνμad≡i∑bcϵabcTνμbcd(20)\Pi_{\nu\mu}^{ad}\equiv i\sum_{bc}\epsilon^{abc}\mathcal{T}_{\nu\mu}^{bcd} \tag*{(20)}

where

Tνμabc=ξνμaCμνbc+Cνμacξμνb.(21)\mathcal{T}_{\nu\mu}^{abc}=\xi_{\nu\mu}^{a}C_{\mu\nu}^{bc}+C_{\nu\mu}^{ac}\xi_{\mu\nu}^{b}. \tag*{(21)}

Correction to transition energy

Next, we examine the velocity operator v~≡U~†(ℏ−1∇kH)U~\tilde{\mathbf{v}}\equiv\tilde{U}^{\dagger}(\hbar^{-1}\nabla_{\mathbf{k}}H)\tilde{U} under the electric field. The off-diagonal components are found to be

v~νμ=iωνμξ~νμ+eℏ(E0⋅Dk)ξνμ,(ν≠μ)(22)\tilde{\mathbf{v}}_{\nu\mu}=i\omega_{\nu\mu}\tilde{\boldsymbol{\xi}}_{\nu\mu}+\frac{e}{\hbar}(\mathbf{E}_0\cdot\mathbf{D}_{\mathbf{k}})\boldsymbol{\xi}_{\nu\mu},\qquad(\nu\ne\mu) \tag*{(22)}

At E0=0\mathbf{E}_0=0, Eq. (22) recovers a well-known relationship between position and velocity operators, vνμ=iωνμξνμ\mathbf{v}_{\nu\mu}=i\omega_{\nu\mu}\boldsymbol{\xi}_{\nu\mu} for ν≠μ\nu\ne\mu. This equation lets us expect that Eq. (22) should include corrections of ξνμ\boldsymbol{\xi}_{\nu\mu} and ωνμ\omega_{\nu\mu} induced by E0\mathbf{E}_0. The former clearly appears in the first term on the right-hand side of Eq. (22). To demonstrate the latter effect, we consider a non-degenerate case. Multiplied by the polarization vector e\mathbf{e}, Eq. (22) gives

e⋅v~nm=i[ωnm+Δωnm(e)+(imag.)](e⋅ξ~nm),(23)\mathbf{e}\cdot\tilde{\mathbf{v}}_{nm}=i[\omega_{nm}+\Delta\omega_{nm}(\mathbf{e})+(\mathrm{imag.})](\mathbf{e}\cdot\tilde{\boldsymbol{\xi}}_{nm}), \tag*{(23)}

with

ℏΔωnm(e)=−eE0⋅Rnm(e).(24)\hbar\Delta\omega_{nm}(\mathbf{e})=-e\mathbf{E}_0\cdot\mathbf{R}_{nm}(\mathbf{e}). \tag*{(24)}

Here, Rnm(e)≡ξnn−ξmm−∇karg⁡(e⋅ξnm)\mathbf{R}_{nm}(\mathbf{e})\equiv\boldsymbol{\xi}_{nn}-\boldsymbol{\xi}_{mm}-\nabla_{\mathbf{k}}\arg(\mathbf{e}\cdot\boldsymbol{\xi}_{nm}) is the well-known shift vector, which measures the spatial movement of an electron cloud accompanying an interband transition m→nm\to n [10]. Given such a spatial shift, the electrostatic potential for the initial and final states should differ by ℏΔωnm(e)\hbar\Delta\omega_{nm}(\mathbf{e}) under E0\mathbf{E}_0. Equation (A12) tells that such a modification in the transition energy is naturally encoded in Eq. (22), including polarization dependence of the shift vector. Even for systems with degenerate bands, appropriate choice of bases resolves the mutually independent pairs of initial and final states, making the above argument valid (see Appendix C). We note that (imag.)(\mathrm{imag.}) in Eq. (23), indicating an imaginary number, is not important because it does not affect the transition probability.

Anomalous velocity and acceleration

The other effects of E0\mathbf{E}_0 are well-known. First, the diagonal component of the modified velocity operator takes the form of

v~νν=1ℏ∇kϵν−eℏE0×Ων,(25)\tilde{\mathbf{v}}_{\nu\nu}=\frac{1}{\hbar}\nabla_{\mathbf{k}}\epsilon_{\nu}-\frac{e}{\hbar}\mathbf{E}_0\times\boldsymbol{\Omega}_{\nu}, \tag*{(25)}

where

Ων=∇k×ξνν−iξνν×ξνν=∑μ≠νΩνμ,(26)\boldsymbol{\Omega}_{\nu}=\nabla_{\mathbf{k}}\times\boldsymbol{\xi}_{\nu\nu}-i\boldsymbol{\xi}_{\nu\nu}\times\boldsymbol{\xi}_{\nu\nu}=\sum_{\mu\ne\nu}\Omega_{\nu\mu}, \tag*{(26)}

is the Berry curvature of the ν\nu-th band. The first and second terms in Eq. (25) correspond to the group and anomalous velocities, respectively [2, 6].

The last effect of E0\mathbf{E}_0 is acceleration. To see this, we complete the formalism by adding an optical field E1\mathbf{E}_1 to the total electric field, E=E0+E1\mathbf{E}=\mathbf{E}_0+\mathbf{E}_1. Equation (A1) is then transformed into

iℏ∂∂t∣ϕ~⟩=(H~−eE1⋅ξ~−ieE1⋅∇k)∣ϕ~⟩.(27)i\hbar\frac{\partial}{\partial t}|\tilde{\phi}\rangle=\left(\tilde{H}-e\mathbf{E}_1\cdot\tilde{\boldsymbol{\xi}}-ie\mathbf{E}_1\cdot\nabla_{\mathbf{k}}\right)|\tilde{\phi}\rangle. \tag*{(27)}

For a density matrix, ρ=∣ϕ~⟩⟨ϕ~∣\rho=|\tilde{\phi}\rangle\langle\tilde{\phi}|, the equation of motion is given by

ρ˙=−iℏ[H~−eE1⋅ξ~,ρ]−eℏ(E⋅∇k)ρ.(28)\dot{\rho}=-\frac{i}{\hbar}[\tilde{H}-e\mathbf{E}_1\cdot\tilde{\boldsymbol{\xi}},\rho]-\frac{e}{\hbar}(\mathbf{E}\cdot\nabla_{\mathbf{k}})\rho. \tag*{(28)}

Compared to the case of E0=0\mathbf{E}_0=0, Eq. (28) includes three changes: (i) correction to energy, H→H~\mathcal{H}\to\widetilde{\mathcal{H}}, (ii) correction to transition dipole moment, ξ→ξ~\boldsymbol{\xi}\to\widetilde{\boldsymbol{\xi}}, and (iii) an additional term, −(e/ℏ)(E0⋅∇k)ρ-(e/\hbar)(\mathbf{E}_0\cdot\nabla_{\mathbf{k}})\rho. The last one is responsible for acceleration of electrons. Equation (28) can be extended to a time-dependent E0\mathbf{E}_0 as long as its frequency lies far below the interband transition frequencies.

We solve Eq. (28) neglecting scattering processes. This is done by putting a time-dependent initial condition,

ρνν(0)=fν+eℏ(A0⋅∇k)fν,(29)\rho_{\nu\nu}^{(0)}=f_\nu+\frac{e}{\hbar}(\mathbf{A}_0\cdot\nabla_{\mathbf{k}})f_\nu, \tag*{(29)}

before the arrival of the optical field E1\mathbf{E}_1. Here, the superscript in parentheses indicates the order with respect to E1\mathbf{E}_1, and fνf_\nu the Fermi-Dirac distribution function for energy ϵν\epsilon_\nu. In reality, scattering processes resist the acceleration by E0\mathbf{E}_0, which should generate a constant distribution function instead of Eq. (29). Scattering processes can also take part in photocurrent generation by themselves [61, 62, 63]. However, it is challenging to incorporate these effects in a comprehensive manner, since there exist complex scattering channels including side jump [64] and skew scattering [65] known in anomalous and spin Hall effects [6, 55]. Rather than elaborating a model that takes into account these complexities, we aim at obtaining a clear physical picture of photovoltaic Hall effect in a simplified situation, which helps comparison with Floquet theory and lays the foundation for future extension. Equation (29) is indeed justified in a time scale shorter than the scattering times, which is relevant to recent ultrafast experiments using terahertz spectroscopy [21, 22, 33, 34, 40, 41].

Calculation of photocurrent

Injection current

Solving Eq. (28) to obtain the nonlinear current requires a lengthy calculation. Here, we only present the final results. The second-order nonlinear current in the presence of a static field is classified into

J(2)=Jjer+Jinj+Jshi+P˙rec+Jdre.(30)\mathbf{J}^{(2)}=\mathbf{J}_{\mathrm{jer}}+\mathbf{J}_{\mathrm{inj}}+\mathbf{J}_{\mathrm{shi}}+\dot{\mathbf{P}}_{\mathrm{rec}}+\mathbf{J}_{\mathrm{dre}}. \tag*{(30)}

The first term, called jerk current, arises from acceleration of photoexcited carriers [44, 66, 67]. The second and third, i.e., injection and shift currents, originate from momentum asymmetry and spatial transfer of photoexcited carriers, respectively [10]. The fourth is the well-known optical rectification, and the last originates from response on the Fermi surface. For monochromatic illumination, J˙jer\dot{\mathbf{J}}_{\mathrm{jer}}, J˙inj\dot{\mathbf{J}}_{\mathrm{inj}}, Jshi\mathbf{J}_{\mathrm{shi}}, and Prec\mathbf{P}_{\mathrm{rec}} become constant [57], while Jdre\mathbf{J}_{\mathrm{dre}} includes terms with different dependencies. In a nonmagnetic material, only Jinj\mathbf{J}_{\mathrm{inj}} and Jdre\mathbf{J}_{\mathrm{dre}} yield the photovoltaic Hall effect. We therefore omit the other terms.

Injection current in the presence of a static field is decomposed into

Jinj=Jinj,0+Jinj,dip+Jinj,ene+Jinj,ano+Jinj,fer.(31)\mathbf{J}_{\mathrm{inj}}=\mathbf{J}_{\mathrm{inj},0}+\mathbf{J}_{\mathrm{inj},\mathrm{dip}}+\mathbf{J}_{\mathrm{inj},\mathrm{ene}}+\mathbf{J}_{\mathrm{inj},\mathrm{ano}}+\mathbf{J}_{\mathrm{inj},\mathrm{fer}}. \tag*{(31)}

The first term corresponds to the usual injection current in the absence of E0\mathbf{E}_0, which follows

J˙inj,0=−e3ℏ2∑k∑ν∑μ>νfνμ(∇kωνμ)∑bcΓνμbcMμνbc,(32)\dot{\mathbf{J}}_{\mathrm{inj},0}=-\frac{e^3}{\hbar^2}\sum_{\mathbf{k}}\sum_\nu\sum_{\mu>\nu}f_{\nu\mu}(\nabla_{\mathbf{k}}\omega_{\nu\mu})\sum_{bc}\Gamma_{\nu\mu}^{bc}M_{\mu\nu}^{bc}, \tag*{(32)}

where fνμ≡fν−fμf_{\nu\mu}\equiv f_\nu-f_\mu and Mμνab≡Tr⁡ξμνaξνμbM_{\mu\nu}^{ab}\equiv\operatorname{Tr}\xi_{\mu\nu}^{a}\xi_{\nu\mu}^{b}. Γνμbc\Gamma_{\nu\mu}^{bc} is defined as Γνμbc≡(1/2)(∂3GνμbGμνc/∂t3)\Gamma_{\nu\mu}^{bc}\equiv(1/2)(\partial^3G_{\nu\mu}^{b}G_{\mu\nu}^{c}/\partial t^3) using

Fνμ≡∫−∞tdt′ eiωνμt′E1(t′),Gνμ≡∫−∞tdt′ Fνμ(t′),(33)F_{\nu\mu}\equiv\int_{-\infty}^{t}\mathrm{d}t'\,e^{i\omega_{\nu\mu}t'}\mathbf{E}_1(t'),\qquad G_{\nu\mu}\equiv\int_{-\infty}^{t}\mathrm{d}t'\,F_{\nu\mu}(t'), \tag*{(33)}

and is equal to

Γνμbc=2πE1bE1c∗δ(ω−ωνμ)+(ν↔μ,b↔c),(34)\Gamma_{\nu\mu}^{bc}=2\pi E_1^bE_1^{c*}\delta(\omega-\omega_{\nu\mu})+(\nu\leftrightarrow\mu,b\leftrightarrow c), \tag*{(34)}

for a monochromatic wave, E1(t)=E1e−iωt+E1∗eiωt\mathbf{E}_1(t)=\mathbf{E}_1e^{-i\omega t}+\mathbf{E}_1^*e^{i\omega t}. In the presence of time-reversal symmetry, Jinj,0\mathbf{J}_{\mathrm{inj},0} can be expressed alternatively as

J˙inj,0=e32ℏ2∑k∑ν∑μ>νfνμ(∇kωνμ)(Γνμ−Ωνμtr),(35)\dot{\mathbf{J}}_{\mathrm{inj},0}=\frac{e^3}{2\hbar^2}\sum_{\mathbf{k}}\sum_\nu\sum_{\mu>\nu}f_{\nu\mu}(\nabla_{\mathbf{k}}\omega_{\nu\mu})(\Gamma_{\nu\mu}-\Omega_{\nu\mu}^{\mathrm{tr}}), \tag*{(35)}

where Ωνμtr≡Tr⁡Ωνμ\Omega_{\nu\mu}^{\mathrm{tr}}\equiv\operatorname{Tr}\Omega_{\nu\mu} and Γνμa≡i∑bcdϵabcdΓνμbc\Gamma_{\nu\mu}^{a}\equiv i\sum_{bcd}\epsilon^{abcd}\Gamma_{\nu\mu}^{bc}. For a monochromatic wave, we have Γνμ=2πi(E1×E1∗)δ(ω−ωνμ)\Gamma_{\nu\mu}=2\pi i(\mathbf{E}_1\times\mathbf{E}_1^*)\delta(\omega-\omega_{\nu\mu}), which is parallel to the propagation direction of light and is proportional to the degree of circular polarization. Equation (1) in the main text derives from Eq. (35).

The second term in Eq. (31) is defined by

Jinj,dip=−e3ℏ2∑k∑ν∑μ>νfνμ(∇kωνμ)∑bcdΓνμbcTμνbcdE0d,(36)\mathbf{J}_{\mathrm{inj},\mathrm{dip}}=-\frac{e^3}{\hbar^2}\sum_{\mathbf{k}}\sum_\nu\sum_{\mu>\nu}f_{\nu\mu}(\nabla_{\mathbf{k}}\omega_{\nu\mu})\sum_{bcd}\Gamma_{\nu\mu}^{bc}T_{\mu\nu}^{bcd}E_0^d, \tag*{(36)}

where Tμνbcd≡Tr⁡TμνbcdT_{\mu\nu}^{bcd}\equiv\operatorname{Tr}T_{\mu\nu}^{bcd}. This contribution arises from the correction in transition dipole moment discussed in Sec. A 2. In the presence of time-reversal symmetry, we obtain an alternative expression,

Jinj,dip=e32ℏ2∑k∑ν∑μ>νfνμ(∇kωνμ)(Γνμ⋅ΔΩνμtr),(37)\mathbf{J}_{\mathrm{inj},\mathrm{dip}}=\frac{e^3}{2\hbar^2}\sum_{\mathbf{k}}\sum_\nu\sum_{\mu>\nu}f_{\nu\mu}(\nabla_{\mathbf{k}}\omega_{\nu\mu})(\Gamma_{\nu\mu}\cdot\Delta\Omega_{\nu\mu}^{\mathrm{tr}}), \tag*{(37)}

where ΔΩνμtr≡Tr⁡ΔΩνμ\Delta\Omega_{\nu\mu}^{\mathrm{tr}}\equiv\operatorname{Tr}\Delta\Omega_{\nu\mu}. Equation (37) unifies the third and fourth terms in Eq. (137) of Ref. [57] in a physically clear form. Equation (5) in the main text derives from Eq. (37).

The third term in Eq. (31) is defined by

Jinj,ene=e4ℏ3∑k∑ν∑μ>νfνμ∑bcd(∇kΓνμbc)SνμbcdE0d,(38)\mathbf{J}_{\mathrm{inj},\mathrm{ene}}=\frac{e^4}{\hbar^3}\sum_{\mathbf{k}}\sum_\nu\sum_{\mu>\nu}f_{\nu\mu}\sum_{bcd}(\nabla_{\mathbf{k}}\Gamma_{\nu\mu}^{bc})S_{\nu\mu}^{bcd}E_0^d, \tag*{(38)}

where Sνμbcd≡Tr⁡Sνμbcd\mathcal{S}^{bcd}_{\nu\mu} \equiv\operatorname{Tr} \mathcal{S}^{bcd}_{\nu\mu} and

Sνμabc≡−i2[ξμνa(Dcξμνb)−(Dcξμνa)ξμνb].(39)\mathcal{S}^{abc}_{\nu\mu} \equiv-\frac{i}{2}\left[\xi^{a}_{\mu\nu}\left(D^{c}\xi^{b}_{\mu\nu}\right)-\left(D^{c}\xi^{a}_{\mu\nu}\right)\xi^{b}_{\mu\nu}\right]. \tag*{(39)}

An identity ∇kΓνμbc=(∂Γνμbc/∂ωνμ)∇kωνμ\nabla_{\mathbf{k}}\Gamma^{bc}_{\nu\mu}=\left(\partial\Gamma^{bc}_{\nu\mu}/\partial\omega_{\nu\mu}\right)\nabla_{\mathbf{k}}\omega_{\nu\mu} suggests that Jinj,ene\mathbf{J}_{\mathrm{inj,ene}} should be related to the field-induced energy shift in the interband transitions. To confirm this, here we consider a non-degenerate case by replacing (ν,μ)→(n,m)(\nu,\mu)\rightarrow(n,m). Since Γnmbc\Gamma^{bc}_{nm} vanishes for nonresonant light, we can adopt rotating wave approximation, which allows us to put Γnmbc=Γnmebec∗\Gamma^{bc}_{nm}=\Gamma_{nm}e^{b}e^{c*}. After some of algebra, we obtain

Jinj,ene=−e3ℏ2∑k∑n∑m>nfnm(∇kωnm)×∑bcΔωnmb(e)∂Γnmbc∂ωnmMnmbc.(40)\mathbf{J}_{\mathrm{inj,ene}}=-\frac{e^{3}}{\hbar^{2}}\sum_{\mathbf{k}}\sum_{n}\sum_{m>n}f_{nm}\left(\nabla_{\mathbf{k}}\omega_{nm}\right)\times\sum_{bc}\Delta\omega_{nm}^{\vphantom{b}}(e)\frac{\partial\Gamma^{bc}_{nm}}{\partial\omega_{nm}}\mathcal{M}^{bc}_{nm}. \tag*{(40)}

As expected, Eq. (40) can be interpreted as a correction to Eq. (32) by the energy shift $\hbar\Delta\omega_{nm}(e)$ determined by the shift vector [Eq. (A13)]. A similar role of the shift vector has been recognized in the context of Landau–Zener tunneling [13, 68, 69]. It is also quite analogous to the anomalous distribution mechanism of the AHE in magnets, where Fermi’s golden rule for impurity scattering is modified by the potential energy difference added by side jump under a bias electric field [70, 71]. Even for degenerate bands, singular value decomposition resolves the shift vector Rνμ,n(e)R_{\nu\mu,n}(e) and the energy change ℏΔωνμ,n(e)\hbar\Delta\omega_{\nu\mu,n}(e) for each path nn, reducing Eq. (6) to a form equivalent to Eq. (40) (see Appendix C). In nonmagnetic materials, we can obtain an alternative expression:

Jinj,ene=e32ℏ2∑k∑ν∑μ>νfνμ∑b(∇kΓνμb)Wνμb,(41)\mathbf{J}_{\mathrm{inj,ene}}=\frac{e^{3}}{2\hbar^{2}}\sum_{\mathbf{k}}\sum_{\nu}\sum_{\mu>\nu}f_{\nu\mu}\sum_{b}\left(\nabla_{\mathbf{k}}\Gamma^{b}_{\nu\mu}\right)W^{b}_{\nu\mu}, \tag*{(41)}

where Wνμa≡−(ie/ℏ)∑bcdϵabcSνμbcdE0dW^{a}_{\nu\mu}\equiv-(ie/\hbar)\sum_{bcd}\epsilon^{abc}\mathcal{S}^{bcd}_{\nu\mu}E^{d}_{0} contains the information of energy shift in a compact form. Jinj,ene\mathbf{J}_{\mathrm{inj,ene}} in this form is equivalent to the second term in Eq. (137) of Ref. [57]. Equation (6) in the main text follows from (B11).

We note that both of Jinj,dip\mathbf{J}_{\mathrm{inj,dip}} and Jinj,ene\mathbf{J}_{\mathrm{inj,ene}} generally appear simultaneously because they are governed by similar tensors, Tνμbcd\mathcal{T}^{bcd}_{\nu\mu} and Sνμbcd\mathcal{S}^{bcd}_{\nu\mu}. Actually, a simple proportionality Sνμbcd=(ℏωνμ/4e)Tνμbcd\mathcal{S}^{bcd}_{\nu\mu}=(\hbar\omega_{\nu\mu}/4e)\mathcal{T}^{bcd}_{\nu\mu} holds for a two-band system with a Hamiltonian linear in k\mathbf{k}, such as the massive Dirac electron model. A convenient relationship Wνμ=−(ωνμ/4)ΔΩνμW_{\nu\mu}=-(\omega_{\nu\mu}/4)\Delta\Omega_{\nu\mu} holds in such a case.

The fourth term in Eq. (31) arises from the anomalous velocity of photoexcited electrons and holes. It obeys

Jinj,ano=−e3ℏ2∑k∑ν∑μ>νfνμ∑bcΓνμbcTr⁡(ΔvννMνμbc−ΔvμμMμνcb),(42)\mathbf{J}_{\mathrm{inj,ano}}=-\frac{e^{3}}{\hbar^{2}}\sum_{\mathbf{k}}\sum_{\nu}\sum_{\mu>\nu}f_{\nu\mu}\sum_{bc}\Gamma^{bc}_{\nu\mu}\operatorname{Tr}\left(\Delta\mathbf{v}_{\nu\nu}\mathcal{M}^{bc}_{\nu\mu}-\Delta\mathbf{v}_{\mu\mu}\mathcal{M}^{cb}_{\mu\nu}\right), \tag*{(42)}

where Δvνν=−(e/ℏ)E0×Ων\Delta\mathbf{v}_{\nu\nu}=-(e/\hbar)\mathbf{E}_{0}\times\mathbf{\Omega}_{\nu} denotes the anomalous velocity, and Mνμab≡ξνμaξμνb\mathcal{M}^{ab}_{\nu\mu}\equiv\xi^{a}_{\nu\mu}\xi^{b}_{\mu\nu}. Similar to the preceding

terms, the rate of current generation Jinj,ano\mathbf{J}_{\mathrm{inj,ano}} becomes a constant for a combination of a static bias field and monochromatic light, which is the reason why it is formally classified into the injection current according to the criteria in Ref. [57]. However, there is also a qualitative difference between this term and the others. As discussed above, Jinj,dip+Jinj,ene\mathbf{J}_{\mathrm{inj,dip}}+\mathbf{J}_{\mathrm{inj,ene}} originates from modification of the transition probability by E0\mathbf{E}_{0}, so that it appears only when the light and bias fields are applied simultaneously. By contrast, in the case of Jinj,ano\mathbf{J}_{\mathrm{inj,ano}}, E0\mathbf{E}_{0} does not affect the excitation process but changes the intraband velocity, so that this current can be generated even when the bias field is applied after a light pulse excites the system. To cover such a persisting response, it is better to recognize Jinj,ano\mathbf{J}_{\mathrm{inj,ano}} as a light-induced AHE originating from time-reversal symmetry breaking by photoexcited carriers, rather than a field-induced CPGE which can only explain the situation with temporally overlapping light and bias fields. Nevertheless, the case with a static bias field allows both interpretations, as indicated in Figure 5. In a nonmagnetic system, one can obtain an alternative expression,

Figure 5. Classification of photovoltaic Hall effect in nonmagnetic materials.

Jinj,anoa=−e42ℏ3∑k∑ν∑μ>νfνμ∑bcdϵabcE0b×Tr⁡(ΩνcΩνμd+ΩμcΩμνd)Γνμd,(43)J^{a}_{\mathrm{inj,ano}}=-\frac{e^{4}}{2\hbar^{3}}\sum_{\mathbf{k}}\sum_{\nu}\sum_{\mu>\nu}f_{\nu\mu}\sum_{bcd}\epsilon^{abc}E^{b}_{0}\times\operatorname{Tr}\left(\Omega^{c}_{\nu}\Omega^{d}_{\nu\mu}+\Omega^{c}_{\mu}\Omega^{d}_{\mu\nu}\right)\Gamma^{d}_{\nu\mu}, \tag*{(43)}

which shows dependence on the helicity of light through Γνμ\Gamma_{\nu\mu}. This form is equivalent to the first term in Eq. (137) of Ref. [57]. A set of tensor indices (abcd)=(xyz)(abcd)=(xyz) matters when light impinges the sample normally and the AHE is detected within the sample plane. For two-band systems, we have Ω1=Ω12\Omega_{1}=\Omega_{12} and Ω2=Ω21\Omega_{2}=\Omega_{21}, which further simplifies calculation. A massive Dirac electron model yields a Berry curvature Ω1,2∼2λσ\Omega_{1,2}\sim2\lambda\sigma in the neighborhood of band gap. Since (σz)2(\sigma_{z})^{2} has a nonzero trace, Eq. (43) does not vanish. If the operator σ\sigma is associated with spin, it can be regarded as a manifestation of inverse spin Hall effect. In graphene, we obtain Jinj,ano=0\mathbf{J}_{\mathrm{inj,ano}}=0 due to a vanishing Berry curvature.

The last term in Eq. (31) is qualitatively different from the others. It is given by

Jinj,fera=−e4ℏ3∑k∑ν∑μ>ν(A0⋅∇kfνμ)∑bcΓμνbcSνμbca.(44)J_{\mathrm{inj,fer}}^{a}=-\frac{e^{4}}{\hbar^{3}}\sum_{\mathbf{k}}\sum_{\nu}\sum_{\mu>\nu}\left(\mathbf{A}_{0}\cdot\nabla_{\mathbf{k}}f_{\nu\mu}\right)\sum_{bc}\Gamma_{\mu\nu}^{bc}S_{\nu\mu}^{bca}. \tag*{(44)}

Physically, this contribution can be interpreted as a shift current generated on the carrier distribution function accelerated by the bias field. A similar role of acceleration in current-induced second harmonic generation has been discussed in literature [72]. Similar to the other terms, Jinj,fer\mathbf{J}_{\mathrm{inj,fer}} vanishes for nonresonant light, shows a constant generation rate J˙inj,fer\dot{\mathbf{J}}_{\mathrm{inj,fer}} for monochromatic illumination in the absence of relaxation, and exhibits helicity dependence in a nonmagnetic case. However, unlike the other terms, it appears only when a Fermi surface exists and is directly involved in interband transitions. This is a rather rare situation in the context of photovoltaic Hall effect, so we do not delve into Jinj,fer\mathbf{J}_{\mathrm{inj,fer}} in more detail.

Contribution by dressed states

The photocurrent carried by light-dressed electrons can be divided into

Jdre=Jdre,0+Jdre,acc+Jdre,ano.(45)\mathbf{J}_{\mathrm{dre}}=\mathbf{J}_{\mathrm{dre},0}+\mathbf{J}_{\mathrm{dre},\mathrm{acc}}+\mathbf{J}_{\mathrm{dre},\mathrm{ano}}. \tag*{(45)}

The first term,

Jdre,0=eℏ∑k∑νTr⁡fν∇kδϵν,(46)\mathbf{J}_{\mathrm{dre},0}=\frac{e}{\hbar}\sum_{\mathbf{k}}\sum_{\nu}\operatorname{Tr}f_{\nu}\nabla_{\mathbf{k}}\delta\epsilon_{\nu}, \tag*{(46)}

with

δϵν=e22ℏ(A1⋅∇k)(A1⋅vνν)+e22ℏ(A1×E1)⋅Ων+ie22ℏ∑μ≠ν∑bc(FνμbF˙μνc−F˙νμbFμνc)Mνμbc,(47)\delta\epsilon_{\nu}=\frac{e^{2}}{2\hbar}\left(\mathbf{A}_{1}\cdot\nabla_{\mathbf{k}}\right)\left(\mathbf{A}_{1}\cdot\mathbf{v}_{\nu\nu}\right)+\frac{e^{2}}{2\hbar}\left(\mathbf{A}_{1}\times\mathbf{E}_{1}\right)\cdot\mathbf{\Omega}_{\nu}+\frac{ie^{2}}{2\hbar}\sum_{\mu\ne\nu}\sum_{bc}\left(F_{\nu\mu}^{b}\dot{F}_{\mu\nu}^{c}-\dot{F}_{\nu\mu}^{b}F_{\mu\nu}^{c}\right)\mathcal{M}_{\nu\mu}^{bc}, \tag*{(47)}

is independent of E0\mathbf{E}_{0} and accounts for the Fermi-surface contribution to the bulk photovoltaic effect [73]. It includes the nonlinear Hall effect by Berry curvature dipole [74] and Berry connection polarizability [45, 46, 47], as well as the metallic jerk current [57, 75]. The second term,

Jdre,acc=e2ℏ2∑k∑νTr⁡(A0⋅∇kfν)∇kδϵν,(48)\mathbf{J}_{\mathrm{dre},\mathrm{acc}}=\frac{e^{2}}{\hbar^{2}}\sum_{\mathbf{k}}\sum_{\nu}\operatorname{Tr}\left(\mathbf{A}_{0}\cdot\nabla_{\mathbf{k}}f_{\nu}\right)\nabla_{\mathbf{k}}\delta\epsilon_{\nu}, \tag*{(48)}

emerges from acceleration of dressed states by E0\mathbf{E}_{0}. It may contribute to the photovoltaic Hall effect when combined with energy-dependent scattering [76], so we have included it in Figure 5. However, we do not focus on it because such an extrinsic effect lies outside the scope of this paper. The last term,

Jdre,ano=−e2ℏ∑k∑νTr⁡fνDk(E0⋅δξνν)−e2ℏ∂∂t∑k∑νTr⁡fν(A0⋅Dk)δξνν,(49)\mathbf{J}_{\mathrm{dre},\mathrm{ano}}=-\frac{e^{2}}{\hbar}\sum_{\mathbf{k}}\sum_{\nu}\operatorname{Tr}f_{\nu}\mathbf{D}_{\mathbf{k}}\left(\mathbf{E}_{0}\cdot\delta\boldsymbol{\xi}_{\nu\nu}\right)-\frac{e^{2}}{\hbar}\frac{\partial}{\partial t}\sum_{\mathbf{k}}\sum_{\nu}\operatorname{Tr}f_{\nu}\left(\mathbf{A}_{0}\cdot\mathbf{D}_{\mathbf{k}}\right)\delta\boldsymbol{\xi}_{\nu\nu}, \tag*{(49)}

describes the light-induced AHE caused by the anomalous velocity of light-dressed electrons. Indeed, for monochromatic illumination that makes the light-induced Berry connection δξνν\delta\boldsymbol{\xi}_{\nu\nu} constant, Eq. (49) is reduced to

Jdre,ano=e∑k∑νTr⁡fν(−eℏE0×δΩν),(50)\mathbf{J}_{\mathrm{dre},\mathrm{ano}}=e\sum_{\mathbf{k}}\sum_{\nu}\operatorname{Tr}f_{\nu}\left(-\frac{e}{\hbar}\mathbf{E}_{0}\times\delta\mathbf{\Omega}_{\nu}\right), \tag*{(50)}

where δΩν=Dk×δξνν\delta\mathbf{\Omega}_{\nu}=\mathbf{D}_{\mathbf{k}}\times\delta\boldsymbol{\xi}_{\nu\nu} denotes the light-induced Berry curvature. The explicit form of δξνν\delta\boldsymbol{\xi}_{\nu\nu} is given by

δξννa=−e22ℏ2(A1⋅Dk)(A1×Ων)a−e2ℏ∑b(A1×E1)bΠννba−ie2ℏ∑μ≠ν∑bc(FνμbF˙μνc−F˙νμbFμνc)Tνμbca+e22ℏ2∑μ≠ν∑bc(GνμbF˙μνc+F˙νμbGμνc)Sνμbca,(51)\delta\xi_{\nu\nu}^{a}=-\frac{e^{2}}{2\hbar^{2}}\left(\mathbf{A}_{1}\cdot\mathbf{D}_{\mathbf{k}}\right)\left(\mathbf{A}_{1}\times\mathbf{\Omega}_{\nu}\right)^{a}-\frac{e}{2\hbar}\sum_{b}\left(\mathbf{A}_{1}\times\mathbf{E}_{1}\right)^{b}\Pi_{\nu\nu}^{ba}-\frac{ie}{2\hbar}\sum_{\mu\ne\nu}\sum_{bc}\left(F_{\nu\mu}^{b}\dot{F}_{\mu\nu}^{c}-\dot{F}_{\nu\mu}^{b}F_{\mu\nu}^{c}\right)T_{\nu\mu}^{bca}+\frac{e^{2}}{2\hbar^{2}}\sum_{\mu\ne\nu}\sum_{bc}\left(G_{\nu\mu}^{b}\dot{F}_{\mu\nu}^{c}+\dot{F}_{\nu\mu}^{b}G_{\mu\nu}^{c}\right)S_{\nu\mu}^{bca}, \tag*{(51)}

where

Πννad=∑bcϵabcdDbCννcd=∑μ≠νΠνμad.(52)\Pi_{\nu\nu}^{ad}=\sum_{bc}\epsilon^{abcd}D^{b}C_{\nu\nu}^{cd}=\sum_{\mu\ne\nu}\Pi_{\nu\mu}^{ad}. \tag*{(52)}

is the Berry curvature polarizability which describes the correction δΩν=ΠνE0\delta\mathbf{\Omega}_{\nu}=\Pi_{\nu}\mathbf{E}_{0} to Berry curvature. One can show that δξνν\delta\boldsymbol{\xi}_{\nu\nu} also describes the optical rectification,

Prec,0=e∑k∑νTr⁡fνδξνν.(53)\mathbf{P}_{\mathrm{rec},0}=e\sum_{\mathbf{k}}\sum_{\nu}\operatorname{Tr}f_{\nu}\delta\boldsymbol{\xi}_{\nu\nu}. \tag*{(53)}

Thus, δξνν\delta\boldsymbol{\xi}_{\nu\nu} can be naturally interpreted as spatial shift of electron wavefunctions under a light field. One can also show that Eq. (51) is equivalent to the prediction by the Floquet formalism.

Singular value decomposition

When the degeneracy of band ν\nu is written as gνg_{\nu}, the number of possible excitation paths from band μ\mu to band ν\nu amounts to gνgμg_\nu g_\mu for an arbitrary choice of bases, as shown in Figure 6(a). Singular value decomposition helps us reduce it to less than min⁡(gν,gμ)\min(g_\nu,g_\mu), as shown in Figure 6(b). The procedure is as follows. A gμ×gνg_\mu\times g_\nu matrix, A≡e∗⋅ξμνA \equiv\mathbf{e}^{*}\cdot\boldsymbol{\xi}_{\mu\nu}, can always be decomposed into A=VΣW†A=V\Sigma W^\dagger, with a gμ×gμg_\mu\times g_\mu unitary matrix VV, a gν×gνg_\nu\times g_\nu unitary matrix WW, and a gμ×gνg_\mu\times g_\nu matrix

Figure 6. (a) Before singular value decomposition (SVD), there are gν×gμg_{\nu}\times g_{\mu} excitation paths available for the interband transition from band μ\mu to band ν\nu. The case of gν=gμ=2g_{\nu}=g_{\mu}=2 is shown. (b) SVD resolves mutually independent pairs of initial and final states, the number of which is less than min⁡(gν,gμ)\min(g_{\nu},g_{\mu}).

Σ=(λ1⋱0λr0).(54)\Sigma= \begin{pmatrix} \sqrt{\lambda_1} & & & \\ & \ddots& & 0 \\ & & \sqrt{\lambda_r} & \\ & & & 0 \end{pmatrix}. \tag*{(54)}

λn\lambda_n is a positive number, and r≤min⁡(gν,gμ)r \leq\min(g_\nu,g_\mu) is called the rank of AA. With this decomposition, the total transition matrix element, i.e., the trace of Mνμ(e)M_{\nu\mu}(\mathbf{e}), is reduced to

∑bcebec∗Mνμbc=Tr⁡A†A=∑n=1rMνμ,n(e),(55)\sum_{bc}e^b e^{c*}M_{\nu\mu}^{bc}=\operatorname{Tr}A^\dagger A=\sum_{n=1}^{r}M_{\nu\mu,n}(\mathbf{e}), \tag*{(55)}

where we have rewritten λn\lambda_n as Mνμ,n(e)M_{\nu\mu,n}(\mathbf{e}). One can see that the interband transition μ→ν\mu\rightarrow\nu is now decomposed into a smaller number of possible paths labeled by n=1,2,⋯ ,rn=1,2,\cdots,r, each having a transition matrix element Mνμ,n(e)M_{\nu\mu,n}(\mathbf{e}). It is straightforward to find its correction

ΔMνμ,n(e)≡2λnRe⁡[e⋅(W†CνμV)nnE0],(56)\Delta M_{\nu\mu,n}(\mathbf{e})\equiv2\sqrt{\lambda_n}\operatorname{Re}\left[\mathbf{e}\cdot\left(W^\dagger C_{\nu\mu}V\right)_{nn}\mathbf{E}_0\right], \tag*{(56)}

from

∑bcdebec∗TνμbcdE0d=∑n=1rΔMνμ,n(e).(57)\sum_{bcd}e^b e^{c*}T_{\nu\mu}^{bcd}E_0^d=\sum_{n=1}^{r}\Delta M_{\nu\mu,n}(\mathbf{e}). \tag*{(57)}

The singular value decomposition is particularly useful for the shift vector. One obtains

∑bcebec∗Sνμbca=−i2Tr⁡[A†(DaA)−(DaA†)A]=∑n=1rMνμ,n(e)Rνμ,na(e).(58)\sum_{bc}e^b e^{c*}S_{\nu\mu}^{bca}=-\frac{\mathrm{i}}{2}\operatorname{Tr}\left[A^\dagger(D^aA)-(D^aA^\dagger)A\right]=\sum_{n=1}^{r}M_{\nu\mu,n}(\mathbf{e})R_{\nu\mu,n}^a(\mathbf{e}). \tag*{(58)}

Here, Rνμ,n(e)≡(ξˉνν)nn−(ξˉμμ)nnR_{\nu\mu,n}(\mathbf{e})\equiv(\bar{\xi}_{\nu\nu})_{nn}-(\bar{\xi}_{\mu\mu})_{nn} is identified as the shift vector accompanying the nn-th path, while ξˉνν≡W†(ξνν+i∇k)W\bar{\xi}_{\nu\nu}\equiv W^\dagger(\xi_{\nu\nu}+\mathrm{i}\nabla_k)W and ξˉμμ≡V†(ξμμ+i∇k)V\bar{\xi}_{\mu\mu}\equiv V^\dagger(\xi_{\mu\mu}+\mathrm{i}\nabla_k)V are Berry connections for the initial and final states, respectively. (58) applied to Eq. (38) lets us recognize ℏΔωνμ,n(e)=−eE0⋅Rνμ,n(e)\hbar\Delta\omega_{\nu\mu,n}(\mathbf{e})=-e\mathbf{E}_0\cdot\mathbf{R}_{\nu\mu,n}(\mathbf{e}) as the field-induced energy shift for the nn-th path. In practice, it is more convenient to first calculate Dνμ≡−e(E0⋅Dk)ξνμD_{\nu\mu}\equiv-e(\mathbf{E}_0\cdot D_k)\xi_{\nu\mu} and B≡e∗⋅DμνB\equiv\mathbf{e}^{*}\cdot D_{\mu\nu} for an arbitrary E0\mathbf{E}_0. According to an identity

Im⁡Tr⁡A†B=∑n=1rMνμ,n(e)ℏΔωνμ,n(e)=∑n=1rλnIm⁡(V†BW)nn,(59)\operatorname{Im}\operatorname{Tr}A^\dagger B=\sum_{n=1}^{r}M_{\nu\mu,n}(\mathbf{e})\hbar\Delta\omega_{\nu\mu,n}(\mathbf{e}) =\sum_{n=1}^{r}\sqrt{\lambda_n}\operatorname{Im}(V^\dagger BW)_{nn}, \tag*{(59)}

we can express

ℏΔωνμ,n(e)=1λnIm⁡(V†BW)nn,(60)\hbar\Delta\omega_{\nu\mu,n}(\mathbf{e})=\frac{1}{\sqrt{\lambda_n}}\operatorname{Im}(V^\dagger BW)_{nn}, \tag*{(60)}

from which the shift vector follows:

Rνμ,n(e)=−ℏe∂Δωνμ,n∂E0.(61)\mathbf{R}_{\nu\mu,n}(\mathbf{e})=-\frac{\hbar}{e}\frac{\partial\Delta\omega_{\nu\mu,n}}{\partial\mathbf{E}_0}. \tag*{(61)}

References

  1. [1]R. Resta, Manifestations of Berry’s phase in molecules and condensed matter, J. Phys.: Condens. Matter 12, R107 (2000).DOI
  2. [2]D. Xiao, M.-C. Chang, and Q. Niu, Berry phase effects on electronic properties, Rev. Mod. Phys. 82, 1959 (2010).DOI
  3. [3]D. Vanderbilt, Berry phases in electronic structure theory (Cambridge University Press, Cambridge, 2018).DOI
  4. [4]R. Karplus and J. M. Luttinger, Hall Effect in Ferromagnetics, Phys. Rev. 95, 1154 (1954).DOI
  5. [5]M.-C. Chang and Q. Niu, Berry curvature, orbital moment, and effective quantum theory of electrons in electromagnetic fields, J. Phys.: Condens. Matter 20, 193202 (2008).DOI
  6. [6]N. Nagaosa, J. Sinova, S. Onoda, A. H. MacDonald, and N. P. Ong, Anomalous Hall effect, Rev. Mod. Phys. 82, 1539 (2010).DOI
  7. [7]F. D. M. Haldane, Model for a Quantum Hall Effect without Landau Levels: Condensed-Matter Realization of the “Parity Anomaly”, Phys. Rev. Lett. 61, 2015 (1988).pubmed.ncbi.nlm.nih.gov/10038961
  8. [8]R. Yu, W. Zhang, H.-J. Zhang, S.-C. Zhang, X. Dai, and Z. Fang, Quantized Anomalous Hall Effect in Magnetic Topological Insulators, Science 329, 61 (2010).
  9. [9]C.-Z. Chang, J. Zhang, X. Feng, J. Shen, Z. Zhang, M. Guo, K. Li, Y. Ou, P. Wei, L.-L. Wang, Z.-Q. Ji, Y. Feng, S. Ji, X. Chen, J. Jia, X. Dai, Z. Fang, S.-C. Zhang, K. He, Y. Wang, L. Lu, X.-C. Ma, and Q.-K. Xue, Experimental Observation of the Quantum Anomalous Hall Effect in a Magnetic Topological Insulator, Science 340, 167 (2013).DOI
  10. [10]J. E. Sipe and A. I. Shkrebtii, Second-order optical response in semiconductors, Phys. Rev. B 61, 5337 (2000).DOI
  11. [11]P. Hosur, Circular photogalvanic effect on topological insulator surfaces: Berry-curvature-dependent response, Phys. Rev. B 83, 035309 (2011).
  12. [12]J. Orenstein, J. E. Moore, T. Morimoto, D. H. Torchinsky, J. W. Harter, and D. Hsieh, Topology and Symmetry of Quantum Materials via Nonlinear Optical Responses, Annu. Rev. Condens. Matter Phys. 12, 247 (2021).DOI
  13. [13]T. Morimoto, S. Kitamura, and N. Nagaosa, Geometric Aspects of Nonlinear and Nonequilibrium Phenomena, J. Phys. Soc. Jpn. 92, 072001 (2023).DOI
  14. [14]F. de Juan, A. G. Grushin, T. Morimoto, and J. E. Moore, Quantized circular photogalvanic effect in Weyl semimetals, Nat Commun 8, 15995 (2017).DOI
  15. [15]A. Avdoshkin, V. Kozii, and J. E. Moore, Interactions Remove the Quantization of the Chiral Photocurrent at Weyl Points, Phys. Rev. Lett. 124, 196603 (2020).DOI
  16. [16]W. J. Jankowski and R.-J. Slager, Quantized Integrated Shift Effect in Multigap Topological Phases, Phys. Rev. Lett. 133, 186601 (2024).DOI
  17. [17]T. Oka and H. Aoki, Photovoltaic Hall effect in graphene, Phys. Rev. B 79, 081406(R) (2009).DOI
  18. [18]C.-K. Chan, P. A. Lee, K. S. Burch, J. H. Han, and Y. Ran, When Chiral Photons Meet Chiral Fermions: Photoinduced Anomalous Hall Effects in Weyl Semimetals, Phys. Rev. Lett. 116, 026805 (2016).DOI
  19. [19]M.-N. Chen and W.-C. Chen, Photoinduced Weyl semimetal phase and anomalous Hall effect in a three-dimensional topological insulator, Chin. Phys. B 30, 110308 (2021).DOI
  20. [20]J. W. McIver, B. Schulte, F.-U. Stein, T. Matsuyama, G. Jotzu, G. Meier, and A. Cavalleri, Light-induced anomalous Hall effect in graphene, Nat. Phys. 16, 38 (2020).
  21. [21]N. Yoshikawa, Y. Hirai, K. Ogawa, S. Okumura, K. Fujiwara, J. Ikeda, T. Koretsune, R. Arita, A. Mitra, A. Tsukazaki, et al., Light-induced anomalous Hall conductivity in massive 3D Dirac semimetal Co3Sn2S2, arXiv:2210.11932.
  22. [22]Y. Hirai, N. Yoshikawa, M. Kawaguchi, M. Hayashi, S. Okumura, T. Oka, and R. Shimano, Anomalous Hall effect of light-driven three-dimensional Dirac electrons in bismuth, arXiv:2301.06072.arxiv.org/abs/2301.06072
  23. [23]Y. Hirai, S. Okumura, N. Yoshikawa, T. Oka, and R. Shimano, Floquet Weyl states at one-photon resonance: An origin of nonperturbative optical responses in three-dimensional materials, Phys. Rev. Research 6, L012027 (2024).DOI
  24. [24]M. W. Day, K. Kusyak, F. Sturm, J. I. Aranzadi, H. M. Bretscher, M. Fechner, T. Matsuyama, M. H. Michael, B. F. Schulte, X. Li, et al., Nonperturbative Nonlinear Transport in a Floquet-Weyl Semimetal, arXiv:2409.04531.arxiv.org/abs/2409.04531
  25. [25]H. Cao, J.-T. Sun, and S. Meng, Floquet engineering of anomalous Hall effects in monolayer MoS2, npj Quantum Mater. 9, 90 (2024).DOI
  26. [26]A. A. Bakun, B. P. Zakharchenya, A. A. Rogachev, M. N. Tkachuk, and V. G. Fleisher, Observation of a surface photocurrent caused by optical orientation of electrons in a semiconductor, Pis’ma Zh. Eksp. Teor. Fiz. 40, 464 (1984) [JETP Lett. 40, 1293 (1984)].
  27. [27]M. I. Miah, Observation of the anomalous Hall effect in GaAs, J. Phys. D 40, 1659 (2007).DOI
  28. [28]C. M. Yin, N. Tang, S. Zhang, J. X. Duan, F. J. Xu, J. Song, F. H. Mei, X. Q. Wang, B. Shen, Y. H. Chen, et al., Observation of the photoinduced anomalous Hall effect in GaN-based heterostructures, Appl. Phys. Lett. 98, 122104 (2011).
  29. [29]K. S. Virk and J. E. Sipe, Optical Injection and Terahertz Detection of the Macroscopic Berry Curvature, Phys. Rev. Lett. 107, 120403 (2011).DOI
  30. [30]J. L. Yu, Y. H. Chen, C. Y. Jiang, Y. Liu, H. Ma, and L. P. Zhu, Observation of the photoinduced anomalous Hall effect spectra in insulating InGaAs/AlGaAs quantum wells at room temperature, Appl. Phys. Lett. 100, 142109 (2012).DOI
  31. [31]N. Okamoto, H. Kurebayashi, T. Trypiniotis, I. Farrer, D. A. Ritchie, E. Saitoh, J. Sinova, J. Mašek, T. Jungwirth, and C. H. W. Barnes, Electric control of the spin Hall effect by intervalley transitions, Nat. Mater. 13, 932 (2014).DOI
  32. [32]S. Priyadarshi, K. Pierz, and M. Bieler, Detection of the Anomalous Velocity with Subpicosecond Time Resolution in Semiconductor Nanostructures, Phys. Rev. Lett. 115, 257401 (2015).DOI
  33. [33]T. Fujimoto, T. Kurihara, Y. Murotani, T. Tamaya, N. Kanda, C. Kim, J. Yoshinobu, H. Akiyama, T. Kato, and R. Matsunaga, Observation of Terahertz Spin Hall Conductivity Spectrum in GaAs with Optical Spin Injection, Phys. Rev. Lett. 132, 016301 (2024).DOI
  34. [34]Y. Murotani, N. Kanda, T. Fujimoto, T. Matsuda, M. Goyal, J. Yoshinobu, Y. Kobayashi, T. Oka, S. Stemmer, and R. Matsunaga, Anomalous Hall Transport by Optically Injected Isospin Degree of Freedom in Dirac Semimetal Thin Film, Nano Lett. 24, 222 (2024).DOI
  35. [35]D. Xiao, G.-B. Liu, W. Feng, X. Xu, and W. Yao, Coupled Spin and Valley Physics in Monolayers of MoS2 and Other Group-VI Dichalcogenides, Phys. Rev. Lett. 108, 196802 (2012).DOI
  36. [36]K. F. Mak, K. L. McGill, J. Park, and P. L. McEuen, The valley Hall effect in MoS2 transistors, Science 344, 1489 (2014).DOI
  37. [37]S. A. Sato, J. W. McIver, M. Nuske, P. Tang, G. Jotzu, B. Schulte, H. Hübener, U. De Giovannini, L. Mathey, M. A. Sentef, et al., Microscopic theory for the light-induced anomalous Hall effect in graphene, Phys. Rev. B 99, 214302 (2019).DOI
  38. [38]S. A. Sato, P. Tang, M. A. Sentef, U. De Giovannini, H. Hübener, and A. Rubio, Light-induced anomalous Hall effect in massless Dirac fermion systems and topological insulators with dissipation, New J. Phys. 21, 093005 (2019).DOI
  39. [39]P. X. Nguyen and W.-K. Tse, Photoinduced anomalous Hall effect in two-dimensional transition metal dichalcogenides, Phys. Rev. B 103, 125420 (2021).DOI
  40. [40]Y. Murotani, N. Kanda, T. Fujimoto, T. Matsuda, M. Goyal, J. Yoshinobu, Y. Kobayashi, T. Oka, S. Stemmer, and R. Matsunaga, Disentangling the Competing Mechanisms of Light-Induced Anomalous Hall Conductivity in Three-Dimensional Dirac Semimetal, Phys. Rev. Lett. 131, 096901 (2023).DOI
  41. [41]T. Fujimoto, Y. Murotani, T. Tamaya, T. Kurihara, N. Kanda, C. Kim, J. Yoshinobu, H. Akiyama, T. Kato, and R. Matsunaga, Light-induced inverse spin Hall effect and field-induced circular photogalvanic effect in GaAs revealed by two-dimensional terahertz Fourier analysis, arXiv:2411.00528 [Phys. Rev. B (to be published)].DOI
  42. [42]H. Watanabe and Y. Yanase, Chiral Photocurrent in Parity-Violating Magnet and Enhanced Response in Topological Antiferromagnet, Phys. Rev. X 11, 011001 (2021).
  43. [43]C. Aversa and J. E. Sipe, Nonlinear optical susceptibilities of semiconductors: Results with a length-gauge analysis, Phys. Rev. B 52, 14636 (1995).pubmed.ncbi.nlm.nih.gov/9980796
  44. [44]B. M. Fregoso, R. A. Muniz, and J. E. Sipe, Jerk Current: A Novel Bulk Photovoltaic Effect, Phys. Rev. Lett. 121, 176604 (2018).DOI
  45. [45]Y. Gao, S. A. Yang, and Q. Niu, Field Induced Positional Shift of Bloch Electrons and Its Dynamical Implications, Phys. Rev. Lett. 112, 166601 (2014).DOI
  46. [46]H. Liu, J. Zhao, Y.-X. Huang, W. Wu, X.-L. Sheng, C. Xiao, and S. A. Yang, Intrinsic Second-Order Anomalous Hall Effect and Its Application in Compensated Antiferromagnets, Phys. Rev. Lett. 127, 277202 (2021).
  47. [47]H. Liu, J. Zhao, Y.-X. Huang, X. Feng, C. Xiao, W. Wu, S. Lai, W. Gao, and S. A. Yang, Berry connection polarizability tensor and third-order Hall effect, Phys. Rev. B 105, 045118 (2022).DOI
  48. [48]A. G. Volosniev, A. S. Kumar, D. Lorenc, Y. Ashour- ishkri, A. A. Zhumekenov, O. M. Bakr, M. Lemeshko, and Z. Alipshchev, Spin-Electric Coupling in Lead Halide Perovskites, Phys. Rev. Lett. 130, 106901 (2023).DOI
  49. [49]Y. Lu, Q. Wang, L. Han, Y. Zhao, Z. He, W. Song, C. Song, and Z. Miao, Spintronic Phenomena and Applications in Hybrid Organic-Inorganic Perovskites, Adv. Funct. Mater. 34, 2314427 (2024).DOI
  50. [50]S. Murakami, N. Nagaosa, and S.-C. Zhang, Dissipationless Quantum Spin Current at Room Temperature, Science 301, 1348 (2003).
  51. [51]X. Dai and F.-C. Zhang, Light-induced Hall effect in semiconductors with spin-orbit coupling, Phys. Rev. B 76, 085343 (2007).DOI
  52. [52]J. L. Yu, Y. H. Chen, Y. Liu, C. Y. Jiang, H. Ma, L. P. Zhu, and X. D. Qin, Intrinsic photoinduced anomalous Hall effect in insulating GaAs/AlGaAs quantum wells at room temperature, Appl. Phys. Lett. 102, 02408 (2013).DOI
  53. [53]B. A. Bernevig, T. L. Hughes, and S.-C. Zhang, Orbitronics: The Intrinsic Orbital Current in p-Doped Silicon, Phys. Rev. Lett. 95, 066601 (2005).pubmed.ncbi.nlm.nih.gov/16090968
  54. [54]D. Go, D. Jo, H.-W. Lee, M. Kläui, and Y. Mokrousov, Orbitronics: Orbital currents in solids, EPL 135, 37001 (2021).
  55. [55]J. Sinova, S. O. Valenzuela, J. Wunderlich, C. H. Back, and T. Jungwirth, Spin Hall effects, Rev. Mod. Phys. 87, 1213 (2015).DOI
  56. [56]J. Ahn, G.-Y. Guo, N. Nagaosa, and A. Vishwanath, Riemannian geometry of resonant optical responses, Nat. Phys. 18, 290 (2022).DOI
  57. [57]B. M. Fregoso, Bulk photovoltaic effect in the presence of a static electric field, Phys. Rev. B 100, 064301 (2019); B. M. Fregoso, Erratum: Bulk photovoltaic effects in the presence of a static electric field [Phys. Rev. B 100, 064301 (2019)], Phys. Rev. B 102, 059901(E) (2020).DOI
  58. [58]Y. Dong, M.-M. Yang, M. Yoshii, S. Matsuoka, S. Kitamura, T. Hasegawa, N. Ogawa, T. Morimoto, T. Ideue, and Y. Iwasa, Giant bulk piezophotovoltaic effect in 3R-MoS2, Nat. Nanotechnol. 18, 36 (2023).DOI
  59. [59]A. V. Parafilo, V. M. Kovalev, and I. G. Savenko, “Photoinduced anomalous supercurrent Hall effect,” Phys. Rev. B 108, L180509 (2023).DOI
  60. [60]E. I. Blount, Bloch Electrons in a Magnetic Field, Phys. Rev. 126, 1636 (1962).DOI
  61. [61]B. I. Sturman, Ballistic and shift currents in the bulk photovoltaic effect theory, Phys. Usp. 63, 407 (2020).
  62. [62]Z. Dai, A. M. Schankler, L. Gao, L. Z. Tan, and A. M. Rappe, Phonon-Assisted Ballistic Current from First-Principles Calculations, Phys. Rev. Lett. 126, 177403 (2021).DOI
  63. [63]P. Zhu and A. Alexandradinata, Anomalous shift and optical vorticity in the steady photovoltaic current, Phys. Rev. B 110, 115108 (2024).DOI
  64. [64]L. Berger, Side-Jump Mechanism for the Hall Effect of Ferromagnets, Phys. Rev. B 2, 4559 (1970).DOI
  65. [65]J. Smit, The spontaneous hall effect in ferromagnetics II, Physica 24, 39 (1958).DOI
  66. [66]G. B. Ventura, D. J. Passos, J. M. Viana Parente Lopes, and J. M. B. Lopes dos Santos, Comment on Jerk Current: A Novel Bulk Photovoltaic Effect, Phys. Rev. Lett. 126, 259701 (2021).DOI
  67. [67]B. M. Fregoso, R. A. Muniz, and J. E. Sipe, Fregoso, Muniz, and Sipe Reply, Phys. Rev. Lett. 126, 259702 (2021).DOI
  68. [68]S. Kitamura, N. Nagaosa, and T. Morimoto, Nonreciprocal Landau-Zener tunneling, Commun. Phys. 3, 63 (2020).DOI
  69. [69]S. Takayoshi, J. Wu, and T. Oka, Nonadiabatic nonlinear optics and quantum geometry — Application to the twisted Schwinger effect, SciPost Phys. 11, 075 (2021).DOI
  70. [70]N. A. Sinitsyn, A. H. MacDonald, T. Jungwirth, V. K. Dugaev, and J. Sinova, Anomalous Hall effect in a two-dimensional Dirac band: The link between the Kubo-Streda formula and the semiclassical Boltzmann equation approach, Phys. Rev. B 75, 045315 (2007).DOI
  71. [71]N. A. Sinitsyn, Semiclassical theories of the anomalous Hall effect, J. Phys.: Condens. Matter 20, 023201 (2008).DOI
  72. [72]K. Takasan, T. Morimoto, J. Orenstein, and J. E. Moore, Current-induced second harmonic generation in inversion-symmetric Dirac and Weyl semimetals, Phys. Rev. B 104, L161202 (2021).DOI
  73. [73]L. Gao, Z. Addison, E. J. Mele, and A. M. Rappe, Intrinsic Fermi-surface contribution to the bulk photovoltaic effect, Phys. Rev. Res. 3, L042032 (2021).DOI
  74. [74]I. Sodemann and L. Fu, Quantum Nonlinear Hall Effect Induced by Berry Curvature Dipole in Time-Reversal Invariant Materials, Phys. Rev. Lett. 115, 216806 (2015).DOI
  75. [75]O. Matsyshyn and I. Sodemann, Nonlinear Hall Acceleration and the Quantum Rectification Sum Rule, Phys. Rev. Lett. 123, 246602 (2019).DOI
  76. [76]M. V. Durnev, Photovoltaic Hall effect in the two-dimensional electron gas: Kinetic theory, Phys. Rev. B 104, 085306 (2021).DOI

Paper details

vlnc:W_8skdgmp6
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