betaFeedback

WorksPhysics and Astronomy

Nonlinear Nonreciprocal Photocurrents under Phonon Dressing

Abstract

Nonlinear optical (NLO) effects have attracted great interest recently. However, by far the computational studies on NLO use the independent particle approximation and ignore many-body effects. Here we develop a generic Green's function framework to calculate the NLO response functions, which can incorporate various many-body interactions. We focus on the electron-phonon coupling and reveal that phonon dressing can make significant impacts on nonlinear photocurrent, such as the bulk photovoltaic (BPV) and bulk spin photovoltaic (BSPV) currents. BPV/BSPV should be zero for centrosymmetric crystals, but when phonons are driven out-of-equilibrium, by for example, a temperature gradient ∇T\nabla T, the optical selections rules are altered and phonon-pumped BPV/BSPV currents can be non-zero in nominally centrosymmetric crystal. Moreover, we elucidate that such NLO responses under non-equilibrium phonon dressing can be nonreciprocal, as the direction of the current does not necessarily get reversed when the direction of the temperature gradient is reversed.

Introduction

Light is a valuable tool for characterizing diverse material properties, including topological quantities such as the Berry curvature [1, 2, 3, 4, 5, 6, 7, 8]. Light can also drive electronic, phononic and nuclear-spin excitations [9, 10] and even trigger ionic [11, 12, 13] and electronic [14, 15, 16] phase transitions. On the other hand, light-matter interactions can be used to detect and generate light. For example, the spontaneous parametric down-conversion in nonlinear optical (NLO) materials can be applied to create entangled photon pairs used in quantum information technologies [17].

To better understand and harness light-matter interactions, it is crucial to obtain the response functions from first principles. Most works hitherto rely on the independent particle approximation (IPA), which assumes that electrons are (nearly) free particles and ignores many-body interactions. This is unsatisfactory since many-body interactions can play a central role in quantum materials [18]. For example, excitonic effects can remarkably enhance NLO responses [19, 20, 21]. Even for conventional materials, it is indispensable to include many-body interactions such as electron-phonon (e-ph) coupling [22], if one aims to accurately calculate the response functions. Therefore, it is highly desirable to have a theoretical framework to systematically incorporate many-body interactions when calculating the response functions.

In this work, we formulate a generic Green’s function framework, which can be used to deal with various many-body interactions. We use e-ph coupling as an example to demonstrate how many-body interactions can make quantitative and even qualitative impacts on light-matter interactions. We will focus on nonlinear photocurrents [23, 24, 25] such as the bulk photovoltaic (BPV) and bulk spin photovoltaic (BSPV) effects, while the methodologies in this work also apply to other NLO responses, including higher-order NLO effects. We start with the temperature-gradient effect on BPV responses. If the phonons are at equilibrium, the IPA can give reasonably good results compared with the many-body formalism. In contrast, when phonons are out-of-equilibrium, rich physics can arise, which cannot be captured if e-ph couplings are ignored. Using a two-temperature model, we illustrate that phonon dressing can effectively break the original spatial symmetries and consequently alter the optical selection rules. We provide a geometrical interpretation for this effect. We also reveal that the nonlinear photocurrent under phonon dressing can be nonreciprocal, in that for nominally centrosymmetric crystals, the direction of the current does not necessarily get reversed when the direction of the temperature gradient is reversed. We also introduce a symmetry-breaking vector, which quantifies and unifies the symmetry breaking from both atomic structures (intrinsic) and phonon dressings (extrinsic).

Green’s Function Framework. First, we briefly introduce the Green’s function framework and demonstrate how many-body interactions can be incorporated. The BPV effect can be described as

ja=σbca(ω)Eb(ω)Ec(−ω)(1)j^{a} = \sigma^{a}_{bc}(\omega)\mathcal{E}^{b}(\omega)\mathcal{E}^{c}(-\omega) \tag*{(1)}

Here E(ω)\mathcal{E}(\omega) is the alternating electric field with frequency ω\omega. jaj^a is the DC charge current. aa and b/cb/c denote the directions of the current and the electric field, respectively. σbca(ω)\sigma^a_{bc}(\omega) is the BPV conductivity and can be expressed as

σbca(ω)=−ie3ω2∫dk(2π)d∫dE2πTr⁡{vaG~bc<(E)}(2)\sigma^a_{bc}(\omega)=-\frac{ie^3}{\omega^2}\int\frac{\mathrm{d}k}{(2\pi)^d}\int\frac{\mathrm{d}E}{2\pi}\operatorname{Tr}\left\{v^a\widetilde{\mathcal{G}}^<_{bc}(E)\right\} \tag*{(2)}

where

G~bc<=Gr(E)vbGr(E+ω)vcG<(E)+Gr(E)vbG<(E+ω)vcGa(E)+G<(E)vbGa(E+ω)vcGa(E)+(b↔c,+ω↔−ω)\widetilde{\mathcal{G}}^<_{bc}= \mathcal{G}^r(E)v^b\mathcal{G}^r(E+\omega)v^c\mathcal{G}^<(E)+\mathcal{G}^r(E)v^b\mathcal{G}^<(E+\omega)v^c\mathcal{G}^a(E)+\mathcal{G}^<(E)v^b\mathcal{G}^a(E+\omega)v^c\mathcal{G}^a(E)+(b\leftrightarrow c,+\omega\leftrightarrow-\omega)

The Feynman diagram for the process above is shown in Figure 1. For simplicity, we did not include terms that may arise from higher-order band dispersions in a tight-binding model. See Supplementary Materials [26] and Refs. [6, 27] for detailed discussions. (b↔c,+ω↔−ω)(b\leftrightarrow c,+\omega\leftrightarrow-\omega) indicates simultaneous exchange between bb, cc and +ω+\omega, −ω-\omega. ee is the electron charge, dd is the dimension of the system, and vv is the electron velocity. G~\widetilde{G} and GG denote Green’s functions with and without light illumination, while the superscripts rr, aa, and << denote retarded, advanced, and lesser Green’s functions, respectively. Both vv and G/G~G/\widetilde{G} are functions of kk-point in the first Brillouin zone, and the kk arguments are occasionally omitted for simplicity. EE is a frequency/energy argument (the Planck constant ℏ\hbar is occasionally set as 1 in the following). The many-body interactions are incorporated into the Green’s function GG, which can be obtained perturbatively with e.g., Feynman diagrams or non-perturbatively with other methods [28].

Feynman diagram and electron propagator under phonon dressing

Figure 1. (a) Feynman diagram of the BPV process. (b) Electron propagator under phonon dressing. Think (thin) solid arrow: dressed (undressed) electron propagator. Wavy line: photon propagator. Dashed arrow: phonon propagator. Double arrow: charge current output.

The Migdal electron self-energy due to e-ph coupling can be expressed as [28, 29, 30]

Σnn′(E,k)=∑m,qνgmnν∗(k,q)gmn′ν(k,q)[nqν+1−fm,k+qE−ϵm,k+q−ωqν+iδp+nqν+fm,k+qE−ϵm,k+q+ωqν+iδp](3)\Sigma_{nn'}(E,k)=\sum_{m,\mathbf{q}\nu}g^*_{mn\nu}(k,q)g_{mn'\nu}(k,q)\left[\frac{n_{q\nu}+1-f_{m,k+q}}{E-\epsilon_{m,k+q}-\omega_{q\nu}+i\delta_p}+\frac{n_{q\nu}+f_{m,k+q}}{E-\epsilon_{m,k+q}+\omega_{q\nu}+i\delta_p}\right] \tag*{(3)}

where gmnν(k,q)≡(ℏ2m0ωqν)12⟨ψm,k+q∣∂qνV∣ψnk⟩g_{mn\nu}(k,q)\equiv\left(\frac{\hbar}{2m_0\omega_{q\nu}}\right)^{\frac{1}{2}}\langle\psi_{m,k+q}|\partial_{q\nu}V|\psi_{nk}\rangle is electron-phonon coupling matrix, with mm, nn as electron band indices, ψ\psi as electron wavefunction, VV as the self-consistent potential, qq, ν\nu and ωqν\omega_{q\nu} as the phonon wavevector, branch, and frequencies, respectively. ϵnk\epsilon_{nk} is the electronic band energy. fnkf_{nk} and nqνn_{q\nu} are electron and phonon occupation numbers. m0m_0 is a mass parameter that makes Σ\Sigma have the unit of energy. Here we adopt the convention in the EPW [30, 31] package and take m0m_0 as the proton mass. Note that if one sticks to the same m0m_0 throughout the calculation process, then the final results would be the same. δp\delta_p corresponds to phonon lifetime and is set as 50 meV. Varying δp\delta_p would not change the essence of our general results (see Figure S6 in Ref. [26]). Here we do not include the static Debye-Waller term in the electron self-energy, whose contributions to the BPV conductivities are small according to our tests. The electron Green’s function can be obtained from the Dyson’s equation [28]

G−1(E,k)=G0−1(E,k)−Σ(E,k)(4)G^{-1}(E,\mathbf{k}) = G_0^{-1}(E,\mathbf{k}) - \Sigma(E,\mathbf{k}) \tag*{(4)}

where G0(E,k)=1E−Hk+iδeG_0(E,\mathbf{k}) = \frac{1}{E-H_{\mathbf{k}}+i\delta_e} is the non-interacting Green’s function, with HkH_{\mathbf{k}} as the electron single-particle Hamiltonian. δe\delta_e represents electron self-energy due to interactions with defects, other electrons, etc. (excluding interactions with phonons). δe\delta_e is taken as 20 meV (corresponding to relaxation time ~0.2 ps), based on recent experimental results [32, 33].

In the following, we will use monolayer WSe2_2 as an example to illustrate the phonon dressing effects on BPV. We use monolayer WSe2_2 because some concepts can be easily demonstrated and visualized in 2D materials. The main conclusions in this work apply to 3D materials as well. As an example, we also studied 3D silicon, which has inversion symmetry. It is shown that temperature difference can effectively break the inversion symmetry and produce a DC photocurrent in silicon (Figure S5 in Ref. [26]). We will focus on the BPV charge photocurrents under linearly polarized light, while similar conclusions should also carry over to circularly polarized light, and spin current under BSPV as well [34].

Temperature Effect on BPV: Phonons in Equilibrium. First, we consider a phonon ensemble with equilibrium occupation nqν=nBE(ωqν,T)n_{\mathbf{q}\nu}=n_{\mathrm{BE}}(\omega_{\mathbf{q}\nu},T), with nBEn_{\mathrm{BE}} as the Bose-Einstein distribution and TT as the temperature. The BPV conductivities σbca(ω,T)\sigma_{bc}^{a}(\omega,T) is a function of TT, due to the temperature-dependence of electron self-energies in Eq. (3). σyyy(ω,T)\sigma_{yy}^{y}(\omega,T) of WSe2_2 at several temperatures are shown in Figure 2a. One can see that at elevated temperatures, some resonance peaks in the σyyy(ω,T)\sigma_{yy}^{y}(\omega,T) vs. ω\omega spectrum disappear, and the magnitude of σyyy(ω,T)\sigma_{yy}^{y}(\omega,T) decreases with TT in a sublinear fashion (Figure 2b). This is because phonon populations increase with temperature, leading to faster e-ph scatterings and thus shorter electron lifetimes. The distributions of the imaginary and real part of the phonon self-energy at band energies Σnn(ϵnk,k)\Sigma_{nn}(\epsilon_{n\mathbf{k}},\mathbf{k}) are shown in Figure 2c, Figure 2d (filled areas), respectively. Generally, Σnn(ϵnk,k)\Sigma_{nn}(\epsilon_{n\mathbf{k}},\mathbf{k}) has larger absolute values and wider distributions at higher temperatures. The mean electron self-energies defined as ⟨Σ⟩≡1N∑kΣnn(ϵnk,k)\langle\Sigma\rangle\equiv\frac{1}{N}\sum_{\mathbf{k}}\Sigma_{nn}(\epsilon_{n\mathbf{k}},\mathbf{k}) with NN as the number of total electron modes, are shown as black curves in Figure 2c, Figure 2d. Im⁡[⟨Σ⟩]\operatorname{Im}[\langle\Sigma\rangle] grows linearly with TT for T≳100T\gtrsim100 K, while Re⁡[⟨Σ⟩]\operatorname{Re}[\langle\Sigma\rangle] is close to 0 and is almost independent of TT.

Phonons in equilibrium: four panels showing photocurrent, temperature dependence, and the imaginary and real parts of electron self-energy

Figure 2. Phonons in equilibrium. (a) σyyy(ω,T)\sigma_{yy}^{y}(\omega,T) of monolayer WSe2_2 at selected temperatures. Colored solid curves are from many-body formalism while the black dashed curve is from IPA. (b) σyyy(ω,T)\sigma_{yy}^{y}(\omega,T) as a function of TT for selected ω\omega. (c) imaginary and (d) real part of electron self-energy due to e-ph coupling. The colored areas represent the distribution of self-energy at selected temperatures (labelled on the xx-axis), while the black curve is the mean electron self-energy.

The calculations above use the many-body formalism and explicitly incorporate e-ph coupling. The BPV conductivities can also be calculated using IPA. In this case, the e-ph coupling effect can be approximated with a mode-independent relaxation time ⟨τ⟩=ℏIm⁡[⟨Σ⟩]\langle\tau\rangle=\frac{\hbar}{\operatorname{Im}[\langle\Sigma\rangle]}. Here we take ℏ/⟨τ⟩=26\hbar/\langle\tau\rangle=26 meV, corresponding to the mean self-energy at T=0T=0 (Figure 2c). σyyy(ω)\sigma_{yy}^{y}(\omega) based on IPA is shown as the black dashed curve in Figure 2a, which in general agrees with the σyyy(ω,T=0 K)\sigma_{yy}^{y}(\omega,T=0\ \mathrm{K}) from the many-body formalism. We also find similar agreements for other components of the σbca\sigma_{bc}^{a} tensor at other temperatures. This indicates that when phonons are in thermal equilibrium, IPA with mode-independent relaxation time can give reasonably good results. However, this is not the case when phonons are out-of-equilibrium, as we will show below.

Symmetry Breaking from Temperature Difference. In the previous section, we considered an equilibrium phonon distribution under uniform temperature. Richer physics can arise when the phonons are out-of-equilibrium, driven by the boundary condition or volumetric pumping by e.g. another photon source. Typically, phonons are out-of-equilibrium when there is a temperature difference – if the temperature at the left boundary T1T_1 is higher than that at the right boundary T2T_2, there will be more phonons traveling to the right. Here we adopt a simplified two-temperature model to describe non-equilibrium phonons – phonons with velocities vqν⋅x^>0\mathbf{v}_{\mathbf{q}\nu}\cdot\hat{\mathbf{x}}>0 have occupations number nqν1=nBE(ωqν,T1)n_{\mathbf{q}\nu}^{1}=n_{\mathrm{BE}}(\omega_{\mathbf{q}\nu},T_1), while those with vqν⋅x^<0\mathbf{v}_{\mathbf{q}\nu}\cdot\hat{\mathbf{x}}<0 have nqν2=nBE(ωqν,T2)n_{\mathbf{q}\nu}^{2}=n_{\mathrm{BE}}(\omega_{\mathbf{q}\nu},T_2). vqν\mathbf{v}_{\mathbf{q}\nu} is the group velocity of the phonon and x^\hat{\mathbf{x}} is the unit vector along transport xx-direction (Figure 3c). The phonon occupations nqν1n_{\mathbf{q}\nu}^{1} and nqν2n_{\mathbf{q}\nu}^{2} are inserted into Eq. (3), yielding the electron self-energy, which is in turn used to calculate the BPV conductivities σbca(ω,T1,T2)\sigma_{bc}^{a}(\omega,T_1,T_2). Here we have made two approximations: 1) the phonon occupations obey an ideal two-temperature model, and 2) we extend the Migdal self-energy to non-equilibrium systems. These two approximations are reasonable when the system is slightly out-of-equilibrium (∣T1−T2∣≪T1\lvert T_{1}-T_{2}\rvert\ll T_{1}). However, to reduce numerical noise, ∣T1−T2∣∼T1\lvert T_{1}-T_{2}\rvert\sim T_{1} is used. The essence of the results should be qualitatively valid. Quantitatively, the phonon dressing effects can be weaker in reality, since the phonon distributions tend to thermalize and should be less extreme than the two-temperature model described above. To rigorously calculate the BPV responses under a continuous temperature gradient, one may solve the Boltzmann transport equation that couples the electron and phonon distributions. But this would greatly complicate both theoretical and computational analyses, and we would like to leave this for future works.

Two-temperature model for monolayer WSe2

Figure 3. Two-temperature model. (a) σxxx(ω,T1,T2)\sigma_{xx}^{x}(\omega,T_{1},T_{2}) of monolayer WSe2_2 in the two-temperature model with T1=300T_{1}=300 K and varied T2T_{2}. (b) atomic structure of monolayer WSe2_2 with Mx\mathcal{M}_{x} symmetry labelled by the dashed line. (c) Illustration of the two-temperature model. Wavy arrows denote phonons. (d) σxxx(ω,T1,T2)\sigma_{xx}^{x}(\omega,T_{1},T_{2}) as a function of T2T_{2} with selected ω\omega and T1=300T_{1}=300 K. (e, f, g) electron self-energy Im⁡[Σnn(E,k)]\operatorname{Im}[\Sigma_{nn}(E,\mathbf{k})] with selected T1T_{1} and T2T_{2}. nn corresponds to the lowest conduction band, and EE is 0.2 eV higher than the conduction band minimum. The black hexagons denote the first Brillouin zone. The color-bars are in the unit of eV.

A prominent effect of the temperature gradient is symmetry breaking. Due to mirror symmetry Mx\mathcal{M}_{x} (Figure 3b) of monolayer WSe2_2, some elements of the BPV tensors, such as σxxx\sigma_{xx}^{x}, are forbidden. When phonons are in equilibrium, Mx\mathcal{M}_{x} will be preserved even if e-ph couplings are considered. However, when more phonons are traveling to the right (T1>T2T_{1}>T_{2}, Figure 3c), +x+x and −x-x will be inequivalent for phonons. The Mx\mathcal{M}_{x} symmetry breaking can be transmitted to the electronic system due to asymmetry scatterings with asymmetric phonons, leading to nonzero σxxx\sigma_{xx}^{x}. This effect is illustrated in Figure 3a, Figure 3d, where we fix T1=300T_{1}=300 K and vary T2T_{2}. One can see that when T2=T1=300T_{2}=T_{1}=300 K, σxxx\sigma_{xx}^{x} is unanimously zero for all light frequencies. In contrast, when T2≠T1T_{2}\ne T_{1}, σxxx\sigma_{xx}^{x} can be non-zero, and the sign of σxxx\sigma_{xx}^{x} is opposite for T2>T1T_{2}>T_{1} and T1>T2T_{1}>T_{2}. This is different from coherently exciting phonons via THz irradiation [35], which triggers transient symmetry breaking from equilibrium.

The Mx\mathcal{M}_x symmetry breaking in the electron system can be directly visualized in the Migdal self-energy. Here we plot Im⁡[Σnn(E,k)]\operatorname{Im}[\Sigma_{nn}(E,\mathbf{k})] as a function of k\mathbf{k}, with nn corresponding to the lowest conduction band, and EE fixed as 0.2 eV higher than the conduction band minimum. When T1=T2=300T_1 = T_2 = 300 K, the Mx\mathcal{M}_x symmetry in Im⁡[Σnn(E,k)]\operatorname{Im}[\Sigma_{nn}(E,\mathbf{k})] is preserved (Figure 3f). This corroborates that Mx\mathcal{M}_x symmetry in the electron system cannot be broken by phonons in equilibrium. On the other hand, when T1≠T2T_1 \ne T_2, the Mx\mathcal{M}_x symmetry in Im⁡[Σnn(E,k)]\operatorname{Im}[\Sigma_{nn}(E,\mathbf{k})] disappears (Figure 3e, Figure 3g). In other words, electrons with wavevector k\mathbf{k} are scattered by phonons at different rates than those with Mxk\mathcal{M}_x\mathbf{k} (Mx⋆^\mathcal{M}_x\hat{\star} denotes the mirror-xx image of ⋆^\hat{\star}). Under light illumination, contributions from electrons at k\mathbf{k} and Mxk\mathcal{M}_x\mathbf{k} would not cancel, resulting in non-zero σxxx\sigma_{xx}^{x}.

The symmetry breaking due to phonon dressings can be interpreted geometrically as well. Electrons can be represented by the localized Wannier functions ∣mR⟩|m\mathbf{R}\rangle, where mm and R\mathbf{R} label the orbital and the unit-cell, respectively. The rate of phonon-assisted electron jumping in real space is pmR→m′R′∝nqν∣⟨m′R′∣∂qνV∣mR⟩∣2p_{m\mathbf{R}\to m'\mathbf{R}'} \propto n_{\mathbf{q}\nu}|\langle m'\mathbf{R}'|\partial_{\mathbf{q}\nu}V|m\mathbf{R}\rangle|^2. Due to Mx\mathcal{M}_x symmetry, one has ∣⟨m′R′∣∂qνV∣mR⟩∣2=∣⟨m′,MxR′∣∂Mxq,νV∣m,MxR⟩∣2|\langle m'\mathbf{R}'|\partial_{\mathbf{q}\nu}V|m\mathbf{R}\rangle|^2 = |\langle m',\mathcal{M}_x\mathbf{R}'|\partial_{\mathcal{M}_x\mathbf{q},\nu}V|m,\mathcal{M}_x\mathbf{R}\rangle|^2. When phonons are in equilibrium, one has nqν=nMxq,νn_{\mathbf{q}\nu} = n_{\mathcal{M}_x\mathbf{q},\nu}, and thus pmR→m′R′=pm,MxR→m′,MxR′p_{m\mathbf{R}\to m'\mathbf{R}'} = p_{m,\mathcal{M}_x\mathbf{R}\to m',\mathcal{M}_x\mathbf{R}'}. However, if nqν≠nMxq,νn_{\mathbf{q}\nu} \ne n_{\mathcal{M}_x\mathbf{q},\nu}, then pmR→m′R′≠pm,MxR→m′,MxR′p_{m\mathbf{R}\to m'\mathbf{R}'} \ne p_{m,\mathcal{M}_x\mathbf{R}\to m',\mathcal{M}_x\mathbf{R}'}, indicating that scatterings between +x+x and −x-x are asymmetrical, and that electrons have tendencies to move in a certain direction (either +x+x or −x-x). Consequently, the free carriers generated by light would have an overall displacement in real space, leading to a net charge current. This geometrical interpretation is similar to that of the shift current [36], where the asymmetrical scatterings come from asymmetrical atomic potentials.

Nonreciprocal Phonon Dressing. One interesting observation from Figure 2 and Figure 3 is the large magnitude of σxxx\sigma_{xx}^{x} in the two-temperature model. When ∣T1−T2∣=200|T_1-T_2| = 200 K, the differences between the self-energies of electrons at k\mathbf{k} and Mxk\mathcal{M}_x\mathbf{k} are on the order of 0.01 eV (Figure 3e, Figure 3g). This is rather small compared with the typical energy scales involved in NLO responses, such as the electron band energies and the light frequency, both of which are ∼1\sim1 eV. Remarkably, the small asymmetry in electron self-energies leads to quite a large σxxx\sigma_{xx}^{x}. Comparing Figure 3a and Figure 2a, one can see that the magnitude of σxxx(ω,T1=100 K,T2=300 K)\sigma_{xx}^{x}(\omega,T_1=100\ \mathrm{K},T_2=300\ \mathrm{K}) can reach ∼30%\sim30\% of that of σyyy(ω,T=100 K or 300 K)\sigma_{yy}^{y}(\omega,T=100\ \mathrm{K}\ \text{or}\ 300\ \mathrm{K}). Note that σyyy\sigma_{yy}^{y} in Figure 2a comes from the intrinsic asymmetry in the atomic structure of WSe2_2. To understand this phenomenon, one can look at the k\mathbf{k}-resolved contributions to the total BPV conductivities, defined as Ibca(k)=−ie3ω2∫dE2πTr⁡{vaG~bc<(E)}I_{bc}^{a}(\mathbf{k})=-\frac{ie^3}{\omega^2}\int\frac{dE}{2\pi}\operatorname{Tr}\{v^a\widetilde{G}_{bc}^{<}(E)\}. Here we fix ω=2\omega=2 eV and T1=T2=300T_1=T_2=300 K. Ixxx(k)I_{xx}^{x}(\mathbf{k}) for σxxx\sigma_{xx}^{x} and Iyyy(k)I_{yy}^{y}(\mathbf{k}) for σyyy\sigma_{yy}^{y} are plotted in Figure 4a and Figure 4b, respectively. Due to the Mx\mathcal{M}_x symmetry, one has Ixxx(k)=−Ixxx(Mxk)I_{xx}^{x}(\mathbf{k})=-I_{xx}^{x}(\mathcal{M}_x\mathbf{k}). In contrast, mirror-yy My\mathcal{M}_y symmetry is broken in WSe2_2, and hence Iyyy(k)≠−Iyyy(Myk)I_{yy}^{y}(\mathbf{k})\ne-I_{yy}^{y}(\mathcal{M}_y\mathbf{k}). However, the difference between Iyyy(k)I_{yy}^{y}(\mathbf{k}) and −Iyyy(Myk)-I_{yy}^{y}(\mathcal{M}_y\mathbf{k}) is barely noticeable in Figure 4b. Quantitatively, integrating the absolute value of Iyyy(k)I_{yy}^{y}(\mathbf{k}) gives ∣∫dk(2π)2Iyyy(k)∣∼0.01×∫dk(2π)2∣Iyyy(k)∣\left|\int\frac{d\mathbf{k}}{(2\pi)^2}I_{yy}^{y}(\mathbf{k})\right|\sim0.01\times\int\frac{d\mathbf{k}}{(2\pi)^2}|I_{yy}^{y}(\mathbf{k})|. This suggests that the My\mathcal{M}_y symmetry breaking in Iyyy(k)I_{yy}^{y}(\mathbf{k}) is only ∼1%\sim1\%, and that most contributions from k\mathbf{k} are canceled by those from Myk\mathcal{M}_{y}\mathbf{k}. As a result, the magnitude of σyyy\sigma_{yy}^{y} is limited.

Nonreciprocal phonon dressing

Figure 4. Nonreciprocal phonon dressing. (a, b) k\mathbf{k}-resolved contributions to the BPV conductivities for (a) σxxx\sigma_{xx}^{x} and (b) σyyy\sigma_{yy}^{y} at ω=2\omega=2 eV and T1=T2=300T_{1}=T_{2}=300 K. The black hexagon is the first Brillouin zone. The color-bars are in the unit of Å3^{3}/A/V2^{2}. (c) σxxx(ω,T1,T2)\sigma_{xx}^{x}(\omega,T_{1},T_{2}) as a function of θ\theta with ω=3\omega=3 eV, T1=300T_{1}=300 K. Pink and green curves correspond to T2=100T_{2}=100 and 300 K, respectively. The green curve is zero for all θ\theta. (d) similar to (c), but for σyyy\sigma_{yy}^{y}. The insets in c (d) are σxxx\sigma_{xx}^{x} (σyyy\sigma_{yy}^{y}) vs. θ\theta plotted with polar axes. (e) An illustration of the symmetry breaking vector.

To further quantify the symmetry break strength, we introduce a symmetry-breaking vector s\mathbf{s}. Note that s\mathbf{s} is a macroscopic quantity. Each electron mkm\mathbf{k} may have its own symmetry condition (e.g., k\mathbf{k} can be a low symmetry point in the Brillouin zone), and s\mathbf{s} corresponds to the “average” or “global” symmetry condition of all electrons. The intrinsic asymmetry in the atomic structure of monolayer WSe2_2 leads to sint=(0,a0,0)\mathbf{s}_{\mathrm{int}}=(0,a_{0},0) with a0a_{0} as a constant, which breaks My\mathcal{M}_{y} and preserves Mx\mathcal{M}_{x}. The temperature difference described above leads to an extrinsic symmetry breaking. Here we generalize the temperature difference to an arbitrary direction. Assuming that the temperature difference is along n^=(cos⁡θ,sin⁡θ,0)\hat{\mathbf{n}}=(\cos\theta,\sin\theta,0), then phonons with vqν⋅n^>0\mathbf{v}_{\mathbf{q}\nu}\cdot\hat{\mathbf{n}}>0 and vqν⋅n^<0\mathbf{v}_{\mathbf{q}\nu}\cdot\hat{\mathbf{n}}<0 have temperature T1T_{1} and T2T_{2}, respectively. The macroscopic symmetry breaking vector sΔT\mathbf{s}_{\Delta T} induced by asymmetric phonons dressing should be parallel to n^\hat{\mathbf{n}}, hence sΔT=(bcos⁡θ,bsin⁡θ,0)\mathbf{s}_{\Delta T}=(b\cos\theta,b\sin\theta,0) with b=b(T1,T2)b=b(T_{1},T_{2}) as a function of both T1T_{1} and T2T_{2}. The total symmetry breaking vector is (Figure 4e)

s=sint+sΔT=(bcos⁡θ,a0+bsin⁡θ,0)(5)\mathbf{s}=\mathbf{s}_{\mathrm{int}}+\mathbf{s}_{\Delta T}=(b\cos\theta,a_{0}+b\sin\theta,0) \tag*{(5)}

The BPV conductivities σbca\sigma_{bc}^{a} should be a function of s\mathbf{s}. Here we fix ω=3\omega=3 eV. When T1=T2T_{1}=T_{2}, one has b=0b=0 and s=(0,a0,0)\mathbf{s}=(0,a_{0},0), hence both σxxx\sigma_{xx}^{x} and σyyy\sigma_{yy}^{y} are independent of θ\theta (green curves in Figures 4c, 4d). When T1≠T2T_{1}\ne T_{2} (T1=300T_{1}=300 K, T2=100T_{2}=100 K here), one has b≠0b\ne0, then σxxx\sigma_{xx}^{x} and

σyyy\sigma_{yy}^{y} would be θ\theta-dependent (pink curves in Figure 4c, Figure 4d). For σxxx\sigma_{xx}^{x}, sx=bcos⁡θs_x = b \cos\theta is relevant, and one has σxxx≈7.1×cos⁡θ\sigma_{xx}^{x} \approx7.1 \times\cos\theta [nm ⋅\cdot μ\muA/V2^2]. As for σyyy\sigma_{yy}^{y}, sy=a0+bsin⁡θs_y = a_0 + b \sin\theta is relevant, and again one has the sinusoidal relationship σyyy≈−12.3+6.1×sin⁡θ\sigma_{yy}^{y} \approx-12.3 + 6.1 \times\sin\theta [nm ⋅\cdot μ\muA/V2^2]. One can deduce that ∣a0b∣∼12.36.1∼2\left|\frac{a_0}{b}\right| \sim\frac{12.3}{6.1} \sim2, indicating that the magnitude of sint\mathbf{s}_{\mathrm{int}} and sΔT\mathbf{s}_{\Delta T} are close. As discussed before, the asymmetry in electron self-energies due to asymmetric phonon dressing is weak (∼\sim100 times smaller than the typical energy scale in optical processes). Therefore, the intrinsic symmetry breaking is weak, there is plenty of room to enhance the BPV conductivities by breaking the symmetries to a greater extent. Potential ways of breaking symmetries may be strain gradient, electric field, magnetic field, etc.

Interestingly, the BPV current under phonon dressing is nonreciprocal [37, 38]. As shown in Figure 4d, the photocurrent does not necessarily change direction when the direction of the temperature difference reverses (θ→π+θ\theta\to\pi+ \theta). This comes from the interference between the asymmetry from 1) extrinsic phonon dressing and 2) intrinsic atomic potential [Eq. (5)]. The intrinsic shift current is the consequence of the so-called shift vector Rint\mathbf{R}_{\mathrm{int}} – upon photoexcitation, the electron wave package can jump in real space by Rint\mathbf{R}_{\mathrm{int}}, owing to scatterings with asymmetric atomic potentials. The average Rint\mathbf{R}_{\mathrm{int}} (over all electron modes) is closely related to the intrinsic symmetry breaking and should be parallel with sint\mathbf{s}_{\mathrm{int}}. The scatterings with asymmetrical phonons, on the other hand, lead to another electron displacement RΔT\mathbf{R}_{\Delta T} in real space as discussed before, resulting in an additional term in the total current. The average RΔT\mathbf{R}_{\Delta T} should align in parallel with sΔT\mathbf{s}_{\Delta T}. Consequently, one has the total current j∝Rint+RΔT∝sint+sΔT\mathbf{j} \propto\mathbf{R}_{\mathrm{int}} + \mathbf{R}_{\Delta T} \propto\mathbf{s}_{\mathrm{int}} + \mathbf{s}_{\Delta T}. When the direction of the temperature difference reverses (sΔT→−sΔT\mathbf{s}_{\Delta T} \to-\mathbf{s}_{\Delta T}), j\mathbf{j} does not necessarily change sign. Note that time-reversal symmetry T\mathcal{T} is broken in our two-temperature model – when there is a temperature difference, the system would thermalize, leading to entropy increase and T\mathcal{T} breaking. There are also other sources of dissipation in the nonlinear photocurrent generation process. For example, under light illumination, some electrons would be excited to the conduction bands. There will be dissipations when these electrons jump back to the valence bands. The important role of dissipation is discussed in Refs. [39, 40].

In summary, we developed a Green’s function formalism to systematically incorporate the many-body interactions in NLO processes. We use e-ph coupling as an example and demonstrate that the phonon dressing effect can make significant impacts on NLO responses. Notably, out-of-equilibrium phonons can lead to symmetry breaking in nominally centrosymmetric crystals and alter the selection rules on NLO processes. We also elucidate that the phonon dressing effect can be nonreciprocal. This work paves the way for future studies on the correlations between many-body theory and light-matter interactions.

Acknowledgments. This work was supported by an Office of Naval Research MURI through grant #N00014-17-1-2661. Haowei Xu acknowledges helpful discussions with Meihui Liu.

References

  1. [1]S.-Y. Xu, Q. Ma, H. Shen, V. Fatemi, S. Wu, T.-R. Chang, G. Chang, A. M. M. Valdivia, C.-K. Chan, Q. D. Gibson, J. Zhou, Z. Liu, K. Watanabe, T. Taniguchi, H. Lin, R. J. Cava, L. Fu, N. Gedik, and P. Jarillo-Herrero, Electrically Switchable Berry Curvature Dipole in the Monolayer Topological Insulator WTe2, Nat. Phys. 2018 14 19, 900 (2018).
  2. [2]Q. Ma, S.-Y. Xu, C.-K. Chan, C.-L. Zhang, G. Chang, Y. Lin, W. Xie, T. Palacios, H. Lin, S. Jia, P. A. Lee, P. Jarillo-Herrero, and N. Gedik, Direct Optical Detection of Weyl Fermion Chirality in a Topological Semimetal, Nat. Phys. 2017 139 13, 842 (2017).
  3. [3]S.-Y. Xu, Q. Ma, Y. Gao, A. Kogar, A. Zong, A. M. Mier Valdivia, T. H. Dinh, S.-M. Huang, B. Singh, C.-H. Hsu, T.-R. Chang, J. P. C. Ruff, K. Watanabe, T. Taniguchi, H. Lin, G. Karapetrov, D. Xiao, P. Jarillo-Herrero, and N. Gedik, Spontaneous Gyrotropic Electronic Order in a Transition-Metal Dichalcogenide, Nat. 2020 5787796 578, 545 (2020).
  4. [4]D. Kaplan, T. Holder, and B. Yan, Nonvanishing Subgap Photocurrent as a Probe of Lifetime Effects, Phys. Rev. Lett. 125, 227401 (2020).
  5. [5]J. Ahn, G.-Y. Guo, and N. Nagaosa, Low-Frequency Divergence and Quantum Geometry of the Bulk Photovoltaic Effect in Topological Semimetals, Phys. Rev. X 10, 041041 (2020).
  6. [6]T. Holder, D. Kaplan, and B. Yan, Consequences of Time-Reversal-Symmetry Breaking in the Light-Matter Interaction: Berry Curvature, Quantum Metric, and Diabatic Motion, Phys. Rev. Res. 2, 033100 (2020).
  7. [7]H. Watanabe and Y. Yanase, Chiral Photocurrent in Parity-Violating Magnet and Enhanced Response in Topological Antiferromagnet, Phys. Rev. X 11, 011001 (2021).
  8. [8]Q. Ma, A. G. Grushin, and K. S. Burch, Topology and Geometry under the Nonlinear Electromagnetic Spotlight, Nat. Mater. 2021 1 (2021).
  9. [9]E. J. Sie, C. M. Nyby, C. D. Pemmaraju, S. J. Park, X. Shen, J. Yang, M. C. Hoffmann, B. K. Ofori-Okai, R. Li, A. H. Reid, S. Weathersby, E. Mannebach, N. Finney, D. Rhodes, D. Chenet, A. Antony, L. Balicas, J. Hone, T. P. Devereaux, T. F. Heinz, X. Wang, and A. M. Lindenberg, An Ultrafast Symmetry Switch in a Weyl Semimetal, Nature.
  10. [10]E. A. Mashkovich, K. A. Grishunin, R. M. Dubrovin, A. K. Zvezdin, R. V. Pisarev, and A. V. Kimel, Terahertz-Driven Light Driven Coupling of Antiferromagnetic Spins to Lattice, Science (80- ). (2021).
  11. [11]J. Shi, Y.-Q. Bie, W. Chen, S. Fang, J. Han, Z. Cao, T. Taniguchi, K. Watanabe, V. Bulović, E. Kaxiras, P. Jarillo-Herrero, and K. A. Nelson, Terahertz-Driven Irreversible Topological Phase Transition in Two-Dimensional MoTe$_2$, ArXiv (2019).
  12. [12]X. Li, T. Qiu, J. Zhang, E. Baldini, J. Lu, A. M. Rappe, and K. A. Nelson, Terahertz Field-Induced Ferroelectricity in Quantum Paraelectric SrTiO3, Science (80- ). 364, 1079 (2019).
  13. [13]J. Zhou, H. Xu, Y. Shi, J. Li, J. Zhou, H. Xu, J. Li, and Y. Shi, Terahertz Driven Reversible Topological Phase Transition of Monolayer Transition Metal Dichalcogenides, Adv. Sci. 8, 2003832 (2021).
  14. [14]J. W. McIver, B. Schulte, F. U. Stein, T. Matsuyama, G. Jotzu, G. Meier, and A. Cavalleri, Light-Induced Anomalous Hall Effect in Graphene, Nature Physics.
  15. [15]H. Xu, J. Zhou, and J. Li, Light-Induced Quantum Anomalous Hall Effect on the 2D Surfaces of 3D Topological Insulators, Adv. Sci. 8, 2101508 (2021).
  16. [16]M. Mitrano, A. Cantaluppi, D. Nicoletti, S. Kaiser, A. Perucchi, S. Lupi, P. Di Pietro, D. Pontiroli, M. Riccò, S. R. Clark, D. Jaksch, and A. Cavalleri, Possible Light-Induced Superconductivity in K3C60 at High Temperature, Nat. 2016 5307591 530, 461 (2016).
  17. [17]C. Zhang, Y. F. Huang, B. H. Liu, C. F. Li, and G. C. Guo, Spontaneous Parametric Down-Conversion Sources for Multiphoton Experiments, Adv. Quantum Technol. 4, 2000132 (2021).
  18. [18]The Rise of Quantum Materials, Nat. Phys. 2016 122 12, 105 (2016).
  19. [19]T. G. Pedersen, Intraband Effects in Excitonic Second-Harmonic Generation, Phys. Rev. B - Condens. Matter Mater. Phys. 92, 235432 (2015).
  20. [20]Y.-H. Chan, D. Y. Qiu, F. H. da Jornada, and S. G. Louie, Giant Exciton-Enhanced Shift Currents and Direct Current Conduction with Subbandgap Photo Excitations Produced by Many-Electron Interactions, Proc. Natl. Acad. Sci. 118, 2021 (2021).
  21. [21]T. Kaneko, Z. Sun, Y. Murakami, D. Golež, and A. J. Millis, Bulk Photovoltaic Effect Driven by Collective Excitations in a Correlated Insulator, (2020).
  22. [22]Z. Dai, A. M. Schankler, G. L. Z. Tan, and A. M. Rappe, Phonon-Assisted Ballistic Current from First-Principles Calculations, Phys. Rev. Lett. 126, 177403 (2021).
  23. [23]R. Von Baltz and W. Kraut, Theory of the Bulk Photovoltaic Effect in Pure Crystals, Phys. Rev. B 23, 5590 (1981).
  24. [24]J. Sipe and A. Shkrebtii, Second-Order Optical Response in Semiconductors, Phys. Rev. B - Condens. Matter Mater. Phys. 61, 5337 (2000).
  25. [25]L. Z. Tan, F. Zheng, S. M. Young, F. Wang, S. Liu, and A. M. Rappe, Shift Current Bulk Photovoltaic Effect in Polar Materials—Hybrid and Oxide Perovskites and Beyond, Npj Comput. Mater. 2016 21 2, 1 (2016).
  26. [26]See Supplemental Material at [URL Will Be Inserted by Publisher] for Derivations of the Green’s Function Framework, and Other Supplementary Figures; Also See Refs. [6,27,28,30,31,34,41–47], (n.d.).
  27. [27]D. E. Parker, T. Morimoto, J. Orenstein, and J. E. Moore, Diagrammatic Approach to Nonlinear Optical Response with Application to Weyl Semimetals, Phys. Rev. B 99, 045121 (2019).
  28. [28]G. D. Mahan, Many-Particle Physics (Springer US, 2000).
  29. [29]F. Giustino, Electron-Phonon Interactions from First Principles, Rev. Mod. Phys. 89, 015003 (2017).
  30. [30]S. Poncé, I. Z. R. Margine, C. Verdi, and F. Giustino, EPW: Electron–Phonon Coupling, Transport and Superconducting Properties Using Maximally Localized Wannier Functions, Comput. Phys. Commun. 209, 116 (2016).
  31. [31]J. Noffsinger, F. Giustino, B. D. Malone, C. H. Park, S. G. Louie, and M. L. Cohen, EPW: A Program for Calculating the Electron–Phonon Coupling Using Maximally Localized Wannier Functions, Comput. Phys. Commun. 181, 2140 (2010).
  32. [32]Q. Wang, J. Li, J. Besbas, C.-H. Hsu, K. Cai, L. Yang, S. Cheng, Y. Wu, W. Zhang, K. Wang, T.-R. Chang, H. Lin, H. Chang, H. Yang, Q. Wang, J. Besbas, Y. Wu, H. Yang, J. Li, L. Yang, S. Cheng, W. Zhang, H. Chang, C. Hsu, H. Lin, K. Cai, K. Wang, and T. Chang, Room-Temperature Nanoseconds Spin Relaxation in WTe2 and MoTe2 Thin Films, Adv. Sci. 5, 1700912 (2018).
  33. [33]H. Wang, C. Zhang, and F. Rana, Surface Recombination Limited Lifetimes of Photoexcited Carriers in Few-Layer Transition Metal Dichalcogenide MoS2, Nano Lett. 15, 8204 (2015).
  34. [34]H. Xu, H. Wang, J. Zhou, and J. Li, Pure Spin Photocurrent in Non-Centrosymmetric Crystals: Bulk Spin Photovoltaic Effect, Nat. Commun. 2021 121 12, 1 (2021).
  35. [35]Y. Shi and J. Zhou, Coherence Control of Directional Nonlinear Photocurrent in Spatially Symmetric Systems, Phys. Rev. B 104, 155146 (2021).
  36. [36]H. Wang and X. Qian, Ferroic-Driven Nonlinear Photocurrent Switching in Time-Reversal Invariant Ferroic Materials, Sci. Adv. 5, eaav9743 (2019).
  37. [37]Y. Tokura and N. Nagaosa, Nonreciprocal Responses from Non-Centrosymmetric Quantum Materials, Nat. Commun. 2018 91 9, 1 (2018).
  38. [38]T. Morimoto and N. Nagaosa, Nonreciprocal Current from Electron Interactions in Noncentrosymmetric Crystals: Roles of Time Reversal Symmetry and Dissipation, Sci. Reports 2018 81 8, 1 (2018).
  39. [39]Y. Michishita and N. Nagaosa, Dissipation and Geometry in Nonlinear Quantum Transports of Multiband Electronic Systems, (2022).
  40. [40]S. S. Tsirkin and I. Souza, Hall vs Ohmic as the Only Proper Generic Partition of the Nonlinear Current, (2021).

Paper details

vlnc:W_mqs3wm5x
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

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