betaFeedback

WorksPhysics and Astronomy

Optical signatures of Rashba–Dresselhaus coupling across Franz–Keldysh and Landau regimes in two-dimensional semiconductors

Abstract

We develop a unified theory of the interband optical response of a two-dimensional semiconductor with simultaneous Rashba and Dresselhaus spin-orbit couplings in electric and magnetic fields. In crossed fields, the in-plane electric field generates an effective spin-orbit drift field that modifies the Landau spectrum and spinor states, while at zero magnetic field the problem maps onto Landau–Zener spin dynamics.At zero external fields, the total joint density of states above the band gap remains unchanged, while characteristic subgap structures appear. Pure Rashba coupling produces an inverse square root singularity, whereas a finite Dresselhaus term gives a step and a logarithmic peak, allowing the two couplings to be distinguished optically. At zero magnetic field, the Franz–Keldysh response shows a crossover from diabatic to adiabatic spin dynamics as the electric field is reduced. In the weak-field Rashba regime, the field-free singularity becomes a finite peak whose height scales as E−1/3\mathcal E^{-1/3} and whose position approaches Eg−ESOE_g-E_{\rm SO}, providing a direct measure of the Rashba energy. In a perpendicular magnetic field, spin-orbit coupling shifts and splits the Landau absorption lines and produces anticrossings without changing their total weights. The Rashba splitting changes from a B\sqrt{B} dependence at weak fields to a linear dependence at strong fields. Finally, in crossed electric and magnetic fields, the electric field broadens the Landau lines and drives a continuous crossover toward the Franz–Keldysh regime, while spin-orbit coupling produces opposite shifts and different widths for the two members of each doublet.

Introduction

Spin-orbit coupling connects the motion of charge carriers with their spin and has therefore become a central mechanism for spin control in low-dimensional semiconductor systems [1, 2, 3, 4, 5, 6]. In two-dimensional semiconductor heterostructures, the dominant linear-in-momentum spin-orbit interactions are commonly the Rashba and Dresselhaus terms. The Rashba interaction originates from an asymmetry of the confining structure or potential, while the Dresselhaus interaction originates from the lack of inversion symmetry of the crystal lattice. Both interactions can coexist and their strengths can be measured and controlled experimentally [7, 8, 9, 10]. The relative strength of these two couplings determines the spin structure of the system and leads to a special symmetry when the Rashba and Dresselhaus coupling strengths become equal, where the persistent spin helix appears [11, 12, 13, 14]. The observation of giant Rashba splittings in polar semiconductors has also shown that the spin-orbit energy can become comparable to the energy scales associated with experimentally accessible electric and magnetic fields [15, 16, 17]. In this regime, the optical response can be strongly affected by all these mechanisms. An electric field modifies the interband absorption through the Franz–Keldysh effect, while a perpendicular magnetic field produces Landau quantization and a discrete magnetoabsorption spectrum [18, 19, 20, 21]. It is therefore important to understand the optical response when Rashba and Dresselhaus couplings act together in the presence of both in-plane electric and perpendicular magnetic fields.

This combined regime is not only of theoretical interest, but can be accessed experimentally in a number of low-dimensional semiconductor platforms. Among these systems, III–V quantum wells are particularly important because their electronic, spin, and optical properties can be engineered within the same heterostructure. InAs and InSbAs wells combine high mobility two-dimensional electron gases with strong spin-orbit coupling and external field control [22, 23, 24], while GaAs/AlGaAs and InGaAs structures allow the confinement, optical transitions, and spin-orbit interaction to be tuned through the layer and interface design [25, 26]. Their direct band gaps and mature growth technology also make these materials important for semiconductor lasers and integrated photonic devices [27, 28]. Halide perovskites provide another important platform because strong spin-orbit effects can coexist with optical activity and electric field control. In these materials, electrically switchable spin structures have been demonstrated at room temperature, while magneto-optical experiments have shown that magnetic fields can strongly modify the optical response [29, 30]. More recently, Rashba–Dresselhaus type spin physics has also been realized in exciton polaritons, which are mixed light and matter states, allowing spin-dependent optical transport and a spin Hall effect at room temperature [31]. These developments are directly relevant for spin transistors, spin-orbit torque memories, optical spin control, and integrated optospin­tronic devices, where electric fields, magnetic fields, spin-orbit coupling, and light must act within the same low-dimensional system [32, 33, 34, 35].

Existing theoretical studies show that the combined Rashba and Dresselhaus spectrum in a perpendicular magnetic field, including the effect of an in-plane electric field, has mainly been studied at the level of Landau energies and spin structure [36, 37, 38, 39]. However, analytical treatments of this problem have often relied on approximate or perturbative solutions, especially when the in-plane electric field is included, and the role of this field as a spin-orbit drift field together with its effect on the crossings, anticrossings, and mixing of the Landau states has received much less attention. Building on this, optical studies have considered either the zero field Rashba–Dresselhaus response [40] or magnetoabsorption in a perpendicular field, mostly for Rashba coupling alone [20, 41]. Optical absorption in crossed electric and magnetic fields has also been studied, but generally without spin-orbit coupling [42, 43, 44], while calculations including both spin-orbit couplings and both fields have been limited to confined structures such as quantum rings [45]. Therefore, the interband optical response of an extended two-dimensional system with simultaneous Rashba and Dresselhaus couplings, an in-plane electric field, and a perpendicular magnetic field remains largely unexplored. In particular, a common description of the Franz–Keldysh response, the magnetoabsorption spectrum, and the crossover between these two regimes is still missing.

In this work, we develop a unified description of the interband optical response of a two-dimensional semiconductor with simultaneous Rashba and Dresselhaus couplings under an in-plane electric field and a perpendicular magnetic field. In § 2, we introduce the model and formulate the optical absorption in terms of the joint density of states. The crossed field spectrum and spinor states are obtained by numerically converged exact diagonalization in § 3.1, where the electric field induced spin-orbit drift term and the resulting level shifts and anticrossings are analyzed. In § 3.2, the zero magnetic field spin dynamics is mapped onto the Landau–Zener problem, including the persistent spin helix limit. The field-free optical response is studied in § 3.3, followed by the Franz–Keldysh electroabsorption in § 3.4 and the magnetoabsorption spectrum in § 3.5. Finally, § 3.6 treats the full cross-field problem. The main conclusions are summarized in § 4, where we also discuss representative material parameters and the experimental accessibility of the predicted effects.

Model of spin–orbit coupled electroabsorption

We consider a 2D crystal in the xyxy plane in the presence of Rashba and Dresselhaus spin-orbit coupling (SOC), subjected to a magnetic field B=(0,0,B)\mathbf{B}=(0,0,B) normal to the plane and an electric field E=(E,0,0)\mathcal{E}=(\mathcal{E},0,0) applied along the xx direction. The crystal is exposed to monochromatic light of angular frequency ω\omega, whose photons induce direct interband transitions from the valence band to the conduction band, as schematically shown in Figure 1.

Image not extracted

Figure 1. Schematic representation of the model. (a) A 2D crystal of effective thickness dd subjected to a perpendicular magnetic field B∥z^\mathbf{B}\parallel\hat{\mathbf{z}} and an in-plane electric field E∥x^\mathbf{E}\parallel\hat{\mathbf{x}}, together with normally incident monochromatic light of angular frequency ω\omega, wave vector q∥z^\mathbf{q}\parallel\hat{\mathbf{z}}, and in-plane polarization e^\hat{\mathbf{e}}. The dashed orbit shows the cyclotron motion, vd=cE×B/B2v_d=c\mathbf{E}\times\mathbf{B}/B^2 denotes the Hall drift velocity, and ⟨σ⟩\langle\sigma\rangle indicates the spin orientation. (b) In-plane spin orientations for Rashba and Dresselhaus couplings on constant-energy contours. (c) Direct interband optical transitions with ℏω=Ecvs(k)\hbar\omega=E_{cv}^{s}(k) for the pure Rashba limit βD=0\beta_D=0, where s=±s=\pm labels the two spin branches, ⊙\odot and ⊗\otimes indicate in-plane spins pointing out of and into the plane of the figure, EgE_g is the band gap, and ESO=m∗αR2/2E_{\mathrm{SO}}=m^*\alpha_R^2/2 is the Rashba spin–orbit energy.

Denoting the complex dielectric function of the crystal by ϵ=ϵ1+iϵ2\epsilon=\epsilon_1+i\epsilon_2 and the electric-field amplitude of the light wave inside the layer by E⊥E_\perp, the time-averaged absorbed power per unit area of a 2D layer with effective thickness dd can be written on the one hand as w=12ωϵ2ϵ0dE⊥2w=\frac{1}{2}\omega\epsilon_2\epsilon_0dE_\perp^2 and on the other hand as w=ℏωwcvw=\hbar\omega w_{cv}, where ϵ0\epsilon_0 is the vacuum permittivity and wcvw_{cv} is the interband transition rate per unit area. Using Fermi’s golden rule for a direct interband dipole transition from the valence band vv to the conduction band cc, we can write the imaginary part of the dielectric function as ϵ2(ω)=πe2∣e^⋅pcv∣2Jcv(ℏω)ϵ0m02ω2d\epsilon_2(\omega)=\frac{\pi e^2|\hat{\mathbf{e}}\cdot\mathbf{p}_{cv}|^2J_{cv}(\hbar\omega)}{\epsilon_0m_0^2\omega^2d}, where e^\hat{\mathbf{e}} is the polarization vector, pcv\mathbf{p}_{cv} is the interband momentum matrix element, m0m_0 is the free electron mass, and JcvJ_{cv} is the joint density of states per unit area and per unit energy. Knowing ϵ2(ω)\epsilon_2(\omega), the absorption coefficient can be obtained as ωϵ2(ω)/(n0c)\omega\epsilon_2(\omega)/(n_0c), where n0n_0 is the background refractive index. The absorbance A(ω)A(\omega) is then given by

A(ω)=πe2ϵ0n0cm02ω2∣e^⋅pcv∣2Jcv(ℏω).(1)A(\omega)=\frac{\pi e^2}{\epsilon_0 n_0 c m_0^2 \omega^2}\left|\hat{\mathbf{e}}\cdot\mathbf{p}_{cv}\right|^2 J_{cv}(\hbar\omega). \tag*{(1)}

We employ the constant-matrix-element approximation and neglect the momentum dependence of pcv\mathbf{p}_{cv}. As can be seen from Eq. (1), the entire dependence of the optical absorption on the SOCs and external fields is therefore contained in JcvJ_{cv}. For noninteracting electrons and holes, the joint density of states can be written as

Jcv(ℏω)=∑ν∣Ψν(r=0)∣2δ(ℏω−Eg−Eν).(2)J_{cv}(\hbar\omega)=\sum_{\nu}\left|\Psi_{\nu}(\mathbf{r}=0)\right|^2\delta\left(\hbar\omega-E_g-E_{\nu}\right). \tag*{(2)}

where EgE_g is the band gap, while EνE_{\nu} and Ψν\Psi_{\nu} are the eigenvalues and normalized eigenfunctions of the electron–hole relative motion, with r=re−rh\mathbf{r}=\mathbf{r}_e-\mathbf{r}_h denoting the relative coordinate. To determine these quantities, we describe the relative motion within the effective-mass approximation by the Hamiltonian

H=Πx2+Πy22m∗−eEx+g2μBBσz+αR[σxΠy−σyΠx]+βD[σxΠx−σyΠy].(3)H=\frac{\Pi_x^2+\Pi_y^2}{2m^*}-e\mathcal{E}x+\frac{g}{2}\mu_B B\sigma_z+\alpha_R\left[\sigma_x\Pi_y-\sigma_y\Pi_x\right]+\beta_D\left[\sigma_x\Pi_x-\sigma_y\Pi_y\right]. \tag*{(3)}

where Π=−iℏ∇−eA\boldsymbol{\Pi}=-i\hbar\nabla-e\mathbf{A}, 1/m∗=1/me∗+1/mh∗1/m^*=1/m_e^*+1/m_h^* defines the reduced effective mass of the electron–hole pair, gg is the effective Landé factor, μB\mu_B is the Bohr magneton, σi\sigma_i are the Pauli matrices, and αR\alpha_R and βD\beta_D are the effective Rashba and Dresselhaus coupling constants, respectively. We take e=−∣e∣e=-|e| throughout. In Eq. (3) the Zeeman and spin-orbit terms act on the relative motion with effective constants. This description reduces to the standard Landau-level and Franz–Keldysh descriptions of interband absorption when the SOC is switched off, and it is exact in the absence of fields when only one of the two bands carries a linear spin splitting.

For the considered field configuration we use the Landau gauge A=B(0,x,0)\mathbf{A}=B(0,x,0). In this gauge, translational invariance along yy is preserved and kyk_y remains a good quantum number. The general state index ν\nu in Eq. (2) can therefore be represented by the level index nn and the conserved wave vector kyk_y. For the Hamiltonian given in Eq. (3), the stationary Schrödinger equation is written as HΨnky(x,y)=EnkyΨnky(x,y)H\Psi_{n k_y}(x,y)=E_{n k_y}\Psi_{n k_y}(x,y), and we write the spinor wave function as Ψnky(x,y)=Ly−1/2eikyyψnky(x)\Psi_{n k_y}(x,y)=L_y^{-1/2}e^{i k_y y}\psi_{n k_y}(x), where LyL_y is the length of the crystal along yy.

Introducing the dimensionless coordinate z=x/aBz=x/a_B, where aB=ℏ/(m∗ωB)a_B=\sqrt{\hbar/(m^*\omega_B)} is the magnetic length and ωB=∣e∣B/m∗\omega_B=|e|B/m^* is the cyclotron frequency, and measuring energies in units of ℏωB/2\hbar\omega_B/2, the Schrödinger equation becomes

{d2dz2−(z−z0)2−2E~z−σzμZ−α[σx(z−z0)+iσyddz]+iβ[σxddz−iσy(z−z0)]+E~nky}ψnky(z)=0,(4)\left\{\frac{d^2}{dz^2}-(z-z_0)^2-2\widetilde{E}z-\sigma_z\mu_Z-\alpha\left[\sigma_x(z-z_0)+i\sigma_y\frac{d}{dz}\right]+i\beta\left[\sigma_x\frac{d}{dz}-i\sigma_y(z-z_0)\right]+\widetilde{E}_{n k_y}\right\}\psi_{n k_y}(z)=0, \tag*{(4)}

where the dimensionless couplings, fields and energies are

α=4m∗αR2ℏωB,β=4m∗βD2ℏωB,E~=∣e∣EaBℏωB,μZ=gμBBℏωB,E~nky=EnkyℏωB/2,z0=−kyaB.(5)\alpha=\sqrt{\frac{4m^*\alpha_R^2}{\hbar\omega_B}},\quad\beta=\sqrt{\frac{4m^*\beta_D^2}{\hbar\omega_B}},\quad\widetilde{E}=\frac{|e|\mathcal{E}a_B}{\hbar\omega_B},\quad\mu_Z=\frac{g\mu_BB}{\hbar\omega_B},\quad\widetilde{E}_{n k_y}=\frac{E_{n k_y}}{\hbar\omega_B/2},\quad z_0=-k_ya_B. \tag*{(5)}

Here z0z_0 is the dimensionless coordinate of the center of the cyclotron orbit, and E~=m∗vdaB/ℏ\widetilde{E}=m^*v_da_B/\hbar is the drift momentum m∗vdm^*v_d, with vd=E/Bv_d=\mathcal{E}/B the Hall drift velocity, in units of ℏ/aB\hbar/a_B. To write Eq. (4) in Hermite form for the spinor components ψnky=(ψn↑,ψn↓)T\psi_{n k_y}=(\psi_{n\uparrow},\psi_{n\downarrow})^T, we introduce η=z−z0+E~\eta=z-z_0+\widetilde{E}, measured from the displaced orbit center, and factor out the Gaussian by writing ψn↑,n↓(z)=e−η2/2φn↑,n↓(η)\psi_{n\uparrow,n\downarrow}(z)=e^{-\eta^2/2}\varphi_{n\uparrow,n\downarrow}(\eta). We then obtain

{d2dη2−2ηddη+En−μZ}φn↑(η)−{α(ddη−E~)−iβ(ddη−2η+E~)}φn↓(η)=0,\left\{\frac{d^2}{d\eta^2}-2\eta\frac{d}{d\eta}+E_n-\mu_Z\right\}\varphi_{n\uparrow}(\eta)-\left\{\alpha\left(\frac{d}{d\eta}-\widetilde{E}\right)-i\beta\left(\frac{d}{d\eta}-2\eta+\widetilde{E}\right)\right\}\varphi_{n\downarrow}(\eta)=0,
{d2dη2−2ηddη+En+μZ}φn↓(η)+{α(ddη−2η+E~)+iβ(ddη−E~)}φn↑(η)=0.(6)\left\{\frac{d^2}{d\eta^2}-2\eta\frac{d}{d\eta}+E_n+\mu_Z\right\}\varphi_{n\downarrow}(\eta)+\left\{\alpha\left(\frac{d}{d\eta}-2\eta+\widetilde{E}\right)+i\beta\left(\frac{d}{d\eta}-\widetilde{E}\right)\right\}\varphi_{n\uparrow}(\eta)=0. \tag*{(6)}

where the reduced energy is En=E~nky+E~(E~−2z0)−1E_n=\widetilde{E}_{n k_y}+\widetilde{E}(\widetilde{E}-2z_0)-1. Eqs. (6) no longer contain z0z_0, so that the reduced energy EnE_n is independent of kyk_y and the entire kyk_y dependence of the spectrum is contained in the definition of EnkyE_{n k_y}, i.e.,

Enky=ℏωB2(En+1)−m∗vd22+∣e∣Exc,(7)E_{n k_y}=\frac{\hbar\omega_B}{2}(E_n+1)-\frac{m^*v_d^2}{2}+|e|\mathcal{E}x_c, \tag*{(7)}

where xc=−kyaB2x_c=-k_ya_B^2 is the center of the cyclotron orbit in the absence of the electric field. Every level is rigidly shifted by the electric potential at the orbit center and lowered by the drift kinetic energy, exactly as in the spin-degenerate crossed-field problem.

Eqs. (6) are solved by expanding the spinor components in normalized Hermite functions hj(η)=Hj(η)/(π2jj!)1/2h_j(\eta)=H_j(\eta)/(\sqrt{\pi}2^j j!)^{1/2},

φn↑(η)=∑j=0∞aj(n)hj(η),φn↓(η)=∑j=0∞bj(n)hj(η),(8)\varphi_{n\uparrow}(\eta)=\sum_{j=0}^{\infty}a_j^{(n)}h_j(\eta),\qquad\varphi_{n\downarrow}(\eta)=\sum_{j=0}^{\infty}b_j^{(n)}h_j(\eta), \tag*{(8)}

so that the full wave function is ψnky(x)=aB−1/2∑j(aj(n),bj(n))Tϕj(η)\psi_{n k_y}(x)=a_B^{-1/2}\sum_j(a_j^{(n)},b_j^{(n)})^T\phi_j(\eta), with the oscillator functions ϕj=e−η2/2hj\phi_j=e^{-\eta^2/2}h_j and the normalization ∑j(∣aj(n)∣2+∣bj(n)∣2)=1\sum_j\left(|a_j^{(n)}|^2+|b_j^{(n)}|^2\right)=1. Using Hj′′−2ηHj′+2jHj=0H_j''-2\eta H_j'+2jH_j=0, Hj′=2jHj−1H_j'=2jH_{j-1} and Hj′−2ηHj=−Hj+1H_j'-2\eta H_j=-H_{j+1}, the projection of Eqs. (6) onto hjh_j yields the three-term recursion

Enaj(n)=(2j+μZ)aj(n)−(α+iβ)E~bj(n)+α2(j+1)bj+1(n)+iβ2jbj−1(n),E_na_j^{(n)}=(2j+\mu_Z)a_j^{(n)}-(\alpha+i\beta)\widetilde{E}b_j^{(n)}+\alpha\sqrt{2(j+1)}b_{j+1}^{(n)}+i\beta\sqrt{2j}b_{j-1}^{(n)},
Enbj(n)=(2j−μZ)bj(n)−(α−iβ)E~aj(n)−iβ2(j+1)aj+1(n)+α2jaj−1(n).(9)E_nb_j^{(n)}=(2j-\mu_Z)b_j^{(n)}-(\alpha-i\beta)\widetilde{E}a_j^{(n)}-i\beta\sqrt{2(j+1)}a_{j+1}^{(n)}+\alpha\sqrt{2j}a_{j-1}^{(n)}. \tag*{(9)}

Thus, the field and SOC dependence of the electroabsorption is determined by the eigenvalues EnE_n and coefficients aj(n)a_j^{(n)} and bj(n)b_j^{(n)} obtained from Eqs. (9), which determine the energies EnkyE_{n k_y} and wave functions Ψnky\Psi_{n k_y} entering Eq. (2). Since the couplings in Eqs. (9) connect only neighboring Hermite indices, the coefficient matrix is a Hermitian block-tridiagonal matrix in which EnE_n appears only on the diagonal, so that the reduced energies and the expansion coefficients follow from a single diagonalization of the truncated matrix rather than from the roots of a secular determinant. Because the truncation error decreases exponentially with the number NN of retained Hermite functions, the lowest levels are obtained to machine precision already for NN of the order of one hundred, and the closed-form limits of the problem are reproduced to the same accuracy. In this sense the solution is exact.

Results

We now present the results that follow from Eqs. (9) and (2). The spin-orbit coupled energy spectrum in crossed fields is discussed first, together with its limiting cases, because it fixes the positions and the shapes of all absorption lines. The case of a vanishing magnetic field, in which the Hermite expansion is no longer applicable, is treated separately. The joint density of states is then obtained for different Rashba and Dresselhaus couplings in the four field configurations, namely without fields, in a pure electric field, in a pure magnetic field, and in crossed fields.

Spin-orbit coupled energy spectrum in crossed fields

It is useful first to emphasize that Eqs. (9) can be written as the eigenvalue problem Hϕn=Enϕn\mathsf{H}\boldsymbol{\phi}_n=E_n\boldsymbol{\phi}_n, where H=2b^†b^+μZσz+2α(b^σ++b^†σ−)+i2β(b^†σ+−b^σ−)−E~(ασx−βσy)\mathsf{H}=2\hat{b}^{\dagger}\hat{b}+\mu_Z\sigma_z+\sqrt{2}\alpha(\hat{b}\sigma_+ + \hat{b}^{\dagger}\sigma_-)+i\sqrt{2}\beta(\hat{b}^{\dagger}\sigma_+ - \hat{b}\sigma_-)-\tilde{\mathcal{E}}(\alpha\sigma_x-\beta\sigma_y), b^\hat{b} and b^†\hat{b}^{\dagger} are the ladder operators of the oscillator functions φj\varphi_j, with b^φj=jφj−1\hat{b}\varphi_j=\sqrt{j}\varphi_{j-1} and b^†φj=j+1φj+1\hat{b}^{\dagger}\varphi_j=\sqrt{j+1}\varphi_{j+1}, while σ±=(σx±iσy)/2\sigma_{\pm}=(\sigma_x\pm i\sigma_y)/2 are the spin raising and lowering operators. As can be seen from the expression for H\mathsf{H}, the first four terms describe the Landau problem with Rashba and Dresselhaus couplings, in which the Rashba term connects the states ∣j,↑⟩\lvert j,\uparrow\rangle and ∣j+1,↓⟩\lvert j+1,\downarrow\rangle, while the Dresselhaus term connects ∣j,↑⟩\lvert j,\uparrow\rangle and ∣j−1,↓⟩\lvert j-1,\downarrow\rangle. The electric field enters only through the last term. This term has the form of a Zeeman coupling to an effective magnetic field lying in the plane and does not change the oscillator index. Restoring dimensions through (ℏωB/2)E~α=m∗vdαR(\hbar\omega_B/2)\tilde{\mathcal{E}}\alpha=m^*v_d\alpha_R, this term can be written as −m∗vd(αRσx−βDσy)-m^*v_d(\alpha_R\sigma_x-\beta_D\sigma_y). From this expression, it is clearly seen that the orbit center of an electron in crossed fields drifts with the Hall velocity vdv_d along −y-y, so that the electron carries the drift momentum m∗vdm^*v_d, and the spin orbit couplings convert this momentum into an effective in plane field, in the same way as a free electron with momentum ℏk\hbar\mathbf{k} experiences the Rashba field ℏαR(k×z^)\hbar\alpha_R(\mathbf{k}\times\hat{\mathbf{z}}). Apart from the classical drift terms of Eq. (7), which are the same for all levels, this drift field is the only way in which the in-plane electric field acts on the spin of the electron. Thus, for an infinite crystal the nontrivial electric field dependence of the spin-orbit-coupled Landau spectrum is generated entirely by the drift field, which mixes neighboring Landau states, modifies the level structure, and displaces and deforms the spinor densities ∣χn(η)∣2=e−η2(∣ϕn↑(η)∣2+∣ϕn↓(η)∣2)|\chi_n(\eta)|^2=e^{-\eta^2}(|\phi_{n\uparrow}(\eta)|^2+|\phi_{n\downarrow}(\eta)|^2), with ∫∣χn(η)∣2dη=1\int|\chi_n(\eta)|^2d\eta=1, which, as shown in § 3.6, determine directly the electroabsorption line shapes.

We next consider several limiting cases in which H\mathsf{H} can be diagonalized in closed form. We first examine the case of pure Rashba coupling without an electric field, β=E~=0\beta=\tilde{\mathcal{E}}=0. In this limit, the recursion decouples into 2×22\times2 blocks formed by aja_j and bj+1b_{j+1}, together with the isolated coefficient b0b_0, and we obtain

Ej±=ℏωB(j+1)±(ℏωB−qμBB2)2+4ESOℏωB(j+1),(10)E_j^{\pm}=\hbar\omega_B(j+1)\pm\sqrt{\left(\frac{\hbar\omega_B-q\mu_B B}{2}\right)^2+4E_{\mathrm{SO}}\hbar\omega_B(j+1)}, \tag*{(10)}

together with the isolated level Eiso=(ℏωB−gμBB)/2E_{\mathrm{iso}}=(\hbar\omega_B-g\mu_B B)/2, where ESO=m∗αR2/2E_{\mathrm{SO}}=m^*\alpha_R^2/2 is the Rashba energy recalled in Figure 1(c). This is the Bychkov–Rashba spectrum [46]. For pure Dresselhaus coupling, α=E~=0\alpha=\tilde{\mathcal{E}}=0, the blocks are instead formed by aj+1a_{j+1} and bjb_j, while a0a_0 remains isolated. The spectrum is obtained from the Rashba result by the replacements α→β\alpha\to\beta and μZ→−μZ\mu_Z\to-\mu_Z, with the isolated level Eiso=(ℏωB+gμBB)/2E_{\mathrm{iso}}=(\hbar\omega_B+g\mu_B B)/2. The Rashba and Dresselhaus spectra are therefore equivalent only for μZ=0\mu_Z=0. In the absence of SOC, the reduced spectrum becomes Ej=2j±μZE_j=2j\pm\mu_Z for any electric field. Using Eq. (7), the corresponding crossed-field Landau levels are

Enky=ℏωB(j+12)±gμBB2−m∗vd22−∣e∣EkylB2,(11)E_{n k_y}=\hbar\omega_B\left(j+\frac{1}{2}\right)\pm\frac{g\mu_B B}{2}-\frac{m^*v_d^2}{2}-|e|\mathcal{E}k_y l_B^2, \tag*{(11)}

which is the result for the spin-degenerate crossed field problem.

It is important to note here that, for a weak electric field the drift field −m∗vd(αRσx−βDσy)-m^*v_d(\alpha_R\sigma_x-\beta_D\sigma_y) can be treated as a perturbation. The field-free part of H\mathsf{H} changes the oscillator index by one whenever it flips the spin, so that each field-free eigenstate contains spin up components of one parity of jj and spin down components of the opposite parity, whereas σx\sigma_x and σy\sigma_y connect components with the same jj. The drift field therefore has no diagonal matrix elements in the field-free eigenbasis, and every nondegenerate level shifts quadratically, δEn∼E~2(α2+β2)\delta E_n\sim\tilde{\mathcal{E}}^2(\alpha^2+\beta^2), i.e., δEn∼(m∗vd)2(αR2+βD2)/ℏωB\delta E_n\sim(m^*v_d)^2(\alpha_R^2+\beta_D^2)/\hbar\omega_B in physical units, with coefficients that are of opposite sign for the two members of a spin-orbit doublet in all cases studied below. Levels that are degenerate or nearly degenerate at E~=0\tilde{\mathcal{E}}=0 are instead split in first order by the matrix element of the drift field between them. This is a spin-orbit induced quadratic Stark effect of the Landau levels. It is absent without spin-orbit coupling, where Eq. (11) contains no level dependent field term.

Having established these analytical limits, we examine the full spectrum from Eqs. (9). Figure 2 shows the numerical results for N=80N=80 Hermite functions, for which the twelve lowest levels are converged to better than 10−1010^{-10}. For the spectra shown as a function of α\alpha, we choose E~=0.4\tilde{\mathcal{E}}=0.4 and

Image not extracted

Figure 2. Spin-orbit coupled energy spectrum in crossed fields. (a) Comparison of the finite and zero electric field spectra for the pure Rashba case. (b) Spectrum in the presence of Dresselhaus and Zeeman terms. (c) Comparison of the effects of the Dresselhaus and Zeeman couplings near the level anticrossings. (d) Dependence of the spectrum on the electric field.

z0=0.2z_0 = 0.2, which satisfy E~2=2E~z0\widetilde{\mathcal{E}}^2 = 2\widetilde{\mathcal{E}}z_0. With this choice, E~=En+1\widetilde{E} = E_n + 1 and the classical drift terms −m∗vd2/2+∣e∣Exc-m^*v_d^2/2 + |e|\mathcal{E}x_c in Eq. (7) cancel.

Figure 2(a) compares, for β=μZ=0\beta= \mu_Z = 0, the finite electric field spectrum obtained by direct diagonalization of Eqs. (9) with the zero field Bychkov–Rashba spectrum of Eq. (10). The finite field results are numerically exact within the convergence accuracy of the Hermite basis. For weak Rashba coupling, the two spectra remain close, but the drift field −m∗vd(αRσx−βDσy)-m^*v_d(\alpha_R\sigma_x-\beta_D\sigma_y) removes the simple zero-field structure. In particular, the isolated level EisoE_{\mathrm{iso}}, which is independent of α\alpha at E~=0\widetilde{\mathcal{E}} = 0, is shifted upward, while the lower state of the first doublet is shifted downward. Since the Rashba levels are nearly spin degenerate at small α\alpha, the drift field produces a first-order splitting of about ±αE~\pm\alpha\widetilde{\mathcal{E}}, or ±m∗vdαR\pm m^*v_d\alpha_R in physical units. When the Rashba splitting becomes larger than the drift energy αE~\alpha\widetilde{\mathcal{E}}, the field-induced shifts become predominantly quadratic. A more pronounced effect appears near the crossings of the zero-field Bychkov–Rashba spectrum. For example, the lower state of the doublet j+2j+2 crosses the upper state of the doublet jj, with the first crossing occurring at α=0.888\alpha= 0.888 for j=0j = 0. Since the two states differ by two oscillator quanta, they are not coupled directly by the drift field. Instead, one state first couples to a state of a neighboring doublet, and this intermediate state then couples to the second crossing state. Since each coupling is proportional to E~\widetilde{\mathcal{E}}, the resulting interaction between the two crossing states is proportional to E~2\widetilde{\mathcal{E}}^2. As a result, each crossing is replaced by a narrow anticrossing. These gaps are small on the overall energy scale of Figure 2(a), but their systematic field dependence provides a clear signature of the drift field. For E~=0.4\widetilde{\mathcal{E}} = 0.4, the three lowest anticrossing gaps are 0.0800.080, 0.0520.052 and 0.0390.039. Since E~2=0.16\widetilde{\mathcal{E}}^2 = 0.16, these values correspond to Δgap/E~2≃0.50\Delta_{\mathrm{gap}}/\widetilde{\mathcal{E}}^2 \simeq0.50, 0.330.33 and 0.240.24, respectively. We repeated the calculation for weaker electric fields down to E~=0.05\widetilde{\mathcal{E}} = 0.05 and found that these ratios remain nearly constant. This confirms that the anticrossing gaps scale quadratically with the electric field, Δgap∝E~2\Delta_{\mathrm{gap}} \propto\widetilde{\mathcal{E}}^2.

The Dresselhaus coupling changes the spectrum more strongly. In contrast to the pure Rashba case, it directly mixes different Landau doublets already at zero electric field, so that the level crossings are replaced by much wider anticrossings. This behavior is clearly seen in Figure 2(b) for β=0.6\beta= 0.6 and μZ=0.5\mu_Z = 0.5, where the simple Bychkov–Rashba structure is strongly modified. The electric field then acts on these already mixed states through the drift field. This Landau level mixing produced by the simultaneous presence of Rashba and Dresselhaus couplings is consistent with the zero-field results of [47, 48] and remains present at finite electric field.

Figure 2(c) separates the effects of the Zeeman and Dresselhaus terms in the energy window 4.5<E~<94.5 < \widetilde{E} < 9, where levels belonging to different doublets approach each other. For β=0\beta= 0 and μZ=0.5\mu_Z = 0.5, the Zeeman term changes mainly the positions of the avoided crossings, while the gaps themselves are still opened only by the drift field and remain small, with minimum separations between about 0.040.04 and 0.230.23. When the Dresselhaus coupling is included with β=0.6\beta= 0.6, it directly mixes the different doublets and the corresponding gaps increase strongly to values between about 0.50.5 and 1.21.2. Thus, the Dresselhaus coupling changes the anticrossing scale from small drift field induced gaps to gaps of order unity, while the Zeeman term mainly shifts their positions. The anticrossing width therefore provides a direct measure of the Dresselhaus coupling.

The dependence on the electric field is shown in Figure 2(d) for α=0.8\alpha= 0.8, β=0.3\beta= 0.3 and μZ=0.3\mu_Z = 0.3. At E~=0\widetilde{\mathcal{E}} = 0, the two lowest levels are nearly degenerate, with En+1=0.61E_n + 1 = 0.61 and 0.650.65. Since the drift field couples these two states directly, their energies separate linearly as the electric field is turned on. In contrast, the higher levels are not directly coupled within each doublet and therefore show mainly quadratic shifts, with the two members bending in opposite directions. As E~\widetilde{\mathcal{E}} increases, the drift field energy E~α2+β2\widetilde{\mathcal{E}}\sqrt{\alpha^2+\beta^2} becomes comparable to the separation between neighboring levels. The mixing between these levels then becomes stronger and produces a sequence of avoided crossings. At still larger fields, the lowest level is governed mainly by the drift field and decreases approximately linearly with E~\widetilde{\mathcal{E}}, approaching the asymptotic behavior −E~α2+β2-\widetilde{\mathcal{E}}\sqrt{\alpha^2+\beta^2}.

Finally, we examine the effect of the electric field on the spinor densities ∣χn(η)∣2=e−η2(∣φn↑(η)∣2+∣φn↓(η)∣2)|\chi_n(\eta)|^2 = e^{-\eta^2}(|\varphi_{n\uparrow}(\eta)|^2 + |\varphi_{n\downarrow}(\eta)|^2). The corresponding densities for the four lowest states are shown in Figure 3. At E~=0\widetilde{\mathcal{E}} = 0, the densities are symmetric about η=0\eta= 0 because the spin components contain oscillator functions with definite parity. A finite electric field introduces the drift field, which breaks this symmetry and changes both the position and the shape of the densities. For E~=1.2\widetilde{\mathcal{E}} = 1.2, the centers of the four lowest states are shifted to ⟨η⟩n=−0.42\langle\eta\rangle_n = -0.42, −0.06-0.06, 0.070.07 and −0.14-0.14 for n=0,…,3n = 0,\ldots,3. Thus, the electric field modifies not only the energy spectrum but also the spatial structure of the spinor states, which, as shown in § 3.6, is directly reflected in the electroabsorption line shapes.

Image not extracted

Figure 3. Spinor densities ∣χn(η)∣2|\chi_n(\eta)|^2 of the four lowest states for E~=0\widetilde{E}=0 and 1.21.2, with α=0.8\alpha=0.8, β=0.3\beta=0.3 and μZ=0.3\mu_Z=0.3. The curves for different nn are vertically shifted for clarity.

Spin dynamics in a pure electric field

We now consider the case B=0B=0, where the Landau levels disappear, and the electric field drives the electron continuously along the kxk_x direction according to ℏ dkx/dt=−∣e∣E\hbar\,dk_x/dt=-|e|E. Since kyk_y remains conserved, the changing kxk_x also changes the Rashba-Dresselhaus spin-orbit field. The main question is therefore whether the spin follows this changing field or remains approximately fixed as kxk_x varies.

For B=0B=0, the Hamiltonian becomes

H=p22m∗+∣e∣Ex+ℏ[(αRky+βDkx)σx−(αRkx+βDky)σy].(12)H=\frac{p^2}{2m^*}+|e|Ex+\hbar\left[(\alpha_R k_y+\beta_D k_x)\sigma_x-(\alpha_R k_x+\beta_D k_y)\sigma_y\right]. \tag*{(12)}

Since the electric potential is linear in xx, it is convenient to work in the kxk_x representation, where x→i∂/∂kxx\to i\partial/\partial k_x. The Schrödinger equation then takes the form

i∣e∣E∂ϕ∂kx=[ε−ℏ2(kx2+ky2)2m∗−HSO]ϕ.(13)i|e|E\frac{\partial\phi}{\partial k_x}=\left[\varepsilon-\frac{\hbar^2(k_x^2+k_y^2)}{2m^*}-H_{\mathrm{SO}}\right]\phi. \tag*{(13)}

To describe the spin part, we introduce g=αR2+βD2g=\sqrt{\alpha_R^2+\beta_D^2} and choose the spin quantization axis along n1=(βD,−αR)/g\mathbf{n}_1=(\beta_D,-\alpha_R)/g. In this basis the spin-orbit interaction becomes

HSO=ℏ(gkx+2αRβDgky)τz+ℏky∣αR2−βD2∣gτx.(14)H_{\mathrm{SO}}=\hbar\left(gk_x+\frac{2\alpha_R\beta_D}{g}k_y\right)\tau_z+\hbar k_y\frac{|\alpha_R^2-\beta_D^2|}{g}\tau_x. \tag*{(14)}

The first term changes continuously with kxk_x, while the second term couples the two spin states. Since the kinetic energy acts equally on both spin components, it only produces a common phase and does not affect the spin dynamics. This phase can therefore be removed. The remaining spinor χ(kx)\chi(k_x) obeys

i∣e∣Edχdkx=−[Λ(kx)τz+Γτx]χ,(15)i|e|E\frac{d\chi}{dk_x}=-\left[\Lambda(k_x)\tau_z+\Gamma\tau_x\right]\chi, \tag*{(15)}

where Λ(kx)=ℏ(gkx+2αRβDky/g)\Lambda(k_x)=\hbar(gk_x+2\alpha_R\beta_Dk_y/g), Γ=ℏky∣αR2−βD2∣/g\Gamma=\hbar k_y|\alpha_R^2-\beta_D^2|/g. Eq. (15) is exactly the Landau–Zener problem. Here the electric field continuously changes kxk_x, while Γ\Gamma determines the coupling between the two spin states.

The probability that the spin remains in the same τz\tau_z state during the electric field driven passage through the crossing is

Pdia=e−πδLZ,δLZ=ℏky2(αR2−βD2)2∣e∣E(αR2+βD2)3/2.(16)P_{\mathrm{dia}}=e^{-\pi\delta_{\mathrm{LZ}}},\qquad\delta_{\mathrm{LZ}}=\frac{\hbar k_y^2(\alpha_R^2-\beta_D^2)^2}{|e|E(\alpha_R^2+\beta_D^2)^{3/2}}. \tag*{(16)}

The parameter δLZ\delta_{\mathrm{LZ}} gives a direct criterion for the spin response to the electric field. For δLZ≪1\delta_{\mathrm{LZ}} \ll1, which occurs for a strong electric field, small kyk_y, or αR≃βD\alpha_R \simeq\beta_D, the electric field drives kxk_x so rapidly that the spin cannot follow the changing spin orbit field. The spin therefore remains approximately fixed and the dynamics is diabatic. In contrast, for δLZ≫1\delta_{\mathrm{LZ}} \gg1, kxk_x changes slowly enough that the spin continuously follows the local spin-orbit field, corresponding to the adiabatic regime.

This result shows that the electric field controls a crossover between frozen spin dynamics and adiabatic spin following. The crossover is governed by the combined effect of E\mathcal{E}, kyk_y, and the relative strengths of the Rashba and Dresselhaus couplings.

It is also useful to express this result in the natural scales of the electric field problem. Introducing ℓ=(ℏ2/2m∗∣e∣E)1/3\ell= (\hbar^2/2m^*|e|\mathcal{E})^{1/3} and Θ=∣e∣Eℓ\Theta= |e|\mathcal{E}\ell, together with a=ℏαR/(ℓΘ)a = \hbar\alpha_R/(\ell\Theta) and b=ℏβD/(ℓΘ)b = \hbar\beta_D/(\ell\Theta), we obtain

δLZ=(kyℓ)2(a2−b2)2(a2+b2)3/2.(17)\delta_{\mathrm{LZ}} = (k_y\ell)^2 \frac{(a^2-b^2)^2}{(a^2+b^2)^{3/2}}. \tag*{(17)}

Since Θ∝E2/3\Theta\propto\mathcal{E}^{2/3}, increasing the electric field drives the system toward the diabatic regime, while decreasing it leads to the adiabatic regime.

A particularly simple limit appears for αR=βD\alpha_R = \beta_D. In this case Γ=0\Gamma= 0 for every kyk_y, so the two spin projections decouple exactly, and the spin does not rotate during the motion. The problem then reduces to two independent Airy equations. Explicitly, the spin-orbit term becomes 2ℏα(kx+ky)τ\sqrt{2}\hbar\alpha(k_x+k_y)\tau with τ=(σx−σy)/2\tau= (\sigma_x-\sigma_y)/\sqrt{2}, and the two projections τ=±1\tau= \pm1 have the dispersions ετ(k)=ℏ2[(kx+τk0)2+(ky+τk0)2]/2m∗−2m∗α2\varepsilon_\tau(\mathbf{k}) = \hbar^2[(k_x+\tau k_0)^2+(k_y+\tau k_0)^2]/2m^* - 2m^*\alpha^2 with k0=2m∗α/ℏk_0 = \sqrt{2}m^*\alpha/\hbar. The displacement along kxk_x is a gauge transformation e−iτk0xe^{-i\tau k_0x} that leaves the probability density unchanged, so that the eigenfunctions are the Airy functions of the spin-degenerate problem with the energy shifted by −2m∗α2+ℏ2(ky+τk0)2/2m∗-2m^*\alpha^2+\hbar^2(k_y+\tau k_0)^2/2m^*. This is the persistent spin helix point, where the spin-orbit field has a fixed direction and the electric field problem becomes exactly solvable.

Finally, a linear electric potential in an infinite crystal does not produce bound states. The spectrum therefore remains continuous, and the main result of this subsection is the electric field-controlled spin dynamics rather than a discrete energy spectrum.

Spin-orbit induced subgap absorption

From this point onward, we focus on the optical response of the model. For this purpose, we first consider the case in which both the electric and magnetic fields are absent, where the eigenfunctions are plane waves and ∣Ψν(0)∣2=1/A|\Psi_\nu(0)|^2 = 1/A, with AA denoting the area of the crystal. Eq. (2) then reduces to Jcv(ℏω)=1A∑k,sδ(Ecvs(k)−ℏω)J_{\mathrm{cv}}(\hbar\omega) = \frac{1}{A}\sum_{\mathbf{k},s}\delta(E_{\mathrm{cv}}^s(\mathbf{k})-\hbar\omega), where s=±s=\pm labels the two spin-split branches, with s=+s=+ denoting the upper branch and s=−s=- the lower branch, as illustrated in Figure 1(c) for the pure Rashba case. Their dispersions are

Ecvs(k,θ)=Eg+ℏ2k22m∗+sℏkg(θ),(18)E_{\mathrm{cv}}^s(k,\theta) = E_g + \frac{\hbar^2 k^2}{2m^*} + s\hbar k g(\theta), \tag*{(18)}

where g(θ)=αR2+βD2+2αRβDsin⁡2θg(\theta) = \sqrt{\alpha_R^2+\beta_D^2+2\alpha_R\beta_D\sin2\theta}, and θ\theta is the polar angle of k\mathbf{k}. It is important to emphasize here that, for s=+s=+, the spin-orbit term increases the energy with kk, so the minimum remains at k=0k=0 and no additional minima or saddle points appear. The situation is different for the lower branch s=−s=-, where the spin-orbit term lowers the energy and produces minima at finite kk. To locate these minima, it is sufficient to consider the dispersion for a fixed direction θ\theta, Ecv−(k,θ)=Eg+ℏ2k2/(2m∗)−ℏkg(θ)E_{\mathrm{cv}}^-(k,\theta)=E_g+\hbar^2k^2/(2m^*)-\hbar kg(\theta). Its minimum occurs at k=m∗g(θ)/ℏk=m^*g(\theta)/\hbar, with the corresponding energy measured from EgE_g is −m∗g2(θ)/2-m^*g^2(\theta)/2. As can be seen, the angular dependence of the band minimum is therefore determined entirely by g(θ)g(\theta). This function reaches its largest value αR+βD\alpha_R+\beta_D at θ=π/4\theta=\pi/4 and 5π/45\pi/4. These directions therefore correspond to the deepest points of the lower branch, with k=m∗(αR+βD)/ℏk=m^*(\alpha_R+\beta_D)/\hbar and energy −m∗(αR+βD)2/2-m^*(\alpha_R+\beta_D)^2/2. In contrast, g(θ)g(\theta) reaches its smallest value ∣αR−βD∣|\alpha_R-\beta_D| at θ=3π/4\theta=3\pi/4 and 7π/47\pi/4. For each of these directions, the energy is still minimized with respect to kk. However, since g(θ)g(\theta) is smallest there, the corresponding minimum energy −m∗g2(θ)/2-m^*g^2(\theta)/2 is higher than for slightly different values of θ\theta. Thus, the energy is a minimum along the radial direction kk but a maximum along the angular direction θ\theta. These points are therefore saddle points, with k=m∗∣αR−βD∣/ℏk=m^*|\alpha_R-\beta_D|/\hbar and energy −m∗(αR−βD)2/2-m^*(\alpha_R-\beta_D)^2/2. We also note that, for pure Rashba coupling, βD=0\beta_D=0, the function g(θ)=αRg(\theta)=\alpha_R becomes independent of θ\theta. The minimum then has the same energy for every direction, so the isolated minima and saddle points merge into the continuous ring-shaped minimum at −ESO-E_{\mathrm{SO}} shown in Figure 1(c).

In light of these observations, we evaluate the JDOS from Eq. (18) by introducing Δ=ℏω−Eg\Delta=\hbar\omega-E_g. The sign of Δ\Delta determines which spin branches contribute to the JDOS. For Δ>0\Delta>0, both branches contribute, and each branch has one radial solution for a given θ\theta. For Δ<0\Delta<0, however, the upper branch does not contribute, while the lower branch has two radial solutions k±=m∗ℏ[g(θ)±g2(θ)+2Δm∗]k_{\pm} = \frac{m^{*}}{\hbar}\left[g(\theta) \pm\sqrt{g^{2}(\theta) + \frac{2\Delta}{m^{*}}}\right], provided that g2(θ)>−2Δ/m∗g^{2}(\theta) > -2\Delta/m^{*}. After the radial integration, the remaining angular integral gives

Jcvs(Δ>0)=m∗2πℏ2[1−s2π∫02πg(θ) dθg2(θ)+2Δ/m∗].(19)J_{cv}^{s}(\Delta> 0) = \frac{m^{*}}{2\pi\hbar^{2}}\left[1 - \frac{s}{2\pi}\int_{0}^{2\pi}\frac{g(\theta)\,\mathrm{d}\theta}{\sqrt{g^{2}(\theta)+2\Delta/m^{*}}}\right]. \tag*{(19)}

while below the gap

Jcv+(Δ<0)=0,(20)J_{cv}^{+}(\Delta< 0) = 0, \tag*{(20)}

and

Jcv−(Δ<0)=m∗2πℏ2∫g2>−2Δ/m∗g(θ) dθg2(θ)+2Δ/m∗.(21)J_{cv}^{-}(\Delta< 0) = \frac{m^{*}}{2\pi\hbar^{2}}\int_{g^{2}>-2\Delta/m^{*}}\frac{g(\theta)\,\mathrm{d}\theta}{\sqrt{g^{2}(\theta)+2\Delta/m^{*}}}. \tag*{(21)}

From Eq. (19), it can be seen directly that for Δ>0\Delta> 0 the total JDOS becomes Jcv=Jcv++Jcv−=m∗/(πℏ2)J_{cv}=J_{cv}^{+}+J_{cv}^{-}=m^{*}/(\pi\hbar^{2}). In other words, within the constant matrix element approximation, the Rashba and Dresselhaus couplings do not change the total optical absorption above EgE_g. This sum rule holds for arbitrary values of the two couplings and is the optical counterpart of the well-known property of the density of states of a two-dimensional Rashba gas, which is constant above the band crossing point [49]. The couplings only redistribute the weight between the two branches, which approach m∗/2πℏ2m^{*}/2\pi\hbar^{2} from opposite sides for Δ→∞\Delta\to\infty. We therefore continue with the case Δ<0\Delta< 0. Since only the lower branch s=−s=- contributes in this region, the lowest possible transition energy is determined by the absolute minimum of this branch, which is Eg−m∗(αR+βD)2/2E_g-m^{*}(\alpha_R+\beta_D)^2/2. The relevant sub-gap region can thus be written as

Eg−m∗(αR+βD)22<ℏω<Eg.E_g-\frac{m^{*}(\alpha_R+\beta_D)^2}{2}<\hbar\omega<E_g.

We first consider the pure Rashba case, βD=0\beta_D=0. In this limit, Eq. (21) can be evaluated analytically and gives

Jcv(Δ)=m∗πℏ2(1+ΔESO)−1/2.(22)J_{cv}(\Delta)=\frac{m^{*}}{\pi\hbar^{2}}\left(1+\frac{\Delta}{E_{\mathrm{SO}}}\right)^{-1/2}. \tag*{(22)}

The ring-shaped minimum of the lower Rashba branch therefore produces an inverse square root van Hove singularity at ℏω=Eg−ESO\hbar\omega=E_g-E_{\mathrm{SO}}, which appears in the optical response as a divergent subgap absorption edge. The situation changes qualitatively when the Dresselhaus coupling is added. The continuous Rashba ring minimum is then broken into two isolated minima and two saddle points. As a result, the inverse square root singularity is replaced by two separate features. A finite step appears at Δ=−m∗(αR+βD)2/2\Delta=-m^{*}(\alpha_R+\beta_D)^2/2, which corresponds to the absolute minima of the lower branch, while a logarithmic singularity appears at Δ=−m∗(αR−βD)2/2\Delta=-m^{*}(\alpha_R-\beta_D)^2/2, which corresponds to the saddle points.

These results are shown in Figure 4. Figure 4(a) presents the total JDOS for several values of βD/αR\beta_D/\alpha_R, while panel (b) shows the s=+s=+ separately and s=−s=- contributions and their sum for βD/αR=0.5\beta_D/\alpha_R=0.5. All curves in Figure 4(a) become identical above the gap, confirming that the total JDOS is independent of the spin-orbit coupling in this region. For pure Rashba coupling, the lower absorption edge contains the inverse square root singularity associated with the ring minimum, shown by the solid blue curve in Figure 4(a). For βD/αR=0.25\beta_D/\alpha_R=0.25 and 0.50.5, shown by the dashed red and dash-dotted green curves, respectively, this singularity is replaced by a finite step with heights 1.25 and 1.06 in units of m∗/πℏ2m^{*}/\pi\hbar^{2}, followed by a logarithmic peak at the saddle point energy. It is important to note that the height of this step can also be obtained analytically by expanding Eq. (21) near the band minimum, which gives

Image not extracted

Figure 4. Joint density of states at zero electric and magnetic fields. (a) Total JDOS for several values of βD/αR\beta_D/\alpha_R. (b) Contributions of the s=±s=\pm branches and their sum for βD/αR=0.5\beta_D/\alpha_R=0.5.

Jcv(Δ→−m∗(αR+βD)22+0)=m∗2πℏ2αR+βDαRβD.(23)J_{cv}\left(\Delta\to-\frac{m^{*}(\alpha_R+\beta_D)^2}{2}+0\right)=\frac{m^{*}}{2\pi\hbar^{2}}\frac{\alpha_R+\beta_D}{\sqrt{\alpha_R\beta_D}}. \tag*{(23)}

This expression gives the values 1.25 and 1.06 for βD/αR=0.25\beta_D/\alpha_R=0.25 and 0.5, respectively, in agreement with Figure 4(a). At βD=αR\beta_D=\alpha_R, the saddle point structure disappears and the JDOS reduces to the spin degenerate step Jcv=(m∗/πℏ2)ΘH(Δ+4ESO)J_{cv}=(m^{*}/\pi\hbar^{2})\Theta_H(\Delta+4E_{\mathrm{SO}}), shown by the dotted brown curve in Figure 4(a). This value also follows directly from the expression above, which gives the spin-degenerate step height m∗/πℏ2m^{*}/\pi\hbar^{2} at βD=αR\beta_D=\alpha_R. In the opposite limit βD→0\beta_D\to0, the step height diverges, continuously recovering the inverse square root singularity of the pure Rashba case in Eq. (22).

The subgap structure therefore contains direct information about both spin-orbit couplings. The position of the lower step determines (αR+βD)2(\alpha_R+\beta_D)^2, while the logarithmic peak determines (αR−βD)2(\alpha_R-\beta_D)^2. Measuring both energies allows αR\alpha_R and βD\beta_D, and therefore their ratio, to be extracted separately from a single optical absorption spectrum. This provides a direct optical method for determining the two spin-orbit couplings without applying any external electric or magnetic field.

Spin-orbit coupled Franz–Keldysh effect

We now examine how the optical response associated with the SOCs discussed in the previous subsection changes in the presence of an electric field. As emphasized in § 3.2, the motion along the xx direction is unbounded and the states form a continuum, while kyk_y remains conserved. Therefore, the JDOS can be written as

Jcv(Δ)=∫dky2πρky(0,Δ),(24)J_{cv}(\Delta)=\int\frac{d k_y}{2\pi}\rho_{k_y}(0,\Delta), \tag*{(24)}

where ρky(x,ε)=−1πIm⁡Tr⁡Gky(x,x,ε+i0)\rho_{k_y}(x,\varepsilon)=-\frac{1}{\pi}\operatorname{Im}\operatorname{Tr}G_{k_y}(x,x,\varepsilon+i0) is the spin-traced local density of states for a fixed kyk_y and GkyG_{k_y} is the retarded Green’s function of the one-dimensional problem at fixed kyk_y, with the energy ε\varepsilon measured from EgE_g. We first emphasize that, in the absence of spin orbit coupling, the motion along the xx direction reduces to the usual problem of a particle in a linear potential, and the corresponding solutions are Airy functions. The local density of states at the origin is then

ρky(0,ε)=2ΘℓAi⁡2(ℏ2ky2/(2m∗)−εΘ),\rho_{k_y}(0,\varepsilon)=\frac{2}{\Theta\ell}\operatorname{Ai}^{2}\left(\frac{\hbar^{2}k_y^{2}/(2m^{*})-\varepsilon}{\Theta}\right),

where ℓ\ell and Θ\Theta are the length and energy scales introduced in § 3.2 and the factor 2 accounts for the two spin states. After integration over kyk_y, the JDOS is

Jcv(Δ)=2m∗πℏ2F2D(ΔΘ),(25)J_{cv}(\Delta)=\frac{2m^{*}}{\pi\hbar^{2}}F_{2\mathrm{D}}\left(\frac{\Delta}{\Theta}\right), \tag*{(25)}

where F2D(η)=∫−η∞Ai⁡2(t)t+η dt=2∫0∞Ai⁡2(u2−η) duF_{2\mathrm{D}}(\eta)=\int_{-\eta}^{\infty}\frac{\operatorname{Ai}^{2}(t)}{\sqrt{t+\eta}}\,dt=2\int_{0}^{\infty}\operatorname{Ai}^{2}(u^{2}-\eta)\,du. For large positive η\eta, F2D→1/2F_{2\mathrm{D}}\to1/2, and therefore the JDOS approaches the field-free value m∗/πℏ2m^{*}/\pi\hbar^{2}. Eq. (25) describes the two dimensional Franz–Keldysh effect. The electric field allows finite absorption below EgE_g, where the absorption decreases rapidly as the photon energy moves further below the gap. Above EgE_g, the absorption oscillates around the field-free value, with the oscillation amplitude gradually decreasing at higher energies. Both effects occur over the characteristic energy scale Θ\Theta.

For finite SOCs, we first consider the persistent spin helix point αR=βD≡α\alpha_R=\beta_D\equiv\alpha. As shown in § 3.2, the two spin projections decouple exactly in this case and the spin-orbit coupling produces only an energy shift. Therefore,

Jcv(Δ)=2m∗πℏ2F2D(Δ+2m∗α2Θ).(26)J_{cv}(\Delta)=\frac{2m^{*}}{\pi\hbar^{2}}F_{2\mathrm{D}}\left(\frac{\Delta+2m^{*}\alpha^{2}}{\Theta}\right). \tag*{(26)}

Both spin projections give the same contribution and the entire Franz–Keldysh spectrum is shifted by 2m∗α22m^{*}\alpha^{2}. This is consistent with the field free result of § 3.3, where the absorption edge is shifted to Δ=−2m∗α2\Delta=-2m^{*}\alpha^{2}.

For general αR≠βD\alpha_R\ne\beta_D the situation is different. The Landau–Zener coupling derived in § 3.2 mixes the two spin states and a closed analytical expression is no longer available. We therefore calculate Eq. (24) numerically using the real-space Green function. The corresponding results are shown in Figure 5. Figure 5(a) shows the combined effect of the Rashba spin-orbit coupling and the electric field for βD=0\beta_D=0, where the strength of the Rashba coupling is described by α=ℏκR/(ℓΘ)\alpha=\hbar\kappa_R/(\ell\Theta). As α\alpha increases, the absorption edge moves downward by ESO=α2Θ/4E_{\mathrm{SO}}=\alpha^{2}\Theta/4, in agreement with the field-free threshold found in § 3.3. At the same time, the first Franz–Keldysh maximum increases with from 1.27 m∗/πℏ21.27\,m^*/\pi\hbar^2 for a=0a=0 to 1.42 m∗/πℏ21.42\,m^*/\pi\hbar^2 for a=1.5a=1.5 and 1.64 m∗/πℏ21.64\,m^*/\pi\hbar^2 for a=2a=2. The increase of this peak reflects the inverse square-root van Hove singularity of the field-free Rashba system.

Image not extracted

Figure 5. Franz–Keldysh electroabsorption with spin-orbit coupling at B=0B = 0. (a) JDOS for pure Rashba coupling. (b) Local density of states Θℓρkyℓ(0,ε)\Theta\ell\rho_{k_y \ell}(0,\varepsilon) for different kyℓk_y\ell. (c) JDOS for different Rashba and Dresselhaus coupling ratios at fixed a2+b2=1.5\sqrt{a^2+b^2}=1.5. (d) JDOS for a fixed material and different electric field strengths.

The role of the Landau–Zener spin dynamics discussed in § 3.2 is shown in Figure 5(b). The solid curves give the exact local density of states, while the dotted curves correspond to the diabatic approximation, where the spin remains frozen as kxk_x changes. For kyℓ=0.3k_y\ell=0.3, where δLZ=0.13\delta_{\mathrm{LZ}}=0.13, the two results are close, as expected in the diabatic regime. At kyℓ=0.8k_y\ell=0.8, where δLZ=0.96\delta_{\mathrm{LZ}}=0.96, visible differences appear, indicating the crossover between the diabatic and adiabatic regimes. For kyℓ=1.4k_y\ell=1.4, where δLZ=2.94\delta_{\mathrm{LZ}}=2.94, the exact result differs strongly from the diabatic one, with the absorption threshold shifted downward by about 1.5Θ1.5\Theta and a different oscillation structure. This shows that the Landau–Zener spin evolution becomes important for a≳1a\gtrsim1, corresponding approximately to Θ≲4ESO\Theta\lesssim4E_{\mathrm{SO}}, and directly modifies the electroabsorption spectrum.

The effect of the Dresselhaus coupling is shown in Figure 5(c). Here the total spin orbit strength a2+b2=1.5\sqrt{a^2+b^2}=1.5 is kept fixed, where b=ℏβD/(ℓΘ)b=\hbar\beta_{\mathrm{D}}/(\ell\Theta), while the ratio of the Rashba and Dresselhaus couplings is varied. The changes in the spectrum are more moderate than in the pure Rashba case. In particular, the first maximum decreases from 1.42 m∗/πℏ21.42\,m^*/\pi\hbar^2 for b=0b=0 to 1.27 m∗/πℏ21.27\,m^*/\pi\hbar^2 at a=ba=b. This moderate change reflects two competing effects. Increasing the Dresselhaus coupling lowers the band minimum, while at the same time it reduces the Landau–Zener spin mixing. At a=ba=b, this mixing vanishes, and the spectrum reduces to the exact shifted Franz–Keldysh result of Eq. (26).

Figure 5(d) shows the field dependence for a fixed material, where the effective mass m∗m^* and the Rashba coupling αR\alpha_{\mathrm{R}}, and therefore ESO=m∗αR2/2E_{\mathrm{SO}}=m^*\alpha_{\mathrm{R}}^2/2, are kept fixed while the electric field is varied. Since Θ∝E2/3\Theta\propto\mathcal{E}^{2/3}, changing E\mathcal{E} changes the ratio Θ/ESO\Theta/E_{\mathrm{SO}} and continuously connects the strong and weak field regimes discussed in § 3.2. For Θ≫ESO\Theta\gg E_{\mathrm{SO}}, the spectrum is close to the ordinary Franz–Keldysh result shifted by ESOE_{\mathrm{SO}}. As the electric field decreases, the spin dynamics becomes increasingly adiabatic and the spectrum approaches the field-free Rashba result of § 3.3. In this limit, the finite electric field replaces the inverse square root singularity at Δ=−ESO\Delta=-E_{\mathrm{SO}} by a finite peak whose width is of order Θ\Theta. As Θ\Theta decreases, this peak becomes higher and moves toward the field-free absorption edge ℏω=Eg−ESO\hbar\omega=E_g-E_{\mathrm{SO}}. For Θ≲ESO/4\Theta\lesssim E_{\mathrm{SO}}/4, its height and position are well described by

Jmax⁡≃1.83(ESOΘ)1/2m∗πℏ2E−1/3,(27)J_{\max} \simeq1.83\left(\frac{E_{\mathrm{SO}}}{\Theta}\right)^{1/2}\frac{m^*}{\pi\hbar^2}\mathcal{E}^{-1/3}, \tag*{(27)}

and

ℏωmax⁡≃Eg−ESO+0.6Θ.(28)\hbar\omega_{\max} \simeq E_g-E_{\mathrm{SO}}+0.6\Theta. \tag*{(28)}

Thus, decreasing the electric field causes the peak to approach the Rashba edge while its height increases as E−1/3\mathcal{E}^{-1/3}. The field dependence of both quantities therefore provides a direct optical measure of ESOE_{\mathrm{SO}}. The corresponding electroabsorption ΔA=A(E)−A(0)\Delta A=A(\mathcal{E})-A(0) is negative in an energy interval of order Θ\Theta above Eg−ESOE_g-E_{\mathrm{SO}} and positive below it, with an amplitude that also grows as E−1/3\mathcal{E}^{-1/3}. This behavior is qualitatively different from the ordinary two-dimensional Franz–Keldysh effect, where the oscillation amplitude is nearly field independent, and from the three-dimensional case, where it scales as E1/3\mathcal{E}^{1/3}. The opposite E−1/3\mathcal{E}^{-1/3} dependence found here originates from the inverse square root van Hove singularity produced by the ring-shaped minimum of the lower Rashba branch.

Spin-orbit coupled magnetoabsorption

We next consider the optical response in the presence of Rashba and Dresselhaus SOCs and a magnetic field, with E=0\mathcal{E}=0. In this case, the states form spin-orbit-mixed Landau levels, and summing the degenerate kyk_y states with the normalized spinor densities of § 3.1 gives

Jcv(ℏω)=12πaB2∑nδ(ℏω−Eg−En),(29)J_{cv}(\hbar\omega)=\frac{1}{2\pi a_B^2}\sum_n\delta\left(\hbar\omega-E_g-E_n\right), \tag*{(29)}

where En=(ℏωB/2)(En+1)E_n=(\hbar\omega_B/2)(\mathcal{E}_n+1). Eq. (29) shows that each Landau level produces an absorption line at ℏω=Eg+En\hbar\omega=E_g+E_n, with the same total weight 1/(2πaB2)1/(2\pi a_B^2), the Landau degeneracy per unit area. An important result is that this weight does not depend on how strongly the Rashba and Dresselhaus couplings mix the spin components. The spin-orbit coupling changes the energies EnE_n and therefore shifts, splits, and produces anticrossings between the absorption lines, but it does not change their total weights. This is the magnetic field counterpart of the JDOS sum rule obtained in § 3.3 and is exact within the constant matrix element approximation. The evolution of these lines with the magnetic field is shown in Figure 6, which gives their positions for a fixed material as a function of ℏωB∝B\hbar\omega_B\propto B, with pure Rashba coupling (dashed lines) and with βD/αR=0.5\beta_D/\alpha_R=0.5 (solid lines). At weak magnetic fields, the absorption lines originate from the field-free edges obtained in § 3.3. These edges are Eg−ESOE_g-E_{\mathrm{SO}} for pure Rashba coupling and Eg−m∗(αR+βD)2/2E_g-m^*(\alpha_R+\beta_D)^2/2 when the Dresselhaus coupling is also present. In the pure Rashba case, the lowest absorption line remains close to Eg−ESOE_g-E_{\mathrm{SO}} up to ℏωB≃2ESO\hbar\omega_B\simeq2E_{\mathrm{SO}} because the lower branch of the jjth Rashba doublet reaches its minimum near −ESO-E_{\mathrm{SO}} when ℏωB≃ESO/(j+1)\hbar\omega_B\simeq E_{\mathrm{SO}}/(j+1). In the pure Rashba case the splitting of a doublet grows as B\sqrt{B} at weak fields and becomes linear in BB at strong fields, where it approaches ℏωB−gμBB\hbar\omega_B-g\mu_BB. The Dresselhaus coupling enhances the crossings between different Rashba doublets into anticrossings, seen as a repulsion between the corresponding absorption lines. It also shifts the weak field lines to lower energies because the band minimum becomes deeper.

Image not extracted

Figure 6. Positions ℏω=Eg+En\hbar\omega= E_g + E_n of the eight lowest magnetoabsorption lines as a function of the magnetic field for a fixed material, in units of ESOE_{\mathrm{SO}}, with μZ=0.3\mu_Z = 0.3.

Spin-orbit coupled electroabsorption in crossed fields

We finally consider the combined effect of Rashba and Dresselhaus SOCs and crossed electric and magnetic fields on the optical absorption. In crossed fields, the electric field removes the kyk_y degeneracy of the Landau levels and makes the absorption depend directly on the spinor density. The JDOS becomes

Jcv(ℏω)=12πaB21∣e∣EaB∑n∣χn(ηn∗)∣2,(30)J_{cv}(\hbar\omega)=\frac{1}{2\pi a_B^2}\frac{1}{|e|\mathcal{E}a_B}\sum_n\left|\chi_n(\eta_n^*)\right|^2, \tag*{(30)}

where ηn∗=E~+0.5(E~n+1−E~2−Δ~)/E~\eta_n^*=\widetilde{\mathcal{E}}+0.5(\widetilde{E}_n+1-\widetilde{\mathcal{E}}^2-\widetilde{\Delta})/\widetilde{\mathcal{E}} and Δ~=(ℏω−Eg)/(ℏωB/2)\widetilde{\Delta}=(\hbar\omega-E_g)/(\hbar\omega_B/2). Thus, the photon energy selects a position ηn∗\eta_n^* for each Landau level, and the absorption strength is determined by ∣χn(ηn∗)∣2|\chi_n(\eta_n^*)|^2. The electric field therefore changes the shape and position of each absorption line while its integrated weight remains unchanged. Figure 7(a) shows this evolution for α=0.8\alpha=0.8, β=0.3\beta=0.3, and μZ=0.3\mu_Z=0.3. At E~=0.15\widetilde{\mathcal{E}}=0.15 the Landau lines are narrow and well separated. Since the line profile follows the oscillator density ϕj2\phi_j^2, a state with oscillator index jj produces j+1j+1 maxima. The two lowest levels, which are nearly degenerate at E~=0\widetilde{\mathcal{E}}=0 as shown in Figure 2(d), strongly overlap and appear as a single absorption line with approximately twice the spectral weight. Increasing E~\widetilde{\mathcal{E}} broadens the absorption lines until the neighboring Landau-level contributions are no longer resolved separately, and at E~=1.2\widetilde{E} = 1.2 the spectrum becomes Franz–Keldysh-like.

Image not extracted

Figure 7. Electroabsorption in crossed electric and magnetic fields for α=0.8\alpha= 0.8, β=0.3\beta= 0.3, and μZ=0.3\mu_Z = 0.3. (a) JDOS for increasing E~\tilde{E}. (b) Comparison between the cases with and without spin-orbit coupling at E~=0.4\tilde{E} = 0.4. (c) JDOS as a function of Δ~\tilde{\Delta} and E~\tilde{E}. (d) Centroid shifts δn\delta_n and reduced widths wn/(2E~)w_n/(2\tilde{E}) of the six lowest absorption lines.

Figure 7(b) compares the total JDOS with and without SOC at E~=0.4\widetilde{E} = 0.4. The dashed curve shows the total spectrum without spin-orbit coupling; the solid curve shows the total spectrum with spin-orbit coupling, and the dotted curves represent the individual spin-orbit coupling mixed Landau-level contributions. Without spin-orbit coupling, each Landau line follows the oscillator density ϕj2\phi_j^2, with centroid ⟨Δ~⟩=2j+1±μZ+E~2\langle\widetilde{\Delta}\rangle= 2j + 1 \pm\mu_Z + \widetilde{E}^2 and width ∣e∣EaBj+1/2\lvert e\rvert E a_B \sqrt{j + 1/2}. When spin-orbit coupling is present, different oscillator states are mixed, which shifts the individual absorption lines and modifies their widths and line profiles. As a result, the first maximum of the total spectrum moves from Δ~=1.4\widetilde{\Delta} = 1.4 without SOC to Δ~=0.87\widetilde{\Delta} = 0.87 with SOC.

Figure 7(c) shows the JDOS as a two dimensional map of photon energy Δ~\widetilde{\Delta} and electric field E~\widetilde{E}, where darker regions correspond to stronger absorption. At small E~\widetilde{E}, the spectrum consists of narrow structures associated with individual Landau levels. As the electric field increases, these structures broaden and spread in energy until neighboring Landau-level contributions can no longer be distinguished separately. At larger E~\widetilde{E}, the electric-field broadening becomes comparable to or larger than the Landau-level spacing, so the individual magnetic absorption lines can no longer be resolved. The spectrum then approaches the B=0B = 0 Franz–Keldysh response. The relevant energy scale is Θ=(E~2/2)1/3ℏωB\Theta= (\widetilde{E}^2/2)^{1/3}\hbar\omega_B, so this crossover occurs when Θ\Theta becomes larger than the Landau-level spacing.

Finally, Figure 7(d) quantifies how SOC changes the individual absorption lines as the electric field increases. The upper panel shows the additional shift δn\delta_n of the center of each absorption line, defined as

δn=⟨Δ~⟩n−[En(0)+1]−E~2.(31)\delta_n = \langle\widetilde{\Delta}\rangle_n - [E_n(0) + 1] - \widetilde{E}^2. \tag*{(31)}

Here En(0)+1E_n(0) + 1 is the line position at zero electric field and E~2\widetilde{E}^2 is the usual classical drift shift. Thus, δn>0\delta_n > 0 corresponds to an additional shift to higher energy, while δn<0\delta_n < 0 corresponds to a shift to lower energy. At weak electric fields, the two members of a spin-orbit mixed doublet are shifted in opposite directions. For example, at E~=0.3\widetilde{E} = 0.3 we obtain δ2=0.047\delta_2 = 0.047 and δ3=−0.049\delta_3 = -0.049, showing that one line moves to higher energy while its partner moves to lower energy. For nondegenerate levels, these additional shifts vary as E~2\widetilde{E}^2, which means that they arise as a second-order response to the electric field. The nearly degenerate lowest pair is different because the electric field mixes these two states already in first order, so their shifts start linearly in E~\tilde{E}. This opposite motion of the two partners is a direct signature of the spin-orbit coupling induced drift field. The lower panel shows the reduced line widths wn/(2E~)w_n/(2\tilde{E}). The dotted horizontal lines give the corresponding widths without SOC, so the deviation of the colored curves from these lines directly measures the effect of spin mixing. SOC changes some of the widths by about 10%. At E~≳0.8\tilde{E} \gtrsim0.8, the field induced anticrossings discussed in § 3.1 exchange the character of the levels, which is reflected in the crossing and rapid change of the width curves.

Conclusions

In this work, we have developed a unified description of the interband optical response of a two-dimensional semiconductor with Rashba and Dresselhaus spin-orbit couplings in electric and magnetic fields. In crossed fields, the in-plane electric field produces an additional spin-orbit drift field. This term changes the spin-orbit-mixed Landau levels, shifts nearly degenerate states, and produces field-induced anticrossings. At zero magnetic field, the same electric field problem can be mapped onto a Landau–Zener problem in momentum space. This allows the strong-field diabatic and weak-field adiabatic regimes to be clearly identified, together with the special persistent spin helix limit.

The optical response shows several characteristic signatures of the spin-orbit interaction. At zero external fields, the total joint density of states above the band gap remains unchanged, while new structures appear below the gap. Pure Rashba coupling produces an inverse square root singularity, whereas the addition of Dresselhaus coupling changes this structure into a step and a logarithmic peak. In an electric field, the Franz–Keldysh effect removes the Rashba singularity and replaces it by a finite absorption peak. In the weak field regime, the height of this peak varies as E−1/3\mathcal{E}^{-1/3} and its position approaches Eg−ESOE_g-E_{\mathrm{SO}} as the field is reduced. These dependences provide a direct optical measure of the spin-orbit energy. In a perpendicular magnetic field, the spin-orbit interaction shifts and splits the magnetoabsorption lines while leaving their total weights unchanged. When both fields are present, the electric field broadens these lines and gradually drives the spectrum toward the Franz–Keldysh regime. At the same time, the two members of a spin-orbit-mixed Landau doublet can acquire opposite shifts and different widths, providing additional optical signatures of the underlying spin-orbit states.

It is also useful to connect these results with experimentally accessible parameters. For a representative giant-Rashba two dimensional semiconductor with μ=0.1m0\mu= 0.1m_0 and ℏαR=0.5\hbar\alpha_R = 0.5 eV Å, the corresponding spin orbit energy is ESO=1.64E_{\mathrm{SO}} = 1.64 meV. At B=5B = 5 T, the spin-orbit-mixed Landau levels are already well separated, while electric fields of a few kV/cm give a Franz–Keldysh energy scale comparable to ESOE_{\mathrm{SO}}. For these parameters, fields between about 0.5 and 3 kV/cm are sufficient to broaden the magnetoabsorption lines and drive the spectrum from separated Landau structures toward a Franz–Keldysh-like response. At zero magnetic field, fields of 1–4 kV/cm replace the field-free Rashba singularity with a measurable subgap absorption peak whose position changes with the electric field. These energy and field scales are relevant for giant-Rashba materials such as polar semiconductors with heavy elements [15]. In conventional III–V heterostructures, where the Rashba energy is much smaller, the corresponding electroabsorption signatures are expected to be weaker [7, 50].

The predicted subgap absorption, its electric field dependence, and the spin-dependent shifts and widths of the crossed field absorption lines provide direct signatures that can be tested by electroabsorption and magneto-optical spectroscopy. This makes epitaxially grown III–V quantum wells a natural platform for experimental verification, where gate electrodes can control the in-plane electric field and optical measurements can follow the resulting changes in the absorption spectrum. Such structures can also be integrated with semiconductor laser and photodetector technologies, opening a possible route toward electrically tunable spin-dependent optical devices.

Acknowledgments

The authors declare that they have no competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. BT is supported by the Turkish Academy of Sciences (TUBA) under Grant No. AD-2026.

Author contributions

B.S.: Conceptualization, Methodology, Formal analysis, Writing – original draft. S.T.: Formal analysis, Validation, Writing – review and editing. B.T.: Conceptualization, Supervision, Writing – review and editing.

Data availability

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

References

References

  1. [1]Z̆utić I, Fabian J and Das Sarma S 2004 Rev. Mod. Phys. 76(2) 323–410 URL https://link.aps.org/doi/10.1103/RevModPhys.76.323DOI
  2. [2]Awschalom D D and Flatté M E 2007 Nature Physics 3 153–159 URL https://doi.org/10.1038/nphys551
  3. [3]Manchon A, Koo H C, Nitta J, Frolov S M and Duine R A 2015 Nature Materials 14 871–882 URL https://doi.org/10.1038/nmat4360
  4. [4]Avsar A, Ochoa H, Guinea F, Özyilmaz B, Van Wees B and Vera-Marun I J 2020 Reviews of Modern Physics 92 021003 URL https://doi.org/10.1103/RevModPhys.92.021003
  5. [5]Bihlmayer G, Noël P, Vyalikh D V, Chulkov E V and Manchon A 2022 Nature Reviews Physics 4 642–659 URL https://doi.org/10.1038/s42254-022-00490-y
  6. [6]Zollner K, Kurpas M, Gmitra M and Fabian J 2025 Nature Reviews Physics 7 255–269 URL https://doi.org/10.1038/s42254-025-00818-4
  7. [7]Nitta J, Akazaki T, Takayanagi H and Enoki T 1997 Phys. Rev. Lett. 78(7) 1335–1338 URL https://link.aps.org/doi/10.1103/PhysRevLett.78.1335DOI
  8. [8]Meier L, Salis G, Shorubalko I, Gini E, Schön S and Ensslin K 2007 Nature Physics 3 650–654 URL https://doi.org/10.1038/nphys675
  9. [9]Nishimura T, Yoshizumi K, Saito T, Iizasa D, Nitta J and Kohda M 2021 Phys. Rev. B 103(9) 094412 URL https://link.aps.org/doi/10.1103/PhysRevB.103.094412DOI
  10. [10]Chander S, Grobecker B, Poshakinskiy A, Anghel S, Mano T, Moore J, Yusa G and Betz M 2026 Phys. Rev. Appl. 25(3) 034064 URL https://link.aps.org/doi/10.1103/m4cv-6cytDOI
  11. [11]Schliemann J, Egues J C and Loss D 2003 Phys. Rev. Lett. 90(14) 146801 URL https://link.aps.org/doi/10.1103/PhysRevLett.90.146801DOI
  12. [12]Bernevig B A, Orenstein J and Zhang S C 2006 Phys. Rev. Lett. 97(23) 236601 URL https://link.aps.org/doi/10.1103/PhysRevLett.97.236601DOI
  13. [13]Koralek J D, Weber C P, Orenstein J, Bernevig B A, Zhang S C, Mack S and Awschalom D D 2009 Nature 458 610–613 URL https://doi.org/10.1038/nature07871
  14. [14]Tkach Y Y 2022 Phys. Rev. B 105(16) 165409 URL https://link.aps.org/doi/10.1103/PhysRevB.105.165409DOI
  15. [15]Ishizaka K, Bahramy M S, Murakawa H, Sakano M, Shimojima T, Sonobe T, Koizumi K, Shin S, Miyahara H, Kimura A, Miyamoto K, Okuda T, Namatame H, Taniguchi M, Arita R, Nagaosa N, Kobayashi K, Murakami Y, Kumai R, Kaneko Y, Onose Y and Tokura Y 2011 Nature Materials 10 521–526 URL https://doi.org/10.1038/nmat3051
  16. [16]Ketterl A S, Andres B, Polverigiani M, Voroshin V, Gahl C, Kokh K A, Tereshchenko O E, Chulkov E V, Shikin A and Weinelt M 2021 Phys. Rev. B 103(8) 085406 URL https://link.aps.org/doi/10.1103/PhysRevB.103.085406DOI
  17. [17]Lee A C, Peng B, Du K, Kung H H, Monserrat B, Cheong S W, Won C J and Blumberg G 2022 Phys. Rev. B 105(16) L161105 URL https://link.aps.org/doi/10.1103/PhysRevB.105.L161105DOI
  18. [18]Miller D A B, Chemla D S and Schmitt-Rink S 1986 Phys. Rev. B 33(10) 6976–6982 URL https://link.aps.org/doi/10.1103/PhysRevB.33.6976DOI
  19. [19]Citrin D S and Hughes S 2000 Phys. Rev. B 61(8) R5105(R)–R5108(R) URL https://link.aps.org/doi/10.1103/PhysRevB.61.R5105DOI
  20. [20]Sercel P C and Efros A L 2023 Phys. Rev. B 107(19) 195436 URL https://link.aps.org/doi/10.1103/PhysRevB.107.195436DOI
  21. [21]Yudin D and Shelykh I A 2016 Phys. Rev. B 94(16) 161404(R) URL https://link.aps.org/doi/10.1103/PhysRevB.94.161404DOI
  22. [22]Lee W B, Kim S B, Kim K W, Lee K J, Koo H C and Choi G M 2021 Phys. Rev. B 104(18) 184412 URL https://link.aps.org/doi/10.1103/PhysRevB.104.184412DOI
  23. [23]Metti S, Thomas C, Xiao D and Manfra M J 2022 Phys. Rev. B 106(16) 165304 URL https://link.aps.org/doi/10.1103/PhysRevB.106.165304DOI
  24. [24]Stier A V, Meining C J, Whiteside V R, McCombe B D, Rashba E I, Grabs P and Molenkamp L W 2023 Phys. Rev. B 107(4) 045301 URL https://link.aps.org/doi/10.1103/PhysRevB.107.045301DOI
  25. [25]Studer M, Salis G, Ensslin K, Driscoll D C and Gossard A C 2009 Phys. Rev. Lett. 103(2) 027201 URL https://link.aps.org/doi/10.1103/PhysRevLett.103.027201DOI
  26. [26]Alekseev P S and Nestoklon M O 2021 Phys. Rev. B 103(19) 195306 URL https://link.aps.org/doi/10.1103/PhysRevB.103.195306DOI
  27. [27]Mobini E, Espinosa D H G, Vyas K and Dolgaleva K 2022 Micromachines 13 ISSN 2072-666X URL https://www.mdpi.com/2072-666X/13/7/991
  28. [28]Baboux F, Moody G and Ducci S 2023 Optica 10 917–931 URL https://opg.optica.org/optica/abstract.cfm?URI=optica-10-7-917
  29. [29]Zhang L, Jiang J, Multunas C, Ming C, Chen Z, Hu Y, Lu Z, Pendse S, Jia R, Chandra M, Sun Y Y, Lu T M, Ping Y, Sundararaman R and Shi J 2022 Nature Photonics 16 529–537 URL https://doi.org/10.1038/s41566-022-01016-9
  30. [30]Kostadinov-Mutzafi A, Tilchin J, Shapiro A, Dirin D N, Kovalenko M V, Tan L Z and Lifshitz E 2022 Phys. Rev. B 106(3) 035303 URL https://link.aps.org/doi/10.1103/PhysRevB.106.035303DOI
  31. [31]Liang J, Wen W, Jin F, Rubo Y G, Liew T C H and Su R 2024 Nature Photonics 18 357–362 URL https://doi.org/10.1038/s41566-023-01375-x
  32. [32]Ahn E C 2020 npj 2D Materials and Applications 4 17 URL https://doi.org/10.1038/s41699-020-0152-0
  33. [33]Dieny B, Prejbeanu I L, Garello K, Gambardella P, Freitas P, Lehndorff R, Raberg W, Ebels U, Demokritov S O, Akerman J, Deac A, Pirro P, Adelmann C, Anane A, Chumak A V, Hirohata A, Mangin S, Valenzuela S O, Onbasli M C, d’Aquino M, Prenat G, Finocchio G, Lopez-Diaz L, Chantrell R, Chubykalo-Fesenko O and Bortolotti P 2020 Nature Electronics 3 446–459 URL https://doi.org/10.1038/s41928-020-0461-5
  34. [34]Yang H, Valenzuela S O, Chshiev M, Couet S, Dieny B, Dlubak B, Fert A, Garello K, Jamet M, Jeong D E, Lee K, Lee T, Martin M B, Kar G S, Séneor P, Shin H J and Roche S 2022 Nature 606 663–673 URL https://doi.org/10.1038/s41586-022-04768-0
  35. [35]Gish J T, Lebedev D, Song T W, Sangwan V K and Hersam M C 2024 Nature Electronics 7 336–347 URL https://doi.org/10.1038/s41928-024-01167-3
  36. [36]Lipparini E, Barranco M, Malet F, Pi M and Serra L 2006 Phys. Rev. B 74(11) 115303 URL https://link.aps.org/doi/10.1103/PhysRevB.74.115303DOI
  37. [37]Papp E and Micu C 2010 Superlattices and Microstructures 48 9–22 ISSN 0749-6036 URL https://www.sciencedirect.com/science/article/pii/S0749603610000960
  38. [38]Suleymanli B, Nakhmedov E, Alekperov O, Tatardar F and Tanatar B 2022 Solid State Communications 342 114582 ISSN 0038-1098 URL https://www.sciencedirect.com/science/article/pii/S0038109821003665
  39. [39]Candido D R, Erlingsson S I, Gramizadeh H, Costa J V I, Weigele P J, Zumbühl D M and Egues J C 2023 Phys. Rev. Res. 5(4) 043297 URL https://link.aps.org/doi/10.1103/PhysRevResearch.5.043297DOI
  40. [40]Li Z, Marsiglio F and Carbotte J P 2013 Scientific Reports 3 2828 URL https://doi.org/10.1038/srep02828
  41. [41]Yang C, Xu W, Zhang J, Zeng Z, Zhang C and Tang C 2008 Physica E: Low-dimensional Systems and Nanostructures 40 1896–1898 ISSN 1386-9477 URL https://www.sciencedirect.com/science/article/pii/S1386947707003682
  42. [42]Spector H N 1964 Physica 30 1917–1924 ISSN 0031-8914 URL https://www.sciencedirect.com/science/article/pii/0031891464900746
  43. [43]Vrehen Q H F 1966 Phys. Rev. 145(2) 675–688 URL https://link.aps.org/doi/10.1103/PhysRev.145.675DOI
  44. [44]Monozon B, Bezchastnov V and Schmelcher P 2019 Superlattices and Microstructures 128 408–420 ISSN 0749-6036 URL https://www.sciencedirect.com/science/article/pii/S0749603618321700
  45. [45]Pourmand S E and Rezaei G 2018 Physica B: Condensed Matter 543 27–31 ISSN 0921-4526 URL https://www.sciencedirect.com/science/article/pii/S0921452618303193
  46. [46]Bychkov Y A and Rashba E I 1984 Journal of Physics C: Solid State Physics 17 6039 URL https://doi.org/10.1088/0022-3719/17/33/015
  47. [47]Schliemann J and Loss D 2003 Phys. Rev. B 68(16) 165311 URL https://link.aps.org/doi/10.1103/PhysRevB.68.165311DOI
  48. [48]Zarea M and Ulloa S E 2005 Phys. Rev. B 72(8) 085342 URL https://link.aps.org/doi/10.1103/PhysRevB.72.085342DOI
  49. [49]Winkler R 2014 Spin-orbit Coupling Effects in Two-Dimensional Electron and Hole Systems Springer Tracts in Modern Physics (Springer Berlin Heidelberg) ISBN 9783662146002
  50. [50]Vurgaftman I, Meyer J R and Ram-Mohan L R 2001 Journal of Applied Physics 89 5815–5875 ISSN 0021-8979 URL https://doi.org/10.1063/1.1368156

Paper details

vlnc:W_g3fj8h4p
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

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