Birefringent Spin-Lasers: Ultrafast Polarization Dynamics

• Birefringent spin-lasers are semiconductor lasers that combine spin-polarized carrier injection with linear cavity birefringence to achieve ultrafast polarization modulation.
• They harness intentional linear mode splitting to drive polarization oscillations significantly faster than intensity changes, reaching frequencies over 200 GHz.
• Strain-engineered quantum wells and elliptical micropillar designs offer precise control over device dynamics and enable versatile modulation schemes.

Birefringent spin-lasers are semiconductor spin lasers in which spin-polarized carrier injection is combined with linear cavity birefringence, so that the two orthogonal linear polarization eigenmodes are frequency split and dynamically coupled. In this class of devices, the transfer of angular momentum from spin-polarized carriers to photons produces circularly polarized emission, but the decisive speed-setting element is the birefringent splitting of the linear modes: polarization oscillations are governed primarily by that splitting rather than by ordinary intensity relaxation. The result is a device category in which polarization can be modulated an order of magnitude faster than intensity, and in suitably engineered structures the relevant oscillation frequencies can exceed $200$ GHz (Xu et al., 2020, Junior et al., 2015).

1. Definition and operating principle

A spin laser differs from a conventional semiconductor laser because the injected carriers are spin-polarized. The gain medium is therefore fed with unequal spin-up and spin-down electron populations, and optical selection rules transfer that imbalance to photons as unequal populations of the two circular helicities, conventionally denoted $S^+$ and $S^-$. In the standard notation used in the field, the helicity-resolved intensities are

$S^{\pm}=|E^{\pm}|^2,$

where $E^\pm$ are the complex amplitudes of the circularly polarized field components (Xu et al., 2020).

The “birefringent” qualifier is not incidental. In these lasers, linear birefringence splits the two orthogonal linearly polarized cavity eigenmodes and couples the polarization dynamics through rapid phase evolution between them. The circular polarization degree,

$P_\textrm{C}=(S^+-S^-)/(S^++S^-),$

can then oscillate much faster than the total intensity because birefringence acts directly on the relative phase between the linear modes, whereas intensity remains governed mainly by carrier-photon relaxation and gain saturation. This is the core distinction between birefringent spin-lasers and both conventional lasers and spin lasers operated without large birefringence (Xu et al., 2020).

A persistent misconception is that birefringence is always detrimental. The literature on birefringent spin-lasers states the opposite for the polarization channel: in ordinary VCSELs, birefringence is often associated with polarization switching, complex nonlinear dynamics, chaos, and instability of the output polarization, but in spin lasers the same anisotropy can be exploited to generate fast polarization oscillations of the emitted light. In that sense, birefringence is the mechanism that turns spin-dependent gain into an ultrafast polarization oscillator rather than merely a spin-polarized emitter (Junior et al., 2015).

2. Dynamical descriptions: from the spin-flip model to real intensity equations

The standard theoretical foundation is the spin-flip model (SFM), which describes the coupled evolution of optical fields, total carrier density, and carrier spin polarization. In the formulation emphasized for birefringent spin-lasers, the state variables include the circularly polarized optical fields $E^\pm$, the total carrier number $N$, the spin population difference $n$, the birefringence $\gamma_p$, the dichroism $S^+$0, the photon lifetime $S^+$1, the electron spin relaxation time $S^+$2, the recombination rate $S^+$3, the linewidth enhancement factor $S^+$4, and spin-resolved injections $S^+$5. A standard assumption is that hole spin imbalance is negligible because hole spin relaxation is much faster than electron spin relaxation (Xu et al., 2020).

A major conceptual development is the rewriting of the SFM into real-valued intensity equations. Instead of using complex helicity amplitudes as the primary variables, the dynamics are expressed in terms of linear-polarization intensities and phase. In the compact 2025 formulation, the real-variable state vector is

$S^+$6

with $S^+$7, and the circularly polarized intensities are reconstructed from the linear components through

$S^+$8

This representation makes explicit that the circular polarization state is determined by both linear-mode intensities and their phase difference (Labinac et al., 3 Aug 2025).

The same work recasts the birefringent spin-laser as a pair of coupled harmonic oscillators. In that analogy, the two oscillators represent the two orthogonal linearly polarized lasing modes, while birefringence supplies the coupling. The approximation is analytically useful because the dynamics separate into two approximate subsystems: a higher-frequency pair $S^+$9 and a lower-frequency pair $S^-$0. The coupling can be treated perturbatively, and a coupling factor

$S^-$1

distinguishes weak from strong coupling. The reported conclusion is that spin-lasers are typically weakly coupled, even though the coupling is sufficient to produce the defining ultrafast polarization dynamics (Labinac et al., 3 Aug 2025).

3. Intensity modulation, polarization modulation, and the role of birefringence

The dynamical distinction between intensity modulation (IM) and polarization modulation (PM) is central. For IM, the total injection is modulated while spin polarization is held fixed: $S^-$2 For PM, the total injection is fixed while the spin polarization of the injection is modulated: $S^-$3 The two drives produce qualitatively different responses (Xu et al., 2020).

For IM, the small-signal response has the standard driven-oscillator form, with resonance frequency and damping determined by injection and saturation parameters rather than by birefringence. The explicit conclusion is that birefringence does not affect IM: numerically, IM response curves for different $S^-$4 lie on top of one another. Physically, birefringence changes the relative phase between the linear modes, not the total intensities themselves (Xu et al., 2020).

For PM, by contrast, birefringence is the dominant control parameter. The reported central result is

$S^-$5

so the polarization-modulation resonance frequency grows approximately linearly with birefringence. In the coupled-oscillator interpretation, PM exhibits two resonances: a lower one associated with the usual relaxation-oscillation scale and a higher one associated with birefringence, with approximate frequency

$S^-$6

This higher-frequency resonance is the origin of the ultrafast polarization channel (Xu et al., 2020, Labinac et al., 3 Aug 2025).

The 2025 oscillator treatment further predicts mixed IM + PM regimes, tunable phase evolution with little change in amplitude, possible restoration of a second peak by adjusting the relative modulation amplitudes, and a Fano-like resonance/antiresonance in the IM response of $S^-$7. A plausible implication is that phase engineering, not only amplitude engineering, becomes a design degree of freedom in birefringent spin-lasers (Labinac et al., 3 Aug 2025).

4. Microscopic gain, cavity design, and strain-engineered birefringence

Beyond rate equations, the microscopic basis of birefringent spin-lasers has been developed for quantum-well VCSELs using an $S^-$8 $S^-$9 description of the active region. In that treatment, the quantum-well Hamiltonian is

$S^{\pm}=|E^{\pm}|^2,$0

where $S^{\pm}=|E^{\pm}|^2,$1 is the multiband $S^{\pm}=|E^{\pm}|^2,$2 Hamiltonian, $S^{\pm}=|E^{\pm}|^2,$3 is the strain term, and $S^{\pm}=|E^{\pm}|^2,$4 is the band-offset potential. Optical gain is obtained from the imaginary part of the dielectric function according to

$S^{\pm}=|E^{\pm}|^2,$5

with helicity-resolved gain components reflecting the spin-dependent selection rules (Junior et al., 2015).

This framework makes two steady-state spin-laser effects explicit. The first is threshold reduction: with spin-polarized injection, one helicity can reach gain threshold at lower total carrier density than in an unpolarized laser. The second is gain asymmetry, $S^{\pm}=|E^{\pm}|^2,$6, which depends not only on spin polarization but also strongly on carrier density and cavity detuning. The cavity therefore selects which part of the spin-dependent gain spectrum is used, so spin polarization alone does not determine device behavior (Junior et al., 2015).

The principal route to strong birefringence in this microscopic picture is uniaxial strain in the active region. Breaking the in-plane symmetry between the $S^{\pm}=|E^{\pm}|^2,$7 and $S^{\pm}=|E^{\pm}|^2,$8 directions modifies both the band structure and the dipole matrix elements. Even very small strains, such as $S^{\pm}=|E^{\pm}|^2,$9 and $E^\pm$0, produce measurable gain anisotropy and a very large birefringence coefficient. The birefringence is defined as

$E^\pm$1

and the reported strain-induced values reach

$E^\pm$2

Increasing the strain by only about $E^\pm$3 can increase $E^\pm$4 by roughly a factor of 3, substantially larger than typical DBR contributions around $E^\pm$5 (Junior et al., 2015).

The same study states that the attainable birefringence is sufficient to generate polarization oscillations above $E^\pm$6 GHz, and notes experiments around $E^\pm$7 GHz. It also reports that spin-polarized electrons change the gain amplitudes slightly but do not significantly alter the birefringence coefficient itself for the small-to-moderate spin polarizations relevant in real devices. In concise form: strain sets the birefringence, while spin injection activates the polarization dynamics that exploit it (Junior et al., 2015).

5. Experimental realization in bimodal high-$E^\pm$8 quantum-dot micropillars

A concrete experimental platform is the bimodal quantum-dot micropillar spin-laser. The reported devices are grown from an AlGaAs/GaAs planar microcavity with a central one-wavelength GaAs cavity, a single InGaAs QD layer at the cavity center, a lower DBR with $E^\pm$9 mirror pairs, and an upper DBR with $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$0 mirror pairs. The pillars are patterned into elliptical cross sections, intentionally breaking rotational symmetry and inducing birefringent splitting of the two orthogonally linearly polarized fundamental modes. The measured cavity quality factor is on the order of $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$1, and the spontaneous-emission coupling factor is $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$2 (Heermeier et al., 2021).

The pumping scheme combines a CW $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$3 nm laser that drives the device above threshold with a ps-pulsed $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$4 nm circularly polarized laser that injects spin-polarized carriers resonantly with wetting-layer states. The circular emission components $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$5 and $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$6 are time-resolved with a streak camera of $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$7 ps resolution, and the polarization dynamics are quantified through

$P_\textrm{C}=(S^+-S^-)/(S^++S^-),$8

After the spin-polarized pulse, both $P_\textrm{C}=(S^+-S^-)/(S^++S^-),$9 and $E^\pm$0 rise and then exhibit oscillatory antiphase behavior with a decay time of $E^\pm$1 ns; the oscillations become clearly visible for delays beyond $E^\pm$2 ps (Heermeier et al., 2021).

For a pillar with about $E^\pm$3 ellipticity, the measured mode splitting is

$E^\pm$4

corresponding to about $E^\pm$5 GHz. Fitting the measured $E^\pm$6 with a damped sinusoid gives a polarization-oscillation frequency of

$E^\pm$7

in excellent agreement with the cavity mode splitting expressed in frequency units. This agreement is the direct experimental confirmation that the polarization dynamics are governed by birefringent mode splitting (Heermeier et al., 2021).

The structural tuning parameter is the pillar ellipticity,

$E^\pm$8

with $E^\pm$9 and $N$0 the short and long ellipse axes. Across the studied families, the splitting increases approximately linearly with ellipticity, reaching about $N$1eV ($N$2 GHz) for the largest ellipticities in the main dataset and up to about $N$3eV ($N$4 GHz) in the broader discussion. Strongly elliptical QD micropillars can exceed $N$5 meV ($N$6 GHz), implying a large potential range for spin-oscillation frequencies. The splitting does not vanish at nominal zero ellipticity, which is attributed to residual stress/strain and elasto-optic birefringence from internal fields (Heermeier et al., 2021).

The device is also a high-$N$7 microlaser with a threshold pump power $N$8 mW. Simulations based on the SFM, using parameters such as $N$9, $n$0 GHz for the measured case and $n$1 to $n$2 GHz in the broader parameter study, reproduce the measured antiphase oscillations and show that $n$3 can reach at least $n$4 GHz. In an intermediate birefringence range, more complex or chaotic polarization behavior appears instead of simple damped oscillations, while sufficiently large birefringence restores clean single-period oscillations (Heermeier et al., 2021).

6. Coupled architectures, bistability, and conceptual boundaries

Birefringent spin-laser dynamics are not limited to single cavities. In laterally coupled pairs of spin-VCSELs, the system can be modeled as two identical circular waveguides of radius

$n$5

separated by an edge-to-edge distance $n$6, with symmetric and antisymmetric optical supermodes. The coupling coefficient is determined by the supermode frequency splitting,

$n$7

An extended SFM with $n$8 independent dynamical variables then captures spin-resolved carriers, circularly polarized optical amplitudes, and spatial and polarization phases (Vaughan et al., 2020).

In this coupled setting, birefringence enters as the rate $n$9, coupling the two circular polarization components inside each laser and directly influencing the polarization phase. High birefringence produces new stability boundaries, broad regions of bistability, and switching between in-phase and out-of-phase polarization states under variations of pump power or pump ellipticity. For $\gamma_p$0, the frequency scale

$\gamma_p$1

gives about $\gamma_p$2 GHz, matching the oscillations observed in the simulations. The reported result that one laser can switch the polarization state of the other through optical coupling identifies birefringence as a control parameter for coupling-mediated optical switching, not only for single-device modulation (Vaughan et al., 2020).

A separate conceptual boundary concerns laser-induced birefringent media that are not spin-lasers. The study of Ag$\gamma_p$3/Na$\gamma_p$4 ion-exchanged silver-doped glass reports a permanent transverse birefringence with a radial optical-axis pattern written by intense laser irradiation, and diagnoses that texture through spin-to-orbital optical angular momentum conversion (STOC) of a probe beam. In the circular basis, the birefringent element acts as

$\gamma_p$5

so the helicity-flipped component acquires orbital angular momentum $\gamma_p$6 per photon. This system is conceptually adjacent because it is a spin-dependent photonic element based on birefringence, but it is not a spin-laser: it concerns laser-induced birefringence in a passive material rather than lasing dynamics governed by carrier spin and birefringent cavity mode splitting (Amjad et al., 2012).

Taken together, these results delimit the field with some precision. Birefringent spin-lasers are not defined merely by polarization anisotropy or circularly polarized output; they are defined by the joint presence of spin-polarized gain and birefringent mode splitting, with the latter setting the characteristic polarization-dynamical timescale. This suggests a coherent device class spanning strained quantum-well VCSELs, high-$\gamma_p$7 QD micropillars, and coupled spin-VCSEL arrays, all organized around the same dynamical principle: ultrafast polarization control by intentionally exploited birefringence (Junior et al., 2015, Heermeier et al., 2021, Vaughan et al., 2020).