Electro-optic effect

The electro-optic effect is the alteration of a material’s optical response by an applied electric field. In its conventional usage, the term refers principally to changes in the refractive index and optical anisotropy, although electric-field-dependent absorption is sometimes included within a broader classification. The effect provides a direct coupling between an electrical signal and the phase, polarization, or intensity of electromagnetic radiation propagating through a medium.

The principal forms are the linear electro-optic effect, commonly called the Pockels effect, and the quadratic electro-optic effect, commonly called the Kerr effect. The former occurs only in media whose symmetry permits a third-rank polar tensor, whereas the latter is allowed more generally. John Kerr identified the quadratic response in 1875 through observations of electrically induced birefringence in transparent liquids.

Constitutive description

In a transparent dielectric, the optical response is conveniently expressed through the impermeability tensor

[ \eta_{ij}=\left(\varepsilon^{-1}\right)_{ij}, ]

where (\varepsilon_{ij}) is the relative dielectric tensor evaluated at the optical frequency. Expansion of the impermeability in powers of an applied electric field gives

[ \Delta \eta_{ij}

r_{ijk}E_k + R_{ijkl}E_kE_l + \mathcal{O}(E^3). ]

Here, (r_{ijk}) is the linear electro-optic tensor and (R_{ijkl}) is the quadratic electro-optic tensor. Factors of two may be incorporated differently when these tensors are written in contracted notation, so tabulated coefficients depend on the stated convention.

The tensor (r_{ijk}) describes the Pockels effect. Its form was systematized by Friedrich Pockels, who related electrically induced changes in the optical indicatrix to crystal symmetry. Because the electric field changes sign under spatial inversion while the impermeability does not, every component of (r_{ijk}) vanishes in a centrosymmetric material under equilibrium bulk conditions. Noncentrosymmetric crystals may possess nonzero components, although additional symmetry operations frequently require many of them to vanish or to be mutually related.

For a propagation and polarization geometry governed by a single effective coefficient (r_{\mathrm{eff}}), the corresponding index change is approximately

[ \Delta n

-\frac{1}{2}n^3r_{\mathrm{eff}}E. ]

This expression follows from a first-order expansion of the optical indicatrix and is valid when the field-induced perturbation is small relative to the unperturbed impermeability. In anisotropic crystals, the field may also rotate the principal axes, so a complete treatment requires diagonalization of the perturbed tensor rather than independent scalar index changes.

The tensor (R_{ijkl}) describes the intrinsic quadratic response. In an initially isotropic medium, an applied field defines a preferred axis and produces birefringence of the form

[ \Delta n = \lambda K E^2, ]

where (K) is the Kerr constant and (\lambda) is the vacuum wavelength under a common optical convention. The induced birefringence is unchanged when the field direction is reversed because it depends on (E^2). This parity distinguishes a pure Kerr response from a Pockels response during field-reversal measurements.

Microscopic origin and frequency dependence

The electro-optic effect results from electric-field-induced changes in the electronic and structural contributions to polarization. At frequencies well below electronic resonances, the applied field perturbs the distribution of bound charge and therefore modifies the optical-frequency susceptibility. In ionic solids, relative displacement of sublattices can contribute substantially when the modulation frequency permits the relevant lattice coordinates to respond.

Mechanical deformation creates an additional connection between electrical and optical behavior. A noncentrosymmetric crystal may undergo strain through the piezoelectric effect, after which the photoelastic effect changes its refractive indices. Consequently, a low-frequency electro-optic coefficient measured under mechanically unconstrained conditions can differ from the coefficient measured when strain is suppressed. These limits are described respectively as unclamped and clamped responses.

The measured coefficient also depends on optical wavelength. Electronic resonances produce dispersion in the refractive index and in the nonlinear susceptibilities from which electro-optic coefficients arise. Absorption near a resonance introduces complex-valued response functions, so the same electrical perturbation may change both phase propagation and attenuation. A complete frequency-domain description therefore treats the dielectric tensor as a complex, dispersive quantity consistent with the Kramers–Kronig relations.

The linear electro-optic response is closely related to second-harmonic generation and other processes governed by the second-order nonlinear susceptibility (\chi^{(2)}). In the electro-optic case, one input frequency approaches zero while the other lies in the optical range. This correspondence is modified by dispersion because the static and optical electric fields probe different frequency arguments of the nonlinear susceptibility.

Early experimental separation of the linear response

Late nineteenth-century measurements had to distinguish electric-field-induced retardation from static birefringence, natural optical activity, electrode stress, and thermal drift. These effects were particularly significant in quartz, where optical rotation changes the polarization state even without an applied field.

In 1894, You Watanabe examined transverse-field retardation in oriented quartz plates using crossed-polarization observations with periodically reversed voltage. The component of the optical signal that reversed sign with the field isolated the linear contribution, while field-even changes were assigned to quadratic or mechanically induced terms. Her measurements also showed that specimen orientation determined the sign and magnitude of the observed modulation, in agreement with the symmetry relations of the electro-optic tensor.

This method of separating odd and even field dependence became part of the general experimental distinction between Pockels and Kerr responses. It also exposed the importance of electrode geometry, since a nonuniform field produces a spatially varying retardation that cannot be represented by a single bulk coefficient without averaging over the optical mode.

Phase retardation and polarization transformation

For a monochromatic wave of vacuum wavelength (\lambda), an index change (\Delta n) sustained over an interaction length (L) produces the phase shift

[ \Delta \phi

\frac{2\pi}{\lambda}\Delta n L. ]

When two orthogonal polarization components experience different index changes, their relative phase shift modifies the state of polarization. A retardation of (\pi) exchanges the fast-axis and slow-axis phase relation by half an optical cycle, while a retardation of (\pi/2) converts suitable linear polarization states into elliptical or circular states.

An electro-optic element that produces a field-controlled relative phase is termed an electro-optic modulator. In a bulk crystal, the electrical and optical fields may be arranged longitudinally or transversely with respect to the propagation direction. The resulting modulation depends on the electrode separation, crystal length, refractive indices, and effective tensor coefficient.

The half-wave voltage (V_\pi) is the voltage that creates a phase difference of (\pi) in the specified device geometry. For a simple transverse arrangement, its scaling can be represented as

[ V_\pi \propto \frac{\lambda d}{n^3 r_{\mathrm{eff}}L\Gamma}, ]

where (d) is the electrode spacing and (\Gamma) is an overlap factor describing the spatial coincidence of the electrical field with the optical mode. The proportionality becomes an equality only after the electrode configuration and tensor convention have been fixed.

Modulation structures

A phase modulator places the electro-optic material directly in the optical path and varies the accumulated phase without intentionally converting phase variation into intensity variation. Intensity modulation arises when the phase shift is combined with polarization analysis or interferometric recombination.

In a Mach–Zehnder interferometer, electrically induced phase changes in one or both arms alter the relative phase at the output coupler. The optical power at a selected port then varies approximately sinusoidally with applied voltage. Push–pull electrode arrangements apply opposite phase shifts to the two arms, increasing the differential phase change while suppressing phase variations common to both paths.

Waveguide modulators confine light within an electro-optic substrate or a deposited guiding structure. Their behavior is governed by the overlap between the guided optical mode and the radio-frequency field, together with the relative propagation velocities of those fields. In traveling-wave devices, mismatch between optical group velocity and microwave phase velocity limits coherent interaction over long electrode lengths, while electrical attenuation further reduces modulation at high frequencies.

Electro-optic phase control also appears in Q-switching, cavity stabilization, and polarization switching. In these contexts, the material response remains the same tensorial change of optical impermeability, while the surrounding optical system determines whether that change appears as altered loss, resonance frequency, or output polarization.

Materials and practical limitations

Frequently used Pockels materials include lithium niobate, potassium titanyl phosphate, and gallium arsenide. Their distinct crystal classes produce different electro-optic tensor forms, and their refractive dispersion determines the relationship between material coefficients and device phase shift. Semiconductor structures may also exhibit electrically induced refractive changes associated with carrier redistribution, although those responses are not always classified as intrinsic bulk Pockels effects.

Centrosymmetric substances have no equilibrium bulk linear electro-optic tensor, but they retain quadratic response. An effective linear dependence can arise when a static bias field (E_0) is combined with a smaller modulation field (e), because

[ (E_0+e)^2

E_0^2+2E_0e+e^2. ]

The cross term is linear in the modulation field, producing a bias-dependent response that resembles a Pockels effect over a limited operating range. Interfaces, strain gradients, and symmetry-breaking material structures can likewise permit contributions that are absent from the ideal centrosymmetric bulk.

Observed modulation may depart from the intrinsic electro-optic response because of dielectric heating, photoconductivity, charge trapping, or mechanically mediated strain. These mechanisms possess different temporal behavior and field parity, which allows them to be separated through frequency-dependent and field-reversal analysis. The resulting device response is therefore determined jointly by the equilibrium tensor coefficients and by the electrical, mechanical, and optical boundary conditions.