Electro-Optic and Waveguide Properties of Reverse-Biased Gallium Phosphidep−nJunctions

Observations of light transmitted along the plane of reverse-biased GaP $p-n$ junctions show the depletion layer to be birefringent and to produce mode confinement of the light. Extensive measurements were made on diodes having a variety of dopings ($N_D\approx {10}^{17} \mathrm{to} {10}^{18}$ ${\mathrm{cm}}^{-3\mathrm{}}$) and several crystalline orientations (${\mathbf{E}}_J\parallel \mathrm{}\left[111\right]\mathrm{}$ or [110] or [100]) with several laser wavelengths (632.8 nm and 1153 nm from a He-Ne laser and 896.5 nm from a GaAs laser). The measurements were mainly of three types: (1) measurement of the intensity on the exit face of the diode versus position, voltage, and polarization state; (2) measurement of the change of phase with junction voltage of each state of polarization by an optical heterodyne technique; (3) measurement of the phase difference between these two polarization states versus voltage and wavelength by optical compensation methods. We interpret these results in terms of a dielectric waveguide mode in which a higher optical dielectric constant exists in a layer whose width is fixed and approximately equal to the zero-bias junction width. The birefringence arises from the linear electro-optic (Pockels) effect within the junction width, which changes with bias. The optical dielectric constants on the $n$ and $p$ sides are assumed slightly different. Agreement with this model is generally satisfactory; a few discrepancies, including an anom alous modulation for ${\mathbf{E}}_{\mathrm{light}}\parallel {\mathbf{E}}_J\parallel \mathrm{}\left[100\right]\mathrm{}$, appear to be best explained by the presence of a small quadratic electro-optic (Kerr) effect. The origin of the higher dielectric constant that gives mode confinement is not known.