Abstract

The use of a cw laser beam detuned from resonance by only a few gigahertz to observe conical emission¹ makes the computational problem tractable (2 CRAY CPU hours per spectrum), leading to a simple physical model of this spatio-temporal instability. From the excellent qualitative agreement between computations and experimental observations, one understands the physics of conical emission as follows: self-trapping² of A₁, probe generation by resonance fluorescence, Raman-gain amplification^3,4 of this probe which produces A₃(ω₃), four-wave mixing³ between A₁(ω₁) and A₃(ω₃) which generates A₄(ω₄ = 2ω₁ - ω₃) and transfers energy back and forth between A₃ and A₄, division of A₄ into A_4L and A_4R by optical-Stark-shifted absorption, division of A₃ into A_3L and A_3R by four-wave-mixing coupling, and formation of the cone from A_4L (and sometimes from the low-energy shoulder of A_4R) by propagation through the spatially-dependent index of refraction prepared by A₁(x,y,z). We both observe and compute that light A₄(ω₄) of a particular frequency may appear both in the cone and in the center, ruling out phase matching as the determinant of the cone angle, unlike many other cones, e.g., two-wavelength conical emission.⁵

© 1989 Optical Society of America