Theory of fiber optic co- and counter-pumped Raman polarizers

Abstract

Fiber optic Raman amplifiers are polarization-dependent by their nature: the state of polarization (SOP) of the outcoming signal beam strongly depends on the SOP of the pump beam. Thus, in isotropic fibers the SOP of the signal acquires the SOP of the pump beam towards the output, because the Raman gain for a signal SOP that is aligned along the pump SOP is two orders of magnitude larger than the Raman gain for the orthogonal signal polarization. However, telecommunication fibers are not isotropic but exhibit a birefringence which stochastically changes along the propagation distance. Moreover, the pump and the signal beams experience a different birefringence. This mismatch has the consequence that SOPs of the two beams do not change in unison when they propagate through the fiber. On average, the signal beam “sees” the two orthogonal polarizations of the pump beam with equal frequence. This situation is characteristic to high-PMD fibers (here, PMD stands for the polarization mode dispersion), namely, with PMD coefficients > 0.05 ps/km1/2. In this case the signal beam experiences an average Raman gain which is half of the maximal Raman gain. Thus a simple scalar model of Raman amplification is appropriate for describing high-PMD fibers, since all polarization-dependent properties of the Raman gain are washed out by fiber birefringence.