arXiv2020
We report on the experimental and numerical observation of polarization modulation instability (PMI) in a nonlinear fiber Kerr resonator. This phenomenon is phased-matched through the relative phase detuning between the intracavity fields associated with the two principal polarization modes of the cavity. Our experimental investigation is based on a 12-m long fiber ring resonator in which a polarization controller is inserted to finely control the level of intra-cavity birefringence. Depending on the amount of birefringence, the temporal patterns generated via PMI are found to be either stationary or to exhibit a period-doubled dynamics. Experimental results are in good agreement with numerical simulations based on an Ikeda map for the two orthogonally polarized modes. Our study provides new insights into the control of modulation instability in multimode Kerr resonators. Modulation instability (MI) is a nonlinear phenomenon characterized by the exponential growth and evolution of periodic perturbations on top of an intense continuous-wave (cw) laser beam [1, 2]. Underpinned by a nonlinearly phase-matched parametric process, it is associated with a transfer of energy from a narrow pump frequency component to a pair of sidebands arranged symmetrically around the pump. In single-pass optical fiber propagation, MI can be naturally phase-matched through a balance between anomalous group-velocity dispersion and Kerr nonlinearity [1, 2]. In contrast, more general phase-matching conditions are possible in the context of passive Kerr resonators, such as fiber ring cavities, because of the crucial role played by the systems' boundary conditions [3-5]. Various configurations of MI have been investigated in that context, including MI in the normal dispersion regime, MI via bichromatic or incoherent driving, as well as competition between MI and Faraday or period-doubled (P2) instabilities [6--12]. Moreover, at variance with single-pass propagation, MI in Kerr resonators can lead to the emergence of stationary periodic (Turing) patterns; such patterns are now understood to be intimately related to temporal cavity solitons and microresonator optical frequency combs [13--16]. Birefringence, and nonlinear coupling between the polarization components of light, is also known to contribute to the phase-matching of parametric processes. This leads to polarization MI (PMI) and the emergence of vector temporal patterns [17-19]. In driven resonators, PMI has only been investigated theoretically so far [20-22], but recent demonstrations of orthogonally-polarized dual comb generation in microresonators are sparking a renewed interest in this process [23]. In this Letter, we report on the direct experimental observation of PMI in a passive Kerr resonator. Our experimental test-bed is based on a normally dispersive fiber ring cavity that incorporates a polarization controller for adjustment of the intra-cavity birefringence. This localized birefringence gives rise to a relative phase detuning between the two orthogonal polarization modes of the cavity, which in turn affects the frequency shift of the PMI sidebands. We also find that birefringence can lead to period-doubled (P2) dynamics, characterized by a two round-trip cycle. Our experimental results are in good agreement with theoretical predictions and numerical simulations based on an iterative two-component Ikeda map. The experimental setup is displayed in Fig. 1(a). It consists of a L = 12-m long passive fiber ring cavity with a finesse F of about 27, mainly built out of spun fiber. To avoid competition with scalar MI [2], we have built a cavity with normal group-velocity dispersion estimated to 2 = 47 ps 2 /km, a value large enough to neglect third-order dispersion. Also, the use of a spun fiber (nearly isotropic) avoids group-velocity mismatch between the polarization components. Additionally, to prevent any additional source of bending-induced birefringence, the fiber was carefully off-spooled and wound directly on our experimental board with a large 50-cm diameter. We estimate that this causes a birefringence $Δ$n no greater than 10 --8 [24], which can be neglected in our study.