Cavity solitons are localized intensity peaks that can form in a homogeneous background of radiation. They are generated by shining laser pulses into optical cavities that contain a nonlinear medium driven by a coherent field (holding beam). The ability to switch cavity solitons on and off and to control their location and motion by applying laser pulses makes them interesting as potential 'pixels' for reconfigurable arrays or all-optical processing units. Theoretical work on cavity solitons has stimulated a variety of experiments in macroscopic cavities and in systems with optical feedback. But for practical devices, it is desirable to generate cavity solitons in semiconductor structures, which would allow fast response and miniaturization. The existence of cavity solitons in semiconductor microcavities has been predicted theoretically, and precursors of cavity solitons have been observed, but clear experimental realization has been hindered by boundary-dependence of the resulting optical patterns-cavity solitons should be self-confined. Here we demonstrate the generation of cavity solitons in vertical cavity semiconductor microresonators that are electrically pumped above transparency but slightly below lasing threshold. We show that the generated optical spots can be written, erased and manipulated as objects independent of each other and of the boundary. Numerical simulations allow for a clearer interpretation of experimental results.

We investigate the relationship between passive mode locking and the formation of time-localized structures in the output intensity of a laser. We show how the mode-locked pulses transform into lasing localized structures, allowing for individual addressing and arbitrary low repetition rates. Our analysis reveals that this occurs when (i) the cavity round-trip is much larger than the slowest medium time scale, namely the gain recovery time, and (ii) the mode-locked solution coexists with the zero intensity (off) solution. These conditions enable the coexistence of a large quantity of stable solutions, each of them being characterized by a different number of pulses per round-trip and with different arrangements. Then, each mode-locked pulse becomes localized, i.e., individually addressable. [11]. The possibility of using LS as information bits for processing information in optical devices [12][13][14] has attracted an increasing interest in the last twenty years. LS have been observed in the transverse section of broad-area semiconductor microcavities injected by a coherent electromagnetic field [15] (passive morphogenesis) and are also termed "cavity solitons." More recently, spatial LS have been observed in laser systems where they arise from spontaneous emission noise [16,17] (active morphogenesis), without requiring an injected field. Because these lasing LS appear in a phase invariant system, their dynamical ingredients and their properties are very different from the LS appearing in injected resonators [18].Recent works have addressed the question whether the concept of LS can be extended to the time domain [19][20][21][22] in the case of optically injected cavities. Here we propose to answer to this question considering a phase invariant system, namely a passively mode-locked laser. Passive mode locking (PML) is an elegant method leading to the emission of pulses much shorter than the cavity round-trip. It is achieved by combining two elements, a laser amplifier providing gain and a nonlinear loss element, usually a saturable absorber (SA). The different dynamical properties of the SA and of the gain create a window for regeneration only around the pulse. PML can be successfully described via the seminal Haus' master equation, which combines the nonlinear Schrödinger equation with dynamical nonlinear gain and losses [23]. In fiber or Ti:sapphire lasers [24], for which the gain and the absorption are respectively much slower and faster than all the other variables, the Haus equation can be approximated by the subcritical cubic-quintic complex Ginzburg-Landau equation where one replaces for simplicity the slowly evolving net gain-which has a typical time scale of Γ −1 ¼ 10 ms in doped fibers-by a constant. This constant must be determined self-consistently as it depends on the number of PML pulses per round-trip, which may be one (fundamental PML) or N h (N-th order harmonic PML). The stability of these different emission states is described by the so-called background stability criterion of PML [25], whic...

We report on fabrication and characterization of single-longitudinal-and transverse-mode semiconductor ring lasers. A bifurcation from bidirectional stable operation to a regime with alternate oscillations of the counterpropagating modes was observed experimentally and is theoretically explained through a two-mode model. Analytical expressions for the onset and the frequency of the oscillations are derived, and L I curves numerically evaluated. Good quantitative agreement between theory and measurements made over a large number of tested devices is obtained.