Fig. 1. Optical properties of the PCF used in the experiments. Blue curve—attenuation, solid red curve—group velocity dispersion with 2ZD point at 1510 nm and a negative dispersion slope. For comparison the GVD of a regular telecom fiber is shown by the dashed curve. Inset—scanning electron microscope image of the PCF; the central guiding core diameter is about 1.2µm.

Fig. 2. Experimental power-dependent X-FROG spectrograms at the output of a 130 cm-long PCF with λ _2ZD=1510 nm. The pump wavelength is 1510 nm, i.e. at the 2ZD. Formation of a soliton-“hole” pair is observed above 10 mW input power. The onset of resonant energy transfer into the spectral band between the soliton and the dispersive wave branch occurs at 41–43 mW. At 45 mW and above the soliton begins to emit Cherenkov radiation at longer wavelengths. The vertical dashed line shows the 2ZD point of the fiber, separating regions of normal and anomalous dispersion. The color scale is logarithmic with the intensity of the SF signal. SF wavelength is related to the fundamental signal and reference pulse wavelengths as λ_SF =λ_sigλ_ref /(λ_sig +λ_ref ). The reference pulse wavelength is equal to the input pulse wavelength, 1510 nm in this case. QuickTime movie, 2.8 MB.

Fig. 3. Modelling of Eq. (2) for the experimental conditions of Fig. 2. X-FROG traces are computed numerically from the complex field amplitude output of Eq. (2) for the parameters of the experiment. The running numbers indicate average input power. The color scale is logarithmic. QuickTime movie 2.6 Mb

Fig. 4. Illustration of the wavevector matching condition (1) for J=1. The full (dashed) lines show the right hand-side (left hand-side) of Eq. (1) as a function of wavelength. Crossings of these lines give the of resonances. (a) illustrates the case when the group velocities of the soliton and the CW-pump are matched, so that the resonance wavelength coincides with the wavelength of theCW-pump and no new frequency is efficiently excited; (b) shows the case when the group velocities of the soliton and the CW-pump become different due to the soliton self-frequency shift. In this case the most efficiently excited and experimentally observed resonance wavelength is located between the soliton and the CW-pump. The excitation efficiency of the resonance marked by a circle in (b) and other resonances, which can be found for J=±1, is not sufficient to be observed under our experimental conditions. The thick grey arrow in (b) indicates the direction of the energy flow observed in numerical modelling.

Fig. 5. Experimental length-dependent X-FROG spectrograms for pumping conditions similar to Fig. 2: The pump wavelength is 1510 nm and the input power is held constant at 55 mW. QuickTime movie, 0.9 MB

Fig. 6. Modelling results of Eq. (2) corresponding to the experimental measurements in Fig. 5. The running numbers indicate propagation distance inside the fiber. QuickTime movie 960 Kb

Fig. 7. Experimental power-dependent X-FROG spectrograms at the output of a 130 cm-long PCF with λ _2ZD =1510nm. The fiber is pumped at 1430 nm, i.e. on the anomalous dispersion side. At lower input powers formation of the primary strong soliton and its Cherenkov radiation are observed. Subsequently, a second weaker soliton is formed which interacts with the continuum emitted by the first soliton. The scattering of the Cherenkov radiation by the second soliton leads to energy transfer into a spectral band located between the solitons and the Cherenkov radiation band. QuickTime movie, 2 MB.

Fig. 8. Numerical X-FROG traces for the power-dependent case of pumping the PCF in the anomalous dispersion region at 1430 nm; the other parameters correspond to those in Fig. 7. The running numbers indicate average input power. QuickTime movie 1.6 Mb