Fig. 1. Calculated (a) GVD and A_eff , and (b) birefringence of the PCF. Inset of (b) shows the SEM image of the PCF.

Fig. 2. Experiment setup. TSL: Ti:sapphire laser, IS: Isolator, P1,P2: Prism pair, M1~M3: Mirrors, H1~H3: Half-wave plates, A1,A2: Polarizers, FM: Flip mirror, PM: Power meter, AC: Autocorrelator, PCF: Photonic crystal fiber, L1,L2:Microscope objectives, Q: Quarter-wave plate, OSA: Optical spectrum analyzer.

Fig. 3. SC generated with input pulses of different powers: (a) mean spectrum, (b) mean ellipticity and (c) polarization correlation. The power values shown in (a) are the average input powers just before the objective L1. λ ₀=800 nm, pulse FWHM ~120 fs. The input pulses are linearly polarized at θ=45° to the slow axis of the PCF. In (a) black (red) lines indicate slow (fast) axis components. The bottom row of the figure plots the numerical simulation results for the 300mW-input case, assuming a coupling efficiency of 40% into the PCF, pulse width FWHM=120 fs and quantum noise in input pulses.

Fig. 4. SC generated with input pulses of different central wavelengths: (a) mean spectrum, (b) mean ellipticity and (c) polarization correlation. The input pulses are linearly polarized at θ=45° to the slow axis of the PCF. The average input power just before the objective L1 is 150 mW. In (a) black (red) lines indicate slow (fast) axis components.

Fig. 5. SC generated with input pulses polarized at different angles to the slow axis of the fiber: (a) mean spectrum, (b) mean ellipticity and (c) polarization correlation. λ ₀=800 nm. The average input power just before the objective L1 is 200 mW. In (a) black (red) lines indicate slow (fast) axis components.