In Ref. [1] we described the role of the higher zero-dispersion wavelength (ZDW) when generating a supercontinuum in a photonic crystal fiber (PCF) with two ZDWs. It was, e.g., shown that the higher ZDW limits how far a soliton can red-shift, before transferring energy to a dispersive wave in the normal dispersion region. For one of the investigated PCF designs (pitch Λ = 1.2 μm, relative air-hole diameter d/Λ = 0.62) it was found that the energy transferred to ~ 1650 nm in the normal dispersion region did not broaden temporally, but instead seemed to form a bright-bright soliton-pair with a co-propagating soliton at ~ 1270 nm in the anomalous dispersion region.

We have recently found that there was an error in calculating the dispersion operator used in the split-step Fourier method, even though we used a sufficiently high number of $\overline{\beta }$ -coefficients. For the fiber with Λ = 1.2 μm and d/Λ = 0.62 the error corresponded to a change of the dispersion profile, as shown in Fig. 1.

It is seen that the error caused the second ZDW to shift slightly to a higher wavelength, but much more importantly that a third ZDW occurs at about 1600 nm. Therefore there is anomalous dispersion at 1650 nm. This explains why the energy transferred from the soliton at ~ 1270 nm to ~ 1650 nm does not broaden temporally; it experiences anomalous dispersion and can therefore form a soliton. Both components of the temporally overlapping bright-bright soliton pair are thus experiencing anomalous dispersion, but are separated spectrally by a small region of normal dispersion.

 

Fig. 1. Dispersion profile for PCF with structural parameters Λ = 1.2 μm and d/Λ = 0.62 (blue) and the dispersion profile incorrectly used in Ref. [1] (green, dashed).

Download Full Size | PDF