|
Even more than 15 years after their first experimental demonstration, supercontinua have remained an extremely active field of research. In particular, photonic crystal fibers enable the broadening a laser source with a few nanometer initial spectral coverage to a supercontinuum that may encompass several octaves. Unfortunately, this extreme broadening does not come without a caveat: the initially coherent laser light loses most of this favorable property, which makes it impossible to compress the white-light pulse train to the bandwidth-limited single-cycle duration that two octaves could theoretically support. More precisely, this problem arises because of a loss of interpulse coherence, i.e., subsequent white-light pulses exhibit dramatically varying amplitude and phase structures. In fact, these variations are so extreme that they do not even follow a Gaussian distribution anymore, which gives rise to the phenomenon of optical rogue waves.
Optical rogue waves are often explained by soliton dynamics. In the light of the above considerations, however, this seems quite paradoxical as solitons are coherent waveforms that evolve in a highly deterministic fashion. Moreover, frequency metrology applications of supercontinua exist which do not seem to be corrupted by a loss of interpulse coherence. In order to resolve this apparent conflict, we propose a new intrapulse coherence definition, which is experimentally verified by fringe contrast measurements in an f-to-2f interferometer. Numerical simulations indicate that intrapulse coherence is typically quite robust in the case of Kerr-dominated spectral broadening whereas it also quickly vanishes in plasma-dominated broadening scenarios, e.g., during filamentation. Interpulse coherence, in contrast, becomes more fragile at the low photon numbers of oscillator sources. As these two types of coherence appear rather independent of each other, situations can arise where interpulse coherence is conserved but intrapulse coherence vanishes and vice versa. Moreover, compressibility into a train of short pulses must not be used to conclude on the robustness of intrapulse coherence.
We believe that this new coherence criterion has important implications, both for frequency metrology as well as for carrier-envelope phase stabilization of lasers. Specifically, in frequency metrology, limitations may arise for the obtainable maximum precision of optical frequency measurements. These limitations may impact the redefinition of the fundamental time and length units replacing the current microwave cesium standard by an optical one.
|
