Single-shot measurements of the ultrabroadband continuum from microstructure fiber

Summary form only given. It has recently become possible to generate ultrabroadband continuum from 400 to 1600 nm with a TEM/sub 00/ spatial mode. This involves simply allowing sub-nanojoule 800 nm pulses to propagate through a short length of newly developed microstructure optical fiber, which can be nearly dispersion-free in the visible. We have made cross-correlation-FROG (XFROG) measurements of the continuum. These measurements nicely reveal the parabolic group delay of the pulse. However, the retrieved spectrum from our XFROG measurements has much more structure than the directly measured spectrum. Also, we observe fine-scale structure in the retrieved trace that is not seen in the measured trace, a clear indication that something is amiss. Usually, such poor agreement indicates that the more difficult measurement -the XFROG measurement - is contaminated with systematic error. A possible explanation for these observations is that variations in the continuum pulse from shot-to-shot, as predicted by theory, are washing out the structure in directly measured spectra and in measured XFROG traces. That the structure information somehow remains in the multishot XFROG trace is possible because some single-shot information is preserved in the XFROG trace, as all FROG traces contain substantial redundancy, and information washed out in one domain may remain in the other. Specifically, fine-scale frequency information is also present in the trace in the form of slow oscillations in delay, which are less likely to be washed out. To verify this hypothesis, we performed the first single-shot spectral measurements of the microstructure-fiber continuum.