Abstract
The attosecond streaking technique is used for the characterization of isolated extreme ultraviolet (XUV) attosecond pulses. This type of measurement suffers from low photoelectron counts in the streaked spectrogram, and is thus susceptible to shot noise. For the retrieval of few-or mono-cycle attosecond pulses, high-intensity streaking laser fields are required, which cause the energy spectrum of above-threshold ionized (ATI) electrons to overlap with that of the streaked photoelectrons. It is found by using the principal component generalized projections algorithm that the XUV attosecond pulse can accurately be retrieved for simulated and experimental spectrograms with a peak value of 50 or more photoelectron counts. Also, the minimum streaking intensity is found to be more than 50 times smaller than that required by the classical streaking camera for retrieval of pulses with a spectral bandwidth supporting 90 as transform-limited pulse durations. Furthermore, spatial variation of the streaking laser intensity, collection angle of streaked electrons and time delay jitter between the XUV pulse and streaking field can degrade the quality of the streaked spectrogram. We find that even when the XUV and streaking laser focal spots are comparable in size, the streaking electrons are collected from a 4 pi solid angle, or the delay fluctuates by more than the attosecond pulse duration, the attosecond pulses can still be accurately retrieved. In order to explain the insusceptibility of the streaked spectrogram to these factors, the linearity of the streaked spectrogram with respect to the streaking field is derived under the saddle point approximation.