The measurement of the autocorrelation trace by two-photon fluorescence allows our experiment to determine the width of the short laser pulse which has passed through a four-pass amplifier after being Q-switched and mode-locked(QSML) in the Nd:YLF master oscillator. The measured pulsewidth with the use of TPF(Two-Photon Fluorescence) method was compared with that by SHG(Second Harmonic Generation) method.
TPF technique has an advantage of measuring second-order autocorrelation function completely with single shot. Experimental method and apparatus appears to be easy but TPF yield shows great sensitivity to alignment, concentration of dye(Rhodamine 6G), beam intensity and polarization. Especially, slight variation of alignment and beam intensity arouses big difference in contrast ratio. TPF pattern exhibits pulsewidth and contrast ratio implying the extent of mode locking and energy fraction of a pulse. Experimental results show that pulsewidth of 44.3ps and contrast ratio of 2.88 were obtained and energy fraction of a single pulse $E_p/E_total$ was found to be 0.62.
Pulse-selecting position in the QSML pulse train gives rise to very different type of TPF pattern which becomes similar to pulse shape when a pulse is selected at the front of the train, and displays substructure and many sharp spikes as the pulse selected from the trailing part of the pulse train. The result like this is attributed to the spectral broadening arising from the self-phase modulation of the intense light pulses in the middle part of the train due to the nonlinear refractive index of the laser rod, accompanying the conversion of FM to AM for strongly frequency -modulated pulses. From the observed TPF pattern it can be presumed that the single pulse extracted in the trailing part of the QSML pulse train generated by a Nd:YLF laser is composed of various spikes showing more or less irregular intensity fluctuations of picosecond duration.
Another pulsewidth-measuring experiment was performed for the comparison of the TPF results by means of noncollinear type I SHG method. The pulse duration obtained using SHG was 46.6ps and pulse shape was determined as $sech^2$ type by fitting the SHG trace. In consideration of amplification of the input pulse by the amplifier, a simulation provides the calculated width of the output pulse after passing the four-pass amplifier, yielding 45.8ps duration. The pulsewidth(44.3ps) observed by TPF technique was found to be slightly shorter than that(46.6ps) by SHG method, which seems because the amplification by the four-pass amplifier and Q-switching have an effect on the pulse duration.