Controlling the duration of XUV high order harmonic pulses

Abstract

We modulate temporally the polarization of a - 30 fs, SO0 nm IR pulse and use it to generate high order harmonics. The harmonic emission can clearly be confined and the XUV pulse duration can be continuously tuned from - 5 - 7 fs to 50 fs. There are curently two ways to generate a single sub-femtosecond pulse via high order harmonic emission (HHG). The first one is to use ultrashort linearly polarized pulses and to confine the harmonic emission to that of the cutoff harmonics. The second one, valid for plateau harmonics also, is to use a relatively long pulse (-15 fs) and to temporally modulate its polarization. By performing two feasibility experiments, we show that by modulating the polarization of a (T = 30 - 35 fs) long pulse, one can continuously control the harmonic pulse duration. The technique used for temporally modulating the ellipticity of a short pulse is simple and robust. By using two quartz plates, we can transform a pulse (of duration T) into a flat top pulse (duration 2 T) which is linearly polarized at the center (t=O) ofthe pulse and elliptically polarized at the begining and at the end of the pulse. The HHG is extremely sensitive to the ellipticity of the fundamental pulse and typically a 10% ellipticity reduces the efficiency by a factor 2. By temporally modulating the ellipticity of the fundamental pu1se;we therefore create a where the ellipticity is smaller than 10% inside which the harmonic emission is confined. In our experimental conditions, the minimum gate width is T / 6 (narrow gate). It can also be increased up to -2s (large gate) by rotating one of the plates. The high order harmonic pulse duration was estimated in two ways. In a first experiment (performed at the CELIA laboratory), the harmonic spectra were recorded and showed a clear dependance on the gate width. For the cutoff harmonics, the spectra broadened as the gate width was decreased as expected when a temporal confinment occurs. In contrast, for the plateau harmonics, the spectral width decreased with the gate width. Also counterintuitive, this is also consistent with a confinement because of the importance of the intensity dependent atomic,dipole phase. In both cases, the spectra were consistent with a confinement of the harmonic emission down to 5-7 fs. We could also observe that a temporal confinment does not always reduce the HHG efficiency and allows to optimize the phase matching. In a second experiment (performed at the Lund Laser Center), we measured directly the duration of harmonics created in Argon by doing a cross correlation of the harmonic pulse with an ultrashort (7-10 fs) 800 nin pulse. The cross correlation signal was the photoelectron peak corresponding to absorption of one harmonic photon plus absorption (or emission) of one photon of the ultrashort probe pulse. Because of a non-collinear geometry, the resolution of this measurement is larger than 10 fs but was sufficient to clearly observe an evolution of the harmonic pulse profile. For instance for the sideband 18 shown on the figure, the duration was 57 fs in the large gate configuration, 43 fs without any gate and 26 fs in the narrow gate situation. The confinment of HHG in the narrow gate configuration is therefore clearly visible, even for the plateau harmonics considered bere.