Dispersion of femtosecond pulses in vacuum beam lines from ambient pressure down to 0.01 mbar

Complete test of publication follows. In chirped pulse amplification laser systems the laser beam propagates many dozens of meters in air, while the compressed short and high field laser pulses may also travel more tens of meters from the compressor to the target in evaluated beam pipes. The possible dispersion of residual air in the beam pipes may cause undesirable lengthening of the laser pulses just prior to the target. This effect is even more severe for few cycle laser pulses. Thus, the accurate knowledge of air dispersion depending on pressure is of high importance. Although in normal laboratory conditions the modified Edlen's dispersion form can be used, but its pressure dependence was deducted from relatively small variations around atmospheric. In this paper we experimentally prove that the pressure dependent part of Edlen's dispersion form is valid for air pressure down to 0.01 mbar. The main experimental apparatus was a spectrally and spatially resolved interferometer, which is basically a Mach-Zehnder interferometer equipped with an imaging spectrograph. It was illuminated by 800 nm, 18 fs pulses from a Ti:S oscillator working at 71 MHz repetition rate. During the measurements the air conditioning system fixed the room temperature and the relative humidity at 23±1 deg C and 35±2 %, respectively. To test our method, first the dispersion of laboratory air at ambient pressure was measured. We have found the specific group delay dispersion (GDD) to be 20.9±1.0 fs2/m, which is extremely close to the theoretical value 20.7±1.2 fs2/m. For the specific third order dispersion (TOD) our measurement gave -10±10 fs3/m, while the calculated value is 10±3 fs3/m. Having convicted about the accuracy of the method, the pressure dependence of air dispersion was measured by varying the pressure in a long vacuum tube between 1 bar and 0.01 mbar. To make the method more sensitive, the total spectral phase shift has been also determined from the measurement. As a result, the total phase shift gradually approached zero, that is, to the expected value at perfect vacuum state, and the dependence of GDD on pressure was proved to follow Edlen's form within the error of the measurement. Besides of dispersion compensation, the knowledge of the pressure where the dispersion of air diminishes could also result in less tight vacuum requirement, which can lead to a cost effective beam pipe design and also to shorter maintenance time (faster airing and evacuating). For instance, 20% lengthening of 10 fs, 5 fs and 1 fs pulses upon propagation of 50 m long beam pipes can happen at pressures of 23 mbar, 5.7 mbar, and 0.23 mbar, respectively. This work was supported by OTKA under grant T047078 and NKFP 1/00007/2005.