High-power femtosecond oscillator generates extreme-UV source

The development of femtosecond (fs) lasers offered new possibilities for time-resolved spectroscopy. It became possible to monitor the nuclear motion of molecules, crystal lattices, and other out-of-equilibrium structures. Photoelectron or x-ray absorption spectroscopy can deliver direct structural information. For these applications, extreme-UV (XUV) sources must provide continuum radiation, sufficient photon flux, and perfectly synchronized ultrashort laser pulses. A promising way to meet all of these requirements is by high-harmonic generation (HHG). Focusing an ultrashort laser pulse into a rare gas results in higher-order harmonics of the fundamental frequency. The number of XUV photons per pulse is relatively small, but they are delivered in a spatially coherent beam. However, good signal-to-noise statistics requires the highest possible repetition rate. HHG is a highly nonlinear process and requires laser pulses with peak intensity >10¹³W/cm², ideally at a repetition rate of several MHz.

The invention of the Kerr-lens mode-locked titanium (Ti): sapphire laser oscillator has benefited a wide range of applications in ultrafast optics. Owing to the availability of high-reflectivity chirped mirrors with tailored dispersion over unprecedented bandwidths, sub-10fs pulses can now be generated with compact, mirror-dispersion-controlled oscillators. However, the output energy of conventional oscillators is limited to a few tens of nJ, which is insufficient for HHG. Recent experiments have resulted in an enhancement of the pulse energy (achieved by placing a telescope inside the oscillator) while reducing the repetition rate to a few MHz. The alternative approach to high peak power uses a passive enhancement cavity. Experimental results have been shown to generate a spectrum up to the 15^th harmonic order (50nm).^1,2

Contrary to previous approaches, we want to realize intracavity HHG, which requires ultrashort pulses inside the cavity. Recently developed long-cavity oscillators operate in the net-positive-dispersion regime, resulting in strongly chirped pulses inside the cavity.³ However, for intracavity HHG, the oscillator must operate in the net-negative-dispersion regime. Figure 1 shows our laser setup. The Ti:sapphire crystal is pumped with up to 18W of green power. To minimize thermal aberrations, it is cooled to −40^°C and put in a small vacuum chamber to avoid ice formation. The cavity is formed by chirped mirrors with a focal length of 7.5m, corresponding to a repetition rate of 20MHz. With two additional curved mirrors, we now have a second focus inside the cavity, allowing nonlinear HHG action to occur.

Figure 1. Typical setup of a 20MHz titanium:sapphire oscillator with a second focal region for an extreme-UV (XUV) source.

When pumping with more than 3W, we can obtain mode-locked operation. The intracavity energy (see Figure 2) increases linearly with pump power. Using a slightly different setup, we obtained for 8W pump power an intracavity energy of 260nJ. Our laser operates in the net-negative-dispersion regime, implying generation of ultrashort pulses inside the resonator. Considering the estimated pulse duration of 100fs, the intensity is sufficient to generate high-order harmonics up to 35eV.

Figure 2. Intracavity pulse energy and corresponding output spectra of the 20MHz oscillator as a function of pump power. a.u.: Arbitrary units.

Before we can measure the XUV emission, we have to solve the problem of XUV-radiation outcoupling, which propagates collinearly with the fundamental laser pulses. Because of high XUV absorption, methods known to work for lower-order harmonics are not suitable. Figure 1 shows a few options to deal with this problem. The simplest solution involves drilling a hole into one of the focusing mirrors, but this limits the output power, degrades the laser-beam profile, and can hinder mode-locked operation. Applying two crossed beams—see Figure 2(c)—seems an elegant solution, but it has not yet been demonstrated in the XUV regime.⁴ The most promising approach is through outcoupling with a thin, transparent plate. The latter is placed into the resonator under the Brewster angle to minimize losses at the fundamental wavelength. Figure 3 shows XUV reflectivities of a few suitable materials. For diamond and gallium phosphide, this can be on the order of 50% over a broad spectral range. The XUV-radiation yield can also be enhanced and spatially separated from the fundamental laser beam by placing a sample with a grating-like structure into the focal region.⁵

Figure 3. Reflectivity of (a) sapphire, (b) quartz, (c) diamond, and (d) gallium phosphide in the XUV regime, assuming an angle of incidence corresponding to the Brewster angle for the fundamental laser wavelength.

In summary, we have demonstrated a long-cavity Ti:sapphire oscillator operating in the net-negative-dispersion regime, providing ultrashort pulses without an additional compressor. The estimated intensity is suitable for generating XUV radiation up to 35eV. We will next add a small vacuum chamber into the resonator that contains a gas jet and two additional focusing mirrors to generate and characterize the pulses.

This study has been sponsored by German Research Council (DFG) grants SP 687/1-3, SE 1911/1-1, TKM B154-09030, and B 715-08008, and Friedrich Schiller University grant ‘ProChance 2009 A1.’