1. Introduction

The recent progress in the development of coherent extreme ultraviolet (EUV) source provided new opportunities in spectroscopy. Especially, sources based on high harmonic generation (HHG) deliver line and/or continuum radiation up to few keV [1,2] with excellent spatial characteristic and repetition rate up to few kHz. These sources have been successfully applied for x-ray absorption spectroscopy [3,4] with sub-20 fs temporal resolution. However, for time-resolved photoelectron spectroscopy there is a need for pulsed EUV sources with MHz repetition rate. The intensity achieved at the output of ultrafast Ti:sapphire lasers is too low for high harmonic generation. Up to now, two approaches for MHz repetition rate EUV sources have been implemented. One of them is based on a laser oscillators combined with an external enhancement cavity to increase the pulse energy sufficient for HHG [5–10]. With this method harmonics up to 19th order [7] (30 eV) have been generated. The other method is based on conventional Ti:sapphire oscillators and plasmonic field enhancement with bowtie [11] and tapered waveguide [12] arrays. With this approach, harmonics up to 43th orders have been generated.

In this paper we demonstrate a new approach, namely the intracavity high harmonic generation in a Ti:sapphire oscillator with repetition rate of 9.4 MHz.

2. Experimental setup

The experimental setup is presented in Fig. 1 . The oscillator is pumped with a frequency-doubled Nd:YVO₄ laser (Coherent Verdi V18), which can provide an output power up to 18 W at 512 nm. The Ti:sapphire crystal has been placed into a small vacuum chamber and cooled down to –40 °C with a double stage thermoelectric cooler. The beam diameter inside the crystal is set to 100 µm by using two focusing mirrors with a focal length of 100 mm. In this configuration the 4-mm-long Ti:sapphire crystal is about 10-times shorter than the Rayleigh-length of the IR beam. The short and long arms of the oscillator cavity are as long as 6.5 m and 9.5 m, respectively. The total length of 16 m corresponds to a repetition rate of 9.4 MHz. Contrary to most of the long cavity oscillators reported in the literature [13,14] our setup has been realized without a Herriot-Type telescope in the long arm. Using only plane folding mirrors increases the beam diameter [15] inside the cavity and hence enables a smaller spot size and higher intensity in a second focus. The group delay dispersion (GDD) in the cavity is controlled by four pairs of broadband chirped mirrors (−250 fs²/bounce, Femtolasers GmbH, and −50 fs²/bounce, Layertec). The negative GDD in the short arm is about −800 fs² and −700 fs² in the long arm. The chirped mirrors compensate for the positive GDD of the air, the Ti:sapphire crystal, two BK7 lenses, two 1-mm-thick BK7 windows of the HHG chamber, and two wedges. The wedges are used for fine tuning the overall GDD in the cavity [16]. In our setup, using BK7 based optics was essential for stable mode-locked operation. Despite the reduced high order dispersion we were not able to achieve stable pulses with fused silica based wedges or windows of the HHG chamber.

 

Fig. 1 Experimental setup of the intracavity HHG source.

Download Full Size | PDF

3. Characteristics of the long cavity oscillator

A small fraction (~1%) of the energy of the laser pulses has been coupled out from the oscillator cavity with a thin glass plate. This output beam is sent into a fiber coupled spectrograph (Ocean Optics 4000 + ) or into an autocorrelator. To realize hard-aperture Kerr-lens mode-locking, a slit has been placed before the end mirror in the long arm. By adjusting the width and position of the slit, we are able to realize stable mode-locked operation.

According to the measured spectrum and autocorrelation curve (Fig. 2 ), the pulse duration is about 17 fs in the second focus, which is very close to its transform limited duration of 14 fs calculated from the measured spectrum. The shape of the spectrum is a clear indication of net negative intracavity dispersion and has a width of 80 nm centered around 790 nm. The width of the slit for stable pulsed operation as well as the intracavity power as a function of the pump power has been measured and is shown in Fig. 3 . According to theory the necessary slit width decreases exponentially with intracavity power, and the intracavity power should increase linearly with pump power [17,18]. In the range of interest the predicted scaling laws can be very well observed as shown in Fig. 3. From these measurements, a beam waist sensitivity parameter of −0.65 has been estimated, which is in the range provided by theory [17] and in good agreement with our calculations. The intracavity power has been calculated from the measured leakage through the end mirror close to the slit. The transmission has been calibrated by a comparison with the output power measured at the ~1% output coupler. As shown in Fig. 3(b) the intracavity IR power increases linearly with the pump power and does not show saturation in the entire range of the measurement. For a pump power of 8 W, the intracavity power reaches about 10 W and the measured intracavity slope efficiency is about 1.8. The intracavity power of 10 W corresponds to pulse energies of about 1 µJ. A further increase of the pump power causes instabilities and mode-locking ceases.

 

Fig. 2 (a) The measured spectrum with a FWHM of 80 nm is a clear indication of net-negative dispersion in the resonator. (b) The autocorrelation curve reveals pulse duration of 17 fs (FWHM) which is close to transform limited pulse duration of 14 fs.

Download Full Size | PDF

 

Fig. 3 (a) Measured and theoretically calculated width of the split vs. pump power for stable hard aperture mode locking. (b) The measured intra cavity power vs. pump power shows no indication of saturation.

Download Full Size | PDF

4. EUV source

To realize intracavity HHG, a second focus has been implemented in the long cavity arm using two BK7 lenses with focal length of f_HHG = 100 mm (Fig. 1). The GDD in the oscillator cavity, in the short arm and in the two parts of the long arm at the sides of the second focus, has been balanced to have a short compressed pulse in the second focus propagating from left to right and a positively chirped one propagating from right to left. This supports an optimal pulse for HHG in the configuration of Fig. 1 and avoids nonlinear effect for the pulse propagating backward. A small vacuum chamber is placed between the two lenses containing the gas jet and 2-µm-thick Nitrocellulose pellicle output coupler for EUV beam. The intensity in the second focus can be easily estimated assuming a spot size given by the image of the slit onto the focal plane. For our setup the focus diameter is given by 2w_j ≈(f_HHG/l_s)d_s, where l_s = 2.8 m the distance between the slit and the lens and ds = 0.4 mm is the width of the slit. These parameters suggest a beam waist of less than 7 µm. Together with the measured intracavity pulse energy beyond 1 µJ and pulse duration of 17 fs, we can estimate a peak intensity above 1x1014 W/cm². This intensity is sufficient for HHG in Xenon gas with a cutoff energy above 30 eV.

Into the focus we put a 1-mm-long Xenon jet and adjusted the backing pressure to 25 mbar. The EUV radiation is coupled out with the 2-µm-thick Nitrocellulose foil. It is placed in Brewster angle (56.3°) minimizing the loss for the laser. However, due to different index of refraction in the EUV, the foil reflects 7-8% (Fig. 4(a) ) [19] in a broad EUV spectral range. After reflection on the pellicle the EUV beam passes through a 100-nm-thick Silicon foil, which has two tasks. Firstly, it separates the vacuum of the generation chamber from the detector which is an EUV sensitive photomultiplier (Channeltron 4751G). For safe operation the detector requires a vacuum better than 2x10⁻⁴ mbar. Secondly, the Silicon foil suppresses the laser radiation and attenuates the EUV radiation (Fig. 4(a)) ensuring that less than one photon per laser pulse reach the detector. The low photon flux is essential for characterizing the EUV radiation in a single photon detection mode. The minimum photon energy for generating secondary electrons in the detector is W_F = 10 eV [20] corresponding to the 7th harmonic. Photons with higher energies produce current peaks with heights proportional to the photon energies. This measurement method is called energy dispersive x-ray (EDX) spectroscopy. The generated current peaks have been measured with an oscilloscope in a peak detection mode and the histogram of the measured peak values delivers direct information about the energy distribution of the detected incident EUV photons.

 

Fig. 4 (a) Transmission of the 25 mbar Xenon gas in the 1-mm-long jet and of the 100-nm-thick Silicon foil together with the reflection of the Nitrocellulose EUV output coupler. (b) Probability for generating a current pulse corresponding to certain photon energies. The count rate is shown for the oscillator operated in cw mode (red dotted line) and when it was mode-locked (blue line). In the cw mode we expect no EUV photons, so the signal represents the background. (c) Signal after subtracting the background and the estimated out-coupled conversion efficiency.

Download Full Size | PDF

A typical measured histogram is shown in Fig. 4(b) for accumulating data over 90 seconds. To distinguish between noise and signal we repeated the measurement with the oscillator operated in cw and pulsed mode, respectively. In the cw mode, we measured a continuous and smooth background signal. For the mode-locked oscillator distinct peaks appeared in the histogram and are attributed to EUV radiation. These peaks can be better seen if the background is subtracted (Fig. 4(c)). From the measured curve, it can be seen that the highest energy EUV photons have been detected between 30 and 40 eV. This measured cutoff energy is in very good agreement with the cutoff predicted from the estimated intracavity laser intensity. The estimated out-coupled conversion efficiency has been also plotted in Fig. 4(c). For the given numbers we have corrected the measured photon flux taking into account the absorption losses caused by the background pressure in the vacuum chamber and the absorption of the Silicon foil, as well as the detector efficiency.

5. Summary

We have realized for the first time, an intracavity high harmonic source relying on a long cavity Ti:sapphire oscillator operated in the net negative dispersion regime providing ultrashort intracavity pulses. The generated high harmonic spectrum extends beyond 30 eV, which is in good agreement with the estimated cutoff energy from the intracavity peak intensity. Our oscillator-based approach is very simple and compact and does not require an external enhancement cavity with perfectly matched length [5–10]. However, in our setup we have to similarly solve the problem of coupling out the EUV beam from the cavity. Compared to the plasmonic field enhancement approach [11,12], the described source permits a much longer interaction length, which is a prerequisite for a higher conversion efficiency. Further optimization of the cavity dispersion predicts the possibility of using higher pump power and reaching higher intracavity pulse energy to extend the cutoff energy even further.