1. Introduction

Yb³⁺-doped materials have several advantages as gain media for high-power femtosecond laser operation. Their absorption bands and broad emission bands enable femtosecond laser operation with direct laser diode (LD) pumping. The simple energy-level scheme of the Yb³⁺ ion (² F _5/2↔² F _7/2 inter-manifold transition) leads to a very small quantum defect and avoids undesirable processes such as excited-state absorption, cross relaxation, and concentration quenching [1] so that highly efficient high power laser operation with small heat load can be realized unless the Yb³⁺-doping concentration and the inversion exceeds some critical value [2]. Since the spectroscopic, thermal, and mechanical properties of Yb³⁺-doped materials strongly depend on the host material, femtosecond lasers based on various kinds of Yb³⁺-doped crystalline materials have been reported in the past decade [3–11]. By adequate choice of the host material and good thermal management, Yb³⁺-doped gain media can be pumped by high-brightness, high-power LDs. Sub-100 fs mode-locked oscillators with average powers above 1 W have been obtained with high-intensity pump LDs [6,12]. Mode-locked oscillators with average powers of up to 80 W and pulse energies beyond the 20 µJ level have been reported based on Yb:YAG thin-disk laser [13,14]. An amplifier system with multi-mJ pulse energy and sub-200 fs pulse duration at high repetition rate based on cryogenic cooling has also been reported [15].

Among the Yb³⁺-doped materials, the isotropic sesquioxides Yb³⁺:RE₂O₃ (RE = Y, Sc or Lu) are very attractive as gain media for high-power femtosecond laser operation, because they have high thermal conductivities, broader fluorescence spectra than Yb:YAG [10], and their isotropic structures avoid anisotropy-problems. Recently 141-W average power mode-locked laser operation with 738 fs pulse duration based on an Yb³⁺:Lu₂O₃ single crystal thin-disk laser has been reported [16]. Sub-100 fs pulses generation based on Kerr lens mode-locked (KLM) Yb³⁺-doped sesquioxide ceramic and mixed sesquioxide single crystal lasers have also been reported [12,17–20]. In this paper we report on high-power broad-stripe LD pumped sub-100 fs KLM laser operation with around 1-W average powers based on Yb³⁺:Sc₂O₃ and Yb³⁺:Lu₂O₃ single crystals.

2. Experimental setup

The experiments were carried out with a Z-shaped astigmatically controlled cavity as shown in Fig. 1 , which is almost the same as in our previous reports [12,17–19]. As gain medium we used Yb³⁺:Sc₂O₃ (2.4 mm thickness, C_Yb = 2.5 at.%) or Yb³⁺:Lu₂O₃ (1.9 mm thickness, C_Yb = 3 at.%) single crystals at Brewster angle. The crystals were grown by the heat exchange method and their optical and thermal properties have been described in refs [10,11]. The folding mirrors (M1, M2) have 100-mm radii of curvature (ROC) and are antireflection coated for wavelengths below 980 nm and high reflection (≥99.9%) coated above 1020 nm. The mirrors M4, M5 having the same coating as M1, a half-wave plate (0th order, λ_center of 980 nm) and a broadband polarization beam splitter (PBS, 620~1020 nm) were also used in order to protect the LD from the leaking laser beam through the folding mirror M2. The half-wave plate rotates the polarization of the leaking laser beam including the wavelength below 1020 nm and the PBS reflects it. With this half-wave plate, the pump laser becomes s-polarized at the gain medium and therefore it has about 31% reflection loss at the Brewster angle. For stable KLM laser operation, a semiconductor saturable absorber mirror (SESAM, BATOP GmbH, 0.3% non saturable loss, 0.4% saturable loss, 120 µJ/cm² saturation fluence, 500 fs recovery time) was used. The laser beam was focused onto the SESAM with a concave mirror M3 (ROC = 400 mm). An SF10 Brewster prism pair (P) was inserted into the resonator to achieve soliton-like mode locked operation. The distance between the prisms was 70 cm for Yb³⁺:Sc₂O₃ and near 60 cm for Yb³⁺:Lu₂O₃. A wedged-plane mirror with 5% transmission was used as output coupler (OC). The pump source was an 8 W broad-stripe LD (Bookham inc, emission area of 1 x 90 µm (sagittal × tangential), λ_center ~975 nm, Δλ ~4 nm), which has good beam quality only in the sagittal plane (beam-quality values M²_sagittal ≅ 1, M²_tangential ≅ 25). The pump beam was focused into the gain medium to an 1/e ² diameter of ~22 × 110 µm (in air). Due to the poor beam quality of the pump beam, the focusing diameter of pump beam at the tangential plane was restricted and the soft aperture Kerr-lens effect (better mode matching between pump mode area and laser mode area by a self-focusing effect [21,22]) is not applicable in the tangential plane. In fact the self-focusing effect would make worse mode matching in the tangential plane and disturb ultrashort pulse generation. To obtain KLM laser operation with the broad-strip LD pumping, we introduce a large Kerr-lens effect only in the sagittal plane and suppress it in the tangential plane. It is well known that the Kerr-lens effect can be enhanced in case of nearly unstable cavity alignment. Therefore we made the cavity nearly unstable only in the sagittal plane and stable in the tangential plane by adjusting the astigmatism. The distance between the concave mirrors (M1, M2) was shortened to make the cavity nearly unstable and the angles of the concave mirrors were adjusted to introduce astigmatism for KLM. The position of the gain medium was also moved (~1 mm) to the right side from the center position of the concave mirrors (M1, M2) and therefore the focusing point of the cavity and positon of the gain material were not the same in the KLM cavity. The estimated laser mode diameters at the center of the gain medium (about 1 mm shift) were ~64 × 140 µm for pump mode, ~120 × 70 µm for CW in KLM cavity, and ~70 × 70 µm for KLM.

 

Fig. 1 Experimental setup of the SESAM supported Kerr lens mode-locked laser. The PBS and half-wave plate were used in order to protect the LD from the leaking laser beam through the folding mirror M2. The inset shows the pump mode profile at the focusing point (in air).

Download Full Size | PDF

3. Results

Before KLM laser operation, the cavity was optimized for high average power operation (stable in the sagittal and the tangential planes). Average powers of above 1 W were obtained from each Yb³⁺:RE₂O₃ laser (Fig. 2(a) ). Next we aligned the cavity for KLM laser operation (nearly unstable in the sagittal plane and stable in the tangential plane) at the maximum pumping power level (Fig. 2(b), ~5.2 W pump power). The KLM did not start automatically in the experiments and therefore we slightly moved the concave mirror M1 to the right side (Fig. 1) to initiate KLM. An 81-fs pulse duration with an average power of 840 mW and a spectral bandwidth of 18.3 nm was obtained from the Yb³⁺:Sc₂O₃ single crystal (Fig. 3(a) -3(b)). The center wavelength was 1045 nm. The measured time-bandwidth product was 0.407 and the repetition rate was ~78 MHz. A 75-fs pulse duration with an average power of 860 mW and a spectral bandwidth of 19.4 nm was obtained from the Yb³⁺:Lu₂O₃ single crystal (Fig. 3(c)-3(d)). The center wavelength was 1037 nm. The measured time-bandwidth product and the repetition rate were 0.403 and ~85 MHz, respectively. The spectral bandwidths of pulses became ~1.5 times broader than their gain bandwidths (11.6 nm for Yb:Sc₂O₃ and 13 nm for Yb:Lu₂O₃). In order to achieve CW suppression [23,24] under our experimental conditions, at least (ignoring non saturable loss) 6% loss modulation is required. However, the SESAM used in the experiment had only 0.4% modulation depth, which is ~15 times smaller than the estimated requirement and therefore we believe that mode locking was mainly sustained by KLM. The pulse trains did not show Q-switching modulation. The maximum incident pump power was about 5.2 W (corrected for 31% reflection loss) in both cases. The optical-to-optical efficiency against the incident pump power was ~16%. Much higher efficiencies could be expected with better mode matching of pump and laser mode. The proper distance between the folding mirrors (M1, M2) to obtain the KLM had to be aligned with an accuracy of only several hundred µm and varied with the pump power. Once the KLM stopped, the average powers strongly decreased (Fig. 2(b)), because the cavity became nearly unstable without a large Kerr-lens effect. The threshold pump power also became twice as large as that for optimized average power (Fig. 2(a)). After the KLM stopped, a significant change of the laser mode profile due to the disappearance of the large Kerr-lens effect was observed (Fig. 4 ). Similar phenomena were also observed in our previous sub-100 fs KLM lasers with ceramic gain materials [12,17–19]. Due to an increase of the coherence length, Fig. 4(b) (after the KLM stopped) shows interference fringes caused by reflection from a neutral density filter placed in front of the CCD camera (~2 mm thickness).We also monitored the long-term KLM stability by a 250 kHz photodiode. By shielding the cavity from air turbulences, the KLM laser operation was stable for longer than 2 hours (Fig. 5(a) ). The small long-term fluctuation seen in Fig. 5(a) was caused by temperature change in the laboratory (period of ~90 min). Without shielding, some instabilities were observed (Fig. 5(b)). At the instability the KLM stopped and immediately restarted automatically. Even though the KLM did not show self-starting with the SESAM, we suppose the SESAM is useful for initiating the KLM and for this automatic restarting. We also performed the experiment without the half-wave plate. The pump laser becomes then p-polarized at the gain medium and therefore its reflection loss was almost disappearing at the Brewster angle. In this case 71-fs pulse duration with an average power of 1.09 W and a spectral bandwidth of 21.3 nm were obtained from the Yb³⁺:Lu₂O₃ single crystal (Fig. 6 ). The observed pulse duration was longer than the transform limited pulse width (~55 fs). This was probably caused by a spatially chirp due to the prism pair in front of the OC and uncompensated high order dispersion. An external compensation would enable transform limited pulse duration [9]. Further pulse shortening by decreasing the negative dispersion of the cavity was limited by the growth of a narrow CW component in the spectra and/or an instant instability [23]. While we could obtain the shorter pulse duration and higher average power without the half-wave plate, we suffered damage of the LD in the experiment. The destruction was observed at the occurrence of instabilities, which also occurred during the optimization of KLM laser operation. Even with the sharp short wavelength pass filters (M4,M5), we have also observed such destructions of LDs many times even in our previous experiments with ceramic materials. This is the reason why we used the additional half-wave plate and the PBS to protect the LD in the experiment. The reason for this damage has not yet been clarified, but we did not observe the destruction of LD in the experiment with the half-wave plate and the PBS.

 

Fig. 2 Average power versus incident pump power. (a) Optimized for high average powers. The nonlinearlities in the curves were caused by the dependence of pumping wavelength on the pump power. (b) Optimized for short pulse durations (green circle). The average power shows steep decreasing after KLM stopped.

Download Full Size | PDF

 

Fig. 3 Autocorrelation traces and spectra are shown. (a)(b) 81 fs pulses from the Yb³⁺:Sc₂O₃ crystal with the average power of 840 mW. (c)(d) 75 fs pulses from the Yb³⁺:Lu₂O₃ crystal with the average power of 860 mW. In the autocorrelation traces, the experimental data (points) and sech²-fitting curve (solid curve) are shown.

Download Full Size | PDF

 

Fig. 4 Mode profiles of the laser beam outputs in case of the Yb³⁺:Lu₂O₃ single crystal laser. (a) During KLM. (b) After the KLM stopped.

Download Full Size | PDF

 

Fig. 5 Measured long-term mode-locking stability of the Yb³⁺:Sc₂O₃ single crystal laser with a pulse duration of 81 fs and 840mW average power. (a) With shielding the cavity and (b) without shielding the cavity.

Download Full Size | PDF

 

Fig. 6 (a) Autocorrelation trace and (b) spectrum of 71 fs pulses from the Yb³⁺:Lu₂O₃ crystal with an average power of 1.09 W. In the autocorrelation trace, the experimental data (points) and sech²-fitting curve (solid curve) are shown.

Download Full Size | PDF

4. Conclusion

In conclusion we have achieved the generation of 81 fs pulses with 840 mW average power at the center wavelength of 1045nm from an Yb³⁺:Sc₂O₃ single crystal and 71 fs pulses with 1090 mW average power and at the center wavelength of 1037nm from an Yb³⁺:Lu₂O₃ single crystal. The observed pulse durations are about three times shorter than those of previous SESAM mode-locked experiments [10,11]. In usual SESAM mode-locked laser operation, the shortest pulse duration was limited by growth of the mode-locking instabilities and/or transition to a multi-pulsing operation [24,25]. In this paper, we have suppressed such limitation phenomenon by help of a large Kerr lens modulations in comparison to the SESAM effect. To obtain a large Kerr-lens effect with broad-stripe LD pumping, in which the pump beam has a good quality only in the sagittal plane, the cavity was aligned nearly unstable in the sagittal plane and stable in the tangential plane by adjusting the astigmatism. We believe that our high-power broad-stripe LD pumped Kerr-lens mode locked cavity can also be applicable for other Yb³⁺-doped materials enabling the generation of sub-100 fs pulses with high average power. We also believe that sub-100 fs mode-locked laser operation with much higher average power will be possible with Yb³⁺-doped sesquioxide single crystals.