1. Introduction

Yb³⁺-doped crystals are attractive gain media for diode pumped ultrashort pulse solid state lasers (SSLs). A striking property of the Yb³⁺-doped gain media is that they have small quantum defects and absence of deleterious effects such as excited state absorption, up-conversion, and concentration quenching. In addition, Yb³⁺-doped crystals have normally broadband absorption and emission spectra. The absorption band of the Yb³⁺-doped crystals matches the emission wavelength of the high brightness InGaAs laser diodes, which makes direct diode pumping possible. So far, various diode pumped Yb-doped ultrafast crystal lasers have been demonstrated, such as Yb³⁺: Sc₂O₃ [1], Yb³⁺: YAlO₃ [2], Yb³⁺: Y₃Ga₅O₁₂ [3].

The main drawback of Yb³⁺-doped gain media is that they are a quasi-three level system. In most crystals the Stark splitting of the ²F_7/2 ground manifold of Yb³⁺ is a few hundreds of cm⁻¹. Recently, it was shown that this drawback could be partly circumvented in the Yb-doped oxyorthosilicate crystals, where due to the large crystal field strength, large Yb^{3+ 2}F_7/2 ground manifold splitting could be obtained [4]. In addition, it was also found that not only broadband gains could be obtained in Yb³⁺ doped oxyorthosilicate crystals, but also the crystals could have large thermal conductivity [5], an essential property required for high power lasers. Of the Yb³⁺-doped oxyorthosilicate gain media, Yb³⁺: Y₂SiO₅ (YSO) [6], Yb³⁺: Gd₂SiO₅ (GSO) [7] and Yb³⁺: Lu₂SiO₅ (LSO) [6] femtosecond lasers that generated stable pulses of 122 fs, 343 fs and 233 fs, respectively, have already been demonstrated. Compared to the more prominent Yb³⁺ doped materials such as Yb³⁺: Y₃Al₅O₁₂ (YAG) [8] and Yb:KY(WO₄)₂ (KYW) [9], they have been demonstrated to generate pulses as short as 540 fs and 71 fs respectively. Hence, Yb³⁺-doped oxyorthosilicate gain media have proven themselves as viable alternatives as femtosecond source generators.

Yb: SSO also belongs to the oxyorthosilicate class of laser crystals. As the Sc³⁺ ion has an ionic radius smaller than that of the Yb³⁺, they experience extremely large crystal fields [10]. The segregation coefficient of Yb³⁺ in SSO is about 0.96 [5]. This implies a high solubility of Yb³⁺ in SSO and as such high doping concentrations can be achieved for Yb: SSO. Stark splitting of the ²F_7/2 ground manifold of the Yb³⁺ ions in the crystal reaches a value of 1027 cm⁻¹ [5], which is among the largest in Yb³⁺ doped crystals. This is comparable to Yb: GSO which has a value of 1061 cm⁻¹ [7] and larger than that of Yb: YSO and Yb: LSO which have reported values 700 cm⁻¹ [6] and 960 cm⁻¹ [6] respectively. Figure 1 . shows the fluorescence spectrum of Yb: SSO which has a FWHM bandwidth of 108 nm, making it an attractive gain medium for ultrashort pulse generation. Other Yb³⁺-doped oxyorthosilicate gain mediums such as Yb: YSO, Yb: LSO and Yb:GSO have fluorescence spectrums with FWHM bandwidth of about 40 nm [6], 90 nm [6] and 80 nm [7] respectively. In addition, Yb: SSO has a relatively large upper level lifetime of 1.64 ms [5], which is much larger than the reported values of 0.67 ms [6], 0.95 ms [6] and 1.1 ms [11] for Yb: YSO, Yb: LSO and Yb: GSO, respectively. Yb: SSO also has a large thermal conductivity of 7.5 Wm⁻¹K⁻¹ [5] which is much larger than the reported values of 3.6 W m⁻¹ K⁻¹ [6] and 5.3 W m⁻¹ K⁻¹ [6] for Yb: YSO and Yb: LSO respectively. Hence, Yb: SSO is an attractive gain medium for high power laser operation. In this paper the authors shall report on the cw and passive mode-locking performances of Yb: SSO and demonstrate that the Yb: SSO is a promising gain medium for both cw and ultra-short pulse laser operation.

 

Fig. 1 Fluorescence spectrum of Yb: SSO.

Download Full Size | PDF

2. Experiment

A schematic of the Yb: SSO laser is shown in Fig. 2 . The Yb: SSO crystal used is b-cut and has dimensions of 2.2 × 3 × 3 mm³ and an Yb³⁺ doping concentration of 5 at. %. The crystal was grown by the Czochralski method and had high transmission coatings at both the pump and laser wavelengths to minimize the Fresnel reflection losses. The Yb: SSO crystal was wrapped in indium foil and placed in a water-cooled copper house, which was maintained at 18 °C throughout the experiment. The crystal was titled to about 6° with respect to the principal axis of the cavity to suppress etalon effects. This also improves the stability of the mode-locking operation. A single emitter diode with central wavelength at (975 ± 3) nm was used as the pump source. The crystal absorbed about 90% of the pump light incident to it. As segregation coefficient of Yb³⁺ in SSO is 0.96 [5] (relatively high), and concentration quenching is absent in Yb³⁺-doped gain media, the doping level of the crystal can be further increased, resulting in larger absorption. When M4 was changed to a high reflection (HR) mirror at the lasing wavelength and with the prism pair removed, the Yb: SSO laser emitted cw radiation. When M4 was replaced by a SESAM and a pair of SF10 prisms was inserted into the cavity, mode-locked pulses were obtained. The positive group velocity dispersion (GVD) introduced by the Yb: SSO crystal was compensated by the prism pair with a tip to tip, L_prism, separation of 41.5 cm. The prism pair was aligned to Brewster’s angle to minimize the cavity losses. It was estimated that the total negative GVD provided by the prism pair was about – 800 fs². The total positive dispersion was estimated to be no more than 500 fs². ABCD matrices analysis showed that the radius of the spot size in the crystal and on the SESAM was 50 μm and 30 μm, respectively. SESAMs from BATOP GmbH were used to initiate the mode locking. SESAMs with central wavelengths at 1040 nm (modulation depth, ΔR = 0.5%, nonsaturable absorption, A = 0.5%, relaxation time, τ = 500 fs and saturation fluence, Φ_sat = 90 μJ cm⁻²), and 1064 nm (ΔR = 1.2%, A = 0.8%, τ = 500 fs and Φ_sat = 90 μJ cm⁻²) were used. While the SESAM with center wavelength at 1040 nm initiated mode locking at 1041 nm, the SESAM with center wavelength at 1064 nm triggered mode locking at 1060 nm and 1077 nm. Light leaking out of M3 was coupled into a low noise photodetector (New Focus 1611-FC-AC) and connected to a 1 GHz digital oscilloscope (Tektronix DP0714) to detect the mode locked signals. The optical spectrum was obtained using a high resolution optical spectrum analyzer purchased from Ando (AQ-6315B), while the pulse width was measured using a commercial autocorrelator (APE, PulseCheck). The autocorrelation traces measured in this work were all obtained with the absorbed pump powers maintained at 4.5 W.

 

Fig. 2 A schematic of the experimental setup. F1 and F2 are a pair of plano-convex lens with 75mm focal length. M1, M2 and M3 have ROCs of −100 mm, −300 mm and −100 mm respectively. While M4. M5 and M6 have ROCs of ∞. L1, L2, L3, L4, L5 and L6 are 23.5 cm, 50 cm, 5 cm, 22 cm, 26.5 cm and 68 cm respectively. L_prism was 41.5 cm.

Download Full Size | PDF

3. Cw laser emission of Yb: SSO

With the prism pair removed and M4 and M6 were replaced with a HR mirror and an output coupler with 5% transmission at (1040 ± 80) nm respectively, cw laser emission was obtained. The corresponding cw output power versus absorbed pump power curve is shown in Fig. 3 . The laser produced a maximum cw output power of 2.73 W at about 1063 nm with a slope efficiency of about 70%. Due to limited pump power, higher cw emission could not be demonstrated. Previously, the maximum cw efficiency reported for Yb: SSO was 45% [5]. The enhanced performance was most likely due to better mode matching in the cavity used in this work. The cw slope efficiency obtained was comparable to that of Yb: YSO (67%) [12], Yb: LSO (62%) [12] and Yb: GSO (75%) [11]. The results reported in this work indicates that Yb: SSO is a promising cw gain medium.

 

Fig. 3 The cw laser emission of the Yb: SSO with an output coupler of 5% transmission. The insert shows the corresponding optical spectrum at the maximum output power.

Download Full Size | PDF

An interesting point to note was that although the emission cross-section at about 1040 nm was the largest, cw emission around this wavelength was not observed unless a prism was added into the cavity. The reason for this was that there was reabsorption by the Yb³⁺ ions around this wavelength [5]. This additional loss suppressed the cw laser oscillation at the 1040 nm emission peak. The reabsorption cross-section of Yb: SSO at around 1040 nm is about 0.03 × 10⁻²⁰ cm² [5] and diminishes towards longer wavelengths. Compared to Yb: YAG, the reabsorption cross-section is negligible [13] around 1040 nm.

4. Passive mode-locking of Yb: SSO

To initiate mode-locking, M4 was replaced by a SESAM and a pair of SF10 dispersion prisms were inserted into the cavity. L_prism was fixed at 41.5 cm throughout the experiment. M5 was changed to an output coupler with 0.4% transmission for wavelengths of (1040 ± 80) nm. This was done to ensure that the intra-cavity power was high so that the SESAM can be easily saturated and that the shortest pulses could be obtained. Mode-locking was observed at 3 different wavelength bands of 1041 nm, 1060 nm and 1077 nm. The pulse trains for all the 3 wavelength bands had a repetition rate of 90 MHz. This corresponded to the cavity repetition rate, indicating that the fundamental mode-locking had occurred at the 3 different wavelengths.

By using a SESAM whose central wavelength was designed to be around 1040 nm, the authors were able to demonstrate the mode-locking of Yb: SSO at 1041 nm. When mode-locked at this wavelength, the laser had an output power of 40 mW. The corresponding autocorrelation trace is shown in Fig. 5(a) in which a pulse width of 145 fs was measured, assuming a sech² shaped pulse. Its corresponding optical spectrum has a FWHM of 8 nm. The resulting time bandwidth product (TBP) was 0.32, which indicates that the pulses are nearly transform-limited. As the SESAM used has the lowest losses at around 1040 nm, mode-locking at 1041 nm was possible despite the presence of reabsorption at the wavelength.

 

Fig. 5 The autocorrelation traces with their corresponding optical spectrums when Yb: SSO was mode-locked at (a) 1041 nm, (b) 1060 nm and (c) 1077 nm. The blue solid line in the autocorrelation traces are sech² fits.

Download Full Size | PDF

M4 was then replaced by a SESAM whose central wavelength was designed at 1064 nm. It was found that the laser could mode-lock at two wavelength bands of 1060 nm and 1077 nm. The laser initially mode-locked at 1060 nm, producing a maximum output power of 52 mW. At this output power, the pulse width as shown in Fig. 5(b) was 144 fs, assuming a sech² shaped pulse. Its corresponding optical spectrum had a FWHM of 8.8 nm. The resulting TBP was 0.34, which indicates that the pulses are nearly transform-limited. By adjusting the position of the output coupler and the position of the SESAM, the laser was able to mode-lock at 1077 nm with a maximum output power of 102 mW. Figure 5 (c) shows the autocorrelation trace obtained at this wavelength. If a sech² shaped pulse is assumed, the pulses have a pulse width of 125 fs. The optical spectrum has a FWHM of 10.4 nm. Therefore, the TBP was 0.34, indicating that the pulses at this wavelength were also transform-limited. Similar to the cw case, the lower output powers at 1041 nm and 1060 nm compared to that obtained at 1077 nm could be attributed to the presence of reabsorption by the Yb³⁺ ions [5] towards shorter wavelengths. The emission at 1077 nm was obtained by tweaking the cavity such that the least amount of losses occurred at this wavelength. That coupled with the fact that no reabsorption occurs at this wavelength resulted in the laser to mode-lock towards the 1087 nm emission peak [5]. The optical to optical efficiencies were about 0.9%, 1.2% and 2.3% when the laser was mode-locked at 1041 nm, 1060 nm and 1077 nm, respectively. The low efficiencies were due to a small output coupler being used and the large losses introduced by the prism pairs. Larger efficiencies are possible should GTI mirrors be used instead.

All the three mode-locked wavelength bands produced TEM₀₀ beams. Higher output powers were possible by switching to output couplers with larger transmissions, however this was done at the expense of obtaining longer pulse widths. It was noticed that the laser had a tendency of emitting double or even multiple pulses. Such a phenomenon has already been reported in Yb: KLu(WO4)₂ (KLuW) [14] and Yb: KY(WO₄)₂ (KYW) [9]. Such a tendency can be suppressed through careful alignment of the output coupler and the position of the SESAM. The suppression of double and multiple pulses however were not very sensitive to the alignment of the dispersion prism pairs. Therefore, the authors attribute the suppression of double and multiple pulsing to cavity alignments rather than intra-cavity dispersion. The cavity was aligned till only one mode-locked pulse was in the cavity. This could be easily identified experimentally through simultaneously monitoring the oscilloscope trace and autocorrelation trace. Since Yb: SSO could separately mode-lock at the 1041nm, 1060nm and 1077 nm wavelength bands, it may be possible through using a broadband saturable absorber such as graphene [15–17] to mode-lock all these bands together to achieve a sub 100 fs laser. Graphene was found to have saturable absorption that extends from visible to mid infrared spectrum. Graphene has already been demonstrated to be able to mode-lock a ceramic Nd: YAG laser emitting at around 1064 nm [15] and fiber lasers at 1550 nm [16,17].

5. Conclusion

In conclusion, the authors have demonstrated a cw Yb: SSO laser with 70% slope efficiency emitting a maximum cw output power of 2.73 W. This demonstrates that Yb: SSO is an efficient cw gain medium. In addition, the authors also demonstrated femtosecond passive mode-locking of Yb: SSO at 1041 nm, 1060 nm and 1077 nm, respectively, with near Fourier transformed-limited pulse widths of 145 fs, 144 fs and 125 fs and average output power of 40 mw, 52 mw and 102 mW. Compared to the other Yb³⁺ doped oxyorthosilicate gain media, Yb: GSO, Yb: LSO and Yb: YSO generated pulses of 343 fs, 233 fs and 122 fs respectively, which are comparable or much longer than those generated by Yb: SSO in this work. The results obtained demonstrate that Yb: SSO is an efficient cw and a promising ultrashort pulse laser gain medium.