Experimental Results and Discussions

Because the third-order nonlinearity and optical properties of Nd:YVO4 and Nd:GdVO4 crystals are similar, the different between Nd:YVO4 and Nd:GdVO4

crystals for the self-mode-locked operation was found to be negligible. Therefore, we just report the experimental results of Nd:YVO4 laser. The optical cavity length was firstly set to be approximately 6.4 cm, corresponding to the FSR of 2.35 GHz. When the cavity alignment was optimized for generating the maximum average output power, the time trace of the output radiation revealed the laser to be in the spontaneous mode-locked state. Figures 2.2-2(a) and 2.2-2(b) show the pulse trains on two different timescales, one with time span of 5 s, demonstrating the amplitude oscillation, the other with time span of 10 ns, demonstrating the mode-locked pulses.

The corresponding power spectrum is shown in Fig. 2.2-2(c). Although some amplitude fluctuation exists under the circumstance of the optimum output power, it can be definitely improved with the fine-tuning of the cavity alignment by monitoring the temporal behavior of the pulse train profile and the width of the power spectrum.

Figures 2.2-3(a)-(c) show the real-time traces and the power spectrum for the case of

Fig. 2.2-2 Pulse trains on two different timescales. (a) Time span of 5 μs, demonstrating the amplitude oscillation. (b) Time span of 10 ns, demonstrating mode-locked pulses. (c) Power spectrum.

500 ns/div

37

minimizing amplitude fluctuation. As shown in Figs. 2.2-3(a) and 2.2-3(b), the full modulation of pulse trains without any CW background indicates the realization of complete mode locking. Excellent performance on self-mode locking indicates that the YVO4
crystal is a promising host medium for efficient SML operation at GHz oscillations. Experimental results reveal that the relative frequency deviation of the power spectrum, v/v, is smaller than 510^5 over day-long operation, where v is the center frequency of the power spectrum and v is the frequency deviation of full width at half maximum. It is worthwhile to mention that the wedge shape of the laser crystal is vital for obtaining a complete stable mode-locked operation. When a laser crystal without a wedge is used in the flat-flat cavity, the pulse trains exhibit incomplete mode locking with CW background to a certain extent. On the other hand, when an oscilloscope with bandwidth less than 500 MHz is used to measure the present temporal characteristics, the result will display like a pure CW laser. Perhaps this is the reason why the phenomenon of self-mode locking in the range of GHz has not been discovered earlier.

Experimental results reveal that the average output power of the stable continuous-wave mode-locking is approximately 90% of the maximum average output power. Figure 2.2-4(a) shows the average output powers versus the incident pump power obtained at a mode-locked frequency of 5.32 GHz with the cavity alignments for maximum output and stable cw mode-locking, respectively. The slope efficiency for the stable mode-locked operation can be seen to be approximately up to 40% with respect to the incident pump power, corresponding to an optical-optical efficiency of 32%. As shown in Fig. 2.2-4(b), the FWHM width of the optical

Fig. 2.2-3 Same as Fig. 2.2-2 for the stable CW mode-locked operation.

39

spectrum is approximately 0.21 nm around the central wavelength of 1064.3 nm.

Figure 2.2-4(c) depicts the real-time traces with time span of 1 ns to measure the temporal duration of the mode-locked pulses. The pulse width can be clearly found to be approximately 50 ps (FWHM) from the real-time trace for the mode-locked frequency in the range of 26 GHz. However, the pulse duration was measured with a homemade autocorrelator and was found to be as short as 7.8 ps assuming a Gaussian-shaped temporal intensity profile, as shown in Fig. 2.2-4(d). The discrepancy comes from the condition that the impulse response of the present detector has a FWHM of 40 ps and the sampling interval of the present digital oscilloscope is 25 ps. Nevertheless, the real-time trace for the temporal behavior of the GHz mode-locked laser is vital for many practical applications such as high-speed electro-optic sampling and telecommunications.

We performed the same experimental procedure for different cavity lengths to investigate the influence of the intracavity power intensity on the performance of the mode locking. We found that the laser system can be easily operated in a stable single-pulse mode-locked regime when the cavity length is approximately shorter than 7.5 cm (the mode-locked repetition rate >2 GHz). For the cavity length longer than 8.5 cm, a single pulse per round trip was usually observed to split into several pulses.

Figure 2.2-5 shows the experimental time traces for the cavity length at 11.3 cm. It is clear that the laser system generally turns to be in a stable multiple-pulse state when the mode-locked frequency is considerably lower than 2 GHz.

Fig. 2.2-4 (a) Average output powers versus incident pump power with the cavity alignments for maximum output and stable CW mode locking, respectively. (b) Corresponding optical spectrum of the mode locking. (c) Mode-locked pulse trains in time span of 1 ns. (d) Autocorrelation trace.

Incident pump power (W)

41

Fig. 2.2-5 Experimental time trace for the multiple-pulse mode-locked operation at the cavity length of 11.3 cm.

1 ns/div

0