Femtosecond Green Light Generation Using a MgO-Doped Periodically Poled Lithium Niobate Crystal Pumped by a Yb-Doped Fiber Laser

Abstract

We demonstrate, in this paper, the creation of a femtosecond green laser achieved from a MgO-doped periodically poled lithium niobate crystal pumped by a home-made Yb-doped MOPA laser system. With an incident fundamental average power of 2 W, 820 mW (41 nJ) is produced at 515 nm with pulse duration of ~540 fs, corresponding to conversion efficiency of 40%. The standard deviation of the 515 nm pulse-to-pulse intensity is calculated as 0.007 (normalized), and the average power keeps stable with a root-mean-square value of 0.18% in an eight hour measurement. The spectral characters of the green laser and laser-induced damage thresholds of the crystal are also investigated. This stable laser system provides a compact and portable laser source for two-photon photopolymerization application.

1. Introduction

Green ultrafast lasers have attracted increasing attention in scientific and industrial fields for various applications, such as dual-comb spectroscopy [1], optical parametric oscillators [2], direct laser writing [3,4], and biomedical imaging [5]. In particular, green lasers with ultrashort pulse duration (~fs) and high repetition rate (~MHz) are ideal light sources for two-photon photopolymerization (TPP), an essential technique for flexible and high-resolution fabrication of three-dimensional nanoscale structures. During the TPP process, femtosecond pulses with appropriate peak power are required to achieve the TPP threshold and a high repetition rate makes it possible to fabricate rapidly over a large area [6]. Moreover, in comparison to lasers at 780 nm generated from Ti: Sapphire lasers or frequency-doubling Er-doped fiber lasers, femtosecond lasers at 515 nm capacitate a smaller focal spot and a higher single photon energy, which could approach to an improved fabrication resolution [7,8,9].

Indeed, there have been various wavelength-conversion methods, such as third-harmonic generation in nematic liquid crystals [10], supercontinuum generation in fused silica [11], self-frequency doubling lasers [12] and optical sum-frequency from dual-wavelength lasers [13]. For reliable and efficient green femtosecond laser generation, the conventional route applies frequency doubling from femtosecond lasers around 1.0 μm. Among those, Yb-doped fiber laser possesses many intrinsic properties, such as robustness, compactness, and low cost.

The way to obtain mode-locked laser in all-fiber based Yb-doped oscillators is diverse. Nonlinear polarization rotation (NPR) based fiber oscillators are sensitive to environment perturbations due to non-polarization-maintaining (PM) fibers, and the non-PM fiber increases the pulse-to-pulse power fluctuation and even degrades the mode-locking stability [14]. The all-PM fiber architecture can be applied in semiconductor saturable absorber mirror (SESAM) or nonlinear amplify loop mirror (NALM) mode-locking mechanism. However, SESAMs suffer from a low optical damage threshold, which is unreliable for long-term operation. In 2016, T. Jiang et al. reported an integrated all-PM NALM oscillator using a non-reciprocal phase shifter, producing pulses with 538 fs duration after de-chirping [15]. Traditionally, large-mode-area photonic-crystal fiber amplifiers with free-space bulky components are utilized to achieve higher average power, which would inevitably degrade the integrability and stability of laser systems [16]. Therefore, a monolithic PM Yb-doped femtosecond fiber laser was developed instead. In 2017, Z. Guo et al. obtained a diffraction-limited laser beam with high polarization extinction radio at Watt-level average power, which is crucial for efficient second harmonic generation (SHG), from a direct fiber-spliced laser system [17,18].

To date, considerable studies on femtosecond SHG have been conducted in all kinds of nonlinear crystals. The birefringence phase match (BPM) technique was widely applied in these experiments. In 2014, N. A. Chaitanya et al. demonstrated a 176 fs, 28.8 nJ, 78 MHz green laser with a BiB₃O₆ (BIBO) crystal, and the corresponding conversion efficiency was 46.5% [19]. In 2016, M. K. Shukla et al. realized a 150 fs, 2.3 W, 78 MHz green laser using a 4-mm-length LiB₃O₅ (LBO) crystal with an incident fundamental average power of 4.8 W [20]. In 2017, C. Hong et al. proposed a monolithic laser system and a β-BaB₂O₄(BBO)-based SHG module, from which ~350 fs, 1.25 W, 17 MHz green laser was produced, and the corresponding conversion efficiency was 26.6% [21]. However, the BPM-based SHG process are inevitably along with the spatio-temporal beam distortion caused by the walk-off effect, which notably degrades the SHG laser beam quality [22].

Quasi-phase-match (QPM) in periodically poled crystals is an alternative way for SHG, such as MgO-doped periodically poled lithium niobate (MgO:PPLN), periodically poled KTiOPO₄ (PPKTP), periodically poled LiTaO₃ (PPLT), etc. Compared to BPM crystals, QPM crystals provide much a higher nonlinear coefficient and any spatial beam walk-off. To date, for green laser generation, most QPM studies have focused on the SHG by continuous wave, nanosecond, or picosecond pulses, and few works have reported the demonstration of watt-level femtosecond green lasers. For example, L. Hong et al. reported a SHG quasi-Q-switched-mode-locked (QML) laser at 532 nm generated from a MgO:PPLN crystal, and the conversion efficiencies are ~53% for quasi-QML pulses and ~44% for continuous ML pulses, respectively. As the pulse duration of the laser was over 120 ps, a 2 cm long crystal produced an average power of ~200 mW at 532 nm [23]. Lagatsky et al. obtained sub-200 fs green femtosecond lasers with over 40% conversion efficiency using a PPLT and PPKTP crystal, respectively, but the average power of the second harmonic light was limited to ~200 mW [24,25]. T.-Y. Jeong et al. proposed a MgO:PPLN crystal based optical parametric oscillator (OPO) delivering visible femtosecond pulses (522–800 nm and 455–540 nm) with a conversion efficiency below 10% [26].

In this paper, we report an all-PM femtosecond fiber laser system followed by an efficient MgO:PPLN SHG module. The fundamental picosecond pulses are generated stably from a NALM mode-locked oscillator. As high as an 820 mW 20 MHz green laser is produced by a MgO:PPLN crystal, corresponding to conversion efficiency of 40%. To ensure the proper operation of the laser system, the laser induced damage threshold (LIDT) of the crystal is also investigated. The standard deviation of the pulse-to-pulse intensity is calculated as 0.007 (normalized), and the root-mean-square value (RMS) of the SHG average power is only 0.18% in an eight hour measurement. As far as we know, this is the first time a nearly Watt-level femtosecond green laser was realized in a QPM system. Such a meaningful achievement proofs a low-cost, compact, and robust system for the feasible application of TPP.

2. Experimental Setup

As depicted in Figure 1a, the femtosecond laser system at 515 nm consisted of two parts: a home-made 20-MHz Yb-doped master-oscillator-power-amplifier (MOPA) laser system and the SHG module. The central wavelength of the MOPA laser system was 1030 nm, and the maximum output power can boost up to 2 W with pulse duration of 280 fs (nearly Fourier transform limited). A 5 mol% MgO:PPLN (CTL Photonics Corp. Ltd., Fujian, China) with 6.3 μm poling period, 3 mm length, and 2 × 0.5 mm² aperture area was adopted in the experiment. Both surfaces of MgO:PPLN crystal were anti-reflection coated at 515 nm and 1030 nm. LEN1 and LEN2 were both 50 mm lenses, but with B-coating (650–1050 nm) and A-coating (400–700 nm), respectively. Two plane mirrors (M1 and M2) with high reflection coating (1020–1120 nm) were used to adjust the position of the focused beam. A polarizing beam splitter (PBS) and a half-wave plate (HWP1) were composed to control the incident fundamental average power. Another half-wave plate (HWP2) was used to adjust the polarization state of incident beam before MgO:PPLN crystal. The MgO:PPLN was placed in an oven for controlling the temperature. Following LEN2, two dichroic mirrors (DM1 and DM2) were employed to separate the 515 nm light and the residual 1030 nm fundamental light. The whole laser system was integrated into a portable home-made aluminum alloy case, to avoid any potential environment disturbance for a long-term power monitoring.

Figure 1b shows the configuration of the fiber laser. The all-PM NALM oscillator was operated in all normal dispersion regime [27], and a 1030 nm fiber Bragg grating with 0.8 nm bandwidth (Raysung Photonics Inc., Shaanxi, China) was employed as a spectral filter. As the pump power increased to ~300 mW, the mode-locked pulses were stably generated with an average power of ~5 mW. The optical spectrum of the oscillator was recorded by an optical spectrum analyzer (Yokogawa, AQ6370C, Shanghai, China) with a resolution of 0.02 nm, shown as the pink curve in Figure 2a. Its central wavelength was 1029.9 nm, and the 3-dB spectral bandwidth was 0.18 nm, corresponding to the transform-limited pulse duration of ~8.8 ps for the Gaussian shape pulse. Then, the average power of the laser was first amplified to ~160 mW by a YDFA. By using a segment of 15-m-length PM single-mode fiber (PM980-XP, Nufern) for spectral broadening, the 3-dB spectral bandwidth extended to 10.3 nm at central wavelength of 1029.9 nm attributed to the self-phase modulation (SPM) effect, shown as the blue curve in Figure 2a.

The main amplifier was composed of a segment of 1.5 m long PCF (DC-135/14-PM-Yb, NKT Photonics, Shenzhen, China), a combiner (AFR Inc., Guangdong, China), and a 9 W diode laser (K976AB2RN-9.00W, BWT Inc., Beijing, China). Figure 2b shows the slope efficiency of the main amplifier (red dot line). With 9 W pump power, the main amplifier produced up to 3.36 W average power, corresponding to a slope efficiency of ~39%. The violet curve in Figure 2c shows the amplified optical spectrum with 3.36-W output power, and no obviously spectral broadening is observed comparing to the blue curve in Figure 2a. The Fourier transform limited duration of that was about 266 fs. A 1030 nm free-space optical isolator (ISO2, AFR Inc.) with 85% transmission efficiency was employed to block the backward light. A pair of transmission gratings of 1250 lines/mm (PCG-1250-10XX-986, Ibsen Photonics, Farum, Denmark) with totally 70% efficiency was used to compress the pulse from 12.5 ps to 280 fs with a 2 W average power for the compressed pulse (also named as incident fundamental laser). The autocorrelation trace of the compressed pulse with its Gaussian fitting curve are shown in the inset of Figure 2c. As shown in Figure 2d, its RMS power instability was 0.44% during four hour measurement.

3. Results and Discussion

Subsequently, the incident fundamental laser was focused on a MgO:PPLN crystal to achieve type-0 (e + e→e) QPM. By using an open-source Gaussian-beam propagation calculator (GaussianBeam), the beam diameter was calculated as 99.16 μm, since the waist of the incident fundamental laser beam was 660 μm (full width at 1/e² maximum). Figure 3a shows the average power (black curve) and conversion efficiency (red curve) of SHG laser versus the incident fundamental average power. By optimizing the temperature of the MgO:PPLN crystal and the position of the focus, the SHG average power at 515 nm could be optimized to 820 mW with an incident power of 2 W (0.92 GW/cm²), corresponding to a SHG conversion efficiency of ~40%. The curves in Figure 3b represent the optimal distance (black curve) between the lens and the MgO:PPLN crystal and the optimal temperature (yellow curve) versus the incident fundamental average power, respectively. The optimal position for the incident power of 0.6 W was set as the zero point, and the SHG conversion efficiency was ~26.4%, shown in Figure 3a. Increasing the incident power from 0.6 to 1.9 W, the optimal distance decreased 2.2 mm. During this process, more thermal effect and nonlinear refractive index variation were expected. Moreover, the optimal temperature decreased from 45 to 39.5 °C as the incident power increased from 0.6 to 1.2 W, and kept around 39 °C at 1.6 W.

Figure 4a shows the optical spectrum of the 515 nm light with an 820 mW average power, recorded by a Fourier transform optical spectrum analyzer (OSA201C, Thorlabs, Shanghai, China) with a resolution of ~6.4 pm. The spectrum roughly matches sinc² fitting curves. Assuming there is no depletion of the fundamental light, the normalized SHG conversion efficiency function can be simplified as:

$$\eta ={\sin \mathrm{c}}^2(\frac{\Delta k_{Q}L}{2}),$$

(1)

where L is the length of the MgO:PPLN crystal. Δk_Q is the QPM wavevector mismatching expressed as:

$$\Delta k_Q=k_H-2k_F-\frac{2\pi }{\Lambda }.$$

(2)

where k_H and k_F are the wavevectors of the SHG and fundamental light, respectively, and Λ is the poling period of the MgO:PPLN crystal. Meanwhile, the equation of pump acceptance bandwidth (Δλ_F) in theory is given as [28]:

$$\Delta {\lambda }_F=\frac{0.4429{\lambda }_F}{L}{\left|\frac{n_H-n_F}{{\lambda }_F}+\frac{\partial n_F}{\partial {\lambda }_F}-\frac{1}{2}\frac{\partial n_H}{\partial {\lambda }_F}\right|}^{-1}.$$

(3)

The n_F (2.15@ 1030 nm) and n_H (2.23@ 515 nm) can be calculated from the latest Sellmeier equation [29]. For the crystal length of 3 mm and the fundamental light at 1030 nm, the pump acceptance bandwidth was calculated as 0.6 nm (0.3 nm at 515 nm). It can be observed that the spectral bandwidth is much boarder than the theorical value, which may be caused by the high pump intensity. Furthermore, we calculated the Fourier transform limited duration of 515 nm pulse as 220 fs, as shown in Figure 4b. Considering an additional group delay dispersion (GDD) introduced by the MgO:PPLN crystal, a pulse stretching equation was employed to estimate the second harmonic pulse duration more accurately:

$$\tau ={\tau }_0\sqrt{1+\left(4\ln 2\frac{\mathrm{GDD}}{{\tau }_0{}^2}\right)}.$$

(4)

where τ and τ₀ are the stretched pulse duration and the initial pulse duration, respectively. The group velocity dispersion (GVD) parameter β₂ of the extraordinary axis in MgO:PPLN crystal at 515 nm is 813.32 fs²/mm [30]. For a 3 mm length MgO:PPLN crystal, its GDD is calculated to be ~2439.96 fs². Therefore, the final stretched pulse duration was calculated as 222.14 fs. The pulse duration of the 515 nm autocorrelated trace that we measured is ~540 fs (Gaussian shape), as shown in Figure 4b. Compared to the fundamental light, the SHG pulse duration is longer, which may be cause by the limited pump acceptance bandwidth of the MgO:PPLN crystal.

Additionally, it is essential to keep the SHG system operating bellow the LIDT of MgO:PPLN crystal, which is associated with the nonlinear refractive index and the characteristics of the fundamental light [31]. In our experiment, with a maximum incident fundamental average power of 2 W (0.92 GW/cm²), the laser induced damage did not occur. To find the LIDT of the MgO:PPLN crystal, another MOPA laser system with 100 MHz repetition rate, 150 fs pulse duration, and up to 5 W average power served as the fundamental laser. The maximum SHG conversion efficiency of the new SHG system kept below ~20%, which was mainly caused by the decreased single pulse energy and rugged spectrum of the fundamental laser. The MgO:PPLN crystal also kept working properly by increasing the incident power to 2 W, corresponding to a pump intensity of 1.73 GW/cm². However, as the incident power increases, laser induced damage cracks were observed on the incident surface of the MgO:PPLN crystal under the pump intensity of 2.16 GW/cm². Figure 5a shows the magnified picture of damage point at the pump intensity of 2.16 GW/cm². Laser ablation might be caused by the thermal ion emission, for the repetition rate of the fundamental light is as high as 100 MHz and the pulse duration is as short as 150 fs [32]. The power stability of the 515 nm light was tested at 2 W incident fundamental light, as shown in Figure 5b. The RMS value of power stability was ~0.68% at an average power of 820 mW for a four hour measurement. Compared to the power stability of incident fundamental light, the RMS value increased ~0.24% attributed to the temperature fluctuation of the MgO:PPLN crystal. The pulse-to-pulse laser intensity was recorded by a fast InGaAs photodiode with 2-GHz bandwidth (Beijing Lightsensing Technologies Ltd., Beijing, China) and a 4-GHz digital oscilloscope (DSO9404A, Keysight Technology, Beijing, China). Nearly 40,000 pulses were recorded in 2 ms. As shown in Figure 5c, the maximum intensity of each pulse is presented as a normalized voltage. The distribution is close to a Gaussian curve with an R² of 0.997 and its standard deviation is about 0.007. The inset in Figure 5d is the outside view of the home-made aluminum alloy case. With a case, the SHG average power kept stable with an RMS of only 0.18% in an eight hour measurement.

4. Conclusions

In conclusion, we have experimentally demonstrated the 515 nm femtosecond laser generated from a MgO:PPLN crystal based SHG module and a monolithic Yb-doped MOPA laser system. The SHG average power reaches 820 mW, and the pulse duration is ~540 fs. The 515 nm femtosecond laser exhibits power stability with an RMS value of 0.18% for an eight hour measurement and a pulse-to-pulse intensity standard deviation of 0.007. Thanks to the all-PM fiber-based laser configuration, the SHG laser system enjoys compact, robust, and low-cost characteristics.