1. Introduction

Since the first demonstration of the magneto-optical trapping (MOT) of strontium atoms [1], ultracold Sr gases have been attracting considerable attention from various fields. One successful application of ultracold strontium atoms is an optical lattice clock [2], which utilizes the ultranarrow ¹ S ₀ - ³ P ₀ transition as a clock transition. The uncertainty of an optical lattice clock is expected to achieve 10⁻¹⁸ [3]. A clock with an uncertainty of 10⁻¹⁸ would enable us to perform more stringent tests of fundamental physical principles [4] and lead to new atomic clock applications including geoid surface measurement, space positioning systems, and space-based very long baseline interferometry (VLBI) [5,6]. A compact transportable optical lattice clock would be required for such applications. The University of Firenze has already demonstrated a novel simplified optical lattice clock experimentally [7].

In the Sr optical lattice clock, the ¹ S ₀-¹ P ₁ transition (461 nm) is employed to cool strontium atoms to a few mK. Since semiconductor lasers at 461 nm with sufficient power for MOT are not commercially available, the conventional method is to frequency double a 922-nm laser using a power-buildup cavity to enhance the doubling efficiency [7–9]. However, the power-buildup cavity must always be locked to the resonance, which could make the whole system bulky. Therefore, an integrated light source is preferable if we are to realize a compact transportable optical lattice clock.

A waveguide structure enables a tightly confined beam to travel long lengths in a nonlinear crystal. Consequently, single-pass frequency conversion with high efficiency is achieved using a nonlinear crystal waveguide. Recently, an output power of 494 mW was achieved at 589 nm with a Zn-doped PPLN (Zn:PPLN) ridge waveguide [10], which is manufactured using a direct bonding technique [11].

In this report, we demonstrate the second-harmonic (SH) generation of 922-nm light using a single-pass Zn:PPLN ridge waveguide. The power obtained at 461 nm was 76 mW for 248 mW of coupled fundamental light. Long-term operation revealed the reliability of the system. We observed the narrowing of the phase-match temperature bandwidth at an intense SH power. We employed the developed light source to demonstrate a first-stage magneto-optical trap using the spin-allowed ¹ S ₀ - ¹ P ₁ transition.

2. Experimental setup

Figure 1 is a schematic diagram of the experimental setup. A 922-nm pump light is generated by a home-made extended cavity laser diode (ECLD) with a Littrow configuration. The frequency drift is typically within ± 1 GHz for several hours. After passing through a 60-dB isolator with an insertion loss of 12%, 24 mW of the light is input into a tapered amplifier (TA). The output of the TA is about 800 mW with a 2.5 A supply current. The light is passed through a 30-dB isolator with an insertion loss of about 15%. The light is passed through a half waveplate and then coupled into a polarization-maintaining fiber. The waveplate is set to maximize the generated SH light. The measured coupling efficiency with the fiber is about 45%. The fiber is directly connected to the pig-tailing fiber of a Zn:PPLN module package, manufactured by NTT Electronics Corp. The fluctuation of the laser frequency and the TA output power is typically within ± 1 GHz and ± 5% for several hours, respectively. The variations in the temperature and humidity of the room are about ± 0.5 °C and ± 2%.

 

Fig. 1 (Color online) Schematic diagram of experimental setup. ECLD, extended cavity laser diode; TA, tapered amplifier; λ/2, half waveplate; PM fiber, polarization-maintaining fiber; Zn:PPLN, zinc-doped periodically poled lithium niobate waveguide; DM, dichroic mirrors. After separation of the SH light from the fundamental light with the dichroic mirrors, the output power is measured. The SH light is also employed for the MOT of strontium atoms.

Download Full Size | PDF

The Zn:PPLN waveguide housed in the module is a ridge type waveguide that is 6.9 μm wide, 6.5 μm thick and 22.0 mm long. Zinc doping is used to increase the resistance to photorefractive damage [12]. The periodicity of the poled structure is 4.2 μm so that the quasi-phase-matching needed for the SH generation of the 922-nm light is satisfied at around 35 °C. The temperature is monitored with a thermistor and controlled by a Peltier device to within 0.01 °C. The end facets of the waveguide are AR-coated for both the fundamental and SH lights, and polished at a 6-degree angle to avoid any etalon effect. The coupling efficiency from the fiber to the PPLN waveguide is about 80% when measured at a temperature far from the phase-matching temperature. The coupling efficiency includes the propagation loss in the waveguide. The propagation loss is 0.1 dB/cm for 1.5-μm light according to the specification. The generated 461 nm SH light is collimated by using an objective lens.

3. Experimental results

We carefully separated the generated SH light from the original fundamental light with three dichroic mirrors and measured the SH power as a function of the coupled fundamental power. As Fig. 2(a) shows, the waveguide generates more than 70 mW of SH power. We theoretically fitted the experimental data based on the pump depletion model [13]. If the waveguide is lossless, the output power of the SH light $P_{out}^{2\omega }$ is given by

 

Fig. 2 (Color online) SH power (a) and conversion efficiency (b) versus the coupled fundamental power, along with the fitted curves based on the pump depletion model. The inset of (b) shows the normalized conversion efficiency dependence on the coupled fundamental power.

Download Full Size | PDF

We investigated the phase-matching curve of the waveguide by monitoring the variation in SH power with waveguide temperature. For a relatively low pump power of 44 mW, the temperature acceptance bandwidth is about 0.5 °C as shown in Fig. 3(a) . The acceptance becomes extremely narrow (0.1 °C) for a pump power of 248 mW as shown in Fig. 3(b). Therefore, careful temperature control is needed to obtain a stable high SH power. As shown in Fig. 3, the nominal phase-matching temperature decreases by 2.2 °C when the pump power is changed from 44 to 248 mW. This is due to the absorption of the light inside the PPLN waveguide.

 

Fig. 3 (Color online) SH power as a function of the temperature of waveguide (A), when the pump power is 44 mW (a) and 248 mW (b).

Download Full Size | PDF

We measured the transmitted power of the fundamental and SH lights. The fundamental and SH powers observed after the waveguide were 17 and 4 mW, respectively, when we injected a weak pump power of 24 mW. On the other hand, we observed an SH power of 60 mW and a fundamental power of 64 mW when we injected a fundamental power 170 mW. Therefore, 43 mW of the power was absorbed and converted into heat, which is considered to cause the large change in the temperature dependence. It should be noted that the absorption was induced by the SH light, since no loss of fundamental power was observed when the injected light polarization was rotated by 90 degrees.

Narrowing the temperature acceptance bandwidth of an intense light requires careful temperature control. However, long-term overnight operation of over 7 hours was realized with a box covering the PPLN module to suppress any temperature fluctuation. It is worth noting that the spatial mode and the direction of the output beam from the waveguide modules were very stable.

4. Discussion and conclusion

From the fitting parameter ${\eta }_0$, we derive the nonlinear coefficient of waveguide (A). The nonlinear coefficient $d_{eff}$ is given by

The pump depletion model with a lossless medium provides the conversion efficiency η as a function of the phase mismatch $\Delta k$,

Using the experimental data in Fig. 2(a), we derived the normalized nonlinear efficiency of waveguide from the theoretical fitting parameter. With this value we calculated the conversion efficiency dependence on the phase mismatch $\Delta k$. For a pump power of 44 mW, the calculated spectrum bandwidth (FWHM) of the main lobe is 11.7 GHz. For a pump power of 248 mW, the calculated value is 9.92 GHz. We assume that the phase mismatch $\Delta k$ is proportional to the temperature. Although the narrowing of the phase-matching bandwidth is explained by the pump depletion model, our experimental data (shown in Fig. 3) reveal that the bandwidth of the main lobe is reduced by about 80%, which cannot be explained by the simple pump depletion model. This is considered to be due to the heat effect caused by the absorption of the light inside the PPLN waveguide. The optical loss increases from 13% to 27% when the SH power is increased from 4 to 64 mW. Since we did not observed any loss in the fundamental light without the SH light, the absorption loss was induced by the SH light. This kind of absorption was observed, for example, in refs [18]. and [19] for green light generation with a PPLN device. Assuming only the fundamental light is absorbed in the waveguide, the absorption loss of the fundamental light is calculated to be 0.11 cm⁻¹ with 64 mW of SH light. This is more than ten times as large as in previous reports [18,19]. This large absorption loss causes saturation and different performance as regards SH generation even with the same material [10]. The thermal inhibition of high-power SH light in PPLN crystals is discussed in [20].

The conversion efficiency obtained for the Zn:PPLN waveguide device is less than that achieved with the power-buildup cavity. For example, the PPKTP crystal with the cavity system generated a blue power of 234 mW from 310 mW of fundamental light with a 75% conversion efficiency [9]. The low conversion efficiency in the present system results from the output saturation that occurs at an input power of more than 100 mW, and this prevents us from obtaining an SH power of greater than 80 mW. Although saturation is often observed for shorter wavelength frequency conversion using a PPLN waveguide [21], the SH generation experiment at 473 nm became saturated at a much higher output power level (about 189 mW) [22]. To investigate this saturation phenomenon, we are currently working on Zn:PPLN waveguides with different lengths. A shorter waveguide provides a wider phase-matching temperature acceptance bandwidth, which relaxes the requirement of the temperature uniformity along the waveguide. Consequently, the saturation effect is expected to be reduced. On the other hand, the interaction length is also reduced. There should be an optimum waveguide length that can provide a sufficient interaction length with a relatively wide phase-matching acceptance bandwidth.

In conclusion, we developed a compact and reliable light source emitting at 461 nm using a Zn:PPLN waveguide for SH generation. We obtained more than 70 mW of SH power at 461 nm for 250 mW of fundamental light. Compact and reliable light sources are indispensable when launching optical clocks into space [5,6]. It should be noted that we successfully performed the magneto-optical trapping of strontium atoms with 60 mW of 461-nm light from the described compact light source. A robust and compact laser source with long-term stability is the key to cold atom experiments, especially with respect to optical clocks.