The diode-pumped alkali laser (DPAL) has been extensively researched as one of the most promising candidates for high power and efficiency lasers [1-3]. The most important advantages of DPALs are high quantum efficiency 95.3% for cesium (Cs), 98.1% for rubidium (Rb), and 99.6% for potassium (K), an excellent beam quality, reduced thermal problems as a gaseous gain medium, and narrow linewidth [4, 5]. Furthermore, the stimulated cross section, which was collisionally broadened by ethane buffer gas, is several orders larger than the cross section of a common solid-state laser [6-8]. Thus, all these properties and features show that DPAL is being regarded as a promising alternative to the scalable high power laser system [9-11].

Unlike other types of lasers, in DPAL the linewidth of the D₂ absorption line of the alkali atom is extremely narrow for optical pumping with full width half maximum (FWHM) less than a few hundreds of MHz, which could be broadened to be ~10 GHz allowed by buffer gas [4, 5, 8]. The broadened linewidth is still narrow for optical pumping compared to commercial laser diodes (LD), and an overlap between the bandwidth of LD and the linewidth of D₂ absorption would severely affect the optical-to-optical efficiency, slope efficiency, and output laser power in the DPAL [12, 13]. For this reason, the short-term frequency stability of the pump laser is not adequate for the DPAL without active stabilization of the pump laser frequency [14, 15]. Therefore, active controlling and stabilizing the frequency of the pump laser can directly serve a benefit in output and stability increase of the DPAL [16].

In this study, we experimentally achieved a continuous-wave Cs vapor laser using frequency locking of the pump laser. We simultaneously monitored the absorption spectra of cesium and the laser output power with scanning of pump frequency around the D₂ line of Cs, and the frequency of the pump laser was fixed at the maximum absorption frequency to result in an effective gain enhancement. Using this scheme, an output power increase of ~0.1 W could be obtained with an output power of 1.4 W, an optical-to-optical efficiency of 49%, and a slope efficiency of 61.5%. Furthermore, in order to optimize the laser efficiency, we varied the reflectivity of output coupler and temperature of the Cs cell in the laser cavity. To the best of knowledge of the authors, this paper is the first time for DAPL operation with a frequency locked pump laser. We believe that this proposed method could be further applied to a scalable high power DPAL.

Figure 1 shows the energy diagram of Cs lasers, which is composed of the ground level ⁶S_1/2, the first excited level ⁶P_1/2, and the second excited level ⁶P_3/2 [4, 5]. As shown in Fig. 1, two resonance transitions of ⁶P_1/2→⁶S_1/2 and ⁶P_2/3→⁶S_1/2 are called the D₁ transition and the D₂ transition, respectively [4, 5]. The transition wavelength of D₁ and D₂ and the quantum efficiency are listed in Table 1. The linewidth of the D₂ transition could be collisionally broadened to achieve spectrally homogeneous transition by an ethane buffer gas [8]. In this study, the D₂ absorption line of the Cs atom was broadened by 500 Torr ethane buffer gas. The pump LD operating at ~852 nm excited the atoms to the ⁶P_3/2 state, which is rapidly relaxed to theupper laser level ⁶P_1/2 state by collision with ethane buffer gas. This could create a population inversion and lasing at 894 nm [4, 5].

The experimental set-up is schematically shown in Fig. 2. In order to simultaneously monitor the absorption of the pump laser and the laser output, two equivalent Cs cells are employed in the pump absorption measurement set-up and the laser cavity set-up. The pump laser was split into two optical paths by a beam splitter, 0.5% of it was used for frequency locking of the pump laser and 99.5% of it was used for the laser. The wavelength tunable Littrow-type external cavity diode laser (ECDL) was used as the single frequency pump source [17-19]. This pump laser could provide wavelength scanning whose mode-hop-free tuning range is 15 GHz with repetition rate of 10 Hz [17-19]. It produces about 2.9 W continuous wave power at 852 nm with a linewidth ~10 kHz.

The right column of Fig. 2. shows the laser cavity structure. The L-shape 20-cm-long laser cavity consisted of a concave high reflecting mirror with a curvature radius of 25 cm whose reflectivity is higher than 99.5% at both λ = 852 nm and 894 nm and a flat output coupler. Optical paths to the pump and the laser beams were separated by a polarization beam splitter (PBS). The 2 cm long sealed cell with Cs vapor and 500 Torr of ethane buffer gas was used as an optical gain medium. The silica cell had circular windows of 2.5 cm diameter, which were anti-reflection coated on both sides for securing a high transmission at the pump wavelength λ = 852 nm and laser output λ = 894 nm. The temperature of the Cs cell was controlled by heater. In order to optimize the laser output, we varied the reflectivity of output coupler and the temperature of the Cs vapor cell.

The frequency locking set up is schematically shown in the center column of Fig. 2. 0.5% of the pump laser power passes through a Cs vapor cell equivalent to that in the laser cavity set-up. The pump laser through the frequency lock Cs cell could be detected by photo detector (PD), which is connected to a built-in oscilloscope in the pump laser driver. By scanning the frequency of the pump laser around the absorption lines of Cs, the absorption spectrum could be obtained, and proportional-integral-derivative (PID) feedback loop locks the frequency of the pump laser near to the maximum absorption wavelength [20-22].

In Fig. 3(a), there are transmission spectra of both pure Cs (red line) without buffer gas and Cs with 500 Torr ethane buffer gas (black line). In this figure, we could clearly see the two absorption peaks of pure Cs due to hyperfine energy level of ground state ⁶S_1/2 F = 3 and F = 4 [23, 24], and broadened absorption peak by buffer gas, whose overall absorption range was not measured due to the limitation of mode-hop-free range (15 GHz) in our pump laser. As shown in Fig. 3(b), the frequency of the pump laser was locked at near the absorption peak to maximize the output power of the laser. Using this frequency locking scheme, we could obtain a power increase of ~0.1 W compared to output power when frequency locking was not applied. Detailed gain dynamics needs to be further investigated in our laser scheme, which is being pursued by the authors.

The reflectivity of the output coupler and the temperature of the Cs vapor mainly have significant influence on the output performance [25-28]. In order to obtain the maximum output of the laser, we experimentally optimized the reflectivity of output coupler and temperature of the Cs cell. The experimental dependence of laser output power on the pump power was measured for the output couplers with reflectivity 30, 20, and 10% in Fig. 4(a). The corresponding measured values of the output power for these output couplers are 0.91, 1.1, and 1.4 W, and the slope efficiency are 38.2, 48, and 61.5%, respectively with 2.9 W of pump power at 130°C. Fei et al. [25] have numerically analyzed that the output power could be increased with up to ~1% of reflectivity output coupler. Further optimization of output power with detailed reflectivity lower than 10% of output coupler needs to be experimentally investigated in our proposed DPAL, which is being pursued by the authors. In order to obtain the maximum output power, the temperature of the Cs cell was also optimized. The atomic density of Cs vapor was determined by the temperature of the Cs cell, and would severely affect the laser output power [25, 26]. If the temperature is too low, the laser oscillation is not started because of lack of gain. And if the temperature is too high, output is reduced because of reabsorption at the gain medium [27, 28]. Figure 4(b) shows the dependence of output power on the temperature of the Cs cell with 10% reflectivity output coupler. We could optimize the temperature of the Cs cell to be 130°C with maximum output power of 1.4 W using the 2.9 W of pump power. The atomic density of Cs vapor at 130°C was estimated to be 8.41 × 10¹³ cm^-3 [29]. Note that these experiments were conducted with the frequency locked pump laser.

In Fig. 5(a), the blue and black lines indicate the output power as a function of pump power in the same optimized cavity when the frequency locking was applied and not applied, respectively. By applying the frequency locking, the optimized output power of 1.4 W was obtained with pump power of 2.9 W, providing the slope efficiency of 61.5% and the optical-to-optical efficiency of 49.2%. Note that when the frequency locking is not applied, the output power was decreased by ~0.1 W with the output power of 1.3 W, providing the optical-to-optical efficiency of 44.5% and the slope efficiency of 54% as shown in of Fig. 5(a). Furthermore, the fluctuation of output power in a few seconds was decreased from ~3% to ~1% by applying the frequency locking. The beam quality of the optimized output power of 1.4 W was measured along the horizontal and vertical axes in Fig. 5(b), and the inset of Fig. 5(b) shows the beam intensity distribution profile, which exhibits a Gaussian distribution [27]. The M² values of horizontal and vertical axes were measured to be ~1.3 and 1.4, respectively.

In conclusion, we proposed, for the first time to the best knowledge of the authors, a diode-pumped Cs laser with the frequency locking of pump laser. In order to maximize and stabilize the output power of the laser, the frequency of pump laser was locked at near the center of the D₂ absorption line of the Cs atom using PID feedback loop. We experimentally investigated the output power increase of ~0.1 W by applying the frequency locking of the pump laser. Furthermore, we experimentally optimized the key parameters including operation temperature and reflectivity of output coupler, and the optimized output power of 1.4 W was achieved using the pump power of 2.9 W, providing a slope efficiency of 61.5% and optical-to-optical efficiency of 49%. We believe that the proposed frequency locking method could be applied to power scaling of DPAL.