1. Introduction

The power output of diode pumped alkali lasers (DPALs) has increased steadily since their invention in 2003 [1]. Initial laser pumped demonstrations [2,3] were followed by diode pumped systems at the several mW average power level [4,5]. These were later followed by a 10 Watt system [6]. More recently, an average power level of 20W was demonstrated in a cesium DPAL [7]. These experiments all used free propagating cells with relatively large volumes compared to the actual gain volume. Heat removal could only be accomplished by conduction through the buffer gas to the walls and then through convection to the outside environment (usually the air inside of an oven). The large distance between the active lasing region and the side walls, and the low thermal conductivity of the cell walls meant that heat could not be removed efficiently from these systems, resulting in a very high temperature rise. This ultimately limited the power of these systems due to thermal deposition and heating of the lasing medium [7]. A high aspect ratio waveguide architecture was proposed [8] in order to allow for heat removal through the small dimension of the waveguide. This architecture has the additional benefit of guiding the pump light through the gain volume, allowing relatively low brightness laser diode bars to be used as a cost-effective and scalable pump source. Concern also has been raised about the inclusion of hydrocarbon gases in the cells. These gases are used because of their high spin-orbit mixing and low quenching cross sections [9]. While there have been promising results using hydrocarbons [10], under certain conditions it has been shown that these gases can break down, causing deposits on the windows, eventually leading to failure [11]. This problem can be eliminated by using only helium as the buffer gas. This has been demonstrated in both rubidium and potassium [11–14]. A Rb-He system is of great interest because of the near-term potential scaling to high average powers due to the availability of laser pump diodes at the appropriate wavelength. While high efficiency can be achieved in potassium using modest pressures of helium (2-3 atm) [15], higher pressures are required in a rubidium laser because of the low spin-orbit mixing cross sections of helium with rubidium [16]. In order to understand the physics of such a system at modest average powers, we have built a hydrocarbon free, Rb waveguide system. This is the first reported system that has both a means to remove large amounts of waste heat from the gain volume and a waveguide structure to allow the use of commercial diode bar stacks over a multi-centimeter gain length. The laser produced 24W of average power using 4 atm of ⁴He and 28W using 2.4 atm of ³He, demonstrating both a diode pumped hydrocarbon-free rubidium laser and a liquid-cooled, non-flowing design.

2. Experimental setup

The laser head consisted of a copper core contained within a vacuum vessel (Fig. 1 ). The vacuum vessel consisted of standard off the shelf 2 3/4” conflat vacuum components. It was attached to a manifold which was connected to a vacuum pump and several gas cylinders. After being evacuated, the cell could be filled with various gas mixtures. The windows were standard sapphire viewports with random crystal orientation. Initial windows were coated with an ion beam sputtered AR coating. These windows showed damage on the inner surfaces after testing. Both uncoated and e-beam coated windows were also tested. The final test system used e-beam coated windows where the coating was only placed on 3 surfaces, with no coating on the inner surface of the pump window. The single pass loss was measured at 14%. The copper core had a hole through the center, which formed the laser volume. The aperture was a rounded rectangle, measuring 2.5 mm high by 6.0 mm wide, and the inner walls were polished. The gain length was 22 mm. The copper core was temperature controlled by a fluid channel that passed through the core and penetrated the vacuum vessel. This was connected to a pump/heater/heat exchanger system that could either heat the system or remove heat deposited from the operating laser. Heater rope was wrapped around the vacuum vessel to provide additional heat. The fluid temperature was 95°C. Cartridge heaters placed around the optical windows insured that the windows were hotter than the rest of the system, preventing alkali condensation on the optical surfaces.

 

Fig. 1 . Experimental setup (left). Laser head detail (right)

Download Full Size | PDF

The system was pumped with a 20 bar laser diode stack that was narrowed with volume bragg gratings (VBGs). The diode stack used a design similar to that found in the literature [17] where there are upper and lower arrays and an aperture located between the two arrays. The aperture allows for the laser resonator to pass through the stack. This provides a way of geometrically separating the pump and laser beams. Two VBGs were used, one each on the upper and lower arrays. The VBGs were held in a temperature controlled frame in front of the diode stack. The temperature could be changed to tune the peak wavelength to match the absorption of rubidium. The angle of the frame could be adjusted to achieve the best wavelength fidelity. The diode stack operated up to 1.28 kW. A typical narrowed diode spectrum is shown in Fig. 2 . At best performance, we could place ~70% of the laser diode power in a 0.35 nm bandwidth. We examined the wavelength of the individual bars using an imaging spectrometer. This showed that the bandwidth fidelity varied from bar to bar, and tuning the VBG angle changed which bars were best locked. From this we can conclude that overall fidelity was limited by the pointing accuracy of the individual laser diode bars in the stack.

 

Fig. 2 Output spectrum of VBG narrowed pump diodes.

Download Full Size | PDF

The laser head was placed in the middle of a 1.2 m long resonator with a flat rear cavity mirror and a 37% reflective 1 m ROC concave output coupler. The stack was run at full power (1.28 kW) for all of the experiments. The resonator operated multimode, filling the laser aperture. The diodes were focused with a combination 125 mm focal length Gradium lens and a 200 mm focal length cylinder lens. This lens pair was used to account for the difference in divergence between the fast and slow axes.

3. Results and discussion

Figure 3 shows the time history of the output power for the laser using ⁴He. One sees the power increase, rollover, and then decrease. This behavior comes from changes in the rubidium density as energy is deposited. Insufficient cooling in the core causes the temperature, and hence the rubidium density, to increase. The initial rubidium density is below optimum. As the temperature increases, the density reaches its ideal level. The density continues to increase past its optimal value, causing the laser power to decrease. These changes occur on the time scale of seconds. The gain volume has a small thermal mass and its temperature change would produce a much faster transient. This leads to the conclusion that this behavior is associated with the copper core heating up. Other experiments, using a helium/methane mixture, showed that we could affect the power rise and fall behavior by adjusting the coolant temperature. By decreasing the temperature, both the rise and fall times were increased. A larger temperature rise was required to reach the optimal temperature and the larger temperature differential between the coolant and the cell provided more heat transfer, slowing the density increase and associated power decline. This confirmed that the temporal behavior is not due to any fundamental issue and more efficient cooling (provided by multiple cooling channels or microchannel coolers) in the core should allow for better temporal stability in the laser power.

 

Fig. 3 Output power as a function of time for DPAL with 4 atm ⁴He. At 20 seconds the laser was switched off.

Download Full Size | PDF

Laser performance as a function of buffer gas pressure is shown in Fig. 4 . The data shown are the maximum power measured during an experimental run. We produced 24W using 4 atm of ⁴He. The power was limited by the small spin-orbit mixing produced by ⁴He, the resonator losses, as well as the reduced effective pump power due to the fact that all the diode energy is not contained within the useful bandwidth. Taking these into account, the results were modeled with a CW alkali laser model [3]. Because of the changing rubidium density discussed above, an effective temperature was used in the modeling. This was estimated to be 139°C for our conditions, based on a fit to the data. Using the ⁴He spin-orbit mixing cross section from the literature [16], we were able to obtain a best fit using a mode overlap efficiency of 62%.

 

Fig. 4 CW output power as a function of buffer gas pressure using diode pump array.

Download Full Size | PDF

Figure 4 also shows results using ³He. As reported in the literature [13], ³He provides increased spin-orbit mixing compared to ⁴He, improving performance. About half of the gas pressure was required to achieve the same output power compared to ⁴He. Using the mode overlap efficiency and the effective temperature from the ⁴He model, we obtained a best fit to the experimental data using a spin-orbit mixing cross section of 2.4 x 10⁻¹⁷ cm². While this number is lower than previously published by approximately a factor of three [13], the ⁴He to ³He spin-orbit mixing cross section ratio is similar. Laser performance is a result of the interaction of many different parameters. In the modeling of these systems, small variations in one parameter can be offset by changes in another parameter. This makes obtaining an accurate absolute value for the spin-orbit mixing cross section difficult. While the data show that internally consistent models produce similar relative results, an experiment directly measuring the cross section [9] should be used to obtain an absolute value.

The overall performance of the diode pumped system was low, only ~2% optical to optical conversion. This was due, in part, to the spectral purity of the diodes and the system losses. However, the small spin-orbit mixing provided by helium was the major factor limiting the performance of this laser. The pressure in these experiments was limited by system design and, in the case of ³He, the quantity on hand. Increased buffer gas pressure should improve performance. In order to verify this, we used our alexandrite laser pumped, low average power system. This system has been described in detail previously [15]. Briefly, this system consisted of a vacuum cell with a 16 cm gain length pumped using a flash lamp pumped alexandrite laser. At the 780 nm pump wavelength for rubidium, the laser bandwidth was 0.2 nm and the pulse width was 250 ns FWHM. The pulse width is sufficient such that the laser reaches optical steady state during the pump pulse, effectively operating CW. In addition to allowing for higher pressure (due to thicker windows), this system eliminated the spectral purity issue, increased the laser length, and reduced the optical losses seen in the high powered system. The results, with 4 mJ of input energy, are shown in Fig. 5 . We obtained ~50% optical to optical conversion using 14 atm of ⁴He or 10 atm of ³He. At these pressures the laser performance is similar to that which can be achieved using 1 atm of methane. These pressures create a challenge in mechanical design as well as controlling optical aberrations. This high-pressure requirement would not occur if we operated the system with potassium as opposed to rubidium. It has been shown that only 2-3 atm of ⁴He are needed for efficient operation of a potassium laser [15]. Potassium has the added benefit of less than 1/4th the quantum defect as rubidium. This fact, coupled with the lower pressure required for efficient operation, could reduce the optical aberrations by over an order of magnitude compared to a hydrocarbon free rubidium laser.

 

Fig. 5 Pulsed output energy as a function of gas species and pressure. Input energy was 4 mJ using alexandrite laser pump.

Download Full Size | PDF

4. Conclusion

In conclusion, we have demonstrated a high average power diode pumped rubidium laser with a scalable architecture. We achieved 24W using 4 atm of ⁴He and 28W using 2.8 atm of ³He. Power was limited by the system losses, the spin-orbit mixing rates provided by the helium buffer gasses, and laser diode bandwidth fidelity. Although overall performance was low (optical to optical efficiency~2%), experiments performed with high gas pressure and an alexandrite laser pump show that we can achieve 50% optical to optical conversion using only helium as the buffer gas. We believe that these results show that static gas alkali lasers can be scaled further. More testing at elevated power will help to determine the limits of these conductive cooled systems.