1. Introduction

Low power, CW Ar⁺-ion lasers are presently used in applications like bioanalysis, graphics and semiconductor inspection. A trend today is to replace these lasers with diode-pumped alternatives, since the Ar⁺-ion lasers are bulky and inefficient. So far most attempts to reach the blue-green wavelength range have been based on frequency doubling of semiconductor lasers. Efficient blue light generation was first obtained at 425 nm by direct doubling of the radiation from a single stripe laser diode in a quasi-phase matched LiNbO₃ waveguide.¹ However, it was found that such units were unstable in output power as the waveguides had a very narrow phase-matching bandwidth (typically 0.2 nm) and the laser diodes drifted in wavelength with time due to mode hops, temperature and mechanical drift and aging.² In many biotech applications the 488 nm Ar⁺-line is used. For this wavelength, particular chemistry is developed and matched to fluorophores with relatively narrow absorption bands. With the development of high power pump diode lasers for Er:Yb-fiber amplifiers at 980 nm it became possible both to reach this wavelength band and to obtain higher frequency doubled power at the same time. These lasers can be wavelength stabilized with pig-tailed fibre Bragg gratings and with such configurations it has been possible to obtain a long term stable blue-green source emitting more than 5 mW in the 488 nm band.² This early work with frequency-doubled semiconductor lasers involved edge-emitting diodes, but recently, optically^3,4,5 and electrically pumped surface-emitting lasers (VCSELs)⁶ have been used with intra-cavity frequency doubling. The bandgap of the InGaAs semiconductor has in both these cases been tailored to emit at around 976 nm to obtain the desired 488 nm wavelength after frequency doubling. Single-frequency blue output with a linewidth down to a few MHz, has been obtained from an intra-cavity configuration⁷, and using master-oscillator-power-amplifier (MOPA) systems^8,9. Also methods based on sum-frequency mixing (SFM) have been exploited, so far mostly to attain yellow radiation^10,11. Although output powers as high as 20 W in the visible wavelength range has been achieved¹⁰, remaining problems are power stability and high-frequency noise. In this paper we present a simple scheme to obtain visible light by sum-frequency mixing radiation from a diode-pumped solid-state laser (DPSSL) and a laser diode (LD) in a periodically poled KTiOPO₄ (PPKTP) crystal. The two mixing lasers are combined for single-pass SFM, which is advantageous for stability and low noise. Since high-power laser diodes are available at a wide range of wavelengths, whereas diode-pumped solid-state lasers are fixed at certain wavelengths, it is thereby possible to obtain essentially any wavelength in the visible spectrum by appropriate choice of lasers. Furthermore, scaling in power is possible by using a high power DPSSL. As a proof of concept we choose to construct the light source in the above-mentioned technologically interesting wavelength band of the Ar⁺-laser.

2. Experiment

The DPSSL was a CW, diode-pumped Nd:YVO₄ laser operating at 1064.3 nm. It was constructed of a Nd:YVO₄ crystal (0.7 atm%, a-cut) AR coated for 808 nm and HR coated for 1064 nm on the incoupling side and AR coated for 1064 nm on the opposite side. A mirror having a radius of curvature of 100 mm and a reflectivity of 95 % was used for outcoupling of the laser light. It was pumped at 808 nm with a 1.4 W fiber-coupled laser diode (Osram GmbH). This gave a maximum output power of 730 mW, corresponding to an efficiency of 52%. The output beam was polarized in the σ-direction and the beam quality was close to Gaussian (M² ~1.5). For collimation of the beam, a lens with a focal length of 150 mm was used, giving a circular beam with a diameter of approximately 1 mm.

To obtain turquoise radiation, a LD (Axcel Photonics) was employed. For 500 mA of drive current a power of 400 mW was emitted in a single transverse mode. The free-running wavelength at this power and at a temperature of 36 °C was 915.7 nm. A laser diode lens (Melles Griot 06GLC001) was used for collimation of the LD beam. Its N.A. was 0.615 and its focal length was 6.5 mm. After collimation, the LD beam was astigmatic with a diameter of 4.8 mm in the x-direction and 1.9 mm in the y-direction. To achieve a better mode-match in the PPKTP crystal, the diode beam was reduced to 1.7 mm in the x-direction with the help of two uncoated cylindrical lenses acting as a telescope. Due to Fresnel reflections from the cylindrical lenses and the beam combiner, the effective power from the LD reaching the PPKTP crystal was 280 mW.

 

Fig. 1. The first version of the experimental set-up.

Download Full Size | PDF

The two lasers were aligned perpendicular to each other and the beams were then combined with a dichroic mirror and focused to a spot size of 14 μm and 16 μm for the LD and 28 μm for the DPSSL into the PPKTP crystal. For comparison, a numerical simulation of the mixing of two focused circular Gaussian beams was done, based on an analysis made by S. Guha and J. Falk¹². The numerical simulation gave that the optimum focused spot sizes in the PPKTP crystal are 23 μm and 25 μm, respectively. Unfortunately, the available optics limited the choice of focused beam widths to the above stated values. The 9 mm long PPKTP crystal had a grating period of 6.99 μm designed for first order quasi-phase matching using the d₃₃ coefficient. Its non-linear coefficient was d_eff = 10 pm/V, measured by SHG of a Ti:Sapphire laser. This value is close to that of an ideal crystal. The crystal had antireflection coating for 914 nm and 1064 nm on one side, and antireflection coating for 914 nm, 1064 nm and 492 nm on the other side. The spectral width of both the LD and the DPSSL was measured with an optical spectrum analyser (OSA), and it showed to be less than the resolution of the OSA, which was 0.06 nm. However, neither the LD nor the DPSSL were single longitudinal mode. As for the DPSSL, three longitudinal modes with an interspacing of 0.1 nm were present.

3. Results

Two versions of the experimental set-up were tested. In the first set-up (see Fig.1), no attempt was made to long-term stabilize the LD:s spectral output. This gave a maximum SFM output power at approximately 492 nm of 4.0 mW. The experimentally achieved output power should be compared to 6.8 mW, calculated from the numerical simulation for two circular Gaussian beams¹². This discrepancy is attributed to the non-ideal overlap with one circular beam and one elliptical beam. The maximum power was reached at a KTP temperature of 36.2 °C and the effective temperature bandwidth (FWHM) was approximately 5.8 °C, see Fig. 2. Regarding the temperature bandwidth, basic QPM theory¹³ gave a numerical value of 3.4 °C. The larger experimental value of 5.8 °C is due to the presence of several longitudinal modes in the two lasers, giving a larger thermal acceptance bandwidth compared to the ideal single mode case.

 

Fig. 2. The turquoise output power as a function of the temperature of the PPKTP crystal. The measured data is represented by the squares and the curve is a sinc-fit of the data. The temperature bandwidth, ΔT = 5.8 °C.

Download Full Size | PDF

The normalized conversion efficiency

was measured to 2.2 %W^-1cm^-1, where P₃ is the SFM power, P₂ the DPSSL power, P₁ the LD power and L the length of the PPKTP crystal. The SFM power is given by P ₃ = γ_SFM P ₁ P ₂, where the parameter γ_SFM is dependent on the non-linear coefficient of the material, the optical frequencies taking part in the non-linear process and the beam focusing conditions¹⁴. Hence if P₁ is fixed and P₂ is varied, the SFM power should vary linearly as a function of P₂. This is also the case as seen from Fig.3, where the SFM power is plotted vs. the DPSSL power. In this way, power scaling is simple, for example by increasing the DPSSL power.

 

Fig. 3. Turquoise CW output power measured for fixed LD power (at 280 mW) and changing the DPSSL power.

Download Full Size | PDF

The short-term power fluctuations were less than 2.5 %. As can be expected, the beam shape was slightly elliptical, but it was nevertheless Gaussian in both directions.

Occasionally it happened that the LD mode hopped resulting in a large instability of the SFM signal. In this case the wavelength shift was large enough to exceed the phase-matching bandwidth of the PPKTP crystal, 0.2 nm¹⁴. The mode hops, typically 0.3 nm, were primarily attributed to uncontrolled back-reflections from the optics and the crystal. In order to tune and long-term stabilize the LD in wavelength, a transmission grating (Spectrogon AB) with 2000 lines/mm was used in a Littrow configuration, i.e., the -1:st order reflection was coupled back into the LD. The transmission grating was placed after the collimating lens and it was oriented with an angle of 66.3° towards the LD to lock the diode at the phase-matching wavelength 915.7 nm. The LD could be tuned for wavelengths between 906 nm and 916 nm where it operated perfectly stable with a linewidth below the resolution bandwidth of the OSA.

The diode pump power reaching the PPKTP crystal was in this case 110 mW, whereas the DPSSL power was 620 mW. A maximum turquoise power of 0.74 mW was then obtained.

The noise of the turquoise light was studied with an electrical spectrum analyser (HP 8560A). It showed a very low noise, but a small peak at 2.3 MHz could be observed (see Fig. 4) which is assumed to originate from relaxation oscillations of the Nd:YVO₄ laser. The signal-to-noise ratio at 2.3 MHz was then 104 dB.

 

Fig. 4. Noise spectra of the turquoise light (upper) and detection system noise (lower). Relaxation oscillations originating from the DPSSL are present at 2.3 MHz.

Download Full Size | PDF

4. Conclusions

In conclusion, a novel scheme to generate visible radiation at a desired wavelength by mixing a diode-pumped solid-state laser with a single-mode laser diode has been investigated. The main advantage of this scheme is the possibility to reach essentially any wavelength in the visible spectrum by appropriate choice of lasers. PPKTP was used for the frequency mixing and the lasers were chosen to give turquoise light at approximately 492 nm. An output power of 4 mW was obtained. It was long-term stabilized by inserting a transmission grating in order to lock the diode laser at quasi-phase matching. In this case, the output power was reduced to 0.74 mW. A signal-to-noise ratio at 2.3 MHz of 104dB was then obtained.

By using a more powerful pump source for the solid-state laser, the turquoise power could be scaled considerably in power. As an example, a 3 W pump diode would increase the turquoise output power to about 14 mW according to our numerical simulation. Thus, if assumed that we reach 60% of this as in our experiments, about 8 mW could be expected experimentally. A double-pass SFM configuration¹⁵ and a longer PPKTP crystal are established methods of increasing the non-linear conversion efficiency. Furthermore, the results would improve if better collimation and focusing optics for the two beams would be used.

It also deserves to be mentioned that this scheme gives a very inexpensive and easy way of modulating the visible radiation, i.e. by modulating the single mode laser diode via the driving current, which is not possible with DPSSL:s.