1. Introduction

Frequency conversion with periodically-poled nonlinear crystals into the visible spectral regions has been part of optical research for the last decade. The development of compact and efficient CW lasers, both in the green and blue spectral region, has accelerated over the last several years^1,2. These lasers are presently replacing the traditional Ar⁺-ion lasers used in applications like bioanalysis, reprographics and semiconductor inspection. Established methods of reaching the blue-green wavelength range are based on frequency doubling of semiconductor lasers, either by direct doubling of 980 nm telecom pump diodes³ or by intra-cavity doubling of optically^4,5,6 or electrically⁷ pumped VCSELs. Also second harmonic generation (SHG) with diode-pumped solid-state lasers (DPSSLs) has been done with impressive results^8,9. Other methods of generating visible light involve sum frequency mixing (SFM) of two infra-red (IR) lasers, such as mixing two DPSSL:s to attain yellow light^10,11,12.

The concept of using a solid-state laser and a laser diode (LD) for the frequency mixing was first reported by Risk et al¹³, where the LD was simultaneously used for pumping of the solid-state laser. A modulated blue laser source by direct modulation of the LD has also been demonstrated^13,14. To be able to optimize each mixing source individually, it is advantageous to employ separate LD:s for pumping and mixing, respectively. Furthermore, the nonlinear conversion efficiency can be improved by using a resonator cavity to enhance one or both of the IR fields. Several experiments employing DPSSLs and LDs in both intracavity^15,16 and extracavity^17,18 resonator configurations have been demonstrated.

Recently, 4 mW of 492-nm power was demonstrated by mixing radiation from a DPSSL and a LD in a periodically-poled KTiOPO₄ (PPKTP) crystal in a single-pass configuration¹⁹. We now report more than 27 mW of 492-nm output power based on SFM in a similar compact configuration. The 492-nm light was generated inside the cavity of the DPSSL, hereby utilising the high circulating power, whereas the LD was single-passed through the PPKTP. This scheme can be used not only for the technologically interesting wavelength band of the Ar⁺-laser, but for obtaining essentially any wavelength in the visible spectrum by appropriate choice of lasers. Even though diode-pumped solid-state lasers are fixed at certain wavelengths, high-power laser diodes are commercially available at a wide range of wavelengths.

2. Experiment

The experimental set-up is depicted in Fig. 1. The LD (Axcel Photonics) used for the sum frequency mixing emitted 400 mW in a single transverse mode at a drive current of 500 mA. The beam was collimated using a laser diode lens (Melles Griot 06GLC001), giving a beam size of 4.8 mm times 1.9 mm. To control the wavelength and stabilize it at the phase-matching wavelength, a transmission grating (TG) was used in a Littrow configuration. The TG (Spectrogon AB) had 2000 lines/mm and was oriented with an angle of 66.3° towards the LD in order to lock the diode at the phase-matching wavelength 915.7 nm. The grating reflectance at 910 nm was 12–15%, whereas the LD had a reflectance of 2.5%, as specified by the manufacturer. A motorized mount (New Focus) was employed for precision alignment of the TG. The distance from the LD to the TG was 6 cm. Tuning of the wavelength by tilting the TG was possible between 906 nm and 916 nm. After the TG, two anti-reflection (AR) coated cylindrical lenses (f₁=25.7 mm and f₂=10 mm) were used for beam shaping, giving a beam diameter of 1.9 mm in both directions. A dichroic mirror (M1, R>95% at 914 nm) was used for controlling the beam position in the PPKTP. To focus the LD beam, a 100 mm AR-coated lens was employed. This gave a spot radius of 46 µm inside the PPKTP crystal. The effective LD power reaching the PPKTP crystal was 170 mW. The power losses were mainly due to beam clipping, where the outer parts of the LD beam were cut because of the large tilt angle in combination with an undersized aperture of the TG mount. Furthermore, the 0:th order reflection loss of the grating accounted for approximately 20 % of the total loss.

A Nd:YVO₄ crystal (0.7 atm%, a-cut, 2 mm-long) was used for the DPSSL laser. It was AR coated for 808 nm and coated for high reflectance at 1064 nm on the incoupling side and AR coated for 1064 nm on the opposite side. A fiber-coupled diode (Osram GmbH) was used for pumping at 808 nm, giving a maximum pump power of 964 mW at a diode temperature of 35°C. The fiber diameter was 100 µm. Using 1:1 imaging, the pump light was focused into the laser crystal. The laser cavity was folded using two dichroic mirrors, M2 and M3. These mirrors were highly reflective (R>99.95%) at 1064 nm for 45°. The reflectivity at 914 nm (45°) was less than 5%. As an output coupler (OC), a mirror with a radius of curvature of 50 mm was employed. Its reflectivity at 1064 nm was specified to 99.95%.

For the non-linear frequency conversion, a 9 mm long PPKTP crystal of thickness 1 mm times 1.5 mm was placed in between the mirrors M2 and M3. The crystal was flux-grown and had a nominal grating period of 6.99 µm designed for first order quasi-phase matching using the d₃₃ coefficient. Domain inversion of the KTP crystal was achieved by electric-field poling. 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 a perfectly poled 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 opposite side. The beam radius of the 1064 nm beam inside the PPKTP was 90 µm. The DPSSL cavity was very compact, the distance from the Nd:YVO₄ crystal to the OC was less than 20 mm.

 

Fig. 1. The experimental set-up.

Download Full Size | PDF

The spectral output of both the LD and the DPSSL was measured with an optical spectrum analyser having a resolution of 0.06 nm. The spectral width of the grating-locked LD radiation was narrower than the resolution of the spectrometer. With the spectrum analyzer three longitudinal modes could be detected, centered at 1064.3 nm and separated by 0.1 nm. This mode separation corresponds to the Nd:YVO₄ crystal length and is attributed to an imperfect AR coating on the output facet of the laser crystal. Possibly several modes could exist around these peaks in the 20 mm long cavity also, but that could not be resolved in spectrum analyzer.

The LD was not single longitudinal mode either, though the total spectral width was narrower than the resolution of the spectrum analyzer. It was calculated that the LD contained approximately 10 longitudinal modes, using that the external cavity of the LD was 6 cm.

3. Results

The DPSSL had a slope efficiency of 24% and emitted a maximum of 23.5 mW, corresponding to 47 W of circulating power inside the laser cavity. Keeping the LD power fixed at 170 mW, the SFM power was measured as a function of the circulating 1064 nm power. A maximum 492-nm output power of 27.4 mW was achieved at a circulating power of 39 W. This corresponds well to the theoretically calculated SFM power of 29 mW, where the experimental focussing conditions and beam overlap were taken into account. Defining the normalized conversion efficiency as

the maximum output power corresponds to ηSFM=0.45 %W^-1cm^-1, where L is the length of the non-linear crystal. The SFM power as a function of circulating 1064 nm power is plotted in Fig.2.

 

Fig. 2. 492-nm CW output power as a function of the circulating 1064 nm power for fixed LD power (at 170 mW).

Download Full Size | PDF

Since the beam width of the DPSSL was twice as large as the width of the LD beam in the PPKTP crystal, it is expected that effectively 60% of the IR power was not used for the nonlinear interaction. This would explain the relatively low conversion efficiency. Saturation in both the SFM power and the normalized conversion efficiency were observed at high IR power, but the slope in the SFM power is nevertheless linear up to about 40 W of circulating power. Using that the absorption coefficient for KTP at this wavelength is in the order of 10^-4 cm^{-1 20}, this would correspond to a few milliwatts of absorbed power. The heat generation in KTP was thus negligible and a thermal lens was not pronounced. The thermal effects in the laser crystal have also been minimal at these relatively low pump powers. This is also confirmed experimentally, since no infrared or SFM beam quality degradation was observed at high circulating power. The saturation apparent in SFM output at the highest intracavity powers can be attributed to the broadening of the longitudinal mode spectrum and can thus not be explained by the thermal-lens related modification of the mismatch in mode overlap.

No gray tracking or power degradation occurred during the experiment, which took place during several hours per day for more than a week.

The slow SFM power fluctuation during a measurement over two hours was typically less than 10% and related to the mechanical stability of the setup. A high-speed Si-photodetector was used for measuring the high-frequency noise of the 492-nm light. A peak-to-peak value of 10% and a RMS value of 2.5% were recorded. The same noise characteristics were observed in the free-running DPSSL laser. Thus we attribute the SFM noise exclusively to the DPSSL laser source. By studying the spectrum of the SFM signal in Fig. 3(a), it is apparent that two modes of oscillation were taking part in the nonlinear interaction. As can be seen from the figure, the overall bandwidth for the 492-nm light was around 0.4 nm. By placing an etalon inside the cavity of the DPSSL the problem of simultaneous SFM of the different longitudinal modes could be overcome and the noise of the SFM signal significantly reduced. In contrast to the blue-green laser sources based on intracavity second harmonic generation where the nonlinear processes lead to longitudinal mode competition and increased noise in the visible output (the so called “green noise” phenomenon²¹), in the case of a purely SFM based visible source the noise can be easily suppressed by stabilising the DPSSL laser and ensuring that the longitudinal mode spectrum fits the bandwidth of the nonlinear crystal.

The maximum power was reached at a KTP temperature of 31.1°C and the effective temperature bandwidth (FWHM) was approximately 5.3°C, see Fig. 3(b). The latter value is larger than the numerically derived²² value of 3.4°C, which can be expected since the multi-mode case gives a larger thermal acceptance bandwidth.

 

Fig. 3. (a). (left) The spectrum of the SFM signal at an output power of 27 mW. The total bandwidth is approximately 0.4 nm. (b). (right) The 492-nm output power as a function of the temperature of the PPKTP crystal. Measured data is represented by the squares and the curve is a sinc-fit of the data. The temperature bandwidth, ΔT_FWHM=5.3 °C.

Download Full Size | PDF

4. Conclusions

In conclusion, we have demonstrated a compact 492-nm laser emitting 27 mW based on sum frequency mixing between a diode-pumped solid-state laser and a single-mode laser diode in an intra-cavity configuration. The laser diode radiation was single-passed through the non-linear crystal, which was PPKTP. Saturation of the output power at 492 nm was pronounced for high circulating power. Noise measurements of the turquoise light gave a RMS value of 2.5%. The main advantage of this scheme is the possibility to reach essentially any wavelength in the visible spectrum by appropriate choice of lasers.