1. Introduction

Continuous-wave (cw) optical parametric oscillators (OPOs) represent versatile sources of tunable single-frequency radiation in spectral regions inaccessible to lasers. Before the development of quasi-phase-matched (QPM) nonlinear materials, operation of cw OPOs in the singly-resonant oscillator (SRO) configuration was difficult due to the high oscillation thresholds (typically several watts) in birefringent crystals [1]. The advent of periodically-poled materials, in particular periodically-poled lithium niobate (PPLN), heralded major breakthroughs in cw OPO technology, bringing the operation threshold of SROs within the reach of moderate to high-power laser pump sources [2]. Due to their high power (watt-level), single-frequency performance, and simplified fine tuning capability, such cw SROs have since been successfully deployed in applications such as photoacoustic spectroscopy [3].

At the same time, due to the persistently high pump thresholds (still several watts in QPM materials), the common approach to the development of cw SROs has been to deploy an optical cavity with the lowest loss at the resonant signal wave to minimize threshold, while providing maximum output coupling for the non-resonant idler wave in order to achieve the highest power extraction. In PPLN, with a large effective nonlinearity (d _eff~17 pm/V) and long interaction lengths (50–80 mm), this condition can be somewhat relaxed and finite signal output coupling can be tolerated [2,4,5]. However, in other QPM materials any small increase in resonant wave coupling is expected to lead to an unacceptable rise in threshold, rendering SRO operation inefficient or beyond the reach of commonly available cw laser sources.

Here we report operation of a green-pumped cw SRO based on MgO-doped periodically-poled stoichiometric lithium tantalate (MgO:sPPLT) under finite signal output coupling and show that major enhancements in the overall performance of the OPO with regard to output power, extraction efficiency, and useful tuning range can be brought about at little or no cost to threshold, internal conversion efficiency (pump depletion), idler output power and idler extraction efficiency. We also report on the spectral characterization of the signal and idler output and observe thermal effects in the MgO:sPPLT crystal induced by the absorption of the pump as well as the resonant signal wave.

2. Experimental setup

The configuration of the output-coupled SRO (OC-SRO) is identical to that described in our earlier work [6], except for the replacement of mirror M₄ (high reflector at the signal) with an output coupler. The OPO is formed in a ring cavity using two concave (r=50 mm) and two plane mirrors [6]. In SRO arrangement, all mirrors have high reflectivity (R>99.5%) for the resonant signal over 840–1000 nm. In the OC-SRO configuration, one of the plane high reflectors is replaced by an output coupler with varying transmission (T=0.71%–1.1%) across the signal wavelength range. All mirrors have high transmission (T=85–90%) for the idler over 1100–1400 nm, thus ensuring SRO operation in both configurations. The nonlinear crystal is MgO:sPPLT (d _eff~10 pm/V). It is 30-mm long, contains a single grating period of Λ=7.97 µm, and is housed in an oven with a temperature stability of ±0.1 °C. The crystal faces have antireflection (AR) coating (R<0.5%) for the signal (800–1100 nm), with high transmission (T>98%) at 532 nm. The residual reflectivity of the coating is 0.6% to 4% per face for the idler (1100–1400 nm). A 500-µm uncoated fused silica etalon (FSR=206 GHz, finesse~0.6) is used at the second cavity waist for frequency selection. The pump source is a frequency-doubled, single-frequency cw Nd:YVO₄ laser at 532 nm, as described previously [6,7].

In order to directly compare the performance of OPO in SRO and OC-SRO configurations, we used identical operating conditions in both cases using the same focusing and mode-matching parameters for the pump and the resonant signal. We used a fairly strong focusing parameter of ξ=2, corresponding to a pump beam radius of w _op=24 µm inside the crystal [7]. The signal beam waist was w _os~31 µm, resulting in optimum mode-matching to pump (b _s=b _p). We performed measurements of idler and signal output power, extraction efficiency, pump depletion, photon conversion efficiency, oscillation threshold, and spectral characteristics in the two configurations.

3. Results and discussion

Figure 1(a) shows the idler and signal output power and the corresponding pump depletion for the SRO with minimal signal coupling versus pump power at the input to the crystal. The idler powers correspond to the output through the second concave mirror after filtering the pump [6,7], whereas the signal powers correspond to the usable output through one of the plane high reflectors. The measurements were preformed at a crystal temperature of 80 °C (signal=966 nm, idler=1184 nm) near the maximum of idler output power where the effects of thermal lensing, crystal coating loss and gain reduction due to degeneracy factor are minimized [6,7]. The idler power reaches a maximum 2.33 W at 8.6 W of pump, corresponding to an extraction efficiency of 27.1%. The usable signal power, however, remains limited for all pump powers, as expected for a SRO with minimum signal coupling. It reaches 104 mW at 8.6 W of pump, resulting in a useful extraction efficiency of 1.2%. The maximum total extraction efficiency of the SRO is thus 28.3% and threshold is reached at 2.41 W of pump power. The SRO pump depletion rises rapidly reaches a maximum of 73% at 4.68 W of pump before the onset of saturation. This effect, which has also been observed in our earlier work [6,7], is attributed to the onset of back-conversion and is qualitatively consistent with theory [8].

 

Fig. 1. Extracted signal power, idler output power, and pump depletion as functions of pump power in (a) SRO, and (b) OC-SRO. The solid and dashed curves are guide for the eye.

Download Full Size | PDF

Figure 1(b) represents the same plots as in Fig. 1 (a), but for the OC-SRO with finite signal coupling, where one of the plane high reflectors is replaced by an output coupler. The signal powers correspond to the usable output through the output coupler, which has 0.98% transmission at 966 nm. The idler power in this case reaches a maximum of 2.53 W at 8.83 W of pump, corresponding to an extraction efficiency of 28.6%, which is comparable to the 28.3% for the SRO in Fig. 1(a). However, the usable signal power simultaneously generated now reaches 945 mW at 8.83 W of pump, representing an extraction efficiency of 10.7%. The maximum total output power is now 3.48 W and the total extraction efficiency is thus 39.3%, representing an enhancement of 11% from the SRO. The pump depletion now reaches a maximum of ~70% at 6.6 W of pump before saturation takes effect. The pump power threshold for the OC-SRO is now ~2.99 W, representing an increase of only 24% (580 mW) over the SRO. It is thus clear that while the OC-SRO exhibits an insignificant increase in pump power threshold, it provides substantial enhancements in usable signal and idler power and total extraction efficiency. At the same time, the OC-SRO maintains similar idler power, idler extraction efficiency, and pump depletion as the SRO.

We also recorded the extracted signal and idler output power and the corresponding pump depletions across the tuning range for both the SRO and OC-SRO. The results for the SRO are shown in Fig. 2(a). The data were obtained for a change in crystal temperature from 71 °C to 245 °C [6] and were recorded near the peak of the signal and idler powers in Figs. 1(a) and 1(b). In the SRO, the idler power varies from 2.46 W at 1167 nm (T=71 °C) to 487 mW at 1427 nm (T=245 °C). The signal power leakage through one of the plane high reflectors, however, remains below 56 mW across most of the signal tuning range from 978 nm to 866 nm before rising to 195 mW at 848 nm due to increase in signal transmission of the plane high reflector. The pump depletion remains close to ~70% over much of the tuning range, decreasing to ~40% towards the extremes of the tuning range at 1427 nm (idler) and 848 nm (signal). The maximum combined power and total extraction efficiency across the tuning range are 2.52 W and 30%, respectively, at 1190 nm (idler) and 962 nm (signal) at 83 °C. The decline in idler power and pump depletion towards the extreme of the tuning range is consistent with the SRO behavior observed previously [6,7], attributed to the increased effects of thermal lensing at higher temperatures (longer idler, shorter signal wavelengths), crystal coating losses, and parametric gain reduction away from degeneracy.

The same plots as in Fig. 2(a), but for the OC-SRO with finite signal coupling is shown in Fig. 2(b). In this case, the idler power varies from 2.57 W at 1167 nm to 2.59 W at 1190 nm, to 480 mW at 1427 nm, with the pump depletion again remaining close to ~70% over most of the tuning range before declining to ~35% at the extreme of the tuning range. Thus, despite the increased signal coupling, the idler power and pump depletion in OC-SRO remain similar to the SRO in Fig. 2(a). Importantly, however, the OC-SRO can now provide substantial signal powers of up to 1.23 W across the tuning range. The signal output varies from 915 mW at 978 nm to 1.23 W at 925 nm, to 278 mW at 848 nm. The maximum total output power and extraction efficiency across the tuning range are now 3.6 W and 40%, respectively, at 1190 nm (idler) and 962 nm (signal) at 83 °C, representing a 1.08 W increase in output power and 10% in extraction efficiency. Notwithstanding the expected decline in the signal power towards shorter wavelengths [6,7], the variation in the extracted signal power closely follows the output coupler transmission across the tuning range, as shown in Fig. 2(c). From the curve, the optimum value of output coupling in the present device is 1.04% at 925 nm.

 

Fig. 2. Extracted signal power, idler output power, and pump depletion across the tuning range for (a) SRO, (b) OC-SRO. (c) Output coupler transmission over the signal tuning range.

Download Full Size | PDF

It is, therefore, clear that the OC-SRO can provide substantial enhancements in the usable signal power by as much as 1.23 W, in total output power by as much as 1.08 W, and in total extraction efficiency by as much as 10% compared to the SRO. The scheme also effectively expands the useful tuning range of the device by 130 nm into the 848–978 nm signal region. Moreover, the improvement in performance is obtained at little or no expense to idler power or pump depletion across the tuning range.

We performed spectral characterization of the OC-SRO signal and idler output using a confocal Fabry-Perot interferometer [9,10] (FSR=1 GHz, finesse=400 at both wavelengths). The measurements were performed near the maximum of signal and idler power in Fig. 1(b) at 966 nm and 1184 nm, respectively. We observed reliable single-frequency operation at both the signal and idler, but we found that the signal exhibited a substantially narrower linewidth than the idler. For measurements of linewidth, we recorded the fringe pattern for the signal and idler at different instants of time and deduced the corresponding FWHM linewidths. Typical transmission fringes are shown in Fig. 3(a) for the signal and in Fig. 3(b) for the idler. The calculated linewidths over three measurements are also shown in table 1, where a linewidth of ~3 MHz can be deduced for the resonant signal. The corresponding idler exhibits a linewidth ~7 MHz, consistent with our previous observations [6], confirming a substantially narrower linewidth for the resonant signal than the non-resonant idler. This result is not unexpected, since the resonance of signal wave leads to spectral confinement by finesse of the optical cavity. On the other hand, the non-resonant idler spectrum can more freely adjust to satisfy the energy conservation condition, thus resulting in broader bandwidth. This property can have useful implications for the utility of OC-SROs in spectroscopic applications. It suggests that by choosing the spectroscopic wavelength of interest as the resonant wave rather than the non-resonant wave it would be possible to enhance the measurement resolution, while still providing substantial output powers.

 

Fig. 3. Fabry-Perot transmission fringes for (a) signal, and (b) idler at a crystal temperature 80°C.

Download Full Size | PDF

Table 1. Measured linewidth (FWHM) of signal and idler.

View Table

Investigation of the OC-SRO output also revealed significant shift in output wavelength as a function of pump power. This effect, which arises from heating of the crystal, has been previously observed in PPLN under strong pumping [5] and attributed to intracavity signal absorption. Here we find that, unlike in PPLN, absorption of pump in the MgO:sPPLT sample also plays an important role in crystal heating and thus the output wavelength shift. At a nominal (oven) temperature of 80 °C, we measured the signal wavelength using a wavemeter (Burleigh, WA-1000, resolution 0.001nm) as a function of pump power. At the maximum pump power of 9.1 W, we recorded a signal wavelength of 966.4 nm, while at a pump power of 4.8 W, a wavelength of 972.9 nm was measured. Using the data, calculations of phase-matching based on Sellmeier equations for stoichiometric LiTaO₃ [11] resulted in a crystal temperature of 84.7 °C at 9.1 W of pump and 80.7 °C at 4.8 W of pump, implying a rise of 4 °C in crystal temperature with increased pump power, as shown in Fig. 4 (a).

In order to separate the contributions of the pump and intra-cavity signal, at the full pump power of 9.1 W, we rotated the pump polarization away from phase-matching until the OC-SRO was just above threshold. Under this condition, we recorded a signal wavelength of 970.5 nm corresponding to a crystal temperature of 82.1 °C as shown in Fig. 4(b), indicating a still significant rise despite negligible circulating signal power, and thus confirming the contribution of the pump to crystal absorption and heating. Other possible contributions to crystal heating may be green-induced infrared or idler absorption. However, the contribution of green-induced infrared absorption is expected to be low in MgO:sPPLT [12] and the contribution due to the absorption of the single pass idler is also expected to be negligible.

 

Fig. 4. Variation of crystal temperature with the (a) increase of pump power and intracavity signal power and (b) increase of intracavity signal power at fixed pump power.

Download Full Size | PDF

4. Conclusions

In conclusion we have demonstrated that by deploying finite output coupling of the resonant wave in a SRO, it is possible to enhance the overall output power, extraction efficiency, useful tuning range and spectral purity with little or no sacrifice to threshold, idler power, or pump depletion. We have also observed improved spectral purity of the resonant signal compared with the non-resonant idler, implying the advantage of exploiting the resonant out-coupled wave for spectroscopic applications. Measurements of crystal heating effects have also confirmed significant contribution of the pump as well as intracvity signal absorption to output wavelength shift with increasing pump powers. The OC-SRO could also offer additional advantages over SRO. In near-degenerate operation, for example, it could provide improved output power in a single beam with a finite spectrum, which could be useful for cascaded pumping of mid-infrared OPOs or in applications were high-power and broadband cw sources are simultaneously required.

This research was supported, in part, by the Ministry of Education and Science of Spain under grant Ref. No. TEC2006–12360.