The widespread availability of quasi-phase-matched (QPM) nonlinear optical materials has, in recent years, allowed the rapid development of low-threshold continuous-wave (cw) optical parametric oscillators (OPOs) as practical sources of tunable narrow-band radiation in the near and mid infra-red. In particular, periodically-poled lithium niobate (PPLN) has allowed the demonstration of devices having exceptional stability [1], extremely low thresholds [2,3] and wide continuous-tuning ranges [4,5]. This class of device shows considerable promise for spectroscopic applications such as gas-sensing [4] and metrology [6]. Despite these successful demonstrations, however, PPLN is, in many ways, a less than ideal material. Set against the advantages of a large nonlinear coefficient, widespread availability and a well-developed material technology are the major problems of significant photorefractive effects and a high susceptibility to thermally-induced refractive index changes (thermal lensing). Preventing the first of these generally requires the crystal to be maintained at elevated temperatures (typically 150–200°C) while the second can result in appreciable changes in cavity geometry for even relatively low levels of absorbed power. Mainly in response to these two problems, the development of alternative QPM materials has been the focus of considerable activity.

The arsenates and phosphates of the family including KTiOPO₄ (KTP) and its isomorphs RbTiOAsO₄ (RTA) and KTiOAsO₄ (KTA) are particularly promising in this context. Despite having a d₃₃ coefficient approximately 60% that of PPLN, the lack of photorefractive effects and a reduced susceptibility to thermal lensing make these materials attractive alternatives for the development of practical cw OPOs. Several cw OPO devices have been demonstrated based on PPKTP [7,8] and PPRTA [9,10]. Of these two materials, PPRTA is extremely attractive having a similar d₃₃ coefficient (≈16pmV^-1) to KTP but an extended infra-red transmission range, with z-polarised transmission falling to 50% at 4.1µm as opposed to 3.3µm in KTP [11]. To date, both reported demonstrations of cw OPOs based on PPRTA have utilised devices with an intracavity geometry, the nonlinear crystal being placed inside the pump laser cavity to exploit the high circulating fields [9,10]. While this approach has a number of advantages, principally the simplicity with which it allows a singly-resonant OPO to be implemented, achieving stable, single-mode operation of the OPO can prove difficult. Pump-enhanced singly-resonant OPOs (PE-SROs) have been shown to be well suited to stabilised operation [1,2] and we describe such a device based on PPRTA here. We believe this to be the first report of a cw PPRTA OPO other than in an intracavity configuration.

A Nd:YVO₄ laser pumped by a 3W broad-area diode laser was used as the OPO pump source. The laser utilised a standing-wave cavity with an etalon to obtain single-mode operation and a piezo actuated output mirror to give fine frequency control. Up to 1W of single-mode output at 1.064µm could be obtained with a continuous tuning range of up to 3GHz. The laser output passed through a 60dB optical isolator before being focussed into the OPO cavity by an 80mm focal length lens. A half wave plate and polariser were used to control the pump power entering the OPO. The OPO cavity was formed by two 30mm radius of curvature mirrors which were coated for a pump wavelength of 1.064µm, a signal of ≈1.55µm and an idler of ≈3.4µm. The input coupler was highly reflecting for the signal (R>99.9%) and idler (R>99%) with ≈6% transmission at the pump while the output mirror was highly reflecting for the pump and signal with ≈94% transmission at the idler. Both faces of the PPRTA crystal were anti-reflection coated (R<0.2%) at the pump, signal and idler wavelengths. The crystal was mounted in an oven at the centre of the cavity and the cavity length adjusted to be near-concentric, giving a beam waist radius of ≈25µm for the pump and ≈30µm for the signal at the centre of the crystal. Control of the OPO cavity length was achieved by the use of a piezo actuator on which the input mirror was mounted.

To maintain the OPO cavity on resonance for the pump, the Pound-Drever-Hall locking scheme was used. The pump beam was passed through a 75MHz lithium niobate phase modulator before entering the optical isolator. Back-reflected light from the OPO cavity was collected on rejection from the optical isolator and focussed onto a high-bandwidth InGaAs photodiode, the output of which was then mixed with a portion of the signal driving the phase modulator to derive an error signal. This was then fed to a P-I servo amplifier, the output of which actuated the OPO input-mirror piezo.

The PPRTA crystal used (Crystal Associates, supplied 1999) measured 20×10×3mm along the x, y and z axes, respectively and was poled in the z direction with a 39.6µm period grating. Previous experience with this crystal had shown it to have somewhat variable performance across its aperture. It was, therefore, important to identify the regions of the crystal likely to yield the lowest OPO threshold. To this end, the crystal was initially investigated as part of an intra-cavity singly-resonant OPO (IC-SRO) [10] in which it was known that threshold could be achieved. This device consisted of a diode-pumped Nd:YVO₄ laser inside the cavity of which the PPRTA crystal was located. The signal-resonant OPO cavity was common with the laser cavity in the region containing the PPRTA crystal and otherwise separated from the laser cavity by a dichroic beamsplitter. Thus, the laser could be operated while OPO operation was suppressed simply by blocking the section of the OPO cavity not common to that of the laser.

To assess the quality of the crystal, the laser output power in the absence of IC-SRO operation and the idler output power with the IC-SRO operating were both recorded as a function of location in the crystal aperture. The results of these measurements are shown in Fig. 1. It can be seen that while the laser output power shows a high degree of consistency over the crystal aperture, the area yielding significant idler output powers is highly localised. Indeed, over most of the crystal aperture the IC-SRO did not operate at all.

 

Fig. 1. Performance mapping across aperture of PPRTA crystal in IC-SRO. Lefthand plot shows laser output power alone, with OPO operation suppressed. Righthand plot shows idler output power of the IC-SRO when operating.

Download Full Size | PDF

Given the largely consistent nature of the laser performance, we conclude that the loss properties of the crystal and its coatings were, generally, similar over the whole aperture. It seems unlikely that this would be the case for the laser (IC-SRO pump) wavelength but not for the IC-SRO signal wavelength except in wholly unusual circumstances. We therefore conclude that the highly localised nature of the region of optimum IC-SRO performance is due to spatial variation in the effective nonlinearity of the PPRTA crystal. Since it seems unlikely that the intrinsic second-order susceptibility of the material should show any spatial variation in magnitude, we attribute this result to the quality of the periodic poling which appears to have only been wholly successful in the region indicated by optimum OPO operation. Although similar measurements in PPRTA by other workers have yielded more homogeneous results [12], it would appear that the periodic-poling process for this material requires further refinement.

Having identified the optimum region of the crystal aperture for OPO operation, the PPRTA crystal was incorporated into the PE-SRO. To measure threshold and output power characteristics, the incident pump power was varied and the idler output power measured together with the pump and signal leakage powers from the output mirror. These measurements were corrected for the output mirror idler transmission (94%) to obtain a measure of the generated idler power and for the measured pump transmission (0.1%) to determine intracavity pump power levels. In a PE-SRO with a double-passed idler, the threshold depends strongly on the relative phases of the interacting waves. In view of this, the pump frequency was varied to maximise the idler output power for each measurement and it was assumed that this achieved optimum phasing. The results of these measurements are shown in Fig. 2. Also shown is the circulating pump power measured with OPO operation suppressed by moving the crystal away from the optimum region indicated in Fig. 1.

 

Fig. 2. Variation of PE-SRO signal (1.560µm) leakage, idler (3.346µm) output powers and circulating pump power with incident pump power. Circulating power was inferred from leakage powers and measured output mirror transmission levels.

Download Full Size | PDF

It can be seen from Fig. 2 that the maximum idler power generated was 87mW, corresponding to 82mW actually extracted from the OPO. Figure 2 also shows that the external threshold for the PE-SRO was around 250mW, corresponding to a circulating pump power of 4.9W. It can be seen that in the absence of OPO operation, the circulating pump power increases linearly with incident power, as would be expected. With the OPO operating, the circulating pump power, characteristically for a PE-SRO [13], clamps at the OPO threshold level. It is interesting to note that the clamped level, and therefore the OPO internal threshold, decreases with increasing incident pump power reaching a level of 2.9W for 900mW incident pump. We attribute this behaviour to changes in the thermal lens in the PPRTA crystal modifying the cavity geometry, as the PE-SRO cavity was optimised for best performance at maximum pump power. The virtually linear variation of idler output with pump power at higher powers, rather than the expected square-root dependence [13], is attributed to the same mechanism. The lowest internal threshold, implied by the clamped intracavity pump power, of 2.9W agrees well with that calculated, based on loss-estimates from manufacturers’ supplied coating performance data and on a d₃₃ coefficient for RTA of 15.8pmV^-1 [12], of around 2.3W, assuming optimum phasing. This result would seem to indicate that, in the regions of the crystal where OPO operation was achievable, the effective nonlinear coefficient was around 89% of the theoretical maximum and the periodic poling was, therefore, of a reasonably high quality. Up to 26.2mW of signal leakage was measured from the output mirror implying a circulating signal field, based on a measured output mirror transmission of 0.04% at 1.56µm, of 65.5W. We believe the apparent saturation, not reflected in the idler, of the signal output at high powers is an experimental artifact rather than a genuine characteristic of the PE-SRO. A short delay generally existed between the measurement of idler output, optimised as described above, and of the signal. We suspect that during this delay the OPO, on some occasions, reverted to non-optimum operating conditions reducing the output power.

Coarse tuning of the PE-SRO output wavelengths was achieved by varying the temperature of the PPRTA crystal as shown in Fig. 3. The theoretical tuning data was calculated from the Sellmeier equations of Fradkin-Kashi et al. [14] and the temperature coefficients of Peltz et al. [12]. The tuning range can be seen to cover 1.525–1.583µm and 3.245–3.520µm for signal and idler, respectively. This was limited by the temperature range used, although sharp increases in threshold at both ends of the measured range indicated that the maximum range, limited by the coating bandwidths, was not significantly greater. It should be noted that the idler range covers a section of the so-called “fingerprint region” for gas detection and also that these idler wavelengths would suffer significant absorption in KTP [11]. The experimental tuning data agrees well with that calculated at lower temperatures, and deviates from the calculated values at higher temperatures, indicating that the temperature coefficients of the PPRTA sample used differ slightly from those given in [12].

 

Fig. 3. Temperature tuning of the PPRTA PE-SRO. Theoretical data was calculated from the Sellmeier equation of [14] and the temperature coefficients of [12].

Download Full Size | PDF

The signal and idler frequencies were fine-tuned by varying the pump laser frequency with the PE-SRO cavity locked to the pump, a method which has been previously analysed for a PE-SRO [5]. The results of such fine-tuning are shown in Fig. 4, the pump and signal frequencies having been monitored on a wavemeter (Agilent model 86120B). It can be seen from Fig. 4 that sections of smooth tuning exist with the signal frequency changing at approximately the wavelength ratio determined rate [5] of Δν_s=0.67Δν_p. The maximum continuous tuning range observed was approximately 0.7GHz, slightly less than the value of ≈1GHz predicted by the analysis of [5]. It can also be seen that mode hops over frequency intervals considerably greater than the 1.7GHz free spectral range of the cavity occur. This behaviour differs from the analysis of [5] which assumes that phase-matching shifts cause mode-hops to occur to an adjacent mode, closer to the optimally phase-matched wavelength and that the cavity modes and phase-matching are the only significant mechanisms for frequency selection. We attribute the large, seemingly unpredictable, mode hops in the device described here to other factors, such as weak parasitic resonances from the crystal faces and optimal phase selection by the double-passed idler, which cause frequency-dependent changes in the OPO threshold and provide stronger mode-selection than phase-matching alone. Given that we calculate the phase-matching bandwidth at these wavelengths to be ≈0.8THz, even a slight change in threshold due to one of these factors would be the dominant mode selection process over intervals of many free spectral ranges.

 

Fig. 4. Fine tuning of PE-SRO output frequencies by pump frequency tuning. Lower plot shows tuning over full 3GHz continuous tuning range of pump laser. Upper plot shows magnified section of signal tuning range where continuous OPO tuning occurred. Free spectral range of the OPO cavity at the signal (FSR_s) is indicated on both plots.

Download Full Size | PDF

The effective frequency stability of the PE-SRO output would be expected to mirror that of the pump laser. When tuned to a region of stable, single-mode operation the OPO typically showed a frequency stability of better than 1GHz (approximately the maximum smooth-tuning range) and a corresponding power stability of around 5% for periods of the order of tens of minutes. Any disturbance or excessive frequency drift of the pump laser, which was not actively stabilised, generally resulted in the OPO mode hopping and showing large fluctuations in output power.

To conclude, we have described the operation of a pump-enhanced continuous-wave OPO based on PPRTA. We believe this is the first demonstration of a cw PPRTA OPO other than in an intracavity configuration. The OPO had an external threshold of 250mW and up to 87mW of idler at ≈3.35µm was generated. Coarse tuning, achieved by varying the temperature of the PPRTA crystal, covered ranges 1.525–1.583µm and 3.245–3.520µm for signal and idler, respectively. The idler range covers part of an important spectral region for gas sensing and molecular spectroscopy over which PPKTP would have exhibited significant absorption. Fine tuning of the OPO, by varying the pump frequency, was somewhat erratic exhibiting large, unpredictable mode-hops. We attribute this to processes such as parasitic resonance effects and optimum phase selection by the double-passed idler providing stronger frequency-dependent mode selection than phase-matching alone. Only a small area of the PPRTA crystal aperture provided sufficient parametric gain for OPO operation due, we believe, to poling not being wholly successful in the majority of the crystal. It would, therefore, appear that the periodic poling process for RTA requires further refinement. In spite of this, the measured threshold indicates that the effective nonlinear coefficient in the poled region was close to the optimum value and we believe that PPRTA holds considerable promise for stable, low-power cw OPOs operating in the near and mid infrared.