1. Introduction

Laser sources with a narrow linewidth in the middle infrared (MIR) are needed for spectroscopic applications to detect, measure, or manipulate small organic molecules via their fundamental ro-vibrational transitions. An optical parametric oscillator (OPO) is often used to convert laser light at near infrared wavelengths to the MIR region. This is a versatile technique that allows wide tunability and narrow linewidth with continuous-wave (cw) operation. A singly resonant OPO (SRO) design [1] can be constructed using a simple and robust setup. It is driven by a pump beam that passes through a nonlinear crystal and creates signal and idler beams, one of which resonates in an optical cavity containing the crystal.

We have recently developed an SRO system which is pumped by a tunable, single-mode cw Ti:sapphire (titanium-doped sapphire) laser [2]. The properties of the MgO-doped, periodically poled LiNbO₃ (PPLN) crystal allow tuning of the idler beam wavelength across two wide ranges in the MIR with pump wavelengths available using the Ti:sapphire laser. If the output is tuned solely by changing the wavelength of the pump beam, the changes in the SRO resonator are minimal. This avoids practical problems related to changing the temperature, fine adjustments of the alignment, or the poling period of the PPLN crystal inside the SRO resonator and allows rapid adjustments of the idler beam wavelength.

The SRO systems generally require a high-power pump laser beam. Previously, it has been shown that the oscillation threshold, i.e., the minimum power of the pump beam required to overcome the losses of the resonator and start oscillation, can be reduced with the help of stimulated emission in doped glass inserted in the beam path of the SRO resonator [3]. This “hybridly-pumped” system demonstrates excellent stability but is tuned by changing the poling period or temperature of the PPLN crystal.

In this work, we demonstrate, to our knowledge, the first use of a semiconductor-based gain mirror to reduce the oscillation threshold of a continuous-wave SRO system. The system is tuned by changing the pump beam wavelength while keeping the signal beam wavelength in resonance and using the narrow bandwidth of the gain mirror. Longer idler beam wavelengths above 3.6 µm in particular can be widely tuned using this approach. The gain mirror provides gain via stimulated emission. Additional components can be inserted into the SRO resonator and the required minimum optical power of the pump laser to start the SRO is reduced. With a high gain setting, the same system without mechanical changes can be viewed as a laser with an intracavity difference frequency generator controlled by an external, tunable pump laser.

2. Experimental details

The layout of the optical setup used in this work is schematically shown in Fig. 1 . The pump beam is provided by a tunable, continuous-wave Ti:sapphire laser (Coherent MBR) with a single mode output beam. The pump beam wavelength range used in the experiments varied between 775 and 855 nm. A Faraday isolator (Conoptics) is used to prevent back-reflections from disturbing the laser. The optical power was adjustable up to about 4 W after the Faraday isolator. The beam was focused into a 50 mm long nonlinear crystal made of periodically poled lithium niobate (PPLN, HC Photonics). The crystal is MgO-doped and anti-reflection coated (reflectivity R<0.6%). It has several 1 mm thick and 1 mm wide channels with poling periods between 21.25 and 23.50 µm. In this work, we used two poling periods, 21.75 and 23.00 µm, to generate the output idler beam in the MIR, whose wavelength could be tuned depending on the poling period near 3 and 4 µm, respectively. The crystal is placed in the center of a 4-mirror standing wave optical cavity. Standing wave cavity was used because the particular gain mirror used requires that the resonating beam is at normal incidence. The crystal holder allows fine-tuning the alignment and temperature control. In this experiment, the temperature was stabilized slightly above room temperature at 24 degrees Celsius with a stability of 100 mK or better. There was no need to tune or change the crystal temperature during the experiments.

 

Fig. 1 The optical setup of the SRO with gain mirror. HR stands for “highly reflective”, MIR for “middle infrared”, DBS for “dichroic beam splitter”, and DBR for “distributed Bragg reflector”. The green arrows indicate the route of the signal beam in the standing-wave cavity. The schematic structure of the gain mirror is shown on the right.

Download Full Size | PDF

Two concave mirrors (Quality Thin Films) with the radius of curvature of 100 mm are transparent and anti-reflection coated for the pump and idler beams but highly reflecting (HR, R>99.5%) for the signal beam that resonates in the cavity. The two mirrors are symmetrically placed around the crystal with a separation of d ₁ = 146 mm. The pump beam enters through one of the concave mirrors. The idler and remaining pump beams exit through the other concave mirror on the opposite side of the nonlinear crystal. The third mirror (Thorlabs, R>99.5%) is a regular HR mirror for the resonating signal beam and is at the distance d ₂ = 264 mm from the curved mirror. The fourth mirror in the resonator is a flat gain mirror capable of amplifying the resonating signal beam via stimulated emission. Note that the geometry of the SRO was not optimized to provide maximum efficiency and idler output power but rather to match the aperture and optical pumping of the gain mirror. The 1/e² diameter of the signal beam at the secondary focus incident on the gain mirror is about 370 µm and the focusing parameter [4] of the beams within the PPLN crystal is 2.5.

The gain mirror was originally designed for vertical external cavity surface emitting laser (VECSEL) operation [5]. The active region contained 10 GaInAs quantum wells organized in five groups located at the cavity standing wave maxima. Beneath the active region, a 30-layer AlAs/GaAs distributed Bragg reflector was located. The active region thickness was designed to be resonant at the wavelength of 1060 nm. A chip diced of the gain mirror was attached to a wedged diamond heatspreader by liquid capillary bonding and the diamond was attached to a water-cooled copper mount. On top of the diamond, a 2-layer antireflective coating was applied. The gain mirror was pumped by a 808 nm laser bar (Limo) coupled to a 200-µm optical fiber, whose output beam was recollimated and focused to a slightly elliptical spot on the gain mirror. After the SRO tests, the pump spot diameter was measured to be around 0.5 mm. Prior to using the gain mirror in SRO, it was tested in a VECSEL setup. With about 15 W of pump power and a 180 µm pump spot diameter, the output of the laser was centered at 1045 nm with about 2 nm FWHM linewidth. In the SRO experiment, the signal beam is expected to be amplified near the VECSEL lasing wavelength.

A lens to recollimate the exiting beams is placed behind the concave mirror at the output of the resonator. In addition, various filters and/or beam splitters are used to select and attenuate either idler, remaining pump, or leaking signal beam exiting the resonator. The wavelengths of the beams are measured using a wavemeter (Exfo WA-1500) and a spectrum analyzer (Exfo WA-650). Part of the signal beam can be coupled to a Fabry-Perot resonator (Melles Griot Superband, free spectral range of 2 GHz, finesse over 300) to estimate the linewidth and check the single-mode operation of the resonator.

3. Measurements and discussion

Our SRO has two distinct idler beam wavelength ranges of operation, short and long. The short one allows the SRO to produce an idler beam near 3 µm while using a pump beam wavelength shorter than 800 nm and the poling period of 21.75 µm. The long one allows a wide range of idler beam wavelengths around 4 µm with a pump beam wavelength longer than 820 nm and the poling period of 23.00 µm.

In the short idler beam wavelength region, the phase matching requires that the signal beam wavelength changes considerably as a function of the pump beam wavelength. Although the mirror and crystal coatings would allow a signal beam wavelength range of several tens of nanometers, at low pump power the gain mirror was observed to provide gain on a narrow range of wavelengths around 1045 nm. When the pump beam is tuned away from the optimal wavelength, a large change of the signal beam wavelength occurs. This increases losses as the effect of the gain mirror vanishes. As a result, the gain mirror assisted tuning range of the idler beam is only from 3.1 to 3.3 µm. Outside of this range the SRO operates only if the passive reflectivity of the gain mirror is sufficiently high to support conventional SRO operation. Fortunately, the absorbance of the PPLN crystal near 3 µm is low. Thus, hundreds of milliwatts of idler output power is obtained in the short idler beam wavelength region.

In the long idler beam wavelength region, the signal wavelength varies little while tuning the pump beam wavelength. Therefore, the gain mirror assisted experimental wavelength range of the idler beam by pure pump tuning is from 3.6 to 4.7 µm. The output power is limited to a few milliwatts at wavelengths longer than 4 µm due to strong absorption of the idler beam in the PPLN crystal. For example, the idler beam power at 4.5 µm stays below 20 mW even at high pump power levels and multimode operation.

The idler output power of the SRO is a function of the pump beam power and the gain mirror pump power. The total output power is proportional to the pump beam power once the oscillation is initiated and can reach up to 250 mW and 110 mW at idler beam wavelengths of 3.10 and 3.94 µm, respectively. The corresponding pump beam wavelengths are 789 and 830 nm, respectively. In both cases, the output power increases and oscillation threshold decreases as the gain mirror pump power is increased. At high pump power, the gain and the idler output power begin to saturate. In addition, the gain bandwidth becomes wider and shifts to favor longer wavelengths. This leads to longer signal beam wavelengths.

Figures 2a and b) show the oscillation thresholds measured for the pump beam power of the SRO corresponding to several short and long idler beam wavelengths, respectively. Optimally, the oscillation threshold is less than 100 mW. We also observed bistability similar to that described in Ref [3]. As seen in Fig. 2a, the gain mirror works well in reducing the oscillation threshold only when the idler wavelength is close to 3200 nm. When the SRO is tuned with the pump beam and keeping the poling period of the PPLN crystal fixed, the signal wavelength is forced to shift by mode-hopping and the gain mirror is no longer able to amplify it optimally. The dashed horizontal line gives the threshold of a corresponding conventional SRO system. The only difference is that the gain mirror is replaced by a conventional, highly reflecting mirror (Thorlabs). The oscillation threshold of the conventional variant of the system is about 3 W because the SRO is not optimized to minimize the threshold. A similar comparison at the long idler beam wavelengths was impossible because in that case the oscillation threshold could not be reached without the gain mirror. In Fig. 2 b), the oscillation threshold is reduced over a large tuning range. The signal wavelength remains within the gain mirror bandwidth for a wide tuning range, because the phase matching condition requires little change in the signal beam wavelength.

 

Fig. 2 a) The oscillation threshold of the SRO with gain mirror operating at short idler wavelengths. The dashed gray line shows the corresponding experimental oscillation threshold level of a conventional SRO, i.e., when the gain mirror is replaced by a regular highly reflecting mirror. b) The oscillation threshold at long idler wavelengths. The corresponding oscillation threshold using a conventional mirror is unknown due to insufficient pump power.

Download Full Size | PDF

It is interesting to note that at high gain mirror pump power the system behaves like a VECSEL with an intracavity difference frequency generator (DFG). Various intracavity DFG and OPO systems have been studied and constructed earlier using, e.g., Ti:Sapphire and dye lasers [6–8]. For example, a cw intracavity DFG using the resonating laser beam as the pump beam with the signal beam being delivered from the outside has been demonstrated [9]. In our SRO, the pump beam is delivered from the outside while the signal beam resonates and is amplified by the gain mirror. Our setup is simple and has good potential for miniaturization.

Figure 3a shows the pump depletion as a function of the pump beam power with various levels of gain mirror pump power when operating the SRO at 3.10 µm idler beam wavelength. Higher gain provided by the gain mirror increases pump depletion. The pump depletion when using a conventional mirror instead of the gain one is shown for comparison. Figure 3b shows pump depletion as a function of the gain mirror pump power when the SRO pump beam power is constant. Note that, although pump depletion reaches high values, single mode operation occurs near the oscillation threshold. When the SRO is in single-mode operation, the linewidth of the resonating signal beam was observed to be 10 MHz or less.

 

Fig. 3 The pump depletion of the SRO with gain mirror operating at 3.10 µm idler wavelength as a function of a) the tunable pump beam power using various gain mirror power levels and b) the gain mirror pump power at a fixed pump beam power level. The pump depletion of the corresponding, conventional SRO without the gain mirror is also shown in a) for comparison.

Download Full Size | PDF

An important property of the SRO system is its tunability. Because of the gain mirror bandwidth of a few nanometers, the signal beam wavelength stability is improved by inserting a single plate birefringent filter and a YAG etalon inside the SRO resonator. Attempts to scan the SRO without the stabilization provided by these components resulted in mode-hopping. The increased losses due to the additional intracavity components are compensated by increasing the gain mirror pump power accordingly to achieve oscillation. As expected, a wide tuning range is achieved by using the poling period of 23.00 µm.

Figure 4 shows the measured idler beam wavelength while scanning the pump beam wavelength. The SRO was used at a low power level and single mode operation was verified. The measurement was not interrupted during the scan. The scan is performed by combining several short, successive scans of the pump laser. No mode-hops of the resonating signal beam were observed. However, the free spectral range of the SRO resonator for the signal beam is about 200 MHz, which is in the order of magnitude of the resolution of our wavemeter. Therefore, the occurrence of the smallest mode-hops to the adjacent modes of the resonator, although not detected by the wavemeter, cannot be completely ruled out. In the middle of the measurement, the pump laser makes a large discontinuous mode-hop due to the properties of the pump laser. We wanted to check the stability of the system during the mid-scan mode-hop, but the system can be adjusted so that the mode-hop occurs at the end of the scan. Therefore, the maximum scanning range is about 220 GHz. A larger scanning range would require that the SRO oscillation is not disturbed or interrupted. As a result, the continuous scanning range of the SRO is probably larger using another pump laser.

 

Fig. 4 Scanning the SRO with the gain mirror. A birefringent filter and etalon were inserted into the cavity to improve stability. The discontinuity is due to an etalon mode-hop in the pump laser.

Download Full Size | PDF

One of the two most important undesired features of the SRO is the low output power, which is related to the reduced input power of the pump beam. There are less pump photons available for the frequency conversion. It should be possible to improve the efficiency at low power levels by optimizing the oscillator cavity and the mode of the pump beam. The second undesired feature is such that the SRO has a tendency for multi-mode operation unless the power levels are near oscillation threshold. The size, shape, and exact location of the 808 nm pump beam spot on the gain mirror surface is known to influence the multi-mode tendency of VECSELs. The pump beam is not at normal incidence to the surface, which makes the projected spot shape slightly elliptical and its optimal size is tricky to determine. It turned out that our pump beam spot size was slightly larger than the resonating signal beam diameter at the gain mirror. In addition, the closer the SRO is to the condition where it can be regarded as intracavity difference-frequency generator, the more prone it is to multi-mode operation.

The key advantage of using a gain mirror in an SRO is that the oscillation threshold can be significantly reduced down to and beyond the point where the system is no longer an SRO but can be considered a laser with intracavity difference-frequency generation. This opens the possibility to use tunable pump lasers which are more compact, robust, and affordable than a Ti:sapphire laser system. It may also be possible to use an electrically pumped gain mirror, which would facilitate the design of a compact SRO system. Another great benefit is that the whole system can be tuned across a large range of wavelengths above 3.6 µm without changes in the resonator, such as temperature tuning or switching of the poling period, as long as the pump laser is tunable across the suitable wavelength range above 800 nm.

4. Summary

We have constructed a cw SRO system that uses a semiconductor gain mirror in the resonator to reduce the oscillation threshold. The system operates across a large range of idler beam wavelengths in the middle infrared and can be tuned by changing the wavelength of the pump beam. Wide scans are possible using wavelength ranges where the phase matching condition of the PPLN crystal allows the signal beam wavelength to be kept nearly constant. For example, a single nonlinear crystal and poling period are sufficient to reach idler wavelengths between 3.6 and 4.7 µm. The system needs no other resonating beams or additional nonlinear or doped bulk crystals. However, various optical components can be added inside the SRO resonator to improve stability during the scans, because the gain mirror can compensate for the additional losses. Although we use a powerful Ti:sapphire laser as the source for the pump beam, this needs not to be the case. A simpler and more affordable pump laser with modest optical output power could be used as well to construct a compact and widely tunable device.