772
OPTICS LETTERS / Vol. 40, No. 5 / March 1, 2015
Continuous-wave whispering-gallery optical parametric oscillator for high-resolution spectroscopy Christoph S. Werner,1 Karsten Buse,1,2 and Ingo Breunig1,* 1
2
University of Freiburg, IMTEK, Georges-Köhler-Allee 102, 79110 Freiburg, Germany Fraunhofer Institute for Physical Measurement Techniques, Heidenhofstraße 8, 79110 Freiburg, Germany *Corresponding author: [email protected] Received December 4, 2014; revised January 20, 2015; accepted January 21, 2015; posted January 21, 2015 (Doc. ID 229013); published February 20, 2015
We achieve a continuous operation of a whispering gallery optical parametric oscillator by stabilizing the resonator temperature T on the mK level and simultaneously locking the pump frequency to a cavity resonance using the Pound–Drever–Hall technique. The millimeter-sized device converts several mW of a pump wave at 1040 nm wavelength to signal and idler waves around 2000 nm wavelength with more than 50% efficiency. Over 1 h, power and frequency of the signal wave vary by < 1% and by < 25 MHz, respectively. The latter can be tuned over 480 MHz without a mode hop by changing T over 120 mK. In order to prove the suitability for high-resolution spectroscopy, we scan the signal frequency across the resonance of a Fabry–Perot interferometer resolving nicely its 10 MHz linewidth. © 2015 Optical Society of America OCIS codes: (190.0190) Nonlinear optics; (190.4970) Parametric oscillators and amplifiers; (230.5750) Resonators. http://dx.doi.org/10.1364/OL.40.000772
Continuous-wave optical parametric oscillators (OPOs) convert monochromatic pump light into two monochromatic and wavelength-tunable light beams (signal and idler). Standard OPOs comprise a nonlinear-optical medium surrounded by a mirror cavity that is resonant either for one, for two, or for all three interacting waves. All of these resonator configurations have been stabilized actively—see [1] and references therein—enabling continuous and mode-hop-free emission over hours. Due to the combination of such a stability with the wavelength flexibility mentioned above, OPOs are nowadays important sources for coherent and nonclassical light. Some years ago, a new kind of OPOs based on whispering gallery resonators (WGRs) was demonstrated [2]. In these devices the three interacting waves propagate due to total internal reflection in a monolithic spheroidally-shaped resonator made of a nonlinear-optical medium. Compared with their mirror-based counterparts, WGR-OPOs have some advantages. They are more compact, the whole cavity measures typically some mm in diameter. There is no need to align any cavity mirrors. Furthermore, the resonator is intrinsically mechanically stable. WGR-OPOs emitting light in the visible, near-, and mid-infrared have been demonstrated [2–4]. Also the emission of nonclassical light has been proven [5,6]. Realizing a WGR-OPO, one faces a triply-resonant system. Thus, a continuous emission requires the active stabilization of the WGR, because the frequency of the pump laser has to be kept at a resonance of the cavity. However, all WGR-OPOs demonstrated so far operate in a noncontinuous mode. The frequency of the pump wave is swept forth and back across a cavity resonance. Occasionally a triple-resonance is hit and oscillation starts. Real continuous-wave operation and stabilization have not been realized so far. In this report, we demonstrate a whispering gallery OPO continuously emitting light around 2 μm wavelength. This is achieved by locking the pump frequency to a cavity resonance as it was done before in experiments with triply-resonant mirror OPOs [7]. We investigate the power and frequency stability, as well as the mode-hop-free tuning of the light emitted. 0146-9592/15/050772-04$15.00/0
Finally, we prove that a WGR-OPO can be employed in high-resolution spectroscopy. In our experiments, a grating-stabilized diode laser emitting at 1040 nm wavelength serves as the pump source. Despite being widely tunable, this laser allows fast current modulation to generate the sidebands necessary for the Pound–Drever–Hall (PDH) locking technique without any additional electro-optic modulator. Coupling of the pump laser light into the WGR is achieved by prism coupling with a rutile prism, while the gap between the WGR and the prism can be adjusted through a piezo translator to change the coupling regime from strong over- to undercoupling at will. The resonator used in our experiment is made of z-cut magnesium-oxide-doped congruent lithium niobate. All three interacting waves are extraordinarily polarized, while quasi-phase matching is achieved through a linear domain pattern similar to the structure employed in the pioneering first demonstration of frequency doubling in a WGR [8]. Although a radial domain structure would decrease the oscillation threshold and certainly enable a more controlled coarse tuning, we use a linear domain pattern because it is easier to fabricate and readily available. We assume that the fine tuning behavior investigated in this report will not suffer significantly from a nonoptimized domain pattern. The periodicity of 26.5 μm quasi-phase matches OPO processes pumped at 1 μm generating signal and idler fields around 2 μm wavelength. In a WGR, the geometrical dimensions of the cavity influence the dispersion relation [9,10] and hence the phase matching conditions. To optimize the process for room temperature, we manufactured the resonator with a radius of approximately 1.55 mm. The unloaded quality factor of the resonator at the pump wavelength was determined to be approximately 5 × 107 . A closed-loop temperature control with a stability better than 1 mK in combination with an air-tight housing around the resonator and prism ensure a good temperature stability. The housing is made of PMMA and has an entrance window made of NBK-7 to maintain the circular beam profile of the pump laser while the signal, idler, and © 2015 Optical Society of America
March 1, 2015 / Vol. 40, No. 5 / OPTICS LETTERS
Fig. 1. Schematic of the experimental setup: the light of an externally stabilized external-cavity diode laser (ECDL) with the pump power P p is coupled into a whispering gallery resonator (WGR) using a prism coupler. An optical low-pass filter (LP) separates the generated signal and idler light from the pump light. The photodiode D1 detects the transmitted pump light and is fed into the Pound–Drever–Hall (PDH) stabilization system. Some portion of the signal and idler light is coupled via a beamsplitter (BS) into a spectrum analyzer. The remaining total signal and idler power P si can be monitored with the photodiode D2 ; alternatively, the light is analyzed by a Fabry–Perot interferometer (FPI).
pump waves leave the housing through the PMMA as illustrated in Fig. 1. The good temperature stability is necessary in our experiment because the signal and idler resonances are not actively stabilized. Here, continuouswave operation is possible only as long as the temperature stays constant. The fact that pump, signal, and idler waves are all polarized along the optic axis of the crystal comes in favor because the temperature dependence of the refractive indices for the three waves is similar resulting in an increased tolerance against temperature fluctuations. A fast photodiode detects the modulated pump wave that, after demodulation, provides the error signal for locking the laser to the WGR resonances [11], which is then fed back to the pump laser via a PID control. That way, the laser can follow a given resonator mode even during the presence of a nonlinear process that leads to distorted pump resonances [12] as well as to distorted error signals. The bandwidth of the whole feedback loop is limited by the tuning speed of the diode laser and is at the order of a few kilohertz, while the sideband modulation frequency is 12.5 MHz. A fiber-coupled grating-based spectrum analyzer monitors the wavelengths of the signal and idler waves. The remaining light can either be directly detected by an infrared-sensitive photodiode for power measurements or coupled into a scanning Fabry–Perot interferometer (FPI) to monitor the frequency modulation of the signal wave. The photodiode is equipped with a filter providing additional attenuation of more than 70 dB at the pump wavelength ensuring that only the total signal and idler power contributes to the measurement. The FPI can be operated in both scanning and nonscanning mode, and is temperature stabilized to serve as a long-term reference. All powers given are corrected for reflection and absorption losses of the optical
773
components and the housing to obtain the powers that are present inside the coupling prism. In the absence of a nonlinear process, we adjusted the gap between the resonator and the coupling prism until the internal losses matched the coupling losses to achieve critical coupling. Here, we observed a coupling efficiency of 60%, which corresponds to the mode overlap between the external pump laser and the mode inside the resonator. The remaining 40% cannot be coupled into the resonator and are lost for the nonlinear process. Since a perfect mode overlap is generally possible, we corrected the available pump power with the obtained mode overlap to calculate the conversion efficiency. We were able to achieve a combined signal and idler power of P si ≈ 5.2 mW at a pump power of P p ≈ 9.5 mW resulting in a conversion efficiency of 55%. This high efficiency is achieved in the regime of strong overcoupling leading to higher efficiencies at the expense of an increased pump threshold of 2.2 mW. In the case of undercoupling, the threshold reduces to 86 μW, while the efficiency is limited to 3.5%. All these experimental results were obtained while scanning the laser over the desired mode to avoid constant relocking and gain adjustment after changing the pump power. The results are depicted in Fig. 2. In the locked state with 9.5 mW pump power, we achieved a long-term power stability of the signal and idler waves better than 1% over the term of 1 h as shown in Fig. 3. A prominent 30 min cycle is seen in the output power, presumably stemming from the air conditioner in the laboratory.
Fig. 2. Generated combined signal and idler power P si versus the pump power P p . The power values shown are corrected for mode overlap, absorption, and reflection losses (orange circles: experimental data, red line: theoretical fit [13]).
Fig. 3. Power stability of the generated signal and idler light over 1 h.
774
OPTICS LETTERS / Vol. 40, No. 5 / March 1, 2015
process is reversible, although we observed a hysteresis effect presumably stemming from the specific locking behavior and the structure of the pump resonances during the nonlinear process that can introduce additional zero crossings in the error signal. A simplified model assuming that the light is guided at the surface of the resonator and neglecting higher spatial modes gives the dispersion relation ν Fig. 4. Transmission of the signal and idler power through a scanning Fabry–Perot interferometer.
To prove single-frequency operation in the locked state, we analyze the signal wave with the scanning FPI. In Fig. 4, the transmission spectrum of the FPI is shown while the parametric process operates at degeneracy. The measured linewidth of the FPI signal is 10 MHz. This corresponds to the predicted finesse achievable based on the mirror losses. Thus, the 10 MHz are only an upper limit of the signal and idler linewidths. Still locked to the same process, we changed the temperature of the WGR while recording the position of the FPI peaks. In Fig. 5, the change of the output frequency Δν and the corresponding temperature profile are shown. By changing the temperature by 120 mK at an absolute temperature of approximately 26.4°C, the output frequency changes by 480 MHz. The tuning rate is limited by our experimental WGR holder representing a large heat capacitance. Changing the temperature further leads to a mode hop of the signal and idler waves, while the pump laser keeps locked to the resonator mode. This
Fig. 5. Frequency shift Δν of the signal and idler waves (top) and the corresponding change ΔT of the resonator temperature (bottom) versus time.
c0 m 1 ; 2π nTRT
(1)
with the resonance frequency ν, the vacuum light speed c0 , the mode number m for the signal/idler mode, the temperature-dependent resonator radius RT, and refractive index nT. With this formula and considering m 9865 for our case, we can estimate the tuning rate of the resonances at the signal and idler wavelength of 2080 nm. At degeneracy, we find 4.7 MHz∕mK using the temperature dependences given in [14,15]. Our experimental results show a tuning rate of approximately 4 MHz∕mK. Taking into account that the model used neglects that changing the temperature forces the pump, signal, and idler waves to operate off of their respective resonance center and that the temperature coefficients given in the literature may differ from those valid for our sample, the prediction and experimental results are in good agreement. To demonstrate the suitability of the light source for high-resolution spectroscopy, we scan the signal wave over a FPI resonance. This time, the FPI is operated in the nonscanning mode, and the resonator temperature is sweeped while recording the transmitted signal power. As depicted in Fig. 6, the resonator mode is easily
Fig. 6. Transmission (top) of the signal and idler power through the Fabry–Perot interferometer while the temperature (bottom) is ramped up and down. The spectrum shows the same mode scanned from both directions. The steps in the temperature graph are an artifact of the update rate.
March 1, 2015 / Vol. 40, No. 5 / OPTICS LETTERS
775
Financial support from the Deutsche Forschungsgemeinschaft is gratefully acknowledged.
Fig. 7.
1-h frequency stability of the signal and idler light.
resolved in great detail even though the whole measurement took over 1 min. The two transmission peaks shown represent the same FPI resonance scanned in both directions. Considering the tuning rate of 4 MHz∕mK, a linewidth of approximately 10 MHz can be derived, matching the linewidth found in Fig. 4. In general, an important feature of a laser system is its frequency stability. To determine the long-term stability, we scan the FPI and record the drift of the WGR resonance for 1 h while keeping the temperature set-point of the WGR constant. As depicted in Fig. 7, the frequency drift is less than 25 MHz. Like the output power, the frequency drift shows a periodicity of 30 min. In conclusion, by locking a pump laser to a WGR resonance made of periodically poled lithium niobate, we were able to achieve true continuous-wave and single-frequency operation of a triply-resonant optical parametric oscillator. Through temperature sweeping of the resonator, the signal and idler frequencies can be tuned over several hundred MHz allowing for highresolution spectroscopy applications. The frequency and power stability matches that of commercially available frequency-converted laser systems.
References 1. M. Ebrahim-Zadeh, in OSA Handbook of Optics IV, M. Bass, ed. (McGraw-Hill, 2010), p. 17.1. 2. J. U. Fürst, D. V. Strekalov, D. Elser, A. Aiello, U. L. Andersen, C. Marquardt, and G. Leuchs, Phys. Rev. Lett. 105, 263904 (2010). 3. T. Beckmann, H. Linnenbank, H. Steigerwald, D. Haertle, K. Buse, and I. Breunig, Phys. Rev. Lett. 106, 143903 (2011). 4. Ch. S. Werner, T. Beckmann, K. Buse, and I. Breunig, Opt. Lett. 37, 4224 (2012). 5. J. U. Fürst, D. V. Strekalov, D. Elser, A. Aiello, U. L. Andersen, Ch. Marquardt, and G. Leuchs, Phys. Rev. Lett. 106, 113901 (2011). 6. M. Förtsch, J. U. Fürst, C. Wittmann, D. V. Strekalov, A. Aiello, M. V. Chekhova, C. Silberhorn, G. Leuchs, and C. Marquardt, Nat. Commun. 4, 1818 (2013). 7. J. Laurat, L. Longchambon, C. Fabre, and T. Coudreau, Opt. Lett. 30, 1177 (2005). 8. V. S. Ilchenko, A. A. Savchenkov, A. B. Matsko, and L. Maleki, Phys. Rev. Lett. 92, 043903 (2004). 9. I. Breunig, B. Sturman, F. Sedlmeir, H. G. L. Schwefel, and K. Buse, Opt. Express 21, 30683 (2013). 10. M. L. Gorodetsky and A. E. Fomin, IEEE J. Sel. Top. Quantum Electron. 12, 33 (2006). 11. R. W. P. Drever, J. L. Hall, F. V. Kowalski, J. Hough, G. M. Ford, A. J. Munley, and H. Ward, Appl. Phys. B 31, 97 (1983). 12. I. Breunig, B. Sturman, A. Bückle, Ch. S. Werner, and K. Buse, Opt. Lett. 38, 3316 (2013). 13. B. Sturman and I. Breunig, J. Opt. Soc. Am. B 28, 2465 (2011). 14. Y. S. Kim and R. T. Smith, J. Appl. Phys. 40, 4637 (1969). 15. O. Paul, A. Quosig, T. Bauer, M. Nittmann, J. Bartschke, G. Anstett, and J. A. L’Huillier, Appl. Phys. B 86, 111 (2007).
OPTICS LETTERS / Vol. 40, No. 5 / March 1, 2015
Continuous-wave whispering-gallery optical parametric oscillator for high-resolution spectroscopy Christoph S. Werner,1 Karsten Buse,1,2 and Ingo Breunig1,* 1
2
University of Freiburg, IMTEK, Georges-Köhler-Allee 102, 79110 Freiburg, Germany Fraunhofer Institute for Physical Measurement Techniques, Heidenhofstraße 8, 79110 Freiburg, Germany *Corresponding author: [email protected] Received December 4, 2014; revised January 20, 2015; accepted January 21, 2015; posted January 21, 2015 (Doc. ID 229013); published February 20, 2015
We achieve a continuous operation of a whispering gallery optical parametric oscillator by stabilizing the resonator temperature T on the mK level and simultaneously locking the pump frequency to a cavity resonance using the Pound–Drever–Hall technique. The millimeter-sized device converts several mW of a pump wave at 1040 nm wavelength to signal and idler waves around 2000 nm wavelength with more than 50% efficiency. Over 1 h, power and frequency of the signal wave vary by < 1% and by < 25 MHz, respectively. The latter can be tuned over 480 MHz without a mode hop by changing T over 120 mK. In order to prove the suitability for high-resolution spectroscopy, we scan the signal frequency across the resonance of a Fabry–Perot interferometer resolving nicely its 10 MHz linewidth. © 2015 Optical Society of America OCIS codes: (190.0190) Nonlinear optics; (190.4970) Parametric oscillators and amplifiers; (230.5750) Resonators. http://dx.doi.org/10.1364/OL.40.000772
Continuous-wave optical parametric oscillators (OPOs) convert monochromatic pump light into two monochromatic and wavelength-tunable light beams (signal and idler). Standard OPOs comprise a nonlinear-optical medium surrounded by a mirror cavity that is resonant either for one, for two, or for all three interacting waves. All of these resonator configurations have been stabilized actively—see [1] and references therein—enabling continuous and mode-hop-free emission over hours. Due to the combination of such a stability with the wavelength flexibility mentioned above, OPOs are nowadays important sources for coherent and nonclassical light. Some years ago, a new kind of OPOs based on whispering gallery resonators (WGRs) was demonstrated [2]. In these devices the three interacting waves propagate due to total internal reflection in a monolithic spheroidally-shaped resonator made of a nonlinear-optical medium. Compared with their mirror-based counterparts, WGR-OPOs have some advantages. They are more compact, the whole cavity measures typically some mm in diameter. There is no need to align any cavity mirrors. Furthermore, the resonator is intrinsically mechanically stable. WGR-OPOs emitting light in the visible, near-, and mid-infrared have been demonstrated [2–4]. Also the emission of nonclassical light has been proven [5,6]. Realizing a WGR-OPO, one faces a triply-resonant system. Thus, a continuous emission requires the active stabilization of the WGR, because the frequency of the pump laser has to be kept at a resonance of the cavity. However, all WGR-OPOs demonstrated so far operate in a noncontinuous mode. The frequency of the pump wave is swept forth and back across a cavity resonance. Occasionally a triple-resonance is hit and oscillation starts. Real continuous-wave operation and stabilization have not been realized so far. In this report, we demonstrate a whispering gallery OPO continuously emitting light around 2 μm wavelength. This is achieved by locking the pump frequency to a cavity resonance as it was done before in experiments with triply-resonant mirror OPOs [7]. We investigate the power and frequency stability, as well as the mode-hop-free tuning of the light emitted. 0146-9592/15/050772-04$15.00/0
Finally, we prove that a WGR-OPO can be employed in high-resolution spectroscopy. In our experiments, a grating-stabilized diode laser emitting at 1040 nm wavelength serves as the pump source. Despite being widely tunable, this laser allows fast current modulation to generate the sidebands necessary for the Pound–Drever–Hall (PDH) locking technique without any additional electro-optic modulator. Coupling of the pump laser light into the WGR is achieved by prism coupling with a rutile prism, while the gap between the WGR and the prism can be adjusted through a piezo translator to change the coupling regime from strong over- to undercoupling at will. The resonator used in our experiment is made of z-cut magnesium-oxide-doped congruent lithium niobate. All three interacting waves are extraordinarily polarized, while quasi-phase matching is achieved through a linear domain pattern similar to the structure employed in the pioneering first demonstration of frequency doubling in a WGR [8]. Although a radial domain structure would decrease the oscillation threshold and certainly enable a more controlled coarse tuning, we use a linear domain pattern because it is easier to fabricate and readily available. We assume that the fine tuning behavior investigated in this report will not suffer significantly from a nonoptimized domain pattern. The periodicity of 26.5 μm quasi-phase matches OPO processes pumped at 1 μm generating signal and idler fields around 2 μm wavelength. In a WGR, the geometrical dimensions of the cavity influence the dispersion relation [9,10] and hence the phase matching conditions. To optimize the process for room temperature, we manufactured the resonator with a radius of approximately 1.55 mm. The unloaded quality factor of the resonator at the pump wavelength was determined to be approximately 5 × 107 . A closed-loop temperature control with a stability better than 1 mK in combination with an air-tight housing around the resonator and prism ensure a good temperature stability. The housing is made of PMMA and has an entrance window made of NBK-7 to maintain the circular beam profile of the pump laser while the signal, idler, and © 2015 Optical Society of America
March 1, 2015 / Vol. 40, No. 5 / OPTICS LETTERS
Fig. 1. Schematic of the experimental setup: the light of an externally stabilized external-cavity diode laser (ECDL) with the pump power P p is coupled into a whispering gallery resonator (WGR) using a prism coupler. An optical low-pass filter (LP) separates the generated signal and idler light from the pump light. The photodiode D1 detects the transmitted pump light and is fed into the Pound–Drever–Hall (PDH) stabilization system. Some portion of the signal and idler light is coupled via a beamsplitter (BS) into a spectrum analyzer. The remaining total signal and idler power P si can be monitored with the photodiode D2 ; alternatively, the light is analyzed by a Fabry–Perot interferometer (FPI).
pump waves leave the housing through the PMMA as illustrated in Fig. 1. The good temperature stability is necessary in our experiment because the signal and idler resonances are not actively stabilized. Here, continuouswave operation is possible only as long as the temperature stays constant. The fact that pump, signal, and idler waves are all polarized along the optic axis of the crystal comes in favor because the temperature dependence of the refractive indices for the three waves is similar resulting in an increased tolerance against temperature fluctuations. A fast photodiode detects the modulated pump wave that, after demodulation, provides the error signal for locking the laser to the WGR resonances [11], which is then fed back to the pump laser via a PID control. That way, the laser can follow a given resonator mode even during the presence of a nonlinear process that leads to distorted pump resonances [12] as well as to distorted error signals. The bandwidth of the whole feedback loop is limited by the tuning speed of the diode laser and is at the order of a few kilohertz, while the sideband modulation frequency is 12.5 MHz. A fiber-coupled grating-based spectrum analyzer monitors the wavelengths of the signal and idler waves. The remaining light can either be directly detected by an infrared-sensitive photodiode for power measurements or coupled into a scanning Fabry–Perot interferometer (FPI) to monitor the frequency modulation of the signal wave. The photodiode is equipped with a filter providing additional attenuation of more than 70 dB at the pump wavelength ensuring that only the total signal and idler power contributes to the measurement. The FPI can be operated in both scanning and nonscanning mode, and is temperature stabilized to serve as a long-term reference. All powers given are corrected for reflection and absorption losses of the optical
773
components and the housing to obtain the powers that are present inside the coupling prism. In the absence of a nonlinear process, we adjusted the gap between the resonator and the coupling prism until the internal losses matched the coupling losses to achieve critical coupling. Here, we observed a coupling efficiency of 60%, which corresponds to the mode overlap between the external pump laser and the mode inside the resonator. The remaining 40% cannot be coupled into the resonator and are lost for the nonlinear process. Since a perfect mode overlap is generally possible, we corrected the available pump power with the obtained mode overlap to calculate the conversion efficiency. We were able to achieve a combined signal and idler power of P si ≈ 5.2 mW at a pump power of P p ≈ 9.5 mW resulting in a conversion efficiency of 55%. This high efficiency is achieved in the regime of strong overcoupling leading to higher efficiencies at the expense of an increased pump threshold of 2.2 mW. In the case of undercoupling, the threshold reduces to 86 μW, while the efficiency is limited to 3.5%. All these experimental results were obtained while scanning the laser over the desired mode to avoid constant relocking and gain adjustment after changing the pump power. The results are depicted in Fig. 2. In the locked state with 9.5 mW pump power, we achieved a long-term power stability of the signal and idler waves better than 1% over the term of 1 h as shown in Fig. 3. A prominent 30 min cycle is seen in the output power, presumably stemming from the air conditioner in the laboratory.
Fig. 2. Generated combined signal and idler power P si versus the pump power P p . The power values shown are corrected for mode overlap, absorption, and reflection losses (orange circles: experimental data, red line: theoretical fit [13]).
Fig. 3. Power stability of the generated signal and idler light over 1 h.
774
OPTICS LETTERS / Vol. 40, No. 5 / March 1, 2015
process is reversible, although we observed a hysteresis effect presumably stemming from the specific locking behavior and the structure of the pump resonances during the nonlinear process that can introduce additional zero crossings in the error signal. A simplified model assuming that the light is guided at the surface of the resonator and neglecting higher spatial modes gives the dispersion relation ν Fig. 4. Transmission of the signal and idler power through a scanning Fabry–Perot interferometer.
To prove single-frequency operation in the locked state, we analyze the signal wave with the scanning FPI. In Fig. 4, the transmission spectrum of the FPI is shown while the parametric process operates at degeneracy. The measured linewidth of the FPI signal is 10 MHz. This corresponds to the predicted finesse achievable based on the mirror losses. Thus, the 10 MHz are only an upper limit of the signal and idler linewidths. Still locked to the same process, we changed the temperature of the WGR while recording the position of the FPI peaks. In Fig. 5, the change of the output frequency Δν and the corresponding temperature profile are shown. By changing the temperature by 120 mK at an absolute temperature of approximately 26.4°C, the output frequency changes by 480 MHz. The tuning rate is limited by our experimental WGR holder representing a large heat capacitance. Changing the temperature further leads to a mode hop of the signal and idler waves, while the pump laser keeps locked to the resonator mode. This
Fig. 5. Frequency shift Δν of the signal and idler waves (top) and the corresponding change ΔT of the resonator temperature (bottom) versus time.
c0 m 1 ; 2π nTRT
(1)
with the resonance frequency ν, the vacuum light speed c0 , the mode number m for the signal/idler mode, the temperature-dependent resonator radius RT, and refractive index nT. With this formula and considering m 9865 for our case, we can estimate the tuning rate of the resonances at the signal and idler wavelength of 2080 nm. At degeneracy, we find 4.7 MHz∕mK using the temperature dependences given in [14,15]. Our experimental results show a tuning rate of approximately 4 MHz∕mK. Taking into account that the model used neglects that changing the temperature forces the pump, signal, and idler waves to operate off of their respective resonance center and that the temperature coefficients given in the literature may differ from those valid for our sample, the prediction and experimental results are in good agreement. To demonstrate the suitability of the light source for high-resolution spectroscopy, we scan the signal wave over a FPI resonance. This time, the FPI is operated in the nonscanning mode, and the resonator temperature is sweeped while recording the transmitted signal power. As depicted in Fig. 6, the resonator mode is easily
Fig. 6. Transmission (top) of the signal and idler power through the Fabry–Perot interferometer while the temperature (bottom) is ramped up and down. The spectrum shows the same mode scanned from both directions. The steps in the temperature graph are an artifact of the update rate.
March 1, 2015 / Vol. 40, No. 5 / OPTICS LETTERS
775
Financial support from the Deutsche Forschungsgemeinschaft is gratefully acknowledged.
Fig. 7.
1-h frequency stability of the signal and idler light.
resolved in great detail even though the whole measurement took over 1 min. The two transmission peaks shown represent the same FPI resonance scanned in both directions. Considering the tuning rate of 4 MHz∕mK, a linewidth of approximately 10 MHz can be derived, matching the linewidth found in Fig. 4. In general, an important feature of a laser system is its frequency stability. To determine the long-term stability, we scan the FPI and record the drift of the WGR resonance for 1 h while keeping the temperature set-point of the WGR constant. As depicted in Fig. 7, the frequency drift is less than 25 MHz. Like the output power, the frequency drift shows a periodicity of 30 min. In conclusion, by locking a pump laser to a WGR resonance made of periodically poled lithium niobate, we were able to achieve true continuous-wave and single-frequency operation of a triply-resonant optical parametric oscillator. Through temperature sweeping of the resonator, the signal and idler frequencies can be tuned over several hundred MHz allowing for highresolution spectroscopy applications. The frequency and power stability matches that of commercially available frequency-converted laser systems.
References 1. M. Ebrahim-Zadeh, in OSA Handbook of Optics IV, M. Bass, ed. (McGraw-Hill, 2010), p. 17.1. 2. J. U. Fürst, D. V. Strekalov, D. Elser, A. Aiello, U. L. Andersen, C. Marquardt, and G. Leuchs, Phys. Rev. Lett. 105, 263904 (2010). 3. T. Beckmann, H. Linnenbank, H. Steigerwald, D. Haertle, K. Buse, and I. Breunig, Phys. Rev. Lett. 106, 143903 (2011). 4. Ch. S. Werner, T. Beckmann, K. Buse, and I. Breunig, Opt. Lett. 37, 4224 (2012). 5. J. U. Fürst, D. V. Strekalov, D. Elser, A. Aiello, U. L. Andersen, Ch. Marquardt, and G. Leuchs, Phys. Rev. Lett. 106, 113901 (2011). 6. M. Förtsch, J. U. Fürst, C. Wittmann, D. V. Strekalov, A. Aiello, M. V. Chekhova, C. Silberhorn, G. Leuchs, and C. Marquardt, Nat. Commun. 4, 1818 (2013). 7. J. Laurat, L. Longchambon, C. Fabre, and T. Coudreau, Opt. Lett. 30, 1177 (2005). 8. V. S. Ilchenko, A. A. Savchenkov, A. B. Matsko, and L. Maleki, Phys. Rev. Lett. 92, 043903 (2004). 9. I. Breunig, B. Sturman, F. Sedlmeir, H. G. L. Schwefel, and K. Buse, Opt. Express 21, 30683 (2013). 10. M. L. Gorodetsky and A. E. Fomin, IEEE J. Sel. Top. Quantum Electron. 12, 33 (2006). 11. R. W. P. Drever, J. L. Hall, F. V. Kowalski, J. Hough, G. M. Ford, A. J. Munley, and H. Ward, Appl. Phys. B 31, 97 (1983). 12. I. Breunig, B. Sturman, A. Bückle, Ch. S. Werner, and K. Buse, Opt. Lett. 38, 3316 (2013). 13. B. Sturman and I. Breunig, J. Opt. Soc. Am. B 28, 2465 (2011). 14. Y. S. Kim and R. T. Smith, J. Appl. Phys. 40, 4637 (1969). 15. O. Paul, A. Quosig, T. Bauer, M. Nittmann, J. Bartschke, G. Anstett, and J. A. L’Huillier, Appl. Phys. B 86, 111 (2007).
