- Email: [email protected]
THIS: a tuneable heterodyne infrared spectrometer
Spectrochimica Acta Part A 58 (2002) 2457– 2463 www.elsevier.com/locate/saa
THIS: a tuneable heterodyne infrared spectrometer Daniel Wirtz *, Guido Sonnabend, Rudolf T. Schieder I. Physikalisches Institut, Uni6ersita¨t zu Ko¨ln, D-50937 Ko¨ln, Germany
Abstract With the Cologne Tuneable Heterodyne Infrared Spectrometer (THIS) we present a newly developed setup of a transportable heterodyne receiver. Competitiveness with regard to sensitivity, was reached for the first time with a semiconductor laser pumped system. Frequency tuneability of the local oscillator (LO) laser over a wide range of wavelengths is thus provided. This allows a variety of molecules, e.g. O3, NH3, CH4, N2O,…. in the earth’s atmosphere, in planetary atmospheres or even in interstellar space to be observed with very high frequency resolution either from aboard the Stratospheric Observatory For Infrared Astronomy (SOFIA) or other ground based telescopes. Besides the good results with tuneable lead salt laser (TDL) operation there’s room to improve: the power provided by such devices is not sufficient for a sensitivity close to the quantum limit. Therefore, first tests with recently developed high power quantum-cascade lasers (QCL) were carried out and further substantial improvement of the system noise temperature seems to be in reach. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Infrared; Heterodyne; Tuneable diode laser; Quantum cascade laser
1. Introduction In recent years IR heterodyne spectroscopy has been demonstrated as a powerful tool for astrophysical and atmospheric studies [1 – 15]. Whenever high spectral resolution combined with high sensitivity is required heterodyne systems are advantageous because of their high optical throughput as compared to direct detection methods like scanning Fabry – Perot techniques for example. Many useful information was gathered in the earth’s atmosphere as well as in the atmospheres of other planets in the solar system or in the dust clouds of carbon stars like IRC+10216 [5,6,8 –11,16]. * Corresponding author
Gas lasers have commonly been used as local oscillators (LO) and the sensitivity of those systems (Tsys : 3000 K) has been shown to come close to the quantum limit of TQL = 1440 K @10 mm [2,4,15]. (As is standard in radio astronomy, we characterize the sensitivity of our system by means of the overall noise temperature Tsys). But the restriction to the fixed laser frequencies is a severe limitation of gas laser systems, which allows observations only within a 15% range around the few laser lines between 9 and 12 mm, for example [2]. This major limitation is overcome by tuneable diode lasers (TDL) which are available in the wavelength range from about 3 to 25 mm1. Unfor1 Specification of the manufacturer, see http://www. lasercomponents.com/wwwe/products/index.html.
1386-1425/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 6 - 1 4 2 5 ( 0 2 ) 0 0 0 6 2 - 8
2458
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
tunately, the sensitivity of TDL pumped systems is usually worse by a factor of 5– 10 compared with gas laser systems [2,4– 14]. This is due to the lack of pump laser power, and the extremely critical response of TDLs to optical feedback. Recently we presented a method to remove such problems by superimposing the laser and the signal beam with a Fabry– Perot diplexer instead of a commonly used beamsplitter. This brought the performance of our system fairly close to that of gas laser pumped systems [14,15]. The future use of newly developed high power quantum cascade lasers (QCL) will improve the system performance even more so that astronomical measurements with tuneable lasers are in reach [17]. Observations with THIS are not only planned on SOFIA but are possible at various optical and IR telescopes. Laboratory and field tests have already been performed at TIRGO (Telescopo InfraRosso del GOrnergrat), the Hainberg solar tower in Go¨ ttingen, IRSOL (Instituto per Ricerche Solari Locarno) and at our institute in Cologne [18]2.
Fig. 1. Scaled view of the optical receiver part with the main components showing the compactness of the setup. The size of the aluminium cube is roughly 60 × 60 ×45 cm3.
device allows to superimpose about 60% of LO power with more than 90% of signal power. A detailed description of the diplexer and the other components can be found in Schmu¨ lling et al. [15]. The frequency analysis of the IF generated by the mixer is done by an 2048 channel acousto
2. The instrument All heterodyne receivers work in a common way: by superimposing the broadband radiation to be analyzed with the radiation of a monochromatic LO on a non-linear detector an intermediate frequency (IF) signal is generated which can be analyzed in real time by a multichannel spectrometer. Fig. 1 shows a scaled view of the receiver part of our instrument. The lightweight cubic aluminum frame contains the folded mirror optics, a LN2 cooled dewar holding the TDL and a fast photovoltaic mercury– cadmium – telluride (MCT) mixer detector. In order to avoid undesired losses of the fairly low laser power the LO and signal beam are superimposed on a confocal Fabry– Perot ring-resonator (diplexer) consisting of two parabolic mirrors and two beam splitters. This 2 A brief description of the Hainberg Solar Tower may be found in: http://www.uni-sw.gwdg.de/geninf/hainberg/ sonnenturm/esonnenturm.html.
Fig. 2. Block diagram of the experimental setup. The confocal Fabry – Perot diplexer superimposes the local oscillator (LO) and infrared beam (from the telescope or load) and is length stabilized by means of a Helium – Neon laser. A tiltable BaF2 plate ensures frequency stability of the LO by controlling the optical feedback. A mercury – cadmium-telluride (MCT) detector generates the intermediate frequency (0 – 2 GHz) which is amplified by a low noise high electron mobility transistor amplifier (HEMT) and delivered to an acousto-optical spectrometer (AOS) for frequency analysis.
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
optical spectrometer (AOS) with a total bandwidth of 1.4 GHz [19]. Fig. 2 shows a schematic setup of our spectrometer. For convenience our system is developed for 10 mm wavelength. Nevertheless the operating wavelength can easily be adjusted by exchanging the LO, the detector, and the diplexer beamsplitters. The stabilization of the TDL is described below as well as some features of the transportable receiver.
2.1. The transportable recei6er Our intention was to design a transportable receiver which can be mounted easily to different types of telescopes (e.g. to Coude´ , Nasmyth or Cassegrain foci of optical or IR telescopes). This was achieved by placing the complete optical setup including the LO and the detector into a light weight cubic frame made of aluminum. Its size is roughly 60× 60 ×45 cm3 and its weight is below 80 kg. All electronic devices needed are mounted in an additional 19 in. rack. Mechanical stability tests of the setup have been performed successfully at the TIRGO telescope at Gornergrat/Switzerland [18] and at the Hainberg Solar Tower in Go¨ ttingen/Germany2. The receiver is remotely controlled through a personal computer that is also used for data acquisition. To adopt to different f-ratio values of different telescopes one off-axis imaging mirror has to be exchanged in the setup.
2.2. The frequency stabilization The ability to perform highest resolution measurements with a heterodyne system is extremely dependent on a precise LO frequency and an efficient control mechanism. To describe how the frequency stabilization works one has to consider the different resonators present in the system. First, the surfaces of the diode laser chip itself form a low quality resonator which gives rise to resonant modes every few per cm. The laser couples to the diplexer and passes it in transmission. The diplexing Fabry– Perot resonator has a mode spacing of 2.5 GHz and a finesse of about 18.
2459
Behind the diplexer the combined signal and LO beam is focussed onto the detector. Due to the partial reflection from the surface of the detector back into the laser a third resonator is created between laser and mixer and due to its length (1.65 m) resonances every 90 MHz result. The increased number of intra-cavity photons in the external resonator reduces the influence of the TDL cavity, but instead the external resonator now determines the operating frequency of the laser while the coherence of the laser emission is significantly improved [20].
2.2.1. The Fabry–Perot diplexer As shown in Schmu¨ lling et al. [15] the diplexer allows drastically improved superposition of LO and signal beam. But since the spacing of the resonator mirrors can fluctuate due to mechanical vibrations or thermal effects, the length of one roundtrip inside the device may vary. Therefore, the diplexer cavity length must be stabilized. For this purpose a frequency stabilized HeNe laser is coupled into the diplexer and the resulting Fabry–Perot fringes are detected by a photodiode. Slight modulation of the diplexer length by means of a piezo-electric transducer enables a lock-in amplifier to generate an error signal from the detected HeNe fringes. This error signal is used to stabilize the diplexer resonance on a HeNe fringe, and it has been verified that the modulation amplitude required detunes the resonance frequency of the diplexer at 10 mm by less than 1 MHz. This is equivalent to the resolution of the AOS and, therefore, negligible. In this way the resonance frequency of the diplexer is directly linked to the frequency accuracy of the HeNe laser which is specified by the manufacturer to 6/D6 :108 within 8 h. 2.2.2. Frequency stabilization of the diode laser As stated above we use optical feedback from the detector to partly control the diode laser. By filtering this feedback the diplexer forces the laser to operate on exactly one longitudinal mode of the external cavity and suppresses fluctuations which would occur due to mode competition between nearby modes. As long as the diode laser oscillates on one of these modes it emits at a fixed
2460
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
Fig. 3. Scheme the frequency stabilization set up (see text for further description).
frequency which is located somewhere within the resonance of the diplexer despite some minor mode pulling effects of the low quality laser resonator itself. To obtain a well defined LO frequency at the center of the diplexer fringe the length of the external resonator has to be controlled as well. At this point we make use of the modulation already employed. Since the diplexer is part of the external resonator modulating its length also modulates the length of the external cavity and the amount of feedback within. An error signal is again generated by a lock-in amplifier using the DC photo current output of the IR detector. This signal is fed to a scanner motor which carries a 2 mm BaF2 flat inserted into the beam path at nearly Brewster angle. Tilting this flat results in a change of the optical path length and, therefore, a change of the resonance frequency of the external resonator. Therefore, maximum DC output of the IR detector is given if the diplexer and the external resonator both transmit with maximum amplitude. This implies, that both cavities are resonant, and with one of them (the diplexer) fixed in frequency the overall frequency is fixed. A scheme of the stabilization is presented in Fig. 3.
3. Measurements To test the performance of THIS a variety of laboratory and field measurements were performed. In this section we present atmospheric observations as well as laboratory tests of the stability of the system and the first measurements with a QC laser as LO.
3.1. Laboratory measurements The ability to perform astronomical measurements at infrared wavelengths requires the capability of continuous integration on the source for several hours. For a heterodyne system it is essential that the LO frequency and the response of the system is stable during this time. The frequency stability was investigated by integrating a strong molecular absorption line of methanol and then fitting the lineposition and linewidth every 10 s. For this the molecular absorption from a gas cell was observed using a hot and ambient temperature source behind the cell. The result of this fit is shown in Fig. 4. As one can see the lineposition and the linewidth show only slight variations which are mostly due to the fairly low signal to
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
Fig. 4. Position and linewidth of a methanol reference line during 1 h of integration. The linewidth fluctuation is due to the bad signal to noise ratio.
noise ratio of each ‘10 s spectrum’. The impact of the noise becomes obvious especially in the linewidth which was of course constant, since the pressure of the methanol was kept constant. For the investigation of stability the AllanVariance method has become a standard method in recent years [21,22]. This procedure allows to evaluate the stability of a radiometer as a function of integration time and the result for our instrument is shown in the plot in Fig. 5. Depicted is a log –log plot of the spectroscopic Allan-Variance against the integration time, which represents the variance of the relative fluctuations of two inde-
2461
Fig. 6. Stratospheric ozone absorptions at 1070 cm − 1. Shown is a spectrum at 1 MHz resolution after 25 min of integration time.
pendent spectrometer channels as a function of integration time on each of the temperature loads. As one can see, up to several seconds of integration time the system behaves purely radiometrically which is indicated by the slope of − 1 in the plot. For longer integration times additional drift noise contributes, and the Allan-minimum is found around 50–60 s. From there on unfavorable drift noise dominates the behavior of the system. As a result one can deduce that integration times per load should not exceed some 10 s in order to avoid drift noise to contribute to the spectra. For those intervals the assumption of pure radiometric performance is guaranteed.
3.2. Field measurements An example spectrum recorded at IRSOL solar observatory in Locarno, Switzerland is shown in Fig. 63. The ground based acquisition of information about stratospheric trace gases is a valuable input to climatological models and a low cost complement to the launch of balloon sondes. The excellent sensitivity and stability of the instrument is clearly indicated by the appearance of weak 16 16 18 O O O lines. Fig. 5. Allan-variance measurement. Shown is a log –log plot of the variance of the measured signal over the integration. The Allan-variance minimum is visible at about 50 s.
3 A brief description of the Istituto Ricerche Solari Locarno may be found in: http://umwelttechnik.mnd.fh-wiesbaden.de/ divers/irsol/.
2462
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
3.3. QC-lasers, a breakthrough One of the main problems of tuneable IR heterodyne systems besides the optical feedback problem is the lack of LO power. The typical few hundred mW of lead salt lasers are not sufficient for optimized operation of the commonly used MCT-detectors. Due to this, a performance like that of gas laser systems was never reached with TDL-pumped mixers. A breakthrough with regards to sensitivity and performance might be achieved with the use of QC-lasers. These devices provide sufficient power ( 10 mW) but are difficult to handle in cw operation due to their much higher heat dissipation. First results with a QCL from Alpeslasers/Switzerland [23,24] are nevertheless very promising and are shown in Fig. 7. The system temperature is improved significantly compared with a typical TDL result by nearly a factor of 3. The increase in noise temperature with IF-frequency is caused by our MCTmixer, the occasional spikes are due to interference from cellular phone transmitters which are hard to suppress, unfortunately. Further investigation regarding the stability and linewidth of those lasers will be performed soon. We expect that due to the fairly high power available from QCLs the overall performance of the receiver should become identical with CO2 laser pumped systems. This requires a somewhat
Fig. 7. System noise temperature with typical TDL (top) and first QCL result (bottom). The system sensitivity is thus significantly improved.
refined setup, which is presently under construction. Next generation QCLs might also allow operation at or close to room temperature.
4. Conclusion We have presented the prototype of a new competitive and tuneable IR heterodyne spectrometer. Such an instrument will be an interesting application for SOFIA and beyond. The detailed study of absorption line profiles from relatively cold matter or the investigation of the molecular composition of stellar surroundings will improve knowledge about distribution of interstellar gas and the formation of stars. With the tuneable system now available a variety of molecular transitions are observable which could not be seen before, particularly of molecules without permanent dipole moments such as H2 or CH4. This permits a lot of promising scientific questions to be answered not only by using SOFIA but also through observations from ground based telescopes. Employing large telescopes like the VLT might even allow investigations beyond our galaxy.
Acknowledgements This study has been supported by the Deutsche Forschungsgemeinschaft special grants SFB 301 (until 1999) and SFB 494 (since 2000). We would like to thank the following people: Dr Bernd Vohwinkel for the development, testing and manufacturing of the low noise HEMT-amplifiers used in our system, our colleagues at the Go¨ ttingen observatory for giving us the opportunity to use the Hainberg solartower and M. Bianda who supported us during our stay at the IRSOL observatory in Locarno, Switzerland. Furthermore we would like to thank AEG for delivering the MCT mixer detector. Parts of the measurements were performed at TIRGO (Gornergrat, Switzerland). TIRGO is operated by CAISMI-CNR Arcetri, Firenze, Italy.
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
References [1] M.M. Abbas, M.J. Mumma, T. Kostiuk, D. Buhl, Appl. Optics 15 (1976) 427 –436. [2] T. Kostiuk, M.J. Mumma, Appl. Optics 22 (1983) 2644 – 2654. [3] R.T. Menzies, in: E.D. Hinkley (Ed.), Laser Heterodyne Detection Techniques in Laser Monitoring of the Atmosphere, Springer, Verlag, 1976, pp. 297 –353. [4] H. Rothermel, H.U. Ka¨ ufl, Y. Yu, Astron. Astrophys. 126 (1983) 387 – 392. [5] A. Delahaigue, D. Courtois, C. Thiebeaux, S. Kalite, B. Parvitte, Infrared Phys. Technol. 37 (1996) 7 –12. [6] M. Koide, M. Taguchi, H. Fukunishi, Geophys. Res. Lett. 22 (4) (1995) 401 –404. [7] T. Kostiuk, K.E. Fast, T.A. Livengood, T. Hewagama, J.J. Goldstein, F. Espenak, D. Buhl, Geophys. Res. Lett. 28 (12) (2001) 2361 –2364. [8] D.W. Peterson, M.A. Johnson, A.L. Betz, Nature 250 (1974) 128 –130. [9] T. Kostiuk, F. Espenak, M.J. Mumma, P. Romani, Infrared Phys. Technol. 29 (2 –4) (1989) 199 – 204. [10] H. Fukunishi, S. Okano, M. Taguchi, T. Ohnuma, Appl. Optics 29 (18) (1990) 2722 –2728. [11] A.L. Betz, M.A. Johnson, R.A. McLaren, E.C. Sutton, Astrophys. J. Lett. 208 (3) (1976) L141 – L144. [12] R.T. Ku, D.L. Spears, Opt. Lett. 1 (3) (1977) 84 – 86. [13] D. Glenar, T. Kostiuk, D.E. Jennings, D. Buhl, M.J. Mumma, Appl. Optics 21 (2) (1982) 253 – 259.
2463
[14] R. Schieder, Infrared Phys. Technol. 35 (2) (1994) 477 – 486. [15] F. Schmu¨ lling, B. Klumb, M. Harter, R. Schieder, B. Vowinkel, G. Winnewisser, Appl. Optics 37 (1998) 5771 – 5776. [16] David Moses Goldhaber, Infrared heterodyne spectroscopy of polyatomic molecules in circumstellarenvelopes Ph.D. Thesis California Univ., Berkeley, (1988) [17] A. Mu¨ ller, M. Beck, J. Faist, U. Oesterle, M. Ilegems, Appl. Phys. Lett. 75 (11) (1999) 1509 – 1511. [18] P. Salinari, ‘‘The TIRGO observatory’’ 2nd ESO Infrared Workshop Proc., Garching, West Germany, 20 – 23 April 1982, 45 – 54, (1982). [19] R. Schieder, V. Tolls, G. Winnewisser, Exp. Astron. 1 (1989) 101 – 121. [20] G.P. Agrawal, IEEE J. Quant. Elect. 20 (5) (1984) 468 – 471. [21] R. Schieder, G. Rau, B. Vowinkel, Characterization and Measurement of System Stability SPIE, 598, In: Instrumentation for submillimeter spectroscopy; Proceedings of the Meeting, Cannes, France, December 5, 6, 1985 Society of PhotoOptical Instrumentation Engineers, pp. 189 – 192, (1986). [22] R. Schieder, C. Kramer, AA 373 (2001) 746 – 756. [23] J. Faist, F. Capasso, D.L. Sivco, C. Sirtori, A.L. Hutchinson, A.Y. Cho, Science 264 (1994) 553 – 556. [24] J. Faist, F. Capasso, C. Sirtori, D.L. Sivco, A. Hutchinson, A.Y. Cho, Appl. Phys. Lett. 67 (21) (1995) 3057 – 3060.
THIS: a tuneable heterodyne infrared spectrometer Daniel Wirtz *, Guido Sonnabend, Rudolf T. Schieder I. Physikalisches Institut, Uni6ersita¨t zu Ko¨ln, D-50937 Ko¨ln, Germany
Abstract With the Cologne Tuneable Heterodyne Infrared Spectrometer (THIS) we present a newly developed setup of a transportable heterodyne receiver. Competitiveness with regard to sensitivity, was reached for the first time with a semiconductor laser pumped system. Frequency tuneability of the local oscillator (LO) laser over a wide range of wavelengths is thus provided. This allows a variety of molecules, e.g. O3, NH3, CH4, N2O,…. in the earth’s atmosphere, in planetary atmospheres or even in interstellar space to be observed with very high frequency resolution either from aboard the Stratospheric Observatory For Infrared Astronomy (SOFIA) or other ground based telescopes. Besides the good results with tuneable lead salt laser (TDL) operation there’s room to improve: the power provided by such devices is not sufficient for a sensitivity close to the quantum limit. Therefore, first tests with recently developed high power quantum-cascade lasers (QCL) were carried out and further substantial improvement of the system noise temperature seems to be in reach. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Infrared; Heterodyne; Tuneable diode laser; Quantum cascade laser
1. Introduction In recent years IR heterodyne spectroscopy has been demonstrated as a powerful tool for astrophysical and atmospheric studies [1 – 15]. Whenever high spectral resolution combined with high sensitivity is required heterodyne systems are advantageous because of their high optical throughput as compared to direct detection methods like scanning Fabry – Perot techniques for example. Many useful information was gathered in the earth’s atmosphere as well as in the atmospheres of other planets in the solar system or in the dust clouds of carbon stars like IRC+10216 [5,6,8 –11,16]. * Corresponding author
Gas lasers have commonly been used as local oscillators (LO) and the sensitivity of those systems (Tsys : 3000 K) has been shown to come close to the quantum limit of TQL = 1440 K @10 mm [2,4,15]. (As is standard in radio astronomy, we characterize the sensitivity of our system by means of the overall noise temperature Tsys). But the restriction to the fixed laser frequencies is a severe limitation of gas laser systems, which allows observations only within a 15% range around the few laser lines between 9 and 12 mm, for example [2]. This major limitation is overcome by tuneable diode lasers (TDL) which are available in the wavelength range from about 3 to 25 mm1. Unfor1 Specification of the manufacturer, see http://www. lasercomponents.com/wwwe/products/index.html.
1386-1425/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 1 3 8 6 - 1 4 2 5 ( 0 2 ) 0 0 0 6 2 - 8
2458
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
tunately, the sensitivity of TDL pumped systems is usually worse by a factor of 5– 10 compared with gas laser systems [2,4– 14]. This is due to the lack of pump laser power, and the extremely critical response of TDLs to optical feedback. Recently we presented a method to remove such problems by superimposing the laser and the signal beam with a Fabry– Perot diplexer instead of a commonly used beamsplitter. This brought the performance of our system fairly close to that of gas laser pumped systems [14,15]. The future use of newly developed high power quantum cascade lasers (QCL) will improve the system performance even more so that astronomical measurements with tuneable lasers are in reach [17]. Observations with THIS are not only planned on SOFIA but are possible at various optical and IR telescopes. Laboratory and field tests have already been performed at TIRGO (Telescopo InfraRosso del GOrnergrat), the Hainberg solar tower in Go¨ ttingen, IRSOL (Instituto per Ricerche Solari Locarno) and at our institute in Cologne [18]2.
Fig. 1. Scaled view of the optical receiver part with the main components showing the compactness of the setup. The size of the aluminium cube is roughly 60 × 60 ×45 cm3.
device allows to superimpose about 60% of LO power with more than 90% of signal power. A detailed description of the diplexer and the other components can be found in Schmu¨ lling et al. [15]. The frequency analysis of the IF generated by the mixer is done by an 2048 channel acousto
2. The instrument All heterodyne receivers work in a common way: by superimposing the broadband radiation to be analyzed with the radiation of a monochromatic LO on a non-linear detector an intermediate frequency (IF) signal is generated which can be analyzed in real time by a multichannel spectrometer. Fig. 1 shows a scaled view of the receiver part of our instrument. The lightweight cubic aluminum frame contains the folded mirror optics, a LN2 cooled dewar holding the TDL and a fast photovoltaic mercury– cadmium – telluride (MCT) mixer detector. In order to avoid undesired losses of the fairly low laser power the LO and signal beam are superimposed on a confocal Fabry– Perot ring-resonator (diplexer) consisting of two parabolic mirrors and two beam splitters. This 2 A brief description of the Hainberg Solar Tower may be found in: http://www.uni-sw.gwdg.de/geninf/hainberg/ sonnenturm/esonnenturm.html.
Fig. 2. Block diagram of the experimental setup. The confocal Fabry – Perot diplexer superimposes the local oscillator (LO) and infrared beam (from the telescope or load) and is length stabilized by means of a Helium – Neon laser. A tiltable BaF2 plate ensures frequency stability of the LO by controlling the optical feedback. A mercury – cadmium-telluride (MCT) detector generates the intermediate frequency (0 – 2 GHz) which is amplified by a low noise high electron mobility transistor amplifier (HEMT) and delivered to an acousto-optical spectrometer (AOS) for frequency analysis.
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
optical spectrometer (AOS) with a total bandwidth of 1.4 GHz [19]. Fig. 2 shows a schematic setup of our spectrometer. For convenience our system is developed for 10 mm wavelength. Nevertheless the operating wavelength can easily be adjusted by exchanging the LO, the detector, and the diplexer beamsplitters. The stabilization of the TDL is described below as well as some features of the transportable receiver.
2.1. The transportable recei6er Our intention was to design a transportable receiver which can be mounted easily to different types of telescopes (e.g. to Coude´ , Nasmyth or Cassegrain foci of optical or IR telescopes). This was achieved by placing the complete optical setup including the LO and the detector into a light weight cubic frame made of aluminum. Its size is roughly 60× 60 ×45 cm3 and its weight is below 80 kg. All electronic devices needed are mounted in an additional 19 in. rack. Mechanical stability tests of the setup have been performed successfully at the TIRGO telescope at Gornergrat/Switzerland [18] and at the Hainberg Solar Tower in Go¨ ttingen/Germany2. The receiver is remotely controlled through a personal computer that is also used for data acquisition. To adopt to different f-ratio values of different telescopes one off-axis imaging mirror has to be exchanged in the setup.
2.2. The frequency stabilization The ability to perform highest resolution measurements with a heterodyne system is extremely dependent on a precise LO frequency and an efficient control mechanism. To describe how the frequency stabilization works one has to consider the different resonators present in the system. First, the surfaces of the diode laser chip itself form a low quality resonator which gives rise to resonant modes every few per cm. The laser couples to the diplexer and passes it in transmission. The diplexing Fabry– Perot resonator has a mode spacing of 2.5 GHz and a finesse of about 18.
2459
Behind the diplexer the combined signal and LO beam is focussed onto the detector. Due to the partial reflection from the surface of the detector back into the laser a third resonator is created between laser and mixer and due to its length (1.65 m) resonances every 90 MHz result. The increased number of intra-cavity photons in the external resonator reduces the influence of the TDL cavity, but instead the external resonator now determines the operating frequency of the laser while the coherence of the laser emission is significantly improved [20].
2.2.1. The Fabry–Perot diplexer As shown in Schmu¨ lling et al. [15] the diplexer allows drastically improved superposition of LO and signal beam. But since the spacing of the resonator mirrors can fluctuate due to mechanical vibrations or thermal effects, the length of one roundtrip inside the device may vary. Therefore, the diplexer cavity length must be stabilized. For this purpose a frequency stabilized HeNe laser is coupled into the diplexer and the resulting Fabry–Perot fringes are detected by a photodiode. Slight modulation of the diplexer length by means of a piezo-electric transducer enables a lock-in amplifier to generate an error signal from the detected HeNe fringes. This error signal is used to stabilize the diplexer resonance on a HeNe fringe, and it has been verified that the modulation amplitude required detunes the resonance frequency of the diplexer at 10 mm by less than 1 MHz. This is equivalent to the resolution of the AOS and, therefore, negligible. In this way the resonance frequency of the diplexer is directly linked to the frequency accuracy of the HeNe laser which is specified by the manufacturer to 6/D6 :108 within 8 h. 2.2.2. Frequency stabilization of the diode laser As stated above we use optical feedback from the detector to partly control the diode laser. By filtering this feedback the diplexer forces the laser to operate on exactly one longitudinal mode of the external cavity and suppresses fluctuations which would occur due to mode competition between nearby modes. As long as the diode laser oscillates on one of these modes it emits at a fixed
2460
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
Fig. 3. Scheme the frequency stabilization set up (see text for further description).
frequency which is located somewhere within the resonance of the diplexer despite some minor mode pulling effects of the low quality laser resonator itself. To obtain a well defined LO frequency at the center of the diplexer fringe the length of the external resonator has to be controlled as well. At this point we make use of the modulation already employed. Since the diplexer is part of the external resonator modulating its length also modulates the length of the external cavity and the amount of feedback within. An error signal is again generated by a lock-in amplifier using the DC photo current output of the IR detector. This signal is fed to a scanner motor which carries a 2 mm BaF2 flat inserted into the beam path at nearly Brewster angle. Tilting this flat results in a change of the optical path length and, therefore, a change of the resonance frequency of the external resonator. Therefore, maximum DC output of the IR detector is given if the diplexer and the external resonator both transmit with maximum amplitude. This implies, that both cavities are resonant, and with one of them (the diplexer) fixed in frequency the overall frequency is fixed. A scheme of the stabilization is presented in Fig. 3.
3. Measurements To test the performance of THIS a variety of laboratory and field measurements were performed. In this section we present atmospheric observations as well as laboratory tests of the stability of the system and the first measurements with a QC laser as LO.
3.1. Laboratory measurements The ability to perform astronomical measurements at infrared wavelengths requires the capability of continuous integration on the source for several hours. For a heterodyne system it is essential that the LO frequency and the response of the system is stable during this time. The frequency stability was investigated by integrating a strong molecular absorption line of methanol and then fitting the lineposition and linewidth every 10 s. For this the molecular absorption from a gas cell was observed using a hot and ambient temperature source behind the cell. The result of this fit is shown in Fig. 4. As one can see the lineposition and the linewidth show only slight variations which are mostly due to the fairly low signal to
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
Fig. 4. Position and linewidth of a methanol reference line during 1 h of integration. The linewidth fluctuation is due to the bad signal to noise ratio.
noise ratio of each ‘10 s spectrum’. The impact of the noise becomes obvious especially in the linewidth which was of course constant, since the pressure of the methanol was kept constant. For the investigation of stability the AllanVariance method has become a standard method in recent years [21,22]. This procedure allows to evaluate the stability of a radiometer as a function of integration time and the result for our instrument is shown in the plot in Fig. 5. Depicted is a log –log plot of the spectroscopic Allan-Variance against the integration time, which represents the variance of the relative fluctuations of two inde-
2461
Fig. 6. Stratospheric ozone absorptions at 1070 cm − 1. Shown is a spectrum at 1 MHz resolution after 25 min of integration time.
pendent spectrometer channels as a function of integration time on each of the temperature loads. As one can see, up to several seconds of integration time the system behaves purely radiometrically which is indicated by the slope of − 1 in the plot. For longer integration times additional drift noise contributes, and the Allan-minimum is found around 50–60 s. From there on unfavorable drift noise dominates the behavior of the system. As a result one can deduce that integration times per load should not exceed some 10 s in order to avoid drift noise to contribute to the spectra. For those intervals the assumption of pure radiometric performance is guaranteed.
3.2. Field measurements An example spectrum recorded at IRSOL solar observatory in Locarno, Switzerland is shown in Fig. 63. The ground based acquisition of information about stratospheric trace gases is a valuable input to climatological models and a low cost complement to the launch of balloon sondes. The excellent sensitivity and stability of the instrument is clearly indicated by the appearance of weak 16 16 18 O O O lines. Fig. 5. Allan-variance measurement. Shown is a log –log plot of the variance of the measured signal over the integration. The Allan-variance minimum is visible at about 50 s.
3 A brief description of the Istituto Ricerche Solari Locarno may be found in: http://umwelttechnik.mnd.fh-wiesbaden.de/ divers/irsol/.
2462
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
3.3. QC-lasers, a breakthrough One of the main problems of tuneable IR heterodyne systems besides the optical feedback problem is the lack of LO power. The typical few hundred mW of lead salt lasers are not sufficient for optimized operation of the commonly used MCT-detectors. Due to this, a performance like that of gas laser systems was never reached with TDL-pumped mixers. A breakthrough with regards to sensitivity and performance might be achieved with the use of QC-lasers. These devices provide sufficient power ( 10 mW) but are difficult to handle in cw operation due to their much higher heat dissipation. First results with a QCL from Alpeslasers/Switzerland [23,24] are nevertheless very promising and are shown in Fig. 7. The system temperature is improved significantly compared with a typical TDL result by nearly a factor of 3. The increase in noise temperature with IF-frequency is caused by our MCTmixer, the occasional spikes are due to interference from cellular phone transmitters which are hard to suppress, unfortunately. Further investigation regarding the stability and linewidth of those lasers will be performed soon. We expect that due to the fairly high power available from QCLs the overall performance of the receiver should become identical with CO2 laser pumped systems. This requires a somewhat
Fig. 7. System noise temperature with typical TDL (top) and first QCL result (bottom). The system sensitivity is thus significantly improved.
refined setup, which is presently under construction. Next generation QCLs might also allow operation at or close to room temperature.
4. Conclusion We have presented the prototype of a new competitive and tuneable IR heterodyne spectrometer. Such an instrument will be an interesting application for SOFIA and beyond. The detailed study of absorption line profiles from relatively cold matter or the investigation of the molecular composition of stellar surroundings will improve knowledge about distribution of interstellar gas and the formation of stars. With the tuneable system now available a variety of molecular transitions are observable which could not be seen before, particularly of molecules without permanent dipole moments such as H2 or CH4. This permits a lot of promising scientific questions to be answered not only by using SOFIA but also through observations from ground based telescopes. Employing large telescopes like the VLT might even allow investigations beyond our galaxy.
Acknowledgements This study has been supported by the Deutsche Forschungsgemeinschaft special grants SFB 301 (until 1999) and SFB 494 (since 2000). We would like to thank the following people: Dr Bernd Vohwinkel for the development, testing and manufacturing of the low noise HEMT-amplifiers used in our system, our colleagues at the Go¨ ttingen observatory for giving us the opportunity to use the Hainberg solartower and M. Bianda who supported us during our stay at the IRSOL observatory in Locarno, Switzerland. Furthermore we would like to thank AEG for delivering the MCT mixer detector. Parts of the measurements were performed at TIRGO (Gornergrat, Switzerland). TIRGO is operated by CAISMI-CNR Arcetri, Firenze, Italy.
D. Wirtz et al. / Spectrochimica Acta Part A 58 (2002) 2457–2463
References [1] M.M. Abbas, M.J. Mumma, T. Kostiuk, D. Buhl, Appl. Optics 15 (1976) 427 –436. [2] T. Kostiuk, M.J. Mumma, Appl. Optics 22 (1983) 2644 – 2654. [3] R.T. Menzies, in: E.D. Hinkley (Ed.), Laser Heterodyne Detection Techniques in Laser Monitoring of the Atmosphere, Springer, Verlag, 1976, pp. 297 –353. [4] H. Rothermel, H.U. Ka¨ ufl, Y. Yu, Astron. Astrophys. 126 (1983) 387 – 392. [5] A. Delahaigue, D. Courtois, C. Thiebeaux, S. Kalite, B. Parvitte, Infrared Phys. Technol. 37 (1996) 7 –12. [6] M. Koide, M. Taguchi, H. Fukunishi, Geophys. Res. Lett. 22 (4) (1995) 401 –404. [7] T. Kostiuk, K.E. Fast, T.A. Livengood, T. Hewagama, J.J. Goldstein, F. Espenak, D. Buhl, Geophys. Res. Lett. 28 (12) (2001) 2361 –2364. [8] D.W. Peterson, M.A. Johnson, A.L. Betz, Nature 250 (1974) 128 –130. [9] T. Kostiuk, F. Espenak, M.J. Mumma, P. Romani, Infrared Phys. Technol. 29 (2 –4) (1989) 199 – 204. [10] H. Fukunishi, S. Okano, M. Taguchi, T. Ohnuma, Appl. Optics 29 (18) (1990) 2722 –2728. [11] A.L. Betz, M.A. Johnson, R.A. McLaren, E.C. Sutton, Astrophys. J. Lett. 208 (3) (1976) L141 – L144. [12] R.T. Ku, D.L. Spears, Opt. Lett. 1 (3) (1977) 84 – 86. [13] D. Glenar, T. Kostiuk, D.E. Jennings, D. Buhl, M.J. Mumma, Appl. Optics 21 (2) (1982) 253 – 259.
2463
[14] R. Schieder, Infrared Phys. Technol. 35 (2) (1994) 477 – 486. [15] F. Schmu¨ lling, B. Klumb, M. Harter, R. Schieder, B. Vowinkel, G. Winnewisser, Appl. Optics 37 (1998) 5771 – 5776. [16] David Moses Goldhaber, Infrared heterodyne spectroscopy of polyatomic molecules in circumstellarenvelopes Ph.D. Thesis California Univ., Berkeley, (1988) [17] A. Mu¨ ller, M. Beck, J. Faist, U. Oesterle, M. Ilegems, Appl. Phys. Lett. 75 (11) (1999) 1509 – 1511. [18] P. Salinari, ‘‘The TIRGO observatory’’ 2nd ESO Infrared Workshop Proc., Garching, West Germany, 20 – 23 April 1982, 45 – 54, (1982). [19] R. Schieder, V. Tolls, G. Winnewisser, Exp. Astron. 1 (1989) 101 – 121. [20] G.P. Agrawal, IEEE J. Quant. Elect. 20 (5) (1984) 468 – 471. [21] R. Schieder, G. Rau, B. Vowinkel, Characterization and Measurement of System Stability SPIE, 598, In: Instrumentation for submillimeter spectroscopy; Proceedings of the Meeting, Cannes, France, December 5, 6, 1985 Society of PhotoOptical Instrumentation Engineers, pp. 189 – 192, (1986). [22] R. Schieder, C. Kramer, AA 373 (2001) 746 – 756. [23] J. Faist, F. Capasso, D.L. Sivco, C. Sirtori, A.L. Hutchinson, A.Y. Cho, Science 264 (1994) 553 – 556. [24] J. Faist, F. Capasso, C. Sirtori, D.L. Sivco, A. Hutchinson, A.Y. Cho, Appl. Phys. Lett. 67 (21) (1995) 3057 – 3060.