High resolution diode laser and heterodyne spectroscopy with applications toward remote sensing

Infrared Phys. Technol. Vol. 35, No. 213, pp. 471486, 1994 ElsevierScienceLtd. Printedin Great Britain 1350~4495/94$6.00 + 0.00

HIGH RESOLUTION SPECTROSCOPY

DIODE LASER AND HETERODYNE WITH APPLICATIONS TOWARD REMOTE SENSING R.

I. Physikalisches

Institut,

Universitat

SCHIEDER

zu KBln, Ziilpicher

Str. 77, D50937

K61n, Germany

(Received 24 June 1993) Abstract-The recent developments of IR heterodyne and high resolution spectroscopy at the University of Cologne are presented. By using an impedance matched HEMT amplifier together with newly developed HgCdTe-mixers very low noise temperatures of 2100 K (DSB, corrected) have been achieved, which is only 50% above the quantum limit. In addition, very broad banded detection of the heterodyne signals was obtained with a 1.4 GHz, 2000 channel acousto-optical spectrometer. First results with a tuneable infrared diode laser as local oscillator have been obtained. For the interpretation of atmospheric absorption signals laboratory measurements of pressure effects are made using infrared diode laser spectroscopy with very high frequency resolution by frequency stabilization of the laser, and very low noise data acquisition by signal averaging methods.

I. INTRODUCTION In radioastronomy detection of molecular line signals by heterodyne reception is the most powerful experimental method, from which detailed information of the physical and chemical conditions in the circumstellar and interstellar environment can be obtained. The structure of the atomic and molecular lines contains detailed information of the dynamics and temperatures of the investigated region as well as of abundances of various species in the interstellar medium. The situation is similar in the field of atmospheric studies, where the molecular line structure can provide information about the concentration of molecules as a function of altitude for example. This is frequently used for studies of the Earth’s atmosphere as well as for investigations of the atmosphere of other planets. At shorter wavelengths one of the problems in heterodyne spectroscopy is that the required bandwidth of the spectrometer becomes rather large. For example, in the mid-IR the frequency width of an astronomical signal of 10 km s-’ velocity distribution requires at least a 1 GHz bandwidth of the spectrometer. This is a very demanding requirement for the back-end. Similar, in the field of atmospheric studies the pressure broadened linewidth of the molecular lines can be as large as several GHz, although the purely Doppler broadened linewidth is of the order of 50 MHz only. The rather complicated structure of atmospheric lines is composed of narrow contributions from low pressure regions located very high in the stratosphere and pressure broadened signals from lower parts of the atmosphere. Thus a very large total bandwidth of the spectrometer and rather high frequency pixel resolution of a small fraction of the Doppler width (a few MHz) is required at the same time. These experimental and technical requirements are very hard to meet, but, on the other hand, the information generally available from IR transitions of molecules or even atoms or atomic irons is so valuable that it is very desirable to investigate the feasibility of infrared heterodyne spectroscopy in the fields of astrophysics as well as of atmospheric research. The well approved “standard” remote sensing technique in the IR, the Fourier transform spectroscopy (FTS) provide at present a spectral resolution of the order of about 150 MHz, which is just not yet sufficient to resolve the Doppler width of the molecules in the IR completely (for an overview of atmospheric studies with FTS see e.g. Ref. (1)). The additional problem of lineprofile distortions due to sidelobes 477

478

R.

SCHIEDER

in the spectra does inhibit all attempts to deconvolute the true molecular lineprofiles with high confidence. Therefore, at highest resolution, heterodyne spectroscopy and Fourier transform spectroscopy complement each other very nicely. A lot of information is also available from pure rotational transitions of molecules in the microwave-, mm-, or submm-region, which can be observed from ground-based observatories in the various atmospheric windows. A good example is e.g. the carbon monoxide molecule, which is detected in planetary atmospheres as well as in galactic or even extragalactic sources from many places around the globe at mm- and submm-frequencies including the KOSMA telescope on Gornergrat. But, there are numerous very important species which have no permanent dipole moment such as CH,, C, H,, C, H4, CO,, etc., so that they are not seen at radiofrequencies. Instead, they can be seen by IR spectroscopy through their rotational-vibrational transitions. In addition, many very light molecules have no rotational transitions within the accessible atmospheric radio windows.

The most

important

species

of this type

are the diatomic

hydrides.

Another

very

important example is the hydrogen molecule, which has only very weak quadrupole transitions in the infrared. The much higher frequencies in the IR as compared to mm- or submm-frequencies have some consequences for the objects being possibly studied. At 10 pm wavelength the minimum system temperature at the quantum limit is about 1440 K (DSB). This number is reduced to 14.4 K at 1 mm wavelength according to the much lower photon energy at 300 GHz. At the same time the brightness temperature of a 200 K blackbody source is only 1 K at 10 pm, but 193 K at 1 mm according to Planck’s law. The penalty is twice, the higher theoretical quantum limit of the detector noise and the reduced emissivity of the sources according to Planck’s law. This means, that an object of 200 K physical temperature can easily be investigated at 300 GHz but is barely visible at 30 THz. This limits the application of high resolution IR-heterodyne sensing (as well as of high resolution FTS) to signals from rather hot sources like HII-regions, circumstellar shells, planetary atmospheres, or some star forming regions. It is also the same matter of sensitivity, that in the IR atmospheric species are observed in absorption of the solar radiation instead of being detected directly in emission.

II.

THE

HETERODYNE

SYSTEM

We have developed a MIR heterodyne spectrometer, which is intended to be used for both atmospheric and astronomical studies. This represents the unified efforts of both laboratory IR spectroscopy and radioastronomy, which are simultaneously developed at Cologne. The basis of any heterodyne system is the mixer element, which is a photovoltaic HgCdTe detector designed for best response at 10pm. It was newly developed in close cooperation with AEG- Telefunken in Heilbronn/Germany. The reason for this new development was to improve the heterodyne efficiency and extend the usable bandwidth as much as possible. It is planned to modify the set-up into a transportable system to be used at infrared telescopes for example. The experimental arrangement of our IR-heterodyne system is shown in Fig. 1. The CO* LO beam is superimposed to the signal beam by means of a 95% reflectivity beamsplitter. Both are focused together onto the mixer with a 40 mm focal length meniscus lens. The mixer detector is integrated into the housing of a nitrogen cooled HEMT amplifier, which has been built by B. Vowinkel of our group. The mixer impedance has been measured as a function of frequency and these data are used for the design of the amplifier in order to achieve optimum coupling. This has the advantage that mixer losses due to mismatch are avoided and that ripple due to standing waves do not occur. In particular, the response at high frequencies is increased, therefore the bandwidth of the mixer can be improved greatly. At present the usable bandwidth of the system is 2 GHz, but there is no doubt that it can be extended to even higher frequencies. Otherwise the Rf-system is identical to the If-processor chain being used for all radioreceivers.

479

High resolution diode laser and heterodyne spectroscopy

The back-end is a 1.4 GHz bandwidth, 2000 channels acousto-optical spectrometer (AOS), which is similar in design to the Cologne built broad band AOSes of 1 GHz bandwidth.“) The extended bandwidth of 1.4 GHz is obtained with the same Bragg-cell that is used for the construction of the AOS for the Submillimeter Wave Astronomy Satellite (SWAS).“’ With this back-end one of the major limitations of most IR-heterodyne systems is removed, because both, bandwidth and resolution are well adapted to the spectroscopic requirements. For IR-heterodyne as well as for radioastronomical work the inherent stability of the back-end is decisive for obtaining the theoretical limits in terms of signal to noise ratio. One of the major stability problems of most AOSes is due to a rather low wavelength and power stability of the GaAs laser diode used as a coherent light source in the AOS. Usually mode competition and mode hopping generate a lot of additional low frequency noise. These are of minor importance with the new Quantum Well laser diode (wavelength of 780 nm, Hitachi) used in our design. It has a very clean spectral output and is almost free of mode hopping within relatively large temperature intervals of 5 to 10°C. At present the system is located in the laboratory at Cologne University. It can be used for absorption spectroscopy using a 1000°C blackbody source as well as a monitor of atmospheric absorptions seen in the sun light. In Fig. 2 a laboratory absorption spectrum of NH3 is shown. As local oscillator (LO) a waveguide CO,-laser was used. The spectrum was observed while chopping between a hot blackbody and an ambient temperature load. The spectrum was calibrated by division through a similar spectrum taken with the absorption cell evacuated. A spectrum of atmospheric absorptions is shown in Fig. 3, where strong absorption features from stratospheric ozone are visible. The best signal to noise ratio seen corresponds to a system temperature of approximately 2500 K (DSB). If one corrects this value for losses at the beamsplitter, the focussing

Optical isolator spectrum

_M___

analyzer

~,____~A~ I

____ -lo:

lFY I

C02-laser

)’

I I HEMT

I

<

_______+_yqqq+F)s I

I

Chopper CP

Vacuum

Mixer dewu

I I

system

Fig. 1. Scheme of the laboratory heterodyne set-up. The CO,-laser can be replaced by a stabilized tunable diode laser. The optical isolator in front of the laser consists of a linear polarizer/Fresnel-rhomb combination with an isolation of 30 dB. The isolator is essential to avoid feedback noise, particularly when using the diode laser as LO. The Fresnel-rhomb is modified with a Brewster-angled input surface. For further details see text.

480

R. SCHIEDER

NH, 2sQ(l ,l) relative to CO, lOP(14), 1.1

I,

I

I

I

integration time approx. 1 h

I

I

I

I

I

0.6

t 600

700 800 900 freq difference (MHz)

Fig. 2. Integrated laboratory heterodyne spectrum of ammonia using a CO,-laser as LO and an AOS as back-end. The spectrum is the result of two measurements, one with molecules in the cell, the second with empty cell. The plot is the ratio of both. A value of 0.5 relative intensity corresponds to 100% absorption in one sideband of the mixer.

lens, and the window on the detector Dewar one finds that out of three polarized IR-photons falling on the mixer two are detected by heterodyne mixing. This is extremely close to the theoretical limits. Therefore, the sensitivity of heterodyne detection is at least comparable to that of other methods including FTS. But this high detectivity is available together with extremely high spectral resolution of 3 x 10’ in our case, which finally gives IR heterodyne sensing an additional advantage. The heterodyne set-up is presently modified using an infrared tuneable diode laser (TDL) as LO. This modification finally provides the full power of infrared heterodyne remote sensing. The tuneability of the laser allows to detect any molecular species instead of being limited by the availability of fixed frequency gas laser transitions. At present preliminary experiments have been performed. Since the diode laser power available is less than 200 p W only, a 50% beamsplitter had to be used as diplexer. This caused a significant reduction of the heterodyne efficiency by about a factor of four. The measured effective noise temperature of the diode laser heterodyne spectrometer was as high as 18,000 K. If the additional loss due to the beamsplitter is taken into account one obtains a theoretical noise temperature of about twice the number found with the CO, laser. The remaining difference is probably due to the rather poor diode laser beam pattern, which has large impact on the imaging onto the mixer. Diode lasers have already been used as LO some years ago (see e.g. Refs (4-7)), and it has been also found by us, that low power and frequency jitter cause significant problems. In the meantime, we think, we have achieved a drastic improvement of laser frequency stability by means of locking to a HeNe-I, stabilized internally coupled Fabry-Perot interferometer (see below). With this device the accuracy of diode laser spectroscopy has developed to better than lOa, which is not too far away from values one is used to in microwave spectroscopy. On the other hand, the lack of power is still rather problematic, but we hope that we will be able to improve the heterodyne efficiency by

High resolution

diode laser and heterodyne

at least a factor of two to three by using a Fabry-Perot and submm-radioreceivers.

spectroscopy

type diplexer,

481

as is applied

The losses due to the wasted power at a 50% beamsplitter

in many mmas diplexer

should be avoided by this method. The rapid frequency jitter of the diode laser was investigated by beating the diode laser with the CO,-laser using the same heterodyne set-up. The fluctuation width of the laser was found to be of the order of 20 MHz as is reported also by other authors (see e.g. Refs (8-10)). This was measured using the AOS as a frequency analyzer for the beat signal. With the AOS as real time spectrum analyzer one has the advantage that a more realistic picture of the frequency distribution is obtained, whereas the information found with a scanning frequency analyzer represents the spectral output taken over rather long periods of time without obtaining a suitable average. We investigated the spectral distribution source in the AOS, thereby converting

within varying time intervals by chopping the GaAs laser the AOS into a time resolving real time spectrometer. With

this method we have been able to obtain rather reliable information of the IR-diode laser frequency spectrum as a function of averaging time. The shortest time investigated was as short as 5 ps, which is already very close to the sampling limits set by the Fourier relation. We found the TDL width reduced to about 10 MHz or even less at time intervals of approximately 100 ps. The laser seems to fluctuate rather slowly with a narrower lineprofile about its average center frequency. This is very encouraging when considering TDLs as LOS for heterodyne experiments, because this time scale is easily handled by simple feedback circuitry. With the resulting LO linewidth of 10 MHz the molecular line signals should become very well resolved.

Atmdspheric

Ozone

Lines

1000.0 Frequency

[MHz]

Fig. 3. Atmospheric ozone lines seen from Cologne against the sun using a CO,-laser (9P(32)). The spectrum is normalized by division through a blackbody spectrum. The unsigned feature in the right is composed of three different ozone lines. The signal is integrated within about 8 min including the blackbody measurement.

482

R.

SCHIEDER

Grating Stabilizer

HgCdTe detector

Multiple-reflection-cell

Photovoltaic

detector

Fig. 4. Stabilized tunable diode laser system. The HeNe-fringes are used as channel markers in the computer (PC). Accordingly, many scans can be coadded without any frequency shift between individual scans. The signal from the reference cell is used for control purposes. The signal is digitized by means of a digital boxcar integrator using a voltage to frequency converter. For further details see text.

III.

THE

TUNEABLE

DIODE

LASER

SYSTEM

Together with remote sensing spectra from atmospheric constituents a detailed knowledge of the precise location of the lines as well as of pressure effects on the line position and structure is required in particular. There is some information available about pressure broadening of molecular lines, but very little is known about pressure shifts because of its small magnitude.“” Both data are essential for a correct interpretation of the atmospheric signals. Using the stabilized TDL we have begun to investigate pressure effects on various IR-transitions of several molecules such as H,O 3(12)C0,(‘3) NH3,(14) etc. For precise data the signal to noise ratio of the spectrometer and the spectra1 resolution has to be very high. Most of the rapid frequency fluctuations of the laser as well as the amplitude noise due to mechanical vibrations have been removed by the construction of a shock isolated coldhead for the cooling system of the laser (see also Refs (4) and (23)). The laser frequency was finally stabilized by a stabilization scheme as shown in Fig. 4. The confocal Fabry-Perot interferometer (icFP1) (‘5,‘6)has a free spectra1 range (FSR) of about 85 MHz. Coupling in and out of the resonator is done by reflection from an internal beamsplitter, which is just an uncoated optical flat made of KBr. Due to this the interferometer can operate at wavelengths from the visible to the lower end of the MIR simultaneously. The diode laser is locked to one transmission minimum of the icFP1 and tuned by changing the optical pathlength of the icFP1 with a second tilting optical flat. At the same time the beam of an Iodine stabilized HeNe-laser at 630 nm is coupled into the icFP1 and its fringes are registered. A change of the optical

High resolution diode laser and heterodyne spectroscopy

483

pathlength of a quarter wavelength of the HeNe-laser generates the next fringe of the visible laser. In the IR the corresponding frequency change is determined by the ratio of the wavelengths of the two lasers, which results in about 5 MHz tuning per HeNe-fringe. This is a perfect raster for any registration of typical molecular lines. In addition since the HeNe-laser wavelength is stabilized to about 10” relative stability, the tuning in the IR can be controlled and reproduced with very high accuracy (about 108). This is a perfect basis for repeating scans across any molecular line several times, i.e. for the application of signal averaging just by counting the HeNe-fringes up and down. An example of signal averaged lineprofile measurements are seen in Fig. 5 as they have been obtained on the aQ(6,3) transition of ammonia. Several spectra of the same line with varying foreign gas pressure of a 60% nitrogen and 40% oxygen mixture are shown. The pressure broadening of this line is very evident. Each data point (channel) in the plot corresponds to one HeNe-fringe. Any atmospheric line must be understood as a superposition of many of such broadened lines. The top line is an example of the fit of one of the spectra to a theoretical pressure broadening function (Galatry-profile, see e.g. Ref (17)). Although it is only barely visible in the plots the pressure shift can also be derived from these measurements, which is about 187 MHz/bar for this particular line and broadening gas mixture. The error in this shift measurement is of the order of a few percent only. Measurements of different mixtures of oxygen and nitrogen are on the way. The reason for the investigation of varying mixtures of these gases is that, during measurements on some H,O transitions we have found indications that pressure shifts of different gases may not be additive. This may be an effect of collisional line mixing. Therefore shift measurements with pure oxygen and pure nitrogen are not sufficient for a good value of the shift parameters in normal air for example. If this proves to be correct for all molecular lines, the shift data for air have to be re-evaluated in many cases. (This is also discussed in the paper given by A. F. Krupnov on this conference.) In addition, also rather

NH, 2aQ(6,3) Mixture ratio: 60 % Nitrogen, 40 % Oxygen

I

I

I

I

1

I

500.0

750.0

(observed-calculated ,60 mbar)

0.95

0.90

0.80 250.0

Channel number Fig. 5. Pressure broadening of an ammonia line. The zero levels of the curves are displaced for better visibility. The upper curve shows the residuals from a fit to a Voigt-profile. The corresponding fit to a Galatry-profile does not provide different results in this case.

484

R. SCHIEDER

Line Width of Hydrogen

S,(3)

50 t Oi 0

200

400

600

800

_ 1000

pressure [mbar] Fig. 6. Pressure dependence of a hydrogen rotational quadrupole line (self-broadening). The solid line is the result of theoretical calculations. Dicke-narrowing is very prominent on this line.

small amounts of water in the gas samples have very large impact on the pressure shifts observed. The very high permanent dipole moment of water produces very large collisional effects. Experiments of this type are also on the way for a precise measurement of pressure effects on forbidden rotational transitions of molecular hydrogen. These transitions are very good probes for studies of the atmospheres of the outer planets, in particular, of possible global oscillations on Jupiter similar to the oscillations as found for the sun. (‘*) Measurements on the &s,(3) quadrupole a transition at 1063 cm-’ have been performed in our laboratory as well as in other laboratories”9.20’ also on other quadrupole transitions of hydrogen. A plot of the observed self-broadening versus pressure is shown in Fig. 6. The data found in our laboratory are in much closer agreement with theoretical calculations than in Ref. (19). As is visible in the plot, the linewidth becomes reduced with increasing pressure, which is a very nice representation of Dicke-narrowing.“‘) This narrowing is a very interesting phenomenon of coupling between Doppler effect and collisions. It demonstrates that in a dense gas the assumption of unperturbed linear movement of the molecules is no more valid, but instead the changes of the path of the molecules have to be taken into account for the line profiles. As is obvious, pressure induced changes in the linewidth as well as pressure shifts are very important processes which influence the line profiles in planetary atmospheres. The column density of the molecular transitions is rather high in most cases, whereas the particle density is very low. In order to do the laboratory experiments under nearly identical conditions, one has to establish ways to see absorptions of sometimes weak lines even under very low pressure conditions. For instance, for the H, quadrupole lines a pathlength of several hundred meters is required, in order to see these transitions with sufficient signal to noise at pressures of a few Torr only. This can be achieved with multiple path absorption cells. The most widely known cell of this type, the White-cell, has long been used in conventional spectroscopy. But this design causes problems due to optical imaging errors. We have adopted the idea of a Herriott-cell,‘22’ which is very well suited for being used with laser beams. Our design is a modified Herriott design, which separates the beam spots on the mirrors much more effectively (see also Ref. (23)). At the same time the volume of

High resolution

Modified

diode laser and heterodyne

Herriott-Cell spherical mirrors focal length: 250

flat’ mirror

485

spectroscopy

Mirrors

cm

inserted

Fig. 7. Modified Herriott-cell beam pattern. The small circles indicate the laser beam spots on the two mirrors. The additional flat mirror in the center of the large spherical mirror generates four different circles for the beam spots. The hole for input and output of the laser beam is located on the right side of the left mirror.

the cell is used more effectively, which helps to minimize the size of such cells quite significantly (see Fig. 7). The cell is designed for pressures up to 5 bar and has a path length of more than 300 m, while the laser beam is reflected 144 times between the mirrors. The advantage of such a cell is that it generates negligible imaging errors, so that there is minimal loss of signal when focusing the beam onto the small detector area.

IV.

FUTURE

OUTLOOK

The joint efforts of IR laboratory spectroscopy and radioastronomy are intended to result in transportable and tuneable IR heterodyne spectrometer with about 3 GHz total bandwidth. Continuous work is on-going in order to improve the diode laser situation. There seems to be some improvement visible with quantum well devices or distributed feedback structures, so that the power and single mode problem may be resolved rather soon. On the other hand, we are also trying to improve the mode quality of the lasers by using external feedback. First encouraging results have been obtained just recently. This technique has been quite successfully applied to GaAs-lasers, and we believe that it will improve the laser quality significantly. There are numerous applications in the fields of high resolution astrophysical spectroscopy, if the heterodyne system is moved to an IR telescope. We consider the laboratory spectroscopy as an essential and integral part of most future applications of heterodyne experiments. For studies of the atmosphere of our own planet as well as of other planets the analysis of collisional effects is very important and will remain so. Collisional line mixing is another example for this kind of work. It is also an important aspect of laboratory experiments to study the energy transfer between different molecular species during collisions. This is probably one of the key points in understanding the energy balance within the interstellar medium. It all serves as a very good example of fruitful collaboration between laboratory spectroscopy, atmospheric research, and astrophysics. Acknow~eerlgemenrs-This work is supported by the Deutsche Forschungsgemeinschaft presentation is the result of the work of many members of the I. Physikalisches Institut.

through

grant

SFB-301.

The

486

R. SCHI~ER

REFERENCES I. See e.g. M. A. H. Smith, C. P. Rinsland, V. M. Devi, L. S. Rothmann and K. N. Rao, Spectroscopy of the Earth’s Atmosphere and Interstellar Medium (Edited by K. N. Rao and A. Weber), p. 153. Academic Press, New York (1992); and references therein. 2. V. Tolls, R. Schieder and G. Winnewisser, Expl Aslron. 1, 101. 3. See e.g. G. J. Melnick et al., Proc. of the 2!%h Litge International Astrophysical Colloquium “From Ground-Based to Space-Borne Submillimeter Astronomy”, Liege, Belgium (ESA SP-3 14) (1990). 4. D. Glenar, T. Kostiuk, D. E. Jennings, D. Buhl and M. J. Mumma, Appl. Oppr.21, 243 (1982). 5. M. A. Frerking and D. J. Muehlner, Appl. Opt. 16, 526. 6. R. T. Ku and D. L. Spears, Opt. Left. 1, 84 (1977). 7. M. Mumma, T. Kostiuk, S. Cohen, D. Buhl and P. C. von Thuna, Nufure 253, 514 (1975). 8. J. Reid, D. T. Cassidy and R. T. Menzies, Appl. Opt. 21, 3961 (1982). 9. J. J. Hillman, D. E. Jennings and J. L. Faris, Appl. Opt. 18, 1808 (1979). 10. T. L. Worchesky, K. J. Ritter, J. P. Sattler and W. A. Riessler, Opt. Leit. 2, 70 (1978). 11. See e.g. M. A. H. Smith, C. P. Rinsland, V. M. Devi, L. S. Rothmann and K. N. Rao, Spectroscopy of the Earth’s Atmosphere and Interstellur Medium (Edited by K. N. Rao and A. Weber), p. 153. Academic Press, New York (1992); and references therein. 12. T. Giesen, R. Schieder, G. Winnewisser and K. M. T. Yamada, J. mokc. Spectr. 153, 406 (1992). 13. N. Anselm, R. Schieder, G. Winnewisser and K. M. T. Yamada, J. molec. Spectr. In press. 14. H.-J. Clar, R. Schieder, G. Winnewisser and K. M. T. Yamada, J. molec. Structure 190, 447 (1988). 15. M. Reich, R. Schieder, H.-J. Clar and G. Winnewisser, Appl. Opt. 25, 130 (1986). 16. H.-J. Clar, R. Schieder and M. Reich, Appl. Oppt.28, 1648 (1989). 17. L. Galatry, Phys. Rev. 122, 1218 (1961). 18. See e.g. B. Mosser, D. Gautier and T. Kostiuk, Icarus 96, 15 (1992). 19. See e.g. J. Reid and A. R. W. McKellar, Phys. Reo. A 18, 224 (1978). 20. V. DeCosmo, H. P. Gush and M. Halpern, Can. J. Phys. 62, 1713 (1984). 21. R. H. Dicke, Phys. Rev. 89, 472 (1953). 22. D. R. Herriott, H. Kogelnik and R. Kompfner, Appl. Opt. 3, 523 (1964). 23. See N. Anselm, T. Giesen, M. Harter, R. Schieder and G. Winnewisser, Monitoring of Gaseous Pollutants by Tunable Diode Lasers (Edited by R. Grisar, H. Bottner, M. Tacke and G. Restelli). Commission of the European Communities Air Pollution Reearch Reports, Proceedings of the International Symposium held in Freiburg, Germany, pp. 231-240. Kluwer Academic, Dordrecht (1991).