The development of new radiation source technolo-gies has had a major impact on the progression of op-tical trace gas detection. At the end of the 1990s, asemiconductor laser that used intersubband transi-tions in a multiple quantum well structure basedon GaAs or InP was developed [1]. These quantumcascade lasers (QCLs) emit infrared radiation in thewavelength range of 3 to 300 μm depending on sub-strate and layer system. Thus, they are able to reachthe strong fundamental rotational–vibrational ab-sorption lines of most relevant gases. In the earlydays, QCLs required cryogenic cooling using liquidnitrogen, and their emission spectrum was multi-mode. Most of today’s QCLs can easily be operatedat room temperature and spectrally emit a singlelongitudinal mode [2]. This mode can be continuouslytuned within the spectral range via its current andtemperature [3]. For use in optical gas analyzers,
the spectral linewidth has to be very narrow in orderto ensure selective detection of the measuredcompound.
QCLs were first employed in photoacoustic (PA)analyzers in 1999 [4]. In the past few years, severalPA investigations with different types of QCL havebeen reported. Setups with pulsed multimode lasers[5], pulsed distributed feedback (DFB) lasers [2], andcontinuous wave DFB lasers [4] were used for gasdetection.
In this study, the spectral resolution of a PA gasdetection system based on a pulsed DFB QCL is in-vestigated. In order to compare the used laser withthe results of earlier publications [6–10] the effect ofQCL operating parameters pulse length and peakcurrent on the spectral emission linewidth is re-searched. QCL linewidth investigations using PAspectroscopy were first reported in [2,11,12]. Toour knowledge, the influence of modulation techni-ques on the spectral emission of a QCL is studiedhere for the first time. Effective linewidths forpulse gate modulation, pulse frequency modulation,and chopper modulation are compared. All the
measurements were performed by means of PAspectroscopy on nitrogen monoxide (NO) absorptionlines at a wavelength of 5:27 μm.
2. Experimental Setup and Theory
A. Distributed Feedback Quantum Cascade Laser
The commercially available pulsed DFB QCL (nano-plus GmbH, Gerbrunn, Germany) used in this studycan be operated near room temperature. The activelayer system ðGa0:47In0:53As=Al0:48In0:52AsÞ wasgrown on an n-doped InP substrate (n ¼ 2×1017 cm−3) by molecular beam epitaxy. The 25 periodsof active zones and injectors have a total thickness of1130nm. They are embedded between a 400nm GaI-nAs cladding, a 40nm AlGaInAs cladding, and anInP contact layer.The wavelength of the QCL’s single mode emission
can be continuously tuned between 5.269 and5:275 μm via its operational temperature with a tun-ing rate of ð0:40� 0:05Þnm=K. Figure 1 shows thewavelength as a function of the temperature. Its ab-solute operational temperature range is −30 °C to−10 °C, and its pulse length and repetition frequencycan be varied up to 100ns and up to 2MHz, respec-tively. Operating the laser in a safe region, the dutycycle must not exceed 2%. The laser chip is mountedin a TO8 housing with an internal thermoelectriccooler and negative temperature coefficient (NTC)thermistor to control the laser temperature. Itsthreshold current is 1:8A at −10 °C, and its maxi-mum peak current is 2:7A. The average emissionpower for 11:2ns pulse duration and 500kHz repeti-tion frequency at 2:1A is shown in Fig. 2 as a func-tion of temperature.
B. Photoacoustic Spectroscopy
PA detection utilizes the fact that the excitation en-ergy of light-absorbing molecules is, at atmospheric
conditions, to a large extent transferred into kineticenergy of the surrounding molecules via inelastic col-lisions. This causes a local pressure increase in theabsorbing gas. If the excitation source is modulated,a sound wave is generated that can be detected by amicrophone and measured phase sensitively using alock-in amplifier [13]. PA spectroscopy has the ad-vantage of producing a signal only when light is ab-sorbed. It is therefore an offset-free technique. ThePA signal S is directly proportional to the absorptioncoefficient α and the laser power P0 [14]:
S ¼ α · P0 · CCELL: ð1Þ
The proportionality factor CCELL depends on thegeometry of the sample cell as well as the beam pro-file, the microphone sensitivity, and the modulationof the light source. With a tunable radiation sourcethe absorption coefficient can directly be measuredas a function of the wavelength αðλÞ.
The sensitivity of a PA sensor can be considerablyimproved by taking advantage of acoustic resonancesof the sample cell. Modulating the laser radiation ata frequency equivalent to an acoustic mode of the cellenables an enormous enhancement of the signal. Thecombination of PA spectroscopy with spectral-selective radiation sources such as QCLs can provide
Fig. 1. Emission wavelength as a function of operationaltemperature.
Fig. 2. Average emission power of the QCL as a function ofoperational temperature for 11:2ns, 500kHz, and 2.1 A.
a simple and sensitive detection system for nearlyevery molecule.The experimental setup shown in Fig. 3 includes a
PA cell with a first longitudinal acoustic mode at2:8kHz. The cell combines a 60mm cylindrical reso-nance tube with a diameter of 6mm and two buffervolumes of 22:5mm in length and 24mm in diameterat its ends [15]. The signal of the electret microphoneis phase sensitively measured with a lock-in ampli-fier (Signal Recovery Model 7265). A PC is used fordata acquisition and to control the lock-in amplifierand the laser temperature. Details of the electronicpulse module and modulation techniques are givenin Subsection 2.C. PA spectra were recorded usinga calibrated mixture of 100ppm NO in nitrogen(6.0) at ambient pressure and temperature. The pres-sure in the cell was measured and was held constantto within �1hPa. The strong fundamental vibra-tional absorption band (v ¼ 1 → v ¼ 0) of NO canbe reached with the laser. Its emission was spectrallyscanned over the absorption lines R5.5 of NO at5.2705 and 5:2715 μm by means of temperature.The minimum transmission after the complete ab-sorption length of 105mm for 100ppm is 98.26%.The pressure-broadened width of the NO lines is0:30nm (FWHM) at atmospheric pressure.
C. Modulation Techniques
For PA measurements the radiation has to be modu-lated. The most common method to modulate a lightsource is a rotating chopper wheel. This technique ispreferred for broadband sources and high power gaslasers but can be applied with any radiation source.Whereas continuous wave diode lasers are usuallydirectly modulated by periodically changing their in-jection current, pulsed diode lasers require a moresophisticated technique.In this study the QCL is driven by a modified
module from directed energy that generates pulseswith pulse lengths from 6 to 100ns and peak cur-rents up to 5 A. The driver module is triggered bya pulse generator and, depending on pulse duration,can be driven with repetition frequencies up to1:5MHz. In order to modulate the pulsed QCL, threedifferent techniques are used. A mechanical modula-tion using a chopper and two electronically per-formed methods: pulse gate modulation and pulsefrequency modulation.For chopper modulation the laser is driven with a
constant pulse length and constant repetition fre-quency. The chopper modulates the QCL with a fre-quency of 260Hz in a nonresonant acoustic mode.The frequency is measured with a photo sensor thatsimultaneously provides the reference for the lock-inamplifier.Figure 4(a) displays the electronic modulation
methods. For pulse gate modulation, the internallock-in oscillator is used as reference and pulsesare produced only during the positive half-wave.During the negative half-wave the laser is switchedoff. The pulse sequence has constant duration and re-
petition frequency. This method represents an ampli-tude modulation of the laser power.
For pulse frequency modulation, the internal lock-in oscillator is also used as reference, however, therepetition rate changes during one period. A proprie-tary developed controller between the lock-in oscilla-tor and the pulse generator enables modulationusing a voltage controlled oscillator (VCO). The repe-tition frequency of the laser pulses equals the VCOfrequency. It produces a center frequency without avoltage at its input, and its frequency increaseswith positive and decreases with negative voltage[Fig. 4(b)]. The amplitude of the lock-in oscillatorcontrols the repetition frequency of the laser. Thismethod represents a frequency modulation thatresults in an alternating average power. For bothelectronic modulation methods, the frequency ofthe lock-in oscillator is tuned to the first longitudinalmode of the PA cell at 2:8kHz.
D. Calculation of Linewidth
To determine the spectral linewidth of the laser, theabsorption spectrum of 100ppm NO in N2 at 293Kand 1013hPa was calculated using the HITRAN da-tabase [16]. This spectrum was convolved with an as-sumed emission profile of the laser and the resultingfunction was compared to the PA measurements. Asimilar method has been successfully applied in[2,11]. The width of the assumed emission profilewas subsequently modified until the convolved func-tion best fit with the measurements. The linewidth ofthe laser was then calculated as the full width athalf-maximum (FWHM) value.
For pulse lengths shorter than 10ns, a Gaussianemission profile results in a good fit with measure-ment. Pulse lengths larger than 30ns, however, re-quire an almost rectangular profile [17–19]. Forpulse durations between 10 and 30ns, neither oneof these profiles resulted in a good agreement with
Fig. 4. (a) Pulse gate modulation and (b) pulse frequencymodulation.
measurement. Therefore, a new profile was createdand used for the evaluation.It was assumed that the QCL spectrally emits a
Gaussian profile. This constantly narrow emission,however, is spectrally shifted due to periodic heatingof the laser. Its high thermal tuning rate of ð0:40�0:05Þnm=K is the cause of this chirp. The primaryheating occurs during the current pulse. The rela-tively high pulse current of 2.1 A and pulse voltageof 13V result in high power dissipation in the activeregion (current density >20kA=cm2) [20]. A secondheating and the resulting spectral shift occurs ifthe laser is electronically modulated (not by chop-per). At modulation frequencies with periods onthe millisecond time scale, additional heat diffusesperiodically through the laser. Depending on the par-ticular modulation technique this second effect canbe more dominant than the heating during the pulse.An example of the new emission profile is demon-
strated in Fig. 5 (only the primary heating during thepulse). The figure shows the Gaussian emission pro-file with a constant linewidth of 0:12nm FWHM atthe beginning of the electrical pulse and the shiftedprofile at the end of the pulse. The solid line showsthe approximated effective emission profile. The ef-fective function achieved very good agreement withthe experimental data for all pulse lenghts.An example of a calculated NO spectrum, a mea-
sured PA spectrum, and the convolved function isshown in Fig. 6. The presented PA spectrum was re-corded with a pulse duration of 11:2ns, a repetitionfrequency of 1:1MHz, and pulse gate modulation.The measured PA spectrum was normalized accord-ing to the average QCL power. The linewidth of thelaser was calculated as 0:41nm, and the convolvedfunction fits very well with the measured PA spectra.All following linewidth calculations for the laser wereperformed the same way.
3. Results
A. Pulse Length
For these measurements pulse lengths varying from6.8 to 30ns and a constant pulse peak current of 2:1Awas used. Spectra of the calibrated NO mixture wererecorded by sweeping the temperature of the laserfrom −25 °C to −18 °C. The duty cycle was held con-stant at 1.3% by changing the repetition frequency ofthe laser pulses. These measurements were per-formed with pulse gate modulation as described inSubsection 2.C.
Figure 7(a) demonstrates exemplary PA spectrafor five different pulse lengths. The spectra were nor-malized according to the QCL power. It is clear thatthe spectral resolution increases with shorter pulselength. The figure displays that the absorption linesof NO can be separated with pulse durations shorterthan 17:5ns. With longer pulses it is not possible.The calculated linewidth as a function of pulse lengthfor nine pulse durations is shown in Fig. 7(b). Thecalculated linewidth increases quite linearly withthe pulse duration. As a first approach the linewidthΔv can be calculated from the pulse length Δt as
Δν ¼ 0:035nmns
·Δt; ð2Þ
which corresponds to a chirp rate of 380MHz=ns.Longer pulses produce higher power dissipationin the active region and cause further shifting of thewavelength.
B. Pulse Peak Current
For these measurements different laser pulse peakcurrents are used with a constant pulse length of17:5ns. The spectra of the calibrated NO mixtureFig. 5. Effective emission profile of the QCL.
Fig. 6. NO absorption spectrum according to HITRAN, measuredPA spectrum, and convolution of NO spectrum and effective laserprofile.
were recorded by sweeping the temperature of thelaser from −25 °C to −18 °C. The repetition rate wasmodified for each current to get the maximum ofthe absorption spectrum to remain at −21 °C. Thiscorresponds to a wavelength of 5:2715 μm. Thepulse gate modulation technique was used for thesemeasurements.Figure 8(a) illustrates the normalized PA spectra
of NO for 1.8 to 2.2 A peak current. It can be noticedthat the spectral resolution increases with decreas-ing laser peak current. Figure 8(b) shows the line-width as a function of the laser peak current. Thedependency looks quite linear. As a first approachthe linewidthΔv can be calculated from peak currentIpeak as
Δν ¼ 0:325nmA
· Ipeak: ð3Þ
This equation is valid only for pulse currents abovethe threshold. Higher currents produce higher powerdissipation in the active region and cause furthershifting of the wavelength.
C. Modulation Technique
The results obtained in Subsection 3.A for pulselengths of 6.8 to 17:5ns performed by pulse gatemod-ulation are compared here to measurements per-formed by pulse frequency modulation and choppermodulation. For pulse frequency modulation the ex-perimental setup described in Subsection 2.C wasused. The center frequency of the VCO was tunedto 450kHz. The lock-in oscillator is used to tunethe frequency of the VCO up to 900kHz with its max-imum amplitude and down to 5kHz with its mini-mum. These modulation parameters were chosenin order to get a considerable change of laser powerand, therefore, large PA signals. Their optimizationis an issue for further investigations. For choppermodulation the laser was driven with a constant dutycycle of 1.3% equivalent to pulse gate modulation. Allthe measurements were taken at a peak current of2.1 A.
In Fig. 9(a) the results of the PA measurement forall three methods are compared for a pulse length of11:2ns. It can be noticed that the spectral resolutionconsiderably depends on the technique of modula-tion. The calculated linewidths for all pulse lengthsare shown in Fig. 9(b).
Fig. 7. (a) Exemplary PA spectra for five different pulse lengthswith 2.1 A peak current. (b) Calculated laser linewidth as a func-tion of pulse length for nine pulse lengths.
Fig. 8. (a) PA spectra for different pulse peak currents with11:2ns pulse length. (b) Calculated laser linewidth as a functionof laser peak current.
It can be observed that the course of the graphs forall three techniques is parallel. The narrowest laserlinewidth is achieved with chopper modulation. Thisis reasonable because the laser temperature is, dueto the constant pulse length and repetition frequency,more stable than for the electronic modulation tech-niques. For pulse gate and pulse frequency modula-tion the changing repetition frequency and the gateswitching of the laser pulses, respectively, causes anadditional shift of the laser temperature. Thisadditional temperature shift occurs with the fre-quency of the lock-in oscillator of 2:8kHz and istoo quick for the temperature controller to compen-sate. The temperature shift for pulse gate modula-tion is higher than for pulse frequency modulationbecause the laser is turned off completely duringthe negative period of the lock-in oscillator. The line-width results for chopper modulation are in goodagreement with previous works where QCL line-widths have been investigated with constant pulsedurations and repetition frequencies [9,12,17].
4. Conclusion
PA investigations of the effective linewidth of apulsed DFB QCL were performed in order to deter-mine optimum operating parameters for high resolu-
tion PA spectroscopy. It is shown that the emissionline profile depends considerably on the laser pulselength and peak current. Lower currents and shorterpulse lengths reduce the linewidth of the QCL.
The technique of modulating the radiation also hasa significant influence on the spectral resolutionof PA measurements. The highest resolution wasachieved with chopper modulation. The electroni-cally performed methods reduce the spectral resolu-tion, with the pulse frequency modulation beingslightly better than the pulse gate modulation.The electronic modulation techniques result in laserlinewidths that are equal to or larger than the line-width of the NO absorption lines. The chopper mod-ulation leads to a slightly narrower linewidth. Forfuture investigation with higher accuracy, measure-ments at reduced pressures could be appropriate.
The results of this study will be used to develop ahigh resolution PA system for the detection of NO.For the simultaneous goal of high detection sensitiv-ity, a compromise will have to be made. The optimi-zation will be an issue for forthcoming research.
References1. J. Faist, F. Capasso, D. L. Sivco, C. Sirtori, A. L. Hutchinson,
and A. Y. Cho, “Quantum cascade lasers,” Science 264,553–556 (1994).
2. M. G. da Silva, H. Vargas, A. Miklós, and P. Hess, “Photoacous-tic detection of ozone using a quantum cascade laser,” Appl.Phys. B 78, 677–680 (2004).
3. D. Weidmann, F. K. Tittel, T. Aellen, M. Beck, D. Hofstetter,J. Faist, and S. Blaser, “Mid-infrared trace-gas sensing witha quasi-continuous-wave Peltier-cooled distributed feedbackquantum cascade laser,” Appl. Phys. B 79, 907–913 (2004).
4. B. A. Paldus, T. G. Spence, R. N. Zare, J. Oomens, F. J. M.Harren, D. H. Parker, C. Gmachl, F. Capasso, D. L. Sivco,J. N. Baillargeon, A. L. Hutchinson, and A. Y. Cho, “Photoa-coustic spectroscopy using quantum-cascade lasers,” Opt.Lett. 24, 178–180 (1999).
5. S. Barbieri, J.-P. Pellaux, E. Studemann, and D. Rosset, “Gasdetection with quantum cascade lasers: an adapted photoa-coustic sensor based on Helmholtz resonance,” Rev. Sci.Instrum. 73, 2458–2461 (2002).
6. A. A. Kosterev, F. K. Tittel, R. Köhler, C. Gmachl, F. Capasso,D. L. Sivco, A. Y. Cho, S. Wehe, and M. G. Allen, “Thermoelec-trically cooled quantum-cascade-laser-based sensor for thecontinuous monitoring of ambient atmospheric carbonmonox-ide,” Appl. Opt. 41, 1169–1173 (2002).
7. R. Jiménez, M. Taslakov, V. Simeonov, B. Calpini, F. Jeanner-et, D. Hofstetter, M. Beck, J. Faist, and H. van den Bergh,“Ozone detection by differential absorption spectroscopy atambient pressure with a 9:6 μm pulsed quantum-cascadelaser,” Appl. Phys. B 78, 249–256 (2004).
8. A. A. Kosterev, F. K. Tittel, C. Gmachl, F. Capasso, D. L. Sivco,J. N. Baillargeon, A. L. Hutchinson, and A. Y. Cho, “Trace-gasdetection in ambient air with a thermoelectrically cooled,pulsed quantum-cascade distributed feedback laser,” Appl.Opt. 39, 6866–6872 (2000).
9. D. D. Nelson, J. H. Shorter, J. B. McManus, andM. S. Zahniser,“Sub-part-per-billion detection of nitric oxide in air using athermoelectrically cooled mid-infrared quantum cascade laserspectrometer,” Appl. Phys. B 75, 343–350 (2002).
10. J. B. McManus, D. D. Nelson, S. C. Herndon, J. H. Shorter,M. S. Zahniser, S. Blaser, L. Hvozdara, A. Muller, M. Giovan-nini, and J. Faist, “Comparison of cw and pulsed operation
Fig. 9. (a) PA spectra for different modulation techniques for11:2ns pulse length. (b) Calculated laser linewidth as functionof pulse length for three differen modulation techniques.
with a TE-cooled quantum cascade infrared laser for detectionof nitric oxide at 1900 cm−1,” Appl. Phys B 85, 235–241 (2006).
11. M. B. Filho, M. G. da Silva, M. S. Sthel, D. U. Schramm,H. Vargas, A. Miklós, and P. Hess, “Ammonia detection byusing quantum-cascade laser photoacoustic spectroscopy,”Appl. Opt. 45, 4966–4971 (2006).
12. D. Hofstetter, M. Beck, J. Faist, M. Nägele, and M. W. Sigrist,“Photoacoustic spectroscopy with quantum cascade distribu-ted-feedback lasers,” Opt. Lett. 26, 887–889 (2001).
13. W. Demtröder, Laser Spectroscopy (Springer-Verlag, 2002).14. Z. Bozóki, M. Szakáll, Á. Mohácsi, G. Szabó, and Zs. Bor,
“Diode laser based photoacoustic humidity sensors,” Sens.Actuators B. 91, 219–226 (2003).
15. A. Miklós, P. Hess, and Z. Bozóki, “Application of acoustic re-sonators in photoacoustic trace gas analysis and metrology,”Rev. Sci. Instrum. 72, 1937–1955 (2001).
16. L. S. Rothman, D. Jacquemart, A. Barbe, D. C. Benner, M.Birk, L. R. Brown, M. R. Carleer, C. Chackerian, Jr., K.Chance, L. H. Coudert, V. Dana, V. M. Devi, J.-M. Flaud,R. R. Gamache, A. Goldman, J.-M. Hartmann, K. W. Jucks,
A. G. Maki, J.-Y. Mandin, S. T. Massie, J. Orphal, A. Perrin,C. P. Rinsland, M. A. H. Smith, J. Tennyson, R. N. Tolchenov,R. A. Toth, J. Vander Auwera, P. Varanasi, and G. Wagner,,“The HITRAN 2004 molecular spectroscopic database,”J. Quant. Spectrosc. Radiat. Transfer 96, 139–204 (2005).
17. T. Beyer, M. Braun, and A. Lambrecht, “Fast gas spectroscopyusing pulsed quantum cascade lasers,” J. Appl. Phys. 93,3158–3160 (2003).
18. T. Beyer, M. Braun, S. Hartwig, and A. Lambrecht, “Linewidthmeasurement of free-running, pulsed, distributed feedbackquantum cascade lasers,” J. Appl. Phys. 95, 4551–4554(2004).
19. B. Grouiez, B. Parvitte, L. Joly, D. Courtois, and V. Zeninari, “Comparison of a quantum cascade laser used in both cw andpulsed modes. Application to the study of SO2 lines around9 μm, ” Appl. Phys. B 90, 177–186 (2008).
20. J. Faist, C. Gmachl, F. Capasso, C. Sirtori, D. L. Sivco,J. N. Baillargeon, and A. Y. Cho, “Distributed feedbackquantum cascade lasers,” Appl. Phys. Lett. 70, 2670–2672(1997).
![Basic Pneumatic Cascade]](https://static.documents.pub/doc/80x56/5414f9ac7bef0a8d128b45bf/basic-pneumatic-cascade.jpg)
