Narrow linewidth Quantum Cascade Lasers as ultra-sensitive probes of molecules

Invited Paper

Narrow linewidth Quantum Cascade Lasers as ultra-sensitive probes of molecules Saverio Bartalini, Simone Borri, Pablo Cancio Pastor, Iacopo Galli, Giovanni Giusfredi, Davide Mazzotti and Paolo De Natale Istituto Nazionale di Ottica (INO) - CNR, Largo E. Fermi 6, 50125 Firenze, Italy, and European Laboratory for Nonlinear Spectroscopy (LENS), via N. Carrara 1, 50019 Sesto Fiorentino (FI), Italy ABSTRACT Recently, we have demonstrated that the “intrinsic” linewidth of Quantum Cascade Lasers (QCLs) can go beyond the radiative lifetime of the upper level. This represents the first demonstration of a sub-radiative linewidth for any laser. The intrinsic linewidth of a QCL can be as narrow as hundreds Hz, paving new ways for ultrasensitive and precise harnessing and detection of molecules. We are working towards full exploitation of such intrinsic properties by designing appropriate phase-lock loops and enhancement-cavities for interaction with molecules. Combination with optical-frequency-comb-synthesizers and appropriate spectroscopic techniques, like saturated-cavity-ring-down-SCAR or polarization spectroscopy can provide unprecedented sensitivity and frequency accuracy for molecular detection. Keywords: Quantum Cascade Lasers, Frequency Metrology, Infrared Spectroscopy

1. INTRODUCTION Intersubband quantum cascade lasers (QCLs) have become well established light sources in the mid-infrared (IR) spectral region, with many applications reporting high-sensitivity trace-gas detection,1, 2 high-resolution spectroscopy and precision frequency metrology.3, 4 Along with the increased sensitivities and precisions of QCL-based spectroscopic measurements, a comprehensive investigation and understanding of the laser frequency stability and noise properties has become more and more necessary. In fact, noise control in laser sources is a powerful tool for boosting the performances of spectroscopic experiments and applications, from the points of view of both sensitivity and precision. On the other hand, this requires a thorough knowledge of the characteristic noise figure and of the physical mechanisms governing the noise. In this way it is possible to predict the benefits that the suppression of noise may produce, and also to calculate the ultimate noise level achievable by a stabilized source. Indeed, the ultimate limiting factor of the sensitivity and precision of spectroscopic measurements is the laser quantum noise, i.e. the phase and amplitude noise of the photons field. This investigation also plays a key role in understanding the inner physics of QCLs. These devices are, in fact, fundamentally different from conventional IR semiconductor lasers and, since their first demonstration, they have been expected to exhibit peculiar noise characteristics. In the following we summarize our recent results on the measurement of QCLs intrinsic linewidth and on the analysis of their frequency noise features. Our observation provides a direct evidence of the leveling of this noise down to a white noise plateau, corresponding to the QCL intrinsic linewidth, and gives the first experimental confirmation of the most recent theoretical predictions. Some interesting applications of such narrow-linewidth sources will be finally discussed. Further author information: S. Bartalini: E-mail: [email protected], Telephone: +39 (0)55 4572500

Quantum Sensing and Nanophotonic Devices VIII, edited by Manijeh Razeghi, Rengarajan Sudharsanan, Gail J. Brown, Proc. of SPIE Vol. 7945, 794505 · © 2011 SPIE · CCC code: 0277-786X/11/$18 · doi: 10.1117/12.871401

Proc. of SPIE Vol. 7945 794505-1 Downloaded from SPIE Digital Library on 05 May 2011 to 193.205.148.189. Terms of Use: http://spiedl.org/terms

2. INTRINSIC LINEWIDTH OF QCLS Since their first appearance in 1994,5 an intrinsic linewidth very close to that provided by the Schawlow-Townes formula,6 with the so-called α-factor7 close to zero, was predicted for QCLs. If, on one hand, this landmark formula is very general and can be applied to any laser system, on the other hand it is often not easy to evaluate accurately, since it involves parameters (such as intra-cavity intensity and cavity losses) that can not be determined with an acceptable precision. Only recently a comprehensive theoretical work,8 based on rate equation analysis of a three-level model of a QCL, has finally unveiled the physics behind the narrowness of QCLs linewidths, leading to a reformulation of the Schawlow-Townes equation in terms of the characteristic parameters of the QCL medium. In particular, this work introduces the novel concept of “effective” coupling of the spontaneous emission. This effect leads to a remarkable prediction: in QCLs the linewidth should overcome the limit set by the S-T formula for conventional bipolar semiconductor lasers, giving much narrower emission features. The prediction can be verified by performing a direct analysis of the QCL frequency-noise power spectral density (FNPSD) of QCL over a wide enough bandwidth. Such an analysis requires a very challenging set-up, including a low-noise current driver (which has not to mask the very tiny laser intrinsic noise), a fast and sensitive detector and acquisition system. For the first time we put together all these ingredients and have been able to measure the full FNPSD of a free-running QCL in the 0-100 MHz frequency bandwidth. In this experiment we used a mid-IR single mode cryogenically cooled QCL emitting at λ =4.3 μm. The frequency noise measurements have been performed by using the side of a Doppler broadened molecular transition of CO2 as a frequency discriminator. This is a well-established method for a low-noise conversion of the laser frequency fluctuations into detectable intensity variations.9 The laser frequency is locked to the half height position of the absorption line, so that the spectrum of the intensity transmitted by this discriminator reproduces the spectrum of the laser frequency fluctuations amplified by the slope of the absorption profile. The signal is detected with a fast 200 MHz bandwidth HgCdTe detector. We measured the QCL FNPSD by using two different current sources: a home-made ultra-low-noise driver and a low-noise commercial one. In Fig. 1 (a), we show the free-running QCL frequency noise from 10 Hz up to 100 MHz. The expected frequency noise contributions arising from the two drivers have been estimated by

Figure 1. a) QCL FNPSD when driven with a commercial and home-made current supply (light grey and orange curves, respectively). The expected current noise contributions from the drivers are also shown (grey and red curves). b) QCL laser power spectrum for both the situations described before.

Proc. of SPIE Vol. 7945 794505-2 Downloaded from SPIE Digital Library on 05 May 2011 to 193.205.148.189. Terms of Use: http://spiedl.org/terms

measuring the current noise on a sample resistance. From such measurements we conclude that ultra-low-noise QCLs current drivers must be used, as can be deduced comparing the orange and the light grey curves at frequencies greater than 100 kHz. From FNPSD, we calculated the laser linewidth (see Refs. 9, 10 for further details) as is shown in Fig. 1 (b) for both cases considered. Both power spectrum profiles can be assumed Gaussian with a very good approximation, with a significant deviation only on the tails. The calculated total FWHM linewidth (over the recorded time-scale of 10 ms) is about 6.1 MHz when using the home-made current driver, and about 9 MHz when using the commercial one. It is noteworthy to say that these values are in good agreement with the independent estimations of the QCL linewidth (over the same time-scales) obtained by the analysis of the CO2 sub-Doppler linewidths (see Ref. 4). Looking at the case in which the home-made current driver is used (orange curve), as its contribution to the QCL frequency noise is almost negligible in the whole frequency range shown, the noise trend is expected to reproduce the “pure” QCL frequency noise figure. Although residual external noise gives rise to the sharp peaks visible throughout the trace, three different noise dependences on frequency are clearly recognizable: a 1/f trend (10 Hz ÷ 100 kHz), a steeper decay due to a cut-off (100 kHz ÷ 10 MHz) and a final asymptotic flattening (10 ÷ 100 MHz) to a value Nw . The laser intrinsic linewidth δν is closely related to the white noise level Nw by the formula δν = π · Nw : the measurement shown in the figure yields an intrinsic linewidth δν = 510 ± 160 Hz. This measurement has been repeated for several current-to-threshold-current ratio (I/Ith ), that is for several output powers. As reported in Ref. 11, not only the hyperbolic behaviour of the Schawlow-Townes formula is qualitatively respected, but also the narrow linewidth predicted by the model with the effective coupling is confirmed, and results more than 3 orders of magnitude smaller than that of the same model accounting only the standard coupling of spontaneous emission to the lasing mode. This is the first experimental evidence of an effect that, among solid state sources, has been predicted only for QCLs: the prevalence of the non-radiative relaxation channel of the upper level on the radiative one. As a consequence the line-broadening effect determined by the spontaneous emission is naturally suppressed, resulting in a much narrower linewidth.

3. HIGH-PRECISION MOLECULAR SPECTROSCOPY WITH QCLS The measurements described above open exciting new perspectives in high-precision spectroscopy, frequency metrology and molecular manipulation with QCLs. Thanks to their good stability and relatively high emission power, QCLs have proven to be suitable devices for high-sensitivity and high-resolution spectroscopy.1, 3, 4, 12–15 The compactness of these sources, as for diode lasers in the NIR, makes them very attractive for development of molecular gas sensors. In addition, as described in the previous section, these lasers exhibit a narrow intrinsic linewidth, which is important for their use in ultra-sensitive detection schemes. Nonetheless, the excess frequency noise due to current drivers, temperature variations or differences between device design and fabrication, make linewidth narrowing and frequency control challenging for QCLs. In the last years, we demonstrated the versatility of QCLs for high-precision OFCS-assisted spectroscopy. In the first experiment,3 a free-running LN2 -cooled DFB QCL operating at 4.4 μm wavelength was referenced to an Optical Frequency Comb Synthesizer (OFCS) by means of a non-linear sum-frequency generation process. The absolute frequency measurement of a Doppler-broadened fundamental ro-vibrational CO2 line gave an overall uncertainty of 2 MHz, mainly limited by the slow frequency fluctuations of the free-running QCL. By implementing a sub-Doppler spectroscopic set-up with low-bandwidth frequency stabilization of the QCL to the Lamb-dip center, the overall uncertainty of the frequency measurement (75 kHz) was improved by more than two orders of magnitudes,13 limited by systematics mainly due to mechanical instabilities: these instabilities introduced amplitude fluctuations of the detected signal and, as a consequence, shifts of the locking point. We estimate that an improvement of both the mechanical stability of the laser assembly and the locking bandwidth up to hundreds of kHz would further decrease the jitter and the uncertainty of the absolute frequency measurement. The measured linewidth can be largely reduced by means of a suitable frequency stabilization of the laser (in principle down to its intrinsic linewidth). To this purpose the saturation Lamb dip of a CO2 line can be used. In Ref. 4 the first-derivative of the Lamb-dip profile, obtained by an FM technique, has been used as feed-back signal. However, the limits imposed by the modulation bandwidth of the laser current prevented to

Proc. of SPIE Vol. 7945 794505-3 Downloaded from SPIE Digital Library on 05 May 2011 to 193.205.148.189. Terms of Use: http://spiedl.org/terms

achieve a locking bandwidth suitable for a narrowing of the laser emission, only allowing a FNPSD reduction in a bandwidth of a few hundred Hz. An improvement could be obtained by using polarization spectroscopy in conjuction with a double-balanced detection:13 this technique produces a sharp dispersive-like signal (see Fig. 2a) with no need for modulation of the laser current. In addition a signal-to-noise (S/N) ratio higher than with FM technique is obtained. Because of the absence of any modulation onto the laser, this signal is

Figure 2. a) Lamb-dip feature of the (011 1 011 0) P(30) CO2 transition obtained with polarization spectroscopy and double-balanced detection. This signal will be used for locking the QCL frequency to the line center. b) Given the experimental FNPSD, the red dashed line (see Ref. 10) sets the optimal loop bandwidth for the maximum linwidth reduction. Depending on the loop gain, it is possible to suppress different portions of noise: region 1 corresponds to a narrowing down to 200 kHz; digging region 2 would lead to a few-hundred-Hz linewidth. Region 3, even if suppressed by the loop, will not produce any further narrowing.

particularly suited for a frequency stabilization of the laser to the molecular line over the bandwidth required for an efficient narrowing. According to the theoretical treatment carried out in Ref. 10, the required bandwidth can be evaluated by looking for the crossing point between the laser FNPSD and the “magic line” plotted in Fig. 2b. The frequency correspondig to this point sets the maximum bandwidth useful for a narrowing of the laser emission. Indeed, the area of the FNPSD falling below that line only contributes to the tails of the laser emission lineshape, but does not affect its linewidth (intended as the full width at half maximum). Applying this picture to the experimental FNPSD of our QCL we can conclude that a loop bandwidth of about 400 kHz and a loop gain of about 50 dB at 100 Hz would allow to “dig” the blue area of the FPNSD (labeled with 1 ) achieving a 200 kHz-linewidth. Beyond this point, improving the loop bandwidth will not produce any further narrowing. On the contrary, an improvement of the loop gain will produce a proportional narrowing, even with a smaller bandwidth (see areas labeled with 2 and 3 ). Up to now, large fluctuations of the offset of the polarization signal, caused by spatial oscillations of the laser beam, have prevented us to close the frequency-locking loop. Indeed, this instabilities are caused by the poor mechanical rigidity of the laser housing, located into a liquid-N2 cryostat and subjected to the unavoidable vibrations of the transfer line. For this reason we are moving towards room-temperature QCLs with much more compact and stable housings, which will definitely solve the problem related to mechanical instabilities. An alternative and promising approach to the narrowing of a QCL emission can be optical injection. The key concept is to use a mid-IR QCL (around 4.5 μm) as an optical amplifier of a narrow-linewidth but low-power radiation produced, for example, by a DFG apparatus.16, 17 The topic is very interesting, mainly because of the variety of advanced applications that could be enabled by such a narrow and intense source (ultra-high-precision

Proc. of SPIE Vol. 7945 794505-4 Downloaded from SPIE Digital Library on 05 May 2011 to 193.205.148.189. Terms of Use: http://spiedl.org/terms

spectroscopy, high-sensitivity trace-gas detection, cold molecules manipulation etc.). Moreover, the investigation of the optical injection mechanism could provide a further insight on the inner physics of QCLs. The success of a direct optical injection locking of QCLs is, in fact, an open question also from the theoretical point of view.

4. FUTURE APPLICATIONS AND PERSPECTIVES The QCLs measured intrinsic linewidth is comparable with the natural linewidth of molecular IR ro-vibrational transitions. Therefore, similarly to what visible/near-IR-emitting diode lasers have represented in the last 20 years for the progress of atomic physics, mid-IR QCLs can become unique tools for investigating and harnessing molecules with unprecedented precision levels.broadening Future applications of narrow-linewidth QCLs, therefore, include ultra-sensitive gas detection and cold molecules trapping. The first can be done by setting up Cavity Ring-Down (CRD) apparatuses based on stabilized (or better injection locked) QCLs. In particular, the novel Saturated-absorption Cavity Ring-down (SCaR) technique18 would largely benefit of the QCLs high-power emission combined with an improved linewidth. As we have recently demonstrated, in fact, this technique merges saturated-absorption resolution with CRD sensitivity, allowing to retrieve, at one, both the empty-cavity background signal and the gas feature. Finally, molecules are the most natural target for such sources, given their IR emission range, and the frontier field of cooling and interrogation of very low temperature molecular samples will probably make use of QCLs, given their unique properties, discussed in this work. Therefore, new physical achievements should be triggered by QCLs and the next step should be unveiling the intrinsic properties of far-IR (THz) emitting QCLs, that could prove crucial for manipulating and interrogating molecules using their rotational transitions.

5. CONCLUSIONS In conclusion, a deep knowledge of QCLs is necessary for a thorough exploitation of their peculiar features in high-resolution spectroscopy, whenever very high detection sensitivity combined with a high discrimination degree and an absolute frequency scale are required. We focused our attention on mid-IR emitting QCLs, that are also much better known than far-IR emitting ones. From the results reported here, it clearly emerges that mid-IR QCLs can become key sources even for the most demanding spectroscopic applications.

REFERENCES [1] Wysocki, G., Lewicki, R., Curl, R. F., Tittel, F. K., Diehl, L., Capasso, F., Troccoli, M., Hofler, G., Bour, D., Corzine, S., Maulini, R., Giovannini, M., and Faist, J., “Widely tunable mode-hop free external cavity quantum cascade lasers for high resolution spectroscopy and chemical sensing,” Applied Physics B: Lasers and Optics 92, 305–311 (2008). [2] Kosterev, A., Wysocki, G., Bakhirkin, Y., So, S., Lewicki, R., Fraser, M., Tittel, F., and Curl, R. F., “Application of quantum cascade lasers to trace gas analysis,” Applied Physics B: Lasers and Optics 90, 165–176 (2008). [3] Bartalini, S., Cancio, P., Giusfredi, G., Mazzotti, D., Borri, S., Galli, I., Leveque, T., Gianfrani, L., and Natale, P. D., “Frequency-comb-referenced quantum-cascade laser at 4.4 μm,” Opt. Lett. 32(8), 988–990 (2007). [4] Borri, S., Bartalini, S., Galli, I., Cancio, P., Giusfredi, G., Mazzotti, D., Castrillo, A., Gianfrani, L., and Natale, P. D., “Lamb-dip-locked quantum cascade laser for comb-referenced IR absolute frequency measurements,” Opt. Express 16, 11637–11646 (2008). [5] Faist, J., Capasso, F., Sivco, D. L., Sirtori, C., Hutchinson, A. L., and Cho, A. Y., “Quantum cascade laser,” Science 264, 553–556 (1994). [6] Schawlow, A. L. and Townes, C. H., “Infrared and optical masers,” Physical Review 112, 1940–1949 (1958). [7] Henry, C., “Theory of the linewidth of semiconductor lasers,” Quantum Electronics, IEEE Journal of 18(2), 259–264 (1982). [8] Yamanishi, M., Edamura, T., Fujita, K., Akikusa, N., and Kan, H., “Theory of the intrinsic linewidth of quantum-cascade lasers: Hidden reason for the narrow linewidth and line-broadening by thermal photons,” IEEE J. Quantum Electron. 44(1), 12–29 (2008).

Proc. of SPIE Vol. 7945 794505-5 Downloaded from SPIE Digital Library on 05 May 2011 to 193.205.148.189. Terms of Use: http://spiedl.org/terms

[9] Elliott, D. S., Roy, R., and Smith, S. J., “Extracavity laser band-shape and bandwidth modification,” Phys. Rev. A 26, 12–18 (1982). [10] Di Domenico, G., Schilt, S., and Thomann, P., “Simple approach to the relation between laser frequency noise and laser line shape,” Appl. Opt. 49(25), 4801–4807 (2010). [11] Bartalini, S., Borri, S., Cancio, P., Castrillo, A., Galli, I., Giusfredi, G., Mazzotti, D., Gianfrani, L., and De Natale, P., “Observing the intrinsic linewidth of a quantum-cascade laser: Beyond the schawlow-townes limit,” Phys. Rev. Lett. 104(8), 083904 (2010). [12] Borri, S., Bartalini, S., Natale, P. D., Inguscio, M., Gmachl, C., Capasso, F., Sivco, D. L., and Cho, A. Y., “Frequency modulation spectroscopy by means of quantum-cascade lasers,” Appl. Phys. B 85, 223–229 (2006). [13] Bartalini, S., Borri, S., and De Natale, P., “Doppler-free polarization spectroscopy with a quantum cascade laser at 4.3 μm,” Opt. Express 17, 7440–7449 (2009). [14] Gagliardi, G., Tamassia, F., Natale, P. D., Gmachl, C., Capasso, F., Sivco, D. L., Baillargeon, J. N., Hutchinson, A. L., and Cho, A. Y., “Sensitive detection of methane and nitrous oxide isotopomers using a continuous wave quantum cascade laser,” Eur. Phys. J. D 19, 327 (2002). [15] Taubman, M. S., Myers, T. L., Cannon, B. D., and Williams, R. M., “Stabilization, injection and control of quantum cascade lasers, and their application to chemical sensing in the infrared,” Spectrochim. Acta 60, 3457–3468 (2004). [16] Galli, I., Bartalini, S., Cancio, P., Giusfredi, G., Mazzotti, D., and De Natale, P., “Ultra-stable, widely tunable and absolutely linked mid-IR coherent source,” Opt. Express 17(12), 9582–9587 (2009). [17] Galli, I., Bartalini, S., Borri, S., Cancio, P., Giusfredi, G., Mazzotti, D., and P., D. N., “Ti:sapphire laser intracavity difference-frequency generation of 30-mw cw radiation around 4.5μm,” Opt. Lett. 35(21), 3616– 3618 (2010). [18] Giusfredi, G., Bartalini, S., Borri, S., Cancio, P., Galli, I., Mazzotti, D., and De Natale, P., “Saturatedabsorption cavity ring-down spectroscopy,” Phys. Rev. Lett. 104(11), 110801 (2010).

Proc. of SPIE Vol. 7945 794505-6 Downloaded from SPIE Digital Library on 05 May 2011 to 193.205.148.189. Terms of Use: http://spiedl.org/terms