A quantum cascade distributed feedback laser operating at 5.2 µm is used to obtain sub-Doppler resolution limited saturation features in a Lamb-dip experiment on the R(13.5)1/2 and R(13.5)3/2 transitions of NO. The dips appear as transmission spikes with full widths of ~4.3 MHz. At this resolution the 73 MHz Λ-doubling of the R(13.5)3/2 line, which is normally obscured by the 130 MHz Doppler broadening, is easily resolved.
©2000 Optical Society of America
Recent measurements with quantum cascade distributed feedback (QC-DFB) lasers have demonstrated the utility of these devices for high-resolution, Doppler-limited spectroscopy in the mid-IR spectral region.1,2 These devices are designed by band-structure engineering and are grown by molecular-beam epitaxy.3 They have two distinct advantages over the conventional direct-gap-emitting Pb-salt tunable diode lasers (TDL) that have been used for high-resolution mid-IR spectroscopy during the last 25 years. The first is that the QC-DFB laser is inherently a single-mode device that can be tuned continuously over long spectral regions with no breaks or mode hops. With a fixed heat sink temperature, continuous CW current scans exceeding 1–2 cm-1 are common with these lasers, and with careful temperature control frequency scans exceeding 10 cm-1 are possible.3 In contrast, commercially available Pb-salt lasers, which employ an internal Fabry-Perot cavity for optical feedback, are multimode devices. A spectrometer is needed to select a single mode for spectroscopic measurements, and continuous frequency scans of any one mode exceeding 1 cm-1 are rare. The second advantage of QC lasers is that they emit much higher powers than Pb-salt lasers, typically by factors of 102 or more. The higher output power not only makes setting up experiments and aligning the laser beam easier, it also opens up the possibility of nonlinear saturation or hole-burning experiments with sub-Doppler spectral resolution. Both TDL’s and QC-DFB lasers have demonstrated linewidths that are more than adequate for sub-Doppler applications.4,5
In this paper we demonstrate for the first time that a free-running QC-DFB laser operated in a rapid scan mode has sufficient beam quality, power, and linewidth to produce sub-Doppler saturation features in a simple Lamb-dip geometry. The purpose of such experiments is to achieve very high spectral resolution (~102 below the Doppler limit), limited only by the linewidth of the laser, the homogeneous linewidth of the transitions under study, and the geometrical constraints of the experiment. A number of sub-Doppler resolution experiments have been reported using TDL’s. Most of these employed a high-power gas laser to saturate a transition, while the TDL was used as a probe,6–8 or they used a molecular-beam geometry to defeat Doppler broadening.9,10 However, at least two sub-Doppler experiments have been reported with TDL’s. Jennings reported a 40–50 MHz wide Lamb dip in NH3,11 and more recently Mürtz et al. reported 3.5 MHz wide Lamb dips in OCS in an elaborate experiment in which the TDL was optically line-narrowed and frequency-offset locked to a CO gas laser.12 In our experiments the same QC laser is used both as pump and probe in a simple Lamb-dip configuration.13 The laser beam (pump) is brought to a soft focus in a gas cell containing a few mTorr of NO. The transmitted beam (probe) is attenuated and refocused back through the same focal volume and then onto a detector. As the laser is tuned through a transition, the detected signal shows the usual Doppler-broadened profile, 130 MHz full width at half maximum (FWHM), in the center of which the Lamb dip appears as a narrow transmission spike. The spike appears because the pump laser is sufficiently strong to perturb the lower and upper-level populations in a narrow portion of the inhomogeneously broadened Doppler profile, and the molecules in this portion interact with the counter-propagating probe beam only at the line center where the Doppler shift is zero.
2. Experimental description
Figure 1 shows the experimental arrangement. The laser emission is collected and collimated with a ZnSe lens L1 (38 mm focal length). It passes through a ZnSe beam splitter (R=0.29, T=0.71) and then is focused into the gas cell with a 150 mm focal length CaF2 lens, L2. The gas cell has a 20 cm path length and CaF2 Brewster-angled windows. After the cell the beam is refocused back through the same volume with a spherical mirror (40 cm radius of curvature), recollimated by L2, reflected off the beam splitter, and finally measured with a LN2-cooled HgCdTe detector. A wire-grid polarizer (WGP) is inserted between the cell and the mirror to provide variable attenuation for the return beam. The laser polarization is perpendicular to the optical table and the beam splitter is anti-reflection (AR) coated on one side for s-polarized light incident at 45°. The lenses are AR coated to minimize spurious “window” fringes in the laser scans, and the WGP is tilted off normal for the same reason. When properly aligned, half of the probe beam was focused back into the QC laser causing large intensity fluctuations. This feedback was reduced (by about ×100) using the WGP.
The fabrication and characterization of QC-DFB lasers similar to the one used here were described in detail by Gmachl et al.3 A DC bias current of 600–800 mA is provided either by an ILX Lightwave LDC-3744B power supply or a Laser Analytics Model L5830-C power supply (modified for the higher impedance of the QC laser). In the case of the ILX supply, a repetitive ramp is applied at 30–60 Hz on top of the DC bias through an analog modulation input, using an HP 8116A signal generator. The Laser Analytics supply provides an internally generated ramp on top of the DC bias current. The tuning rate is roughly 170 MHz/mA (0.0057 cm-1/mA). The detector signal is amplified by an EG&G model PA-101 preamp and displayed in real-time on a Tektronix TDS784D digitizing oscilloscope, sampling the data at 25–100 kHz.
In order to produce observable Lamb dips in a saturation experiment, the Rabi frequency νR≡µE/h (where µ is the transition dipole, E is the electric field strength in the focal volume, and h is Plank’s constant) must be comparable to or exceed the relaxation rate of the states being excited. For vibrational excitation of a diatomic molecule the appropriate relaxation rate is given by the pressure-broadening coefficient, which for our NO transitions is ~2.5 kHz/mTorr.14 The laser power is 30 mW, which is focused to a spot size of 8×10-3 cm2, yielding an optical electric field of ~54 V/cm. The transition moment, which includes the vibrational15 and rotational16 matrix elements is 1.6×10-3 Debye. These values yield an estimated νR of 45 kHz, which should be adequate to produce Lamb dips with the sample pressure below 20 mTorr.
The NO lines were identified in a separate experiment using a slight modification of the setup in Fig. 1. The mirror was replaced by a second detector-lens combination that measured the single-pass transmission of the cell. The beam splitter was rotated 90° and a solid Ge etalon (free spectral range 978.7 MHz) was inserted in the path before the first detector-lens pair. The signals from the two detectors were then recorded simultaneously as the laser was current-tuned over a pair of lines. Figure 2 shows a typical scan covering the lines studied here. From these data the relative positions of the lines are measured to an accuracy ±0.0003 cm-1, which is sufficient to assign them based on their known frequencies.17 The line on the left has Ω=½, where Ω is the projection of the electronic angular momentum onto the internuclear axis, and its Λ-doubling splitting exceeds the Doppler width, so that it appears as a doublet. The line on the right (Ω=3/2) appears as a singlet, since its Λ-doubling splitting is much less than the Doppler width.
Figure 3 shows spectra from the Ω=½ lines in Fig. 2. The Lamb-dip signals are robust enough to be seen in real time on a single sweep of the laser, but the data here are averaged several hundred times to improve the signal-to-noise ratio. The peak absorption on the broad lines is ~5%, and the Lamb dips are about 1–2% of that, which implies a transmission change of order 10-3.
The total scan length in Fig. 3 is ~1 GHz, and the tuning rate is not constant, increasing slightly throughout the scan. In order to obtain an accurate frequency scale, we assumed a quadratic frequency dependence vs. time (or current) and fit the parameters with two constraints: the Lamb dips must be separated by the known Λ-doubling (276.8 MHz) and the Doppler-broadened lines must have the same widths. When these constraints were satisfied, the broadened lines not only had the same widths, but the widths were precisely the expected 130 MHz Doppler width. The frequency scale is determined in this manner, and the derivative spectrum in the lower panel is obtained from the difference between adjacent channels in the upper panel, divided by the fitted frequency difference between the channels.
Figure 4 shows Lamb dips in the derivative spectrum of the Ω=3/2 line on the far right of Fig. 2. This line is also a Λ-doubled pair, but the 73 MHz splitting between the e and f components is completely obscured in Fig. 2 by the 130 MHz Doppler broadening. The Lamb-dip technique, however, clearly resolves the two components.
Figure 5 shows the f Lamb dip in Fig. 3, after subtracting the Doppler profile, as obtained with the Laser Analytics supply, which consistently produced dips ~40% narrower than those obtained with the ILX supply (used for the data in Figs. 2–4). The Lamb-dip line shape fits well to a Gaussian, not a Lorentzian, suggesting that the broadening mechanism is statistical in nature. The 4.3 MHz width measured here is not due to pressure broadening, since it is pressure independent up to at least 20 mTorr and the expected pressure-broadened widths are much smaller.13 The width is also not due to optical feedback, which increased the noise in the spectra, not the Lamb-dip width. It is also not due to the intrinsic laser linewidth, which is ~10’s of kHz.5 Finally, the width is not due to the effects of beam divergence and transit time, which we estimate using Shimoda’s formula to contribute only ~0.6 MHz to the linewidth.18 In a separate experiment we estimated the laser linewidth, when excited only with the dc supply (no sweeping), by measuring the noise increase over a few ms period when the laser was alternately set on the side of and just off the line. From the slope of the Doppler edge, the noise increase indicates a frequency fluctuation of 2–3 MHz. We conclude that the additional broadening likely comes from fluctuations in the drive electronics.
In summary we demonstrate for the first time saturation Lamb-dip spectroscopy with a QC-DFB laser. Lamb dips are observed on four transitions in NO showing spectral resolution a factor of 30 below the Doppler limit.
We are indebted to B.D. Poindexter for considerable help in setting up the experiments. The work performed at Bell Laboratories was supported in part by DARPA/US Army Research Office under Contract No. DAAG55-98-C-0050.
References and Links
1. S.W. Sharpe, J.F. Kelly, J.S. Hartman, C. Gmachl, F. Capasso, D.L. Sivco, J.N. Baillargeon, and A.Y. Cho, “High-resolution (Doppler-limited) spectroscopy using QC-DFB lasers,” Opt. Lett. 23, 1396 (1998). [CrossRef]
2. A.A. Kosterev, R.F. Curl, F.K. Tittel, C. Gmachl, F. Capasso, D.L. Sivco, J.N. Baillargeon, A.L. Hutchinson, and A.Y. Cho, “Methane concentration and isotopic composition measurements with a mid-infrared quantum-cascade laser,” Opt. Lett. 24, 1762 (1999). [CrossRef]
3. F. Capasso, C. Gmachl, D.L. Sivco, and A.Y. Cho“Quantum cascade lasers,” Phys. World12, 27 (1999);C. Gmachl, F. Capasso, A. Tredicucci, D.L. Sivco, J.N. Baillargeon, A.L. Hutchinson, and A.Y. Cho, “High-power, cw, current-tunable, single-mode QC-DFB lasers at 5.2 and 7.95 µm,” Opt. Lett.25, 230 (2000).
4. C. Freed, J.W. Bielinski, and W. Lo, “Fundamental linewidth in solitary, ultranarrow output PbS1-xSex diode lasers,” Appl. Phys. Lett. 43, 629 (1983). [CrossRef]
5. R.M. Williams, J.F. Kelly, J.S. Hartman, S.W. Sharpe, M.S. Taubman, J.L. Hall, F. Capasso, C. Gmachl, D.L. Sivco, J.N. Baillargeon, and A.Y. Cho, “Kilohertz linewidth from frequency-stabilized mid-IR QC lasers,” Opt. Lett. 24, 1844 (1999). [CrossRef]
7. R. Guccione-Gush, H.P. Gush, R. Schieder, K. Yamada, and G. Winnewisser, “Doppler-free two-photon absorption of ammonia using a carbon dioxide and diode laser,” Phys. Rev. A 23, 2740 (1981). [CrossRef]
8. W.H. Weber and R.W. Terhune, “Sub-Doppler infrared-infrared double resonance spectroscopy of NH3 using Stark tuning and a diode laser,” J. Chem. Phys. 78, 6437 (1983). [CrossRef]
9. T.E. Gough, R.E. Miller, and G. Scoles, “Sub-Doppler resolution IR spectroscopy of supersonic molecular beams of NO,” J. Mol. Spectrosc. 72, 124 (1978). [CrossRef]
10. A.S. Pine and K.W. Nill, “Molecular-beam TDL sub-Doppler spectroscopy of Λ-doubling in NO,” J. Mol. Spectrosc. 74, 43 (1979). [CrossRef]
11. D.E. Jennings, “Saturated absorption in NH3 using a TDL,” Appl. Phys. Lett. 33, 493 (1978). [CrossRef]
12. M Mürtz, M. Schaefer, T. George, J.S. Wells, and W. Urban, “Optically stabilized tunable diode-laser system for saturation spectroscopy,” Appl. Phys. B 60, 31 (1995). [CrossRef]
13. A.E. Siegman, Lasers, (University Science Books, Mill Valley, 1986) Chap. 30.
14. J. Ballard, W.B. Johnston, and B.J. Kerridge, J.J. Remedios, “Experimental spectral line parameters in the 1–0 band of NO,” J. Mol. Spectrosc. 127, 70 (1988). [CrossRef]
15. A.G. Maki and J.S. Wells, Wavenumber Calibration Tables From Heterodyne Frequency Measurements, NIST Special Pub.821, 1991.
16. C.H. Townes and A.L. Schawlow, Microwave Spectroscopy (McGraw-Hill, New York, 1955) p. 96.
17. L.S. Rothman, R.R. Gamache, R.H. Tipping, C.P. Rinsland, M.A.H. Smith, D.C. Benner, V.M. Devi, J.-M. Flaud, C. Camy-Peyret, A. Perrin, A. Goldman, S.T. Massie, L.R. Brown, and R.A. Toth, “The HITRAN molecular database: 1991 and 1992,” J. Quant. Spectrosc. Radiat. Transfer 48, 469 (1992). [CrossRef]
18. K. Shimoda, High-Resolution Laser Spectroscopy, ed. K. Shimoda, (Springer, New York, 1976) pp.11–49. [CrossRef]