Feasibility of using a mid-Infrared tunable sampled-grating distributed Bragg reflectors quantum cascade laser for high resolution multicomponent trace gas spectroscopy is demonstrated. By controlling the driving currents to the front and back sections of the laser, we were able to tune a pulsed 4.55 µm laser over a frequency range a of 30 cm−1 with high resolution, accuracy and repeatability. The laser was applied to absorption spectroscopy of ambient and reduced pressure (150 Torr) air in a 205 meters multi-pass Herriott cell, and by using standard LSQ fitting to a spectral database of these trace gases (HITRAN), the concentrations of nitrous oxide, carbon monoxide, and water vapor were retrieved.
© 2015 Optical Society of America
Detecting and quantifying the concentration of trace gases in ambient atmosphere, indoor environments and in industrial processes has been of great interest to the scientific and industrial community and its associated technologies have significantly improved in accuracy, cost, and speed over the past few years with advances in semiconductor lasers. For example, tunable single mode diode laser absorption spectroscopy provides high spectral resolution which enables species selective detection through the measurement of molecular rotational-vibrational lines.
Since the mid-IR region corresponds to the “fingerprint” region of many gases due to the strong absorption cross section of fundamental vibrational lines of the molecules, quantum cascade lasers (QCLs) are ideal for spectroscopic applications. Today, QCLs can emit high power on the order of hundreds of milliwatts at room temperature. In addition, QCLs can be designed to have a very broad spectral gain bandwidth with Full Width at Half Maximum (FWHM) as large as 300 cm−1, making them useful as a single wavelength tunable source  if a wavelength selective element is incorporated. A single mode QCL such as a distributed feedback (DFB) QCL  is an ideal source for gas sensing in the mid-IR wavelength range. Despite their strong performance and reliability, DFB QCLs are limited by their relatively small tuning range (of a few wavenumbers) which is determined by the thermal rollover of the laser. An external cavity (EC) QCL , on the other hand, is the leading technique for achieving a widely tunable laser source in the mid-IR which is needed for detecting multiple species or large molecules. However, in an EC QCL, the laser and the optical arrangement can be complex, and an external cavity may be susceptible to environmental and other operating conditions. A recent approach undertaken by some groups is to employ DFB QCL Arrays  which consist of multiple lasers on the same chip. However, combining the light beams from different facets requires specialized optics that are difficult to align. This design is also bulky since the size is proportional to the number of devices in the array.
Telecommunication applications, where the need for tunable sources is critical, have driven the development of novel semiconductor laser technologies. For example, to implement dense wavelength division multiplexing (DWDM), there is a need for wavelength tunable laser diodes [5,6]. Sampled-grating distributed Bragg reflector (SG-DBR) structures were developed and have successfully addressed this need. Several groups have taken advantage of SG-DBR performance and its commercial off-the-shelf availability in near-IR for spectroscopy of molecular trace gases despite the weak vibrational absorption line-strength in this region of the spectrum [7,8].
In this work, a broadly tunable source SG-DBR QCL [9–11] is employed because it combines the compactness of the DFB QCL (monolithic) and potentially the wide tuning range of the EC QCL. The simplicity and the potential low cost of the device motivate the development and demonstration of a SG-DBR QCL based system as a candidate in differential absorption based multispecies trace gas sensing and monitoring applications. Despite the recent fabrication of these compact monolithic widely tunable devices in the mid-IR [9–11], their application in mid-IR spectroscopy in gas sensing has not been demonstrated due to the complexity of controlling the tuning parameters to achieve high resolution and yet broad range tuning. Kalchmair et al.  performed a transmission spectroscopy measurement through a 400 μm thick sheet of Polyurethane. We extend this work by showing continuous tuning of the laser and its application to measure fine spectral features of molecular gases. Further, with the complex output pulse spectrum from these lasers, their suitability for high spectral resolution gas spectroscopy is not obvious. This is the first demonstration of the SG-DBR QCL to perform direct absorption spectroscopy to sense multispecies in air by a careful manipulation of both the front and back DBR currents.
2. Experiment setup and wavelength measurement
A SG-DBR QCL is a monolithic device that has four sections: gain, phase, front and rear sampled grating DBR, which all shares a common active core made of quantum cascade stages and can be independently biased as illustrated in Fig. 1. The SG-DBR QCL source used is designed to operate in pulse mode and to cover the spectral range of 2150-2250cm−1. The SG-DBR laser has a side mode suppression ratio comparable to DFB lasers ~30 dB as shown in reference . More information about the characterization of the SG-DBR QCLs used in this work has been published elsewhere . Here, we focus on showing the laser’s capability for high spectral resolution spectroscopy over a relatively wide frequency range.
The measurement setup depicting the optical arrangement for absorption spectroscopy is shown in Fig. 2. An ILX pulsed current source LDP-3830 was used to bias the laser gain section. A low impedance cable connects the output of the LPB-386 pulse board to the laser which is mounted on a custom designed PCB board. A 175 ns current pulse of 900 mA is used to run the laser which has a peak power of 28.6 mW and a pulse repetition rate of 20 kHz. A programmable triple channel DC power supply is used to apply DC currents to the front and back DBR sections of the laser. An Arroyo Instruments (5230) TEC controller is used to keep the laser sub-mount temperature to 20°C. A Tektronics oscilloscope DPO 4054 (2.5Gs/S) is employed to acquire and display measurement data. National Instrument LabVIEW software is used to control the scope and the triple DC power supply to automate all data acquisition.
Two Vigo Systems MCT detectors (100MHz BW) are used in the set up. The first, MCT1, is used to record the fringe pattern when a Ge etalon with FSR = 0.048 cm−1 is placed in the optical path. The shift of the fringe pattern was used for relative wavelength calibration of the laser. The second, MCT2, is used to measure the transmission of the laser pulse through a 205 meters effective path length Aerodyne multi-pass Herriot cell. By normalizing the signal measured when the laser beam traverses the cell filled with air to the signal measured when the cell is filled with Nitrogen the resulting absorbance is calculated using the Beer’s Law as shown in Eq. (1).
The laser was operated with a fixed injection current pulse of 900 mA applied to the gain section at a repetition rate of 20 kHz and with pulse duration of 175ns. In Fig. 3, laser pulse characteristics at zero heater bias currents obtained by different means are plotted. Figure 3(a) shows the pulse shape when the light detected is passed through a 1-inch Ge Fabry-Pérot etalon with FSR (Free Spectra Range) 0.048cm−1 (1.5GHz) while Fig. 3(b) displays a time average of the spectrum as obtained with a Fourier transform infrared (FTIR) spectrometer. The fringe spacing allows for calibration of the relative frequency, including the observed chirp, from the time domain pulse shape. The intra-pulse rapid frequency down-chirp shown in Fig. 3(a) is due to the heating of the laser during the current pulse. The intra-pulse variation of the laser intensity reveals the complexity of the laser pulse spectrum during chirping with two noticeable mode-hops within a single pulse. The mode-hop free tuning range between hops corresponds to 6½ x FSR or ∆vm = 0.3 cm−1 as indicated by the number of peaks in the transmission curve through the etalon. The QCL mode spacing, as expected, matches with the cavity mode spacing; where L = Lgain + Lphase = 4.5 µm is the cavity length, ng is the group index of refraction. Figure 3(b) shows the laser spectrum measured with a FTIR of 0.5 cm−1 resolution. This initial spectrum before tuning the laser is used to convert later relative tuning range obtained from Fabry-Pérot etalon to absolute frequency.
A nonlinear chirp rate is measured varying from 234 MHz.ns−1 to 170 MHz.ns−1. This reduction of chirp rate leads to a resolution limit of about 45MHz based on Eq. (2) where C is a constant equal to 0.886 for a rectangular pulse . However, to be clear, this is a limit and not the actual resolution of the system, which depends on a number of other parameters.
3. Tuning the SG-DBR laser
When a pulsed current is used to drive a laser, the temperature of the device is intrinsically affected by heat dissipation. The emitted wavelength is red shifted because an increase in temperature results in an increase of index of refraction of the gain medium. This chirp phenomenon can also be achieved by applying a current ramp or a saw-tooth pattern to either the gain or phase section of the laser, albeit at a slower rate.
A SG-DBR laser wavelength is tuned using the Vernier effect . The periodic modulations of the Bragg grating result in a periodic reflectivity spectrum. Since the periodicity of the grating of the front Lf and back Lb DBR are slightly different, this imposes a slightly different reflectivity peak spacing of Δλf for front DBR (dash) and of Δλb for back DBR (dash dot) as illustrated in Fig. 4. The small difference in the reflection peaks (combs) guarantees that only one set of reflectance peaks is aligned at any given time represented by the solid lines Fig. 4(a), (b) and (c). The periodic vertical dotted lines represent the cavity modes and the aligned cavity mode(s) with DBR resonances is(are) represented in a vertical solid line to show the lasing mode(s).
The illustration in Fig. 4 shows the evolution of the lasing mode as a pulse current is applied to the gain section while the front and back currents remained constant. The temperature variation in the gain section will lead to an increase of reflective index, which in turn change the effective optical length. Changing the effective optical length will change the resonance condition in the cavity that will shift the cavity modes. In Fig. 4(a) Single mode lasing happens at the dominant mode N. In Fig. 4(b), the cavity modes are slightly red shifted and at this point there is competition between mode N and mode N-1. As the longitudinal mode spectrum is further red shifted, as displayed in Fig. 4(c), the mode N-1 becomes dominant and the laser frequency resets. However this thermal tuning of the cavity gives a maximum tuning range of Δλm, equal to 0.3 cm−1 with the laser used here, and results to step like chirp as shown in Fig. 4(d).
In Fig. 4, the tuning is the result of thermal chirp due to operating the gain section with a current pulse, now we explore what happen when a DC current is applied to the DBR sections. By injecting current into the Bragg reflectors, again the refractive index increases due to Ohmicheating, thereby the optical period of the Bragg gratings are changed hence the resonance wavelength is changed. In Fig. 5 three set of plots are graphed denoted by Reference (blue), Fringe (green) and Transmission (red). Reference signals are obtained with detector MCT1 in Fig. 2 or by MCT2 with the gas cell filled with N2. Fringe signals are obtained when the etalon is inserted in front of MCT1 and Transmission signals are detected by MCT2 representing the signals from the gas cell with air. The resulting signals due to incremental change (few milliamps) to either the front or back DBR current settings are shown; the currents are changed by a small amount such that the new chirp spectrum has notable overlap with the previous one (see below). The dotted lines characterize signals of first current settings while solid lines show signals after changing current applied to DBR with new current settings.
The mode-hops locations can be spotted clearly either in the Reference signals by the local minima, as expected from illustrations in Fig. 4(e), or by the periodicity interruption on the Fringe pattern. Vertical dotted and solid lines are used to show the cavity mode hops for the two DBR current settings. The spectrum range between the mode-hops or vertical lines (as mentioned in Fig. 4(d) is equal to Δλm; the cavity mode spacing. The arrows labeled as A illustrate how locations of mode-hops are shifted between successive current settings. This time domain shift corresponds to a shift of the tuning range in the frequency domain to the red (lower frequencies), there is also a smaller shift that may be due to thermal crosstalk and/or nonlinear chirp rate in the QCL cavity resulting in the shift indicated by the arrows labeled as B in the time domain that is indicative of a correspondingly small shift ΔνShift in the frequency domain. A similar shift is also observed in the absorption spectrum of the N2O line (in this example at 2197.7 cm−1) and shown by the arrows labeled as C in the transmission signal. The resulting coarse tuning of the chirp spectrum of the laser to the red is a combination of these two spectral shifts. The new frequency chirp range of the laser is calibrated relative to the previous one using the spectral shift in the etalon fringes that corresponds to shift B in the time domain. By combining the fine intra-pulse frequency down chirp, shown in Fig. 4, together with coarse thermal tuning of the Bragg reflectors explained above, continuous tuning may be achieved.
Figure 6 shows the tuning characteristics of the laser as the DC bias currents, driving the front and back Bragg mirrors, are tuned. In Fig. 6(a) and 6(b), the relative frequency tuning and heating, resulting from increasing the currents to the front and back heater sections, are plotted respectively. In both figures, the power dissipated by the DBR sections is a quadratic function of the drive current as expected from the Ohmic heating effect. Also, the heating and tuning curves have nearly the same functional dependence on the current, showing the strong correlation between the relative change in laser frequency and Ohmic heating. The 3-D stem plot in Fig. 6(c) shows relative changes in laser frequency as the front and back bias currents are tuned. This is the coarse tuning of the DBR laser, which combined with the intra-pulse frequency chirp results in a fine resolution of the continuous tuning as demonstrated by the differential absorption spectra discussed below.
In order to accurately assess the spectral stability and repeatability of the device, the laser was tuned to two different N2O absorption features at 2198.7 cm−1 (If = 46 mA, Ib = 48 mA) and 2197.7cm−1 (If = 76 mA, Ib = 75 mA) every 15 minutes for 24 hours. This was carried out for air at low pressure 150 Torr (to sharpen the lines and increase the steepness of the slope at the wings). The slope of the absorption features was used as a frequency discriminator to transform frequency variation of the laser into intensity variation . All the lineshape measurements were normalized first to account for intensity fluctuations. Each time, the system was powered up, and the laser was tuned to each of the lines, and then the system is powered off again. The frequency fluctuations at the two lines, Fig. 7(a) and 7(b) at 2198.7 cm−1 and 2197.7cm−1 respectively, indicate a maximum standard deviation of 0.03 GHz or 0.001 cm−1. This relatively small value is evidence that the laser pulse shape (and line shape) is stable and very repeatable over time which is an indication that signal and reference measurements can be done at different times given that the conditions of operation are the same for each measurement.
4. Spectroscopy experiment results
As mentioned above, a laser source capable of narrowband tuning over a broad spectral range is necessary for spectroscopy applications of multiple target species (e.g. trace molecular gases). In this section, we perform a spectroscopy experiment and analyze the resultant absorption spectra. A multivariate least square analysis method was used to retrieve the concentration of different gases [15,16]. Since the measured absorption spectrum is composed of a linear superposition of the individual species’ spectra, Eq. (1) can be rewritten as follows:16,17]. To account for spectral broadening in the chirp process due to the instrument function, an additional convolution is needed for the low pressure absorption spectra. The LSQ solution vector , corresponding to the best fit to the measured data, and is given as
Besides obtaining the individual species concentrations, the LSQ formalism allows us to determine the concentration uncertainties under the assumption that the residual error is random and uncorrelated. The statistical uncertainty corresponding to one standard deviation (1σ) is calculated by taking the square root of the diagonal element of the covariance matrix as shown in Eq. (5).
The term in Eq. (5) corresponds to the root mean square of the fit residuals where m x n is the dimension of matrix.
To demonstrate multi-species detection we used air as test subject. These experiments were conducted in the laboratory during December 2014. Labview was used to automatically apply the current values, plotted in Fig. 6, from a lookup table to their corresponding section on the laser. The system was used to perform gas spectroscopy over a range of 10 cm−1 at 4.55 μm. To tune over this range, 63 settings of front and back bias currents were scanned and data taken at each setting was averaged for 512 pulses. The total scan over the 63 settings takes about 3 minutes. This is in part because we conservatively allowed 2 seconds after each current setting for the system to stabilize. We are investigating how this time may be reduced. The raw data is processed to calculate the absorbance at each setting.
Figure 8 shows the spectrum of measured data and Least square fit (shifted vertically for clarity) to the measurements using data from HITRAN database following the formalism above over the tuning range from 2189.5 cm−1 to 2199.5 cm−1 under standard temperature and pressure over a distance of 205 m. Strong N2O transitions dominate this tuning region, however different gas absorption features with minimal interferences can be spotted at 2194.4 cm−1 and 2190 cm−1 respectively for H2O and CO. There is clearly a good match between measured data and the predicted spectra from HITRAN database. At the bottom of the figure, the residual spectrum, obtained by measured spectrum with least square fit, shows the accuracy of the fitting. Moreover, the alignment of the peaks throughout the entire tuning range is an indication that our method adopted here to tune the laser wavelength is very consistent.
The least square analysis is applied to the measured data and the result for each of the species concentration is summarized in Table 1.
For a better demonstration of the system resolution, a reduced pressure measurement was performed and results are plotted in Fig. 9. In contrast to ambient pressure, high selectivity is obtained with reduced pressure due to reduction in pressure broadening of the absorption lines.
Unlike the ambient retrieval, the measured features at 150 Torr (Table 2) are broader than the trace gas spectral lineshapes at this pressure because the instrument function due to the laser linewidth and receiver bandwidth (detector and digitizer combined). As a result, in this case, it is necessary to convolve the instrument function with the absorption lines from the HITRAN database before the LSQ fitting. Assuming a Gaussian function as an approximation of the instrument function, the best fit was obtained for a Gaussian instrument function with FWHM of 0.07 cm−1; which is what is expected for our instrument function (limited by detector bandwidth). The fit is plotted along with measured data in Fig. 9. As can be seen in this figure, fine absorption features are resolved, thus showing the potential application of SG-DBR QCL for high resolution spectroscopy.
5. Extended tuning
In order to see how the laser would perform at wavelength away from the center of the gain spectrum, we carried out spectroscopy at three different windows. For each window, we first pre-biased the front and or back DBR heater currents to get to a specific wavelength before continuously tuning the laser wavelength following the procedure outlined in section 3. We repeated the experiments on ambient air in the 205 meter Herriot Cell described in the previous section. First, the initial frequency (zero bias current on both DBR sections) is shifted towards higher frequency from 2199 cm−1 to 2214 cm−1 (15 cm−1 shift) by applying a pre-bias current of 78 mA to the back DBR. The spectrum obtained after continuously tuning the laser frequency for 8 cm−1 (2214-2206 cm−1) is shown in Fig. 10(a). By pre-biasing the front DBR with 89 mA, the initial wavelength is down-shifted this time to 2183 cm−1 (corresponding to 14 cm−1 frequency shift) before continuously tuning the laser frequency by 7 cm−1 (2183-2176 cm−1). The corresponding spectrum is plotted in Fig. 10(b). Finally, a pre-bias of 109 mA to back DBR allows the initial laser frequency to be down shifted to 2160 cm−1 (a shift of 33 cm−1). Then, the laser frequency is continuously tuned for 5 cm−1 (2160-2155 cm−1) and the resulting spectrum shows water vapor and carbon monoxide absorptions as plotted in Fig. 10(c).
The continuous tuning range in each spectral window in Fig. 10 is less than 10 cm−1 because of the pre-bias current needed to shift the lasing frequency to a different window. The higher the pre-bias value the smaller the continuous tuning range since the maximum currents applied to the grating sections is the same for each of the three window. The results in Fig. 10, added to the range already shown in Fig. 8, constitutes a cumulative of 30 cm−1 coverage with a single SG-DBR QCL.
In this work, we demonstrate the potential of employing a broadly tunable SG-DBR QCL in laser gas absorption spectroscopy for multispecies detection in the mid-IR. We have demonstrated that the pulsed SG-DBR QCL is repeatable and highly stable, and we have successfully utilized this laser to achieve high resolution tuning over a broad frequency range of 30 cm−1 by controlling the front and back Bragg reflector heater currents. The results obtained from ambient and reduced pressure experiments illustrate that SG-DBR QCLs have the potential of becoming a compact high-resolution broadband tunable laser source in the mid-IR for spectroscopic and sensing applications.
References and links
1. R. Maulini, A. Mohan, M. Giovannini, J. Faist, and E. Gini, “External cavity quantum-cascade laser tunable from 8.2 to 10.4μm using a gain element with a heterogeneous cascade,” Appl. Phys. Lett. 88(20), 201113 (2006). [CrossRef]
2. F. Xie, C. G. Caneau, H. P. LeBlanc, N. J. Visovsky, S. Coleman, L. C. Hughes, and C. Zah, “High-temperature continuous-wave operation of low power consumption single-mode distributed-feedback quantum-cascade lasers at λ – 5.2 μm,” Appl. Phys. Lett. 95, 091110 (2009).
3. A. Hugi, R. Terazzi, Y. Bonetti, A. Wittmann, M. Fischer, M. Beck, J. Faist, and E. Gini, “External cavity quantum cascade laser tunable from 7.6 to 11.4 µm,” Appl. Phys. Lett. 95(6), 061103 (2009). [CrossRef]
4. B. G. Lee, M. A. Belkin, C. Pflugl, L. Diehl, H. A. Zhang, R. M. Audet, J. MacArthur, D. P. Bour, S. W. Corzine, G. E. Hufler, and F. Capasso, “DFB Quantum Cascade Laser Arrays,” IEEE J. Quantum Electron. 45(5), 554–565 (2009). [CrossRef]
5. H. Kobrinski, M. P. Vecchi, M. S. Goodman, E. L. Goldstein, T. E. Chapuran, J. M. Cooper, M. Tur, C. Zah, and S. G. Menocal Jr., “Fast wavelength-switching of laser transmitters and amplifiers,” IEEE J. Sel. Areas Comm. 8(6), 1190–1202 (1990). [CrossRef]
6. B. Glance, U. Koren, C. A. Burrus, and J. D. Evankow, “Discretely- tuned x-frequency laser for packet switching applications based on WDM,” Electron. Lett. 27(15), 1381–1383 (1991). [CrossRef]
7. K. Boylan, V. Weldon, D. McDonald, J. O’Gorman, and J. Hegarty, “Sampled grating DBR laser as a spectroscopic source in multigas detection at 1.52-1.57 μm,” IEEE Proc. Optoelect. 148(1), 19–24 (2001).
8. D. Weidmann, A. A. Kosterev, F. K. Tittel, N. Ryan, and D. McDonald, “Application of a widely electrically tunable diode laser to chemical gas sensing with quartz-enhanced photoacoustic spectroscopy,” Opt. Lett. 29(16), 1837–1839 (2004). [CrossRef] [PubMed]
9. S. Slivken, N. Bandyopadhyay, S. Tsao, S. Nida, Y. Bai, Q. Y. Lu, and M. Razeghi, “Sampled grating, distributed feedback quantum cascade lasers with broad tunability and continuous operation at room temperature,” Appl. Phys. Lett. 100(26), 261112 (2012). [CrossRef]
10. T. S. Mansuripur, S. Menzel, R. Blanchard, L. Diehl, C. Pflügl, Y. Huang, J. H. Ryou, R. D. Dupuis, M. Loncar, and F. Capasso, “Widely tunable mid-infrared quantum cascade lasers using sampled grating reflectors,” Opt. Express 20(21), 23339–23348 (2012). [CrossRef] [PubMed]
11. A. S. Diba, F. Xie, C. Caneau, H. LeBlanc, S. Coleman, and C. Zah, “Wavelength Tuning of Sampled-Grating DBR Quantum Cascade Lasers,” in Conference on Lasers and Electro-Optics 2012, OSA Technical Digest (online) (2012), paper CF3K.3. [CrossRef]
12. S. Kalchmair, R. Blanchard, T. S. Mansuripur, G. M. de Naurois, C. Pfluegl, M. F. Witinski, L. Diehl, F. Capasso, and M. Loncar, “High tuning stability of sampled grating quantum cascade lasers,” Opt. Express 23(12), 15734–15747 (2015). [CrossRef] [PubMed]
13. M. T. McCulloch, E. L. Normand, N. Langford, G. Duxbury, and D. A. Newnham, “Highly sensitive detection of trace gases using the time-resolved frequency downchirp from pulsed quantum-cascade lasers,” J. Opt. Soc. Am. B 20(8), 1761–1768 (2003). [CrossRef]
14. V. Jayaraman, Z.-M. Chuang, and L. A. Coldren, “Theory, design, and performance of extended tuning range semiconductor laser with sampled grating,” IEEE J. Quantum Electron. 29(6), 1824–1834 (1993). [CrossRef]
15. T. L. Myers, R. M. Williams, M. S. Taubman, C. Gmachl, F. Capasso, D. L. Sivco, J. N. Baillargeon, and A. Y. Cho, “Free-running frequency stability of mid-infrared quantum cascade lasers,” Opt. Lett. 27(3), 170–172 (2002). [CrossRef] [PubMed]
16. D. M. Halland, R. G. Easterling, and D. A. Vopicka, “Multivariate least-squares methods applied to the quantitative spectral analysis of multicomponent sample,” Appl. Spectrosc. 39(1), 73-84 (1985).
17. U. Platt and J. Stutz, Differential Optical Absorption Spectroscopy, (Springer, 2008).