Open Access
Add to BibSonomy
Abstract
Midwave infrared interband-cascade light-emitting devices (ICLEDs) have the potential to improve the selectivity, stability, and sensitivity of low-cost gas sensors. We demonstrate a broadband direct absorption CH4 sensor with an ICLED coupled to a plastic hollow-core fiber (1 m length, 1500 µm inner diameter). The sensor achieves a 1σ noise equivalent absorption of approximately 0.2 ppmv CH4 at 1 Hz, while operating at a low drive power of 0.5 mW. A low-cost sub-ppmv CH4 sensor would make monitoring emissions more affordable and more accessible for many relevant industries, such as the petroleum, agriculture, and waste industries.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Low-cost optical gas sensors (< $1000 USD) commonly use a thermal blackbody infrared (IR) emitter as the light source [1,2]. Midwave IR (mid-IR) light-emitting-diodes (LEDs) are under active consideration as alternatives, because they consume little electrical power and can be modulated at high frequencies—where there is less 1/f noise—without a mechanical chopper. Multiple research groups and companies have reported gas sensing with mid-IR LEDs [3–7], and the technology has incrementally improved [8]. Nonetheless, clear superiority over commercial sensors incorporating thermal IR emitters had not been demonstrated [2,9], partly because available mid-IR LEDs could not generate competitive optical powers, even over comparable spectral bandwidths in the mid-IR [8].
However, this shortcoming has been overcome recently by advances in the development of interband cascade light emitting devices (ICLEDs) [10–14]. An ICLED utilizes the active core of an interband cascade laser (ICL) [14], but with no upper and lower optical cladding layers. Since there is also no cavity to provide feedback, an ICLED emits incoherent broadband radiation. At room temperature (RT), ICLEDs with 400 µm mesa diameters recently produced 2.9 mW of continuous wave (cw) output power at a peak wavelength of 3.1 µm [13], 1.4 mW at 4.2 µm [15], and 0.5 mW at 4.7 µm [16]. There is a lack of efficient mid-IR LEDs with output powers > 1 mW [8]. Furthermore, no alloy-based LED has been reported to operate cw at RT at any λ ≥ 4.5 µm. ICLEDs can match the mid-IR output of thermal IR emitters—such as the one used in [17]—while also having wall-plug efficiencies at least as high as the best alloy-based mid-IR LEDs [14].
In addition to high output power, ICLEDs have other advantages. Stages emitting at different wavelengths can be combined for multi-band or broadband output [18]. ICLEDs can be electrically modulated at high frequencies [19]. ICLEDs should also operate cw at RT even in the longwave-IR (LWIR), although output power would be lower. Because of these advantages, we expect ICLEDs to provide optimal mid-IR and LWIR light sources for low-cost gas sensing.
ICLEDs have been applied previously to CH4 sensing [19,20], with the best 1σ detection limit being 3.6 parts-per-million by volume (ppmv) for an averaging time of 1 second. This was achieved using photoacoustic spectroscopy by Zheng et al., who additionally designed a compact and portable CH4 detection unit [19]. The most sensitive 1σ detection limit at 1 Hz for low-cost sensors employing thermal IR emitters has been ∼1 ppmv [9], while other groups have reported detection limits of 2-3 ppmv CH4 [2,21,22]. However, these sensitivities are marginal for numerous atmospheric sampling applications, where typical enhancements above the background CH4 mole fraction may be < 1 ppmv.
Here we report an ICLED-based CH4 sensor with a 1σ noise equivalent absorption of 0.17 ppmv CH4 at 1 Hz. A plastic, hollow-core fiber is used to increase optical path length for enhanced sensitivity. A multi-pass cell would also have been a viable option for our sensor. However, unlike multi-pass cells, hollow-core waveguides have the advantages of easy light handling, potential for integration with light sources, small sample volume, and possibility of miniaturization. Many groups have successfully employed hollow-core fibers as gas cells to increase path length [1,3,23]. In this study, we demonstrate the first implementation of a hollow-core-fiber-coupled ICLED. We also investigate the interference from ambient water vapor, which has not been presented in detail for sub-ppmv CH4 measurements made by low-cost sensors with broadband light sources. If CH4 is to be detected at sub-ppmv concentrations with a broadband light source centered around 3.27 µm, accurate correction must be made for the significant absorption by water vapor.
2. Experimental setup
2.1 Overview and optomechanical setup
Figure 1 shows a diagram (a) of the optomechanical setup for the ICLED CH4 sensor and a photograph (b) of the sensor inside a gas calibration chamber. Optical cage systems and lens tubes are used for system construction and alignment. A 2-lens system couples light from the ICLED into the hollow-core fiber. A plano-convex ZnSe lens (f = 15 mm, d = 25.4 mm, Rocky Mountain Instrument Co.) is used to collimate light from the ICLED, which is approximately Lambertian. A plano-convex CaF2 lens (f = 100 mm, d = 25.4 mm, Thorlabs, LA5817) focuses the collimated light and couples it into the 1.5-mm-inner-diameter hollow-core fiber (OptoKnowledge Systems, Inc.). The CaF2 lens and one end of the hollow-core fiber are connected together with a lens tube and form a gas cell. The other end of the hollow-core fiber is secured in front of another ZnSe lens, identical to the first lens. The fiber output is focused onto a HgCdTe (MCT) photodetector with a 3-stage thermoelectric cooler (TEC) (PVI-3TE-3.4, MIP, Vigo System S.A.). The total optical path length is 1.21 ± 0.02 m. A second MCT detector serves as a reference to quantify output power fluctuations from the ICLED itself. A 45:55 pellicle beamsplitter (CM1-BP145B4, Thorlabs), placed after the first ZnSe lens, splits light between the reference detector and the hollow-core fiber. The fiber was coiled into two circular loops with radii of 6.5 cm. This length was constrained by the size of the gas calibration chamber, as shown in Fig. 1(b). Bending the fiber causes extra transmission losses proportional to the inverse of the bending radius, up to a critical radius [23]. The critical bending radius of the fiber, according to the manufacturer, is 5 cm [24].
Fig. 1. (a) Diagram of the experimental setup (not drawn to scale). (b) Photograph of the setup in a gas calibration chamber with the lid removed. The labels 1-8 denote the following: (1) purge gas line for the chamber, (2) ICLED connected to a lens on an X-Y mount, (3) beamsplitter, (4) reference MCT detector, (5) lens mounted in an SM1-threaded lens tube, (6) inlet for the hollow core fiber gas cell, (7) hollow core fiber secured at the ends by SMA connectors and kinematic mounts, (8) MCT detector for measuring absorption signals.
2.2 Gas-flow system
For controlled concentration measurements, the sensor was operated inside a custom, 72 cm × 20 cm × 14 cm, o-ring-sealed aluminum chamber, pictured in Fig. 1(b). Thermal mass flow controllers delivered mixtures of N2 and CH4 through the hollow-core fiber gas cell at a head pressure of 1.03 atm and constant total flow rate of 600 standard cubic centimeters per minute (sccm) with 1% accuracy. The chamber was purged with 6 standard liters per minute (SLM) of dry N2 to remove exhaust gas from the cell out of the path of the reference detector. CH4 concentrations were varied from 0 to 20 ppmv CH4 by using dry N2 to dilute a gas standard of 20 ppmv CH4 balanced with N2 (Airgas, Inc.). The concentration of the gas standard was determined to be 19.971 ± 0.001 ppmv CH4 with a cavity-enhanced trace gas analyzer (LI-7810, LI-COR, Inc.). The LI-7810 was also used as a reference instrument for comparison and was placed in series before the chamber containing the ICLED CH4 sensor. The LI-7810 was calibrated with a reference gas cylinder containing 1872 ± 3 parts-per-billion by volume (ppbv) CH4 from the National Oceanic and Atmospheric Administration (NOAA) Global Monitoring Division. H2O concentration within the gas calibration chamber was verified with a dew point mirror hygrometer (RH Systems, 373LX), connected to the exhaust port of the chamber.
2.3 ICLED structure and characteristics
The ICLED wafer was grown by molecular beam epitaxy on a Riber Compact 21T system, using procedures similar to those reported previously [25]. The structure consists of a GaSb (100) substrate with doping density n < 5 × 1016 cm-3 (to maximize light emission from the substrate side), an n-GaSb buffer layer, an 18-stage interband cascade active core, and a 20-nm-thick n+-InAs top contact layer. The active core is partitioned into three groups of stages with 6 stages in each group. As described in [13], the groups are separated by n-InAs / AlSb superlattices, with net thicknesses adjusted such that each group is centered on an antinode of the optical field for the peak emission wavelength at normal incidence when the device is mounted epitaxial-side-down on a reflecting metal contact (i.e., a metal mirror). Above the active stages, an n+-InAs0.91Sb0.09 spacer layer with a thickness of 497 nm places the topmost group a distance 3λ/4 away from the metal mirror.
The ICLED has a circular mesa with diameter 200 µm. Apart from a negligible wavelength-dependent optical interference effect that occurs due to the differing refractive indices of various layers in the epitaxial structure, the incoherent emission from the ICLED has a Lambertian angular distribution. Figure 2 shows the power-current (L-I) curve (a) and emission spectrum (b) of the ICLED at 25 ˚C. The output power reaches 480 µW at a current of 65 mA. Figure 2(b) also shows absorptance (1 – I / I0) for 1% H2O by volume and 1 ppmv CH4 for a path length of 1.21 m—the path length of the sensor. Molecular absorption data at 0.01 nm resolution was downloaded from spectraplot.com, which uses HITRAN 2012 data [26].
Fig. 2. (a) Continuous-wave output power vs. input current at an operation temperature of 25 °C (b) Normalized emission spectra for the ICLED in room air at an ICLED temperature of 25 °C. Nearly 100% of light in the optical path of the ICLED sensor should be absorbed for wavelengths in the 2.7 µm band of H2O, according to calculations using HITRAN data. From 3 to 3.5 µm, up to 15% of light will be absorbed by some H2O lines. Figure 2(b) has a maximum y-axis value of 3% so that absorption from 1 ppmv CH4 would be visible on the plot.
In addition to H2O absorption lines, the operating wavelength region of the ICLED contains lines (not shown) for other hydrocarbons, such as ethane (C2H6) and propane (C3H8). In most atmospheric sampling applications, interference from other hydrocarbons is negligible because CH4 would be orders of magnitude higher in concentration. However, in some applications, gas samples can contain very high amounts of non-methane hydrocarbons, so issues with cross-sensitivity would be of concern. The significance of other hydrocarbons needs to be evaluated on a case-by-case basis depending on the measurement environment.
2.4 Operation of ICLED & data acquisition
The ICLED and its attached thermoelectric cooler (TEC) were driven by an Arroyo 6310 ComboSource Laser & TEC controller and a 10 kΩ thermistor. The ICLED operated at 25 °C for all experiments, and the temperature of the thermistor was stable to within 0.002 ˚C. An NI-6361 data acquisition card (National Instruments Corp.) sent square-wave modulation signals to the Arroyo laser controller to drive the ICLED from 20 mA to 120 mA at a frequency of 10 kHz and duty cycle of 50%. The average voltage of the ICLED was 7.21 ± 0.01 V, resulting in a drive power of only 0.5 W. The same NI-6361 simultaneously acquired analog voltage signals from both photodetectors and digitized them at a sample rate of 100 kHz averaged to 1 Hz. A digital lock-in amplifier, implemented in LabVIEW, extracted signals at 10 kHz.
3. Results and discussion
3.1 CH4 sensitivity
Figure 3 shows the ICLED sensor results for varying concentrations of CH4. After purging the system with dry N2, mixtures of CH4 and N2 with concentrations ranging from 0 to 20 ppmv CH4 were delivered to the fiber. The concentration was changed incrementally by 5 ppmv CH4 once every 3 minutes. Figure 3(a) shows a time series of the raw analog voltage signal acquired from the preamplifier of the photodetector at the fiber outlet (i.e., the active channel) normalized by the reference channel. The normalized signal is merely the active voltage signal divided by the reference voltage signal; this normalization method is similar to methods used in [22,27]. The normalized signal is scaled by the median of the active signal so that it is not dimensionless and has physical units (millivolts). The root-mean-square (RMS) noise of the normalized signal was approximately 1/3 lower than the RMS noise of the active signal.
Fig. 3. (a) Time series of readings from the ICLED sensor. For each 3-minute interval, only 1 min and 50 sec of data are used, due to the time required for equilibration of the concentration in the hollow-core fiber. (b) Absorbance vs. concentration, as obtained from data in the shaded regions of Fig. 3(a). 95% confidence intervals (CI) for the slope and y-intercept are (6.49e-05, 6.54e-05) and (1.8e-05, 2.4e-05) respectively. Error bars show one median absolute deviation (MAD). Horizontal error bars for the diluted concentrations of CH4 are negligible; this was verified with the LI-7810 trace gas analyzer. (c) Time series of Fig. 3(a) converted to ppmv CH4, where the blue points are data from the ICLED sensor. Orange denotes the gas mixture concentrations delivered to the fiber, which are derived from the flow rates of the mass flow controllers. (d) Correlation plot of the ICLED sensor and reference instrument. 95% CI for the slope and y-intercept are (1.004, 1.015) and (-0.10, 0.01) respectively. A time series for the reference instrument data is not shown, since it is nearly identical to the orange data in Fig. 3(c).
The data in Fig. 3(a) are used to calculate absorbance based on the Beer-Lambert law, absorbance = - log(I / I0), where I is simply the normalized signal and I0 is chosen as the median of the first concentration measurement step in Fig. 3(a)—when the ICLED sensor is measuring dry N2. Dividing the data in Fig. 3(a) by I0, the median of the first measurement step, and taking the negative logarithm of the result yields the data shown in Fig. 3(b). The gas-exchange time between the hollow-core fiber gas cell and the gas-flow system was approximately 70 seconds. These 70-second intervals are excluded from absorbance calculations and are demarcated by the unshaded regions in Fig. 3(a). Converting voltage to CH4 concentration using the linear fit in Fig. 3(b) yields the time-dependent CH4 concentrations shown in Fig. 3(c). A correlation plot of the ICLED sensor and the LI-7810 reference instrument is shown in Fig. 3(d).
3.2 Allan deviation analysis
Figures 4(a) and 4(b) show drift-corrected time series data and 10-minute Allan plots, respectively, for 10 ppmv CH4 maintained for 3 hours, while Figs. 4(c) and 4(d) show corresponding plots for dry N2. At a constant concentration of 10 ppmv CH4, the Allan deviation at 1 Hz was 0.173 ± 0.002 ppmv CH4, derived from a noise equivalent absorbance (NEA) of 1.13e-5 ± 0.02e-5 and the sensitivity of 6.52e-5 from Fig. 3(b). About 4 minutes of signal averaging yielded the minimum deviation of 0.11 ± 0.04 ppmv CH4 (0.7e-5 ± 0.3e-5 NEA). At 0 ppmv CH4, the 1 Hz deviation was the same as at 10 ppmv CH4. With 4 minutes of signal averaging, the Allan deviation under dry N2 conditions was 0.13 ± 0.07 ppmv (0.8e-5 ± 0.5e-5 NEA). Figure 4 shows that the measurement noise was not white-noise-limited by comparing our data to a τ-1/2 relationship, where τ denotes averaging time in seconds. After approximately 10 minutes, longer averaging times did not result in lower Allan deviations, and long-term drift started to dominate over short-term noise.
Fig. 4. (a) Time series of data for a constant flow of 10 ppmv CH4 for 3 hours. (b) Allan plot of the data from Fig. 4(a); the blue shaded region indicates standard error of the mean (SEM). (c) Corresponding time series of data for a constant flow of dry N2 for 3 hours. (d) Allan plot of the data from Fig. 4(c). 100 minutes of data are used for each 10-minute Allan plot. Data beyond 100 minutes are not shown because averaging times greater than 10 minutes yield Allan deviations greater than the Allan deviation at an averaging time of 1 second, for Fig. 4(b). The red lines in Figs. 4(a) and 4(c) show linear fits to the data. The red dotted lines denote a τ-1/2 relationship, which the data should follow if the measurement noise is white noise.
The amplitude of the signal drifted about 0.001% per hour, which was approximately equal to 0.14 ppmv CH4 hr-1. To correct for this drift, we first tried normalizing the active channel by the reference channel. However, normalization only removed some short-term noise (about 1/3) and did not correct long-term drift. This indicates that the main source of drift in the current system is not from the ICLED. Further observation showed that the transmission of light through the fiber was sensitive to external perturbations (e.g., physical contact and vibrations), which could be induced by ambient temperature changes and fluctuations in the gas flow. Lightly touching the fiber by hand resulted in changes ranging up to 1% of the original reading of the signal. Although we had tried to reinforce the fiber fixture, the dependence of signal amplitude on changes in physical stress could not be fully removed. Hence, single-point intensity calibrations of zero absorption were performed once hourly by flowing pure N2 into the fiber for 10 minutes. The data were then normalized using the discrete calibration points. With this correction method, signal drift was reduced to 0.09 ppmv CH4 hr-1. The data shown in Fig. 4 have been corrected with this method.
Part of the remaining drift was due to changes in the temperature of the sample gas, which has not been accounted for in this study (e.g., a change of 1 K in room temperature is equivalent to a signal change of 0.03 ppmv CH4 in 10 ppmv CH4, simply by the ideal gas law). Another portion of the remaining drift could have come from interfering H2O absorption signals as H2O concentration changed within the gas cell. This is further discussed in Section 3.3. One potential solution for reducing system drift would be to use a dual-channel detector with two different optical bandpass filters (one for background intensity measurements, the other for absorption measurements), such as in a traditional non-dispersive infrared (NDIR) absorption setup.
3.3 Interferences from changes in humidity and temperature
We estimate the sensitivity to water vapor by comparing the signal strength before and after purging ambient air from the system, as shown in Fig. 5(a). One experiment using the LI-7810 trace gas analyzer measured a background water vapor concentration of 12,100 ± 60 ppmv H2O, where 60 ppmv was the standard deviation for one hour of data. According to Fig. 5(a), the change in absorbance per ppmv H2O is approximately 5.74e-6. The change in absorbance per ppmv CH4 according to Fig. 3(b) is 6.52e-5. Therefore, CH4 sensitivity is approximately 6.52e-5 / 5.74e-6 = 11.4 times greater than H2O sensitivity. Without correcting for the water vapor cross-sensitivity, a change of approximately 2.3 ppmv H2O is equivalent to changing the CH4 concentration by 0.2 ppmv. Therefore, ambient fluctuations in humidity must be considered if the ICLED sensor is to reach sub-ppmv CH4 sensitivity in the atmosphere.
Fig. 5. (a) Data used for estimating the sensitivity to H2O. Absorbance calculated from the ICLED sensor signal (blue points) corresponds with the left y-axis. Orange datapoints correspond with the right y-axis and show measurements of H2O from reference instruments. Both y-axes are plotted on log scales. (b) Time series of the ICLED sensor response to temperature changes corresponding to 0.01 kΩ (∼0.02 ˚C), shown on linear scales. Absorbance calculated from the ICLED sensor signal (blue points) corresponds with the left y-axis. Thermistor readings from the Arroyo Laser & TEC controller are shown in orange and correspond with the right y-axis. A PID control loop was used for the TEC controller.
We also investigated how temperature variations would affect CH4 detection sensitivity, since the output of the ICLED is known to vary with temperature [13]. In Fig. 5(b), the Arroyo Laser & TEC controller and a thermistor with 10 kΩ resistance at 25 ˚C were used to vary the operating temperature of the ICLED. For absorbance calculations, I0 was defined as the median of the measurement step with the highest signal, which was the third measurement step starting at 120 seconds in Fig. 5(b). We see that changes as small as 10 Ω (∼0.02 ˚C) affect the signal significantly, since these variations would be equivalent to 3 ppmv CH4 changes. Therefore, sub-ppmv CH4 detection requires that the ICLED operating temperature be controlled to within ∼0.01 ˚C.
3.4 ICLED lifetime
The ICLED in this study operated for > 1000 hours with no noticeable decline in output power. The device was also modulated between 20 mA and its maximum cw operating current of 120 mA, at a frequency of 10 kHz and duty cycle 50%, for over 700 hours with no observable change in performance. We expect the ICLED lifetimes to be at least as long as for interband cascade lasers, i.e., in the tens of thousands of hours range [28,29].
4. Implications
We have demonstrated the feasibility of using ICLEDs and hollow-core fibers for sensitive CH4 detection. ICLEDs have the advantage of producing in-band cw output powers in the mW range with low drive powers. Hollow-core fibers have the advantage of easily extending optical path length in the mid-IR while serving the dual purpose of a gas cell. The observed noise equivalent absorption of 0.17 ppmv CH4 at 1 Hz is sufficient for measuring CH4 near sources such as petrochemical infrastructure, agricultural activities, and wastewater treatment plants. The performance of the ICLED CH4 sensor can likely be improved further by increasing mechanical stability, accounting for temperature changes of the sample gas, and increasing the length of the hollow-core fiber (as long as the signal-to-noise ratio is not compromised).
This study also investigated how water vapor interference affects sub-ppmv CH4 measurements made by a low-cost, broadband, IR source operating under real-world conditions. For detecting CH4 at its lower-explosive-limit (LEL) concentration of 50,000 ppmv, and for certain applications (e.g., combustion processes, biogas plants) where CH4 concentrations are greater than 1000 ppmv, water vapor absorption is negligible. However, simulated absorption based on data from HITRAN and our experimental data both confirm that water vapor cannot be ignored when sub-ppmv sensitivity is required. Even the narrowest commercially-available optical bandpass filters (∼20 nm spectral width) would not eliminate enough of the H2O lines around 3.27 µm to make H2O absorption negligible. Consequently, a fielded ICLED sensor with sub-ppmv sensitivity will need to account for water vapor absorption. One option is to perform routine calibrations. By determining the relationship between the sensor signal and known concentrations of H2O, interfering H2O absorption signals can be removed. Known concentrations can be derived from gas standards or from collocated measurements by a separate instrument. These methods are already being applied at scale to fielded low-cost sensors and could be adapted to the ICLED sensors [30]. Besides calibration, other options such as real-time correction (i.e., a multichannel detection scheme) or sample drying could also remove interferences from H2O. However, regardless of the method used to account for H2O, routine calibration of an ICLED sensor would still be necessary in order to correct for signal drift.
A further possibility for minimizing the parasitic H2O absorption, while also improving the overall system performance, is to exploit recent advances in GaSb-based resonant cavity structures that dramatically narrow the linewidth of both incoherent emitters and detectors. For example, a resonant-cavity mid-IR LED (RCLED) with non-cascade active design displayed 85 times higher enhancement of the peak emission intensity at λ = 4.5 µm, with a linewidth of ∼70 nm rather than > 1 µm [31]. Similarly, a resonant cavity infrared detector (RCID) with only 5 active quantum wells displayed ∼40 nm linewidth at λ = 4.0 µm, coupled with a specific detectivity at room temperature higher than any commercially available photovoltaic device operating at that wavelength [32]. In each case, more reflective top and bottom mirrors to provide a resonant cavity with a higher quality factor could further narrow linewidth and increase peak emission intensity or absorption. In addition to possible improvements to selectivity and sensitivity, the use of resonant cavity structures may be used to reduce sensor footprint if RCLEDs and RCIDs were integrated together. Such integrated devices could be used with hollow-core fiber waveguides to provide miniature, high-sensitivity, low-cost sensors that can be produced at scale.
With any of the aforementioned solutions to increase sensor performance, systems incorporating ICLEDs or RCLEDs may be expected to provide small-footprint, low-power, high-sensitivity, and high-accuracy CH4 detection near activities where enhancements above background concentrations are < 1 ppmv.
Funding
Office of Naval Research; Office of the Dean for Research, Princeton University.
Acknowledgments
The authors would like to acknowledge James McSpiritt of the Atmospheric Chemistry and Composition Group at Princeton University for customization of the sensor calibration chamber, and Michael Soskind of the PULSe Group at Princeton University for providing ray tracing simulations to assist with coupling the hollow-core fiber.
Disclosures
The authors declare no conflicts of interest.
References
1. J. Hodgkinson and R. P. Tatam, “Optical gas sensing: a review,” Meas. Sci. Technol. 24(1), 012004 (2013). [CrossRef]
2. W. T. Honeycutt, M. T. Ley, and N. F. Materer, “Precision and Limits of Detection for Selected Commercially Available, Low-Cost Carbon Dioxide and Methane Gas Sensors,” Sensors 19(14), 3157 (2019). [CrossRef]
3. S. H. Huang, Y. J. Huang, and H. C. Chui, “Trace methane sensor using mid-infrared light emitting diode in hollow-core fiber,” Sens. Actuators, B 282, 599–602 (2019). [CrossRef]
4. P. Mahbub, A. Noori, J. S. Parry, J. Davis, A. Lucieer, and M. Macka, “Continuous and real-time indoor and outdoor methane sensing with portable optical sensor using rapidly pulsed IR LEDs,” Talanta 218, 121144 (2020). [CrossRef]
5. V. Wittstock, L. Scholz, B. Bierer, A. O. Perez, J. Wöllenstein, and S. Palzer, “Design of a LED-based sensor for monitoring the lower explosion limit of methane,” Sens. Actuators, B 247, 930–939 (2017). [CrossRef]
6. B. A. Matveev and G. Y. Sotnikova, “Midinfrared Light-Emitting Diodes Based on А3В5 Heterostructures in Gas-Analyzer-Equipment Engineering: Potential and Applications in 2014–2018,” Opt. Spectrosc. 127(2), 322–327 (2019). [CrossRef]
7. D. Popa and F. Udrea, “Towards integrated mid-infrared gas sensors,” Sensors 19(9), 2076 (2019). [CrossRef]
8. A. Krier, E. Repiso, F. Al-Saymari, P. J. Carrington, A. R. J. Marshall, L. Qi, S. E. Krier, K. J. Lulla, M. Steer, C. MacGregor, C. A. Broderick, R. Arkani, E. O’Reilly, M. Sorel, S. I. Molina, and M. De La Mata, “Mid-infrared light-emitting diodes,” in Mid-Infrared Optoelectronics, E. Tournié and L. Cerutti, eds. (Elsevier, 2020), pp. 59–90. [CrossRef]
9. C. Hummelgård, I. Bryntse, M. Bryzgalov, JÅ Henning, H. Martin, M. Norén, and H. Rödjegård, “Low-cost NDIR based sensor platform for sub-ppm gas detection,” Urban Clim. 14, 342–350 (2015). [CrossRef]
10. R. Q. Yang, C. H. Lin, S. J. Murry, S. S. Pei, H. C. Liu, M. Buchanan, and E. Dupont, “Interband cascade light emitting diodes in the 5-8 μm spectrum region,” Appl. Phys. Lett. 70(15), 2013–2015 (1997). [CrossRef]
11. N. C. Das, K. Olver, F. Towner, G. Simonis, and H. Shen, “Infrared (3.8 μm) interband cascade light-emitting diode array with record high efficiency,” Appl. Phys. Lett. 87(4), 041105 (2005). [CrossRef]
12. J. Abell, C. S. Kim, W. W. Bewley, C. D. Merritt, C. L. Canedy, I. Vurgaftman, J. R. Meyer, and M. Kim, “Mid-infrared interband cascade light emitting devices with milliwatt output powers at room temperature,” Appl. Phys. Lett. 104(26), 261103 (2014). [CrossRef]
13. C. S. Kim, W. W. Bewley, C. D. Merritt, C. L. Canedy, M. V. Warren, I. Vurgaftman, J. R. Meyer, and M. Kim, “Improved mid-infrared interband cascade light-emitting devices,” Opt. Eng. 57(1), 011002 (2017). [CrossRef]
14. J. Meyer, W. Bewley, C. Canedy, C. Kim, M. Kim, C. Merritt, and I. Vurgaftman, “The Interband Cascade Laser,” Photonics 7(3), 75 (2020). [CrossRef]
15. C. D. Merritt, C. S. Kim, M. Kim, C. L. Canedy, W. W. Bewley, M. V. Warren, I. Vurgaftman, and J. R. Meyer, “Effects of ion bombardment on interband cascade laser structures,” Proc. SPIE 11288, 112881N (2020). [CrossRef]
16. I. Vurgaftman, C. L. Canedy, C. S. Kim, M. Kim, C. D. Merritt, W. W. Bewley, S. Tomasulo, and J. R. Meyer, “Interband Cascade Lasers,” in Conference on Lasers and Electro-Optics (CLEO) (2020), paper STh1E.6.
17. S. D. Smith, A. Vass, P. Bramley, J. G. Crowder, and C. H. Wang, “Comparison of IR LED gas sensors with thermal source products,” in IEE Proc.: Optoelectron. 144(5), 266–270 (1997). [CrossRef]
18. R. J. Ricker, S. R. Provence, D. T. Norton, T. F. Boggess, and J. P. Prineas, “Broadband mid-infrared superlattice light-emitting diodes,” J. Appl. Phys. 121(18), 185701 (2017). [CrossRef]
19. H. Zheng, M. Lou, L. Dong, H. Wu, W. Ye, X. Yin, C. S. Kim, M. Kim, W. W. Bewley, C. D. Merritt, C. L. Canedy, M. V. Warren, I. Vurgaftman, J. R. Meyer, and F. K. Tittel, “Compact photoacoustic module for methane detection incorporating interband cascade light emitting device,” Opt. Express 25(14), 16761–16770 (2017). [CrossRef]
20. L. Ch’ien, Y. Wang, A. Shi, X. Wang, J. Bai, L. Wang, and F. Li, “Noise suppression: Empirical modal decomposition in non-dispersive infrared gas detection systems,” Infrared Phys. Technol. 108, 103335 (2020). [CrossRef]
21. W. Ye, Z. Tu, X. Xiao, A. Simeone, J. Yan, T. Wu, F. Wu, C. Zheng, and F. K. Tittel, “A NDIR Mid-Infrared Methane Sensor with a Compact Pentahedron Gas-Cell,” Sensors 20(19), 5461 (2020). [CrossRef]
22. M. Dong, C. Zheng, S. Miao, Y. Zhang, Q. Du, Y. Wang, and F. K. Tittel, “Development and Measurements of a Mid-Infrared Multi-Gas Sensor System for CO, CO2 and CH4 Detection,” Sensors 17(10), 2221 (2017). [CrossRef]
23. J. A. Harrington, Infrared Fibers and Their Applications (SPIE, 2004).
24. Optoknowlede Systems Inc. (Guiding Photonics), “Mid-Infrared Fiber Patch Cables,” https://optoknowledge.com/documents/fliers/Flyer_OKSI_PatchCables_2018_V1.pdf.
25. C. L. Canedy, C. S. Kim, M. Kim, D. C. Larrabee, J. A. Nolde, W. W. Bewley, I. Vurgaftman, and J. R. Meyer, “High-power, narrow-ridge, mid-infrared interband cascade lasers,” J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.--Process., Meas., Phenom. 26(3), 1160–1162 (2008). [CrossRef]
26. C. S. Goldenstein, V. A. Miller, R. Mitchell Spearrin, and C. L. Strand, “SpectraPlot.com: Integrated spectroscopic modeling of atomic and molecular gases,” J. Quant. Spectrosc. Radiat. Transfer 200, 249–257 (2017). [CrossRef]
27. S. E. Aleksandrov, G. A. Gavrilov, A. A. Kapralov, S. A. Karandashov, B. A. Matveev, G. Y. Sotnikova, and N. M. Stus, “Portable optoelectronic gas sensors operating in the mid-IR spectral range (λ=3 5 μm),” Proc. SPIE 4680, 188–194 (2002). [CrossRef]
28. S. Forouhar, C. Borgentun, C. Frez, R. M. Briggs, M. Bagheri, C. L. Canedy, C. S. Kim, M. Kim, W. W. Bewley, C. D. Merritt, J. Abell, I. Vurgaftman, and J. R. Meyer, “Reliable mid-infrared laterally-coupled distributed-feedback interband cascade lasers,” Appl. Phys. Lett. 105(5), 051110 (2014). [CrossRef]
29. I. E. Trofimov, C. L. Canedy, C. S. Kim, M. Kim, W. W. Bewley, C. L. Merritt, I. Vurgaftman, J. R. Meyer, and L. T. Le, “Interband cascade lasers with long lifetimes,” Appl. Opt. 54(32), 9441–9445 (2015). [CrossRef]
30. A. C. Lewis, E. von Schneidemesser, R. E. Peltier, S. C. Lung, R. L. Jones, C. Zellweger, A. Karppinen, M. Penza, T. Dye, C. Hüglin, Z. Ning, R. Leigh, D. Hagan, O. Laurent, G. Carmichael, G. Beig, R. C. Cohen, E. Cross, D. Gentner, M. Gerboles, S. Khan, J. Kroll, P. Mudu, X. Q. Carceller, G. Ruggeri, K. Smith, and O. Tarasova, Low-Cost Sensors for the Measurement of Atmospheric Composition: Overview of Topic and Future Applications, WMO-No. 1215 (World Meteorological Organization (WMO), 2018).
31. F. A. Al-Saymari, A. P. Craig, Q. Lu, A. R. J. Marshall, P. J. Carrington, and A. Krier, “Mid-infrared resonant cavity light emitting diodes operating at 4.5 µm,” Opt. Express 28(16), 23338–23353 (2020). [CrossRef]
32. C. L. Canedy, W. W. Bewley, C. D. Merritt, C. S. Kim, M. Kim, M. V. Warren, E. M. Jackson, J. A. Nolde, C. A. Affouda, E. H. Aifer, I. Vurgaftman, and J. R. Meyer, “Resonant-cavity infrared detector with five-quantum-well absorber and 34% external quantum efficiency at 4 μm,” Opt. Express 27(3), 3771–3781 (2019). [CrossRef]
