Open Access
Add to BibSonomy
Abstract
We demonstrate the sensitive detection of dimethyl methylphosphonate (DMMP, a hydrogen-bond (HB) basic phosphonate ester) using additional optical loss induced in an interband cascade laser with top optical cladding layer replaced by an exposed sensing window coated by a HB acidic sorbent layer. Thin coatings of the sorbents HCSFA2 and oapBPAF were deposited on the sensing window to allow reversible capture and concentration of DMMP for optical interrogation. Analyte levels down to 0.1 mg/m3 (∼20 ppb) were tested and successfully detected by monitoring the laser’s threshold or its output power at a fixed bias as a function of DMMP delivery concentration.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
On-chip chemical sensing systems operating in the midwave infrared (MWIR) “fingerprint” region are expected to find broad applicability in trace chemical detection by exploiting ultra-compactness and minimal operational power for long battery lifetimes [1]. However, it is quite challenging for an on-chip device to attain high sensitivity due to the small spatial overlap of the ambient gas with an optical mode propagating in a semiconductor waveguide. Here we report the parts per billion (ppb) detection of DMMP, a nerve-agent simulant, by employing a custom sorbent layer to enhance the DMMP interaction with the gain mode in an interband cascade laser (ICL) [2–4]. This was achieved by replacing the ICL’s top cladding layer with a thin coating of a HB acidic sorbent to provide strong absorption at low DMMP concentrations in air. Although the present experiments employed an off-chip InSb detector to monitor the output from the ICL’s end facet, a planned next generation of the device will realize full monolithic integration in an ultra-compact package via the electrical I-V characteristics alone or, alternatively, by incorporating an interband cascade detector (ICD) next to the laser cavity [5].
2. Design
The ICL, which was designed for emission near λ ≈ 3.0 µm at room temperature, was grown on a GaSb substrate using procedures described previously [6]. The GaSb-based layering design is similar to that of a conventional ICL [3,4], except that the top separate confinement layer (SCL) is thinner (230 nm) and the top cladding layer is omitted, leaving only a 110 nm-thick superlattice transition region above the SCL. This maximizes optical overlap of the lasing mode with the sorbent layer deposited on top.
The front and side view schematics of the fabricated device are shown in Fig. 1. Wide (60 µm) ICL ridges were fabricated using photolithography and wet etching. The phosphoric acid-based wet etch was designed to stop below the bottom GaSb SCL, within the InAs/GaSb superlattice (SL) bottom cladding layer. A 300-nm-thick Si3N4 layer was deposited by plasma-enhanced chemical vapor deposition (PECVD), after which the top contact and sorbent print area (described below) were etched back by photolithography followed by an SF6-based inductively-coupled plasma (ICP) etch. Approximately 100 nm of SiO2, which was deposited by sputtering to block occasional pinholes in the Si3N4 film, was patterned by photolithography and lifted off the top contact and intended sorbent print area. On top of the 60-µm-wide ridge, a 20-µm-wide sensing window was left open for sorbent deposition. Top metal contact regions of 10 µm width were defined along both sides of the sensing window, with the outer 10 µm on each side covered by Si3N4 isolation layers. The ridge sidewalls were corrugated, with peak-to-valley amplitude ≈ 3.5 µm and period 10 µm, in order to suppress lasing in less stable modes that bounce off the sidewalls [7]. Parts of the ridge away from the sensing window were metallized and then electroplated with ≈ 4 µm of Au, excluding the sorbent printing area. The electroplating was patterned so as to leave non-plated gaps of ≈ 50 µm to facilitate cleaving into individual laser cavities. Ti (20 nm)/Pt (150 nm)/Au (100 nm) was deposited by e-beam evaporation as the top metal contact, and Cr (30 nm)/Sn (40 nm)/Pt (150 nm)/Au (100 nm) was deposited on the bottom of the substrate as the bottom contact. Each device was cleaved to a cavity length of 2 mm and mounted epitaxial-side-up on a copper mount. A high reflection (HR) coating comprising 200 nm Al2O3 topped with 100 nm Au was deposited on each back facet, while the front facet was uncoated.
Fig. 1. Schematics of the sensing device: (a) Front cross-sectional view of the ICL with top clad replaced by a sorbent coating on the sensing window; (b) Side view of the cavity with length 2 mm and width 60 µm, on which the coated sensing window has length 1.88 mm and width 20 µm.
3. Sorbent inkjet printing
We employed a Jetlab4 inkjet station from Microfab Technologies Inc. with a 50 µm dispensing nozzle tip to deposit two different sorbent layers, HCSFA2 [8] and oapBPAF [9], from solutions of the sorbents onto the sensing windows of the ICL ridges. A 0.05% to 0.5% solution of HCSFA2 in n-butanol or 0.05% to 5% of oapBPAF in n-butanol was used to deposit sorbent onto the devices. The sorbent solution selected for each inkjet deposition depended on the target thickness of the sorbent. Thicknesses in the range 600 nm to 2µm were deposited on the sensing windows for evaluation in our vapor tests. The thicknesses were estimated by counting the number of drops deposited, the drop size, the sorbent solution concentration, and taking into account the surface area covered. 90 µm droplets (from the 50 µm nozzle) of sorbent solution were generated at the inkjet nozzle and deposited along the length of the sensing window at spacing intervals of 0.1 mm. This inkjet process was repeated several times, depending on the target thickness. Figure 2(a) shows a top-view micrograph of 7 ICLs (labeled A1 to G1 from right to left), onto which oapBPAF was deposited on devices A1 (0.6 µm thick), D1 (1.2 µm), and F1 (1.8 µm). The micrograph images in Figs. 2(b-d) show the ends of three ridges: (b) N1 onto which 2 µm of HCSFA2 was deposited; (c) I2 (No sorbent); and (d) A1 (0.6 µm of oapBPAF). The diffraction patterns in (b) and (d) indicate variations in the polymer thicknesses. None of the investigated combinations of drop size per burst and number of bursts eliminated these thickness variations. Simulations project that for these thicknesses of the sorbent layer, evanescent coupling of the propagating lasing mode to the sorbent containing the concentrated analyte is ≈ 0.4%.
Fig. 2. (a) Top view micrograph of 7 ICL cavities (labeled A1-G1 from right to left) with 2 mm length and 60 µm width, each with a top sensing window of 1.8 mm length and 20 µm width. Inkjet printing was used to coat oapBPAF sorbent on the A1 (0.6 µm thick), D1 (1.2 µm), and F1 (1.8 µm) devices; (b-d) Microscopy images of the ends of cavities with: (b) 2 µm thick HCSFA2 sorbent; (c) No sorbent; and (d) 0.6 µm thick oapBPAF sorbent. The device in (d) is A1 from (a).
The custom HB acidic sorbent coatings HCFAS2 and oapBPAF are designed to reversibly sorb analytes for detection applications [8–12]. This is principally enabled by intermolecular binding between the target analyte and the sorbent hydroxyl (-OH), which in the present case strongly affects the MWIR absorption spectrum [10–12]. The differential spectral absorption (with and without the analyte) at selected MWIR wavelengths can isolate the hydroxyl stretch shift from the free hydroxyl position to that associated with analyte binding. The peak position for the sorbent hydroxyl during analyte binding allows a preliminary identification of the analyte class or chemical type [10]. Many hazardous chemicals induce a relatively strong spectral shift of the sorbent hydroxy feature, whose peak position scales with the HB basicity of the analyte molecule [10–12]. Specifically, the free hydroxyl sorbent absorption feature at ∼ 3614 cm−1 (2.767 µm) is perturbed when the target analyte species binds and shifts the sorbent hydroxyl to a lower wavenumber [10–12]. The sorbent oapBPAF is based on a bisphenol structure and by design provides a large number of “free” HB acidic hydroxyl sites [9,12]. HCSFA2 is a carbosilane polymer with pendant hexafluoroisopropyl (HFIP) HB acidic groups, but provides a lower number of free hydroxyl sites due to its self-intermolecularly hydrogen bonded structure [8–12]. By invoking the spectral changes of the sorbent as the measurand in this work, we are moving beyond simply measuring the vapor-driven mass changes for sorbent-coated resonant devices such as surface acoustic wave (SAW) [13] or quartz crystal microbalance (QCM) devices, or by using them to trap and release chemicals in preconcentrator applications [14].
4. Optical properties of sorbent and ICL
Figure 3 illustrates that the ICL emission wavelength of λ ≈ 3.0 µm (3350 cm−1) falls within the strong characteristic absorption signature of the hydroxyl-bound DMMP, which spans 2.86-3.33 µm (3000-3500 cm−1). Conversely, neither of the sorbents used in the present investigation strongly absorbs at the ICL emission wavelength, as seen in Fig. 3 for the case of oapBPAF.
Fig. 3. Absorbance spectra of the oapBPAF sorbent and oapBPAF after saturation with DMMP vapor, plotted with a typical pulsed ICL emission spectrum at T = 25°C.
5. Experimental setup and measurement
DMMP vapor for testing was provided by a custom vapor generator shown schematically in Fig. 4. The system provides controlled concentrations ranging from 0.1 ppb to 1000 ppm, depending on the chemical vapor pressure. The main section feeds clean air with 0% relative humidity (RH) to a mass flow controller (MFC), labeled D, at rates ranging from 0 to 20 L/min. Humidified air up to 90% RH can be added to the dry air source. Several MFCs and mass flow meters (MFMs) in the chemical mixing section monitor the flow at each point and precisely regulate the chemical concentration input to the final mixing stage of the main section. Nitrogen supplied via the MFC labeled C passes through a bubbler filled with the chemical (DMMP in this case). The bubbler incorporates a fritted tube from Aceglass Inc., and is immersed into the DMMP to provide small, uniform bubbles of the vapor. Two “bubblers” are used in parallel to prevent back-up of pressure to the MFC, with the second one unfilled to ensure vapor without aerosol drops produced at the output of the vapor generator. At the mixing stage, the saturated vapor mixes with the clean air input at the MFC labelled D. Here two ports control the flow of vapor to the 1st mixing stage: (1) an adjustable valve and (2) a 3-way Lee valve. The adjustable valve controls flow from the other port that is monitored by the MFM E. This monitored flow enters the 2nd mixing stage when the first 3-way Lee valve is turned off. Dry air is also fed to MFC F to provide dilution flow for the second mixing stage. An adjustable valve controls flow to the final mixing stage that is monitored by MFM G. When the 3-way valve of the second mixing stage is turned on, the concentration flowing to the 2nd stage is directed to the final mixing stage. The final concentration can be computed from the concentration at the second mixing stage and the main diluent air flow (normally set at 20 L/min). An Arduino microprocessor-based system controls the valves, MFCs and MFMs. Custom user-friendly software (written in Visual Studio) monitors the flows, computes the final concentration, provides timing for turning the vapor on and off, and generates a final test report. The system can provide up to 5 chemicals independently, or combine any of the 5 chemicals at the outlet consisting of a 4-inch diameter stainless steel cup, which was passivated with the commercial silcosteel process by SilcoTek Inc.
Fig. 4. Schematic of the vapor generator constructed by NRL Code 6365. The system consists of main dilution and 2-stage mixing dilution of the chemical concentration. All air and nitrogen lines are temperature-controlled by a water bath (-15 to 100 °C.) Concentrations ranging from 0.1 ppb to 1000 ppm can be delivered, depending on the chemical properties. For the present case of DMMP, the range is 0.1 ppb to 20 ppm. Up to 5 independent chemicals can be added to the system.
Figure 5(a) is a schematic of the measurement setup. Figures 5(b) and 5(c) show 7 ICL sensing devices on a copper heat sink that is mounted on a thermoelectric (TE) cooling stage. Figure 5(d) illustrates the position of the sensing device, which was placed a few inches inside and along the axis of the 4”-diameter stainless steel cup where DMMP vapor is delivered to the device in a 20 l/min airflow at a controlled humidity level. The pulsed light-current (L-V) characteristics of the ICL with sorbent coating on the sensing window are monitored to determine the dependence of laser emission on DMMP concentration. While the present epi-up-mounted wide ridges cannot reach the lasing threshold at room temperature under cw excitation, future improvements in the ICL performance, processing precision, and design of the sensing window will enable cw operation. Above threshold, current injected through the metal contacts spreads laterally over the sensing window, so that the inner portion of the ridge lases before the metal-covered outer portion with high loss from the metallization. This was verified by increasing the width of the sensing window to 40 µm, at which point two lasing thresholds were observed. The initial reduced slope indicated lasing under the metal, while the second higher threshold resulted from lasing in the sensing window that received less current injection. Light emitted from the ICL output facet is collimated with a 1” lens through a 2.547 µm wavelength bandpass filter, and collected by a cooled InSb detector via a tube that filters out unwanted bounces of the laser emission.
Fig. 5. (a) Schematics of the ICL Sensor measurement setup with vapor generator; (b) ICL Sensor sample with 7 devices (2 mm cavity length) mounted on the copper heat sink; (c) copper heat sink mounted on the thermoelectric (TE) cooler stage and; (d) positioned inside the 4-inch diameter stainless steel tube as the outlet for the vapor generator. The InSb detector is aligned and positioned to collect laser light emitted from the ICL Sensor device.
6. ICL sensing of the DMMP concentration
The experiments employed two selected sorbents, separately, to concentrate DMMP present in the air flowing through the stainless steel tube delivery system. The DMMP diffuses through the relatively thick sorbent coatings into the region actively probed by the evanescent field of the light in the semiconductor waveguide, where it is detected. Figures 6(a)-(c) show a series of results for Device N1, on which a 2-µm-thick coating of HCSFA2 was deposited. Figure 6(a) illustrates that the laser L-V characteristics progressively shift with time as DMMP flowing with a concentration of 0.86 mg/m3 continues to be sorbed. In particular, the lasing threshold increases as the DMMP accumulating in the sorbent increases the internal loss in the ICL waveguide. While the increased loss should also result in a decrease of the slope efficiency, this effect is much less obvious on the scale of the figure. The shift largely plateaued after 71 minutes, the longest duration shown in the figure. However, we see from Fig. 6(b) that the elevated lasing threshold is entirely reversible after the DMMP flow is terminated and replaced by clean air to sweep away DMMP while leaving the sorbent. Figure 6(c) plots the threshold voltage extracted from the L-V curves, both during DMMP flow (red points) and after it was terminated (cyan).
Fig. 6. Time-dependent L-V characteristics for an ICL pulse length of 300 ns and repetition rate 5 KHz, measured during and after DMMP flow: (a) Shift of the L-V curves for Device N1 (coated with a 2 µm thick layer of HCSFA2) at 25°C over a period of 71 minutes following exposure to DMMP with concentration 0.86 mg/m3; (b) Recovery over a period of 1 hour 20 minutes, following termination of the DMMP flow; (c) Threshold voltage (as determined from the L-V curves) vs. time, measured during (red) DMMP flow and after termination (cyan); (d) Shift of the L-V curves for Device A1 (coated with 0.6 µm of oapBPAF) at 23.5°C, following exposure to DMMP with concentrations 0.81 (for 30 minutes, red curves) and 6.0 mg/m3 (next 30 minutes, blue curves). (e) Recovery over a period of 1 hour following termination of the DMMP flow; (f) Threshold voltage vs. time during flow of the DMMP with concentrations 0.81 (red) and 6.0 mg/m3 (blue), and then after termination (cyan).
Figure 6(d) shows the analogous response of Device A1, which was coated with a 0.6 µm thick layer of the oapBPAF sorbent, except that in this case two different DMMP concentrations were tested in succession. At the concentration 0.81 mg/m3, the response plateaued after about 10 minutes. However, increasing the DMMP concentration to 6.0 mg/m3 led to a further shift of the curve that continued for another 15 minutes. Figure 6(e) shows that this device also recovered when the flow of DMMP was replaced with pure air. We see from Fig. 6(f) that recovery of the threshold voltage to its original value was essentially complete after about 45 minutes. The more rapid response for this device may be attributed to the physicochemical properties of the oapBPAF functional polymer sorbent, as well as the thinner sorbent coating. Incomplete control over the thickness and uniformity of the sorbent layers deposited with the present inkjet system may also have played a role. With the time for anlayte diffusion being proportional to the sorbent thickness squared, areas where the nonuniform coating was thicker would have responded more slowly than thinner areas.
Figure 7 illustrates that sensitivity to DMMP can also be attained with HCSFA2 by monitoring the variation of laser output intensity at a fixed bias voltage, rather than the lasing threshold shift as in Fig. 6. Figure 7(a) illustrates sensitivity to DMMP concentrations as low as 0.10 mg/m3, which corresponds to ∼20 ppb. The laser output from a given device initially decreases monotonically with increasing DMMP concentration, although the relation is less systematic after a longer interval. We also observe that the decrease is slightly less rapid for Device N1 at concentration 0.22 mg/m3 than for M1 at concentration 0.10 mg/m3, even though the intended sorbent deposition properties were the same for both devices. While most results in the figure were acquired at minimal RH (< 2%), some experiments were repeated at RH = 50% and 80%. The observed decrease of the laser intensity following the introduction of DMMP is comparable, i.e., independent of humidity.
Fig. 7. Time-dependent decrease of the normalized laser output (InSb detector response) at fixed ICL bias voltage near threshold during the flow of DMMP: (a) Device M1 (coated with a 2-µm-thick layer of HCSFA2, solid points) for two DMMP concentrations and Device N1 (nominally identical coating, open points) for 6 concentrations; (b) Devices A1 (blue), D1 (red), and F1 (green) coated with the oapBPAF thicknesses specified above at concentrations 0.8 mg/m3 for 30 minutes, followed by 6.0 mg/m3 for 30 minutes. In (b), filled points represent room humidity (RH) < 2%, whereas the open points for A1 represent RH = 80%.
The filled points in Fig. 7(b) show analogous results at minimal RH for Devices A1 (0.6 µm sorbent thickness), D1 (1.2 µm), and F1 (1.8 µm), which were coated with oapBPAF. Again, a higher DMMP concentration (≈ 6 mg/m3) was introduced after the laser intensity saturated at the lower concentration (0.8-0.9 mg/m3). In this experiment, the intensity for Device A1 at RH = 80% (open points) decreased even more rapidly at similar DMMP concentration. Note also that the time required to reach saturation increased with increasing thickness of the sorbent layer. If we compare the time-dependent trends in Figs. 6 and 7, we see that monitoring the laser intensity at fixed bias yields a more rapid unambiguous detection of the DMMP.
Figure 8 illustrates the results of a further investigation of humidity effects on the sensitivity to DMMP. To obtain the time-dependent threshold voltages shown in Fig. 8(a), Device N1 (HCSFA2 sorbent) was first exposed only to nominally dry air. Then RH = 80% was introduced for 30 minutes, followed by the additional introduction of DMMP at concentration 0.88 mg/m3. We see that while absorption by the humid air increased the lasing threshold to saturation at 4.30 V after 30 minutes, the device remained sensitive to DMMP (with the threshold increasing further to 4.35 V) despite the high humidity. Figure 8(b) shows results of a similar experiment using Device A1 (coated with oapBPAF), where in this case two different DMMP concentrations were successively introduced. This device appeared less sensitive to the high humidity, whereas it maintained high sensitivity to DMMP. From prior experience, we have observed that HB acidic sorbent coatings more than about a micron thick can result in an undesirable trapping of water. However, for thinner coatings less than a micron this is substantially mitigated because water diffusion in and out of the sorbent occurs fast enough to substantially reduce the accumulation.
Fig. 8. ICL threshold voltage vs. time for various combinations of DMMP and humidity: (a) Device N1 (coated with HCSFA2) with nominally clean air, followed by the introduction of RH = 80%, and then also DMMP at concentration 0.9 mg/m3; (b) Device A1 (coated with oapBPAF) with clean air followed by the sequential introduction of RH = 80%, DMMP at 0.81 mg/m3, and finally DMMP at 6.0 mg/m3.
Figure 9(a) illustrates the L-V characteristics of Device A1 at DMMP concentrations ranging from 0.11 to 5.1 mg/m3. Figure 9(b) represents the normalized output power as a function of time and vapor concentration. Figure 9(b) shows that the response saturated at each given concentration. This typically required about 10 minutes, after which the concentration was increased to a higher level. Finally, at 5.1 mg/m3 the sorbent appeared to be fully saturated. Following termination of the DMMP, the laser intensity recovered to its original value within about 30 minutes of clean airflow.
Fig. 9. (a) L-V curves of Device A1 at DMMP concentrations ranging from 0 to 5.1 mg/m3; (b) Time variation of the normalized intensity emitted by the same device following the introduction of successively higher DMMP concentrations over a period of two hours, followed by recovery after termination of the DMMP flow.
7. Summary and conclusions
These results demonstrate that the L-V characteristics of an interband cascade laser with a top sensing window coated by a sorbent layer can be sensitive to the presence and concentration of hazardous chemicals in the gas phase. In the present case, thin layers of the functionalized sorbents HCSFA2 and oapBPAF were deposited to capture and concentrate DMMP vapor. Analyte levels as low as 20 ppb were tested and detected by monitoring either the laser’s threshold voltage or output power at fixed voltage as a function of DMMP concentration. Although a given device’s response to DMMP concentration was nonlinear, it was reproducible. Furthermore, the sorbent’s response to DMMP was reversible by switching to clean air, leading to recovery of the voltage or intensity shift. This recovery allowed each device to provide reproducible results following repeated recycling of the DMMP concentration, with 8 cycles demonstrated. In related work with the sorbent HCSFA2 we have demonstrated thousands of cycles [15].
The relative responses of different devices to DMMP concentration were not always systematic based on their nominal sorbent compositions and thicknesses, as is apparent from Fig. 7. This is probably related to poor control over the sorbent uniformity produced by the inkjet printing process used to coat the sorbents. Future experiments will employ thinner coatings deposited with an upgraded inkjet printer and more dilute solutions to provide more uniform depositions. A thinner, more uniform sorbent coating will also allow much faster responses to the presence of an analyte such as DMMP. We also note that since the ICL threshold voltage or output intensity at constant bias can be affected by platform temperature as well as humidity or the presence of some other ambient chemical, protocols must be developed to distinguish those scenarios from detection of a chemical of interest. These might include the use of a reference device that is not exposed to the vapor, has a different sorbent, or which operates at a different wavelength.
The ICLs coated with both sorbents exhibited reversible response to the chemical warfare agent simulant DMMP at DOD relevant concentrations down to 20 ppb. In these experiments, the response time was faster and the sensitivity to humidity smaller for oapBPAF, although that was probably due in large part to the thinner oapBPAF films that were deposited. Both sorbents provided sensitivity to DMMP concentrations as low as ≈ 0.1 mg/m3 (20 ppb), and the present experiments did not attempt to determine the sensitivity limit. The most sensitive responses in Figs. 7 and 9 were obtained by monitoring the laser output intensity at fixed voltage, rather than by monitoring the threshold voltage. The present proof-of-concept experiments employing very limited sorbent and laser ridge parameters cannot provide definitive conclusions concerning the optimal sorbent composition or thickness. We do predict much faster signal kinetics for DMMP exposure with thinner films, with the time to diffuse into a sorbent layer being proportional to the sorbent thickness squared. The most significant noise source that will ultimately limit the sensitivity is likely to be fluctuation of the environmental and/or laser temperature. These can be calibrated in part by adding a reference ICL with sensing window that is not coated by a sorbent.
Although the present experiments employed an off-chip InSb detector to quantify the ICL L-V response, on-chip interband cascade detectors formed by reverse biasing the ICL gain stages outside of the laser cavity can be incorporated in future work to perform the detection process in an integrated, monolithic fashion. Alternatively, a sufficient sensitivity to internal loss contributed by analyte molecules captured by the sorbent coating may be attained simply by monitoring the differential I-V characteristics near threshold [16,17], which would eliminate the dedicated detector entirely. The entire sensor would then consist of a short-cavity, low-threshold-power laser with two HR-coated facets. A related, approach has been discussed for sensing with quantum cascade lasers by Phillips et al. [18,19]. Preliminary testing shows that the first-harmonic I-V signal of an ICL sensor can easily detect changes in the internal loss induced by temperature shifts < 0.1°C. By confining the entire sensor package to a single chip (plus drive electronics), the size, weight, power, and cost (SWaPc) of the resulting system can be extremely low. For example, a cw ICL with sorbent-coated sensing window could be powered by batteries, and ultimately operate with drive power < 200 mW for a sensing laser plus reference laser. This will be ideal for implementation on extremely small platforms such as wearable badges, micro-UAVs, or miniature robotic platforms that respond reversibly to chemical threats at concentrations relevant to DoD.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data supporting the results presented in this paper are not publicly available but may be obtained from the authors upon reasonable request.
References
1. J. R. Meyer, C. S. Kim, M. Kim, et al., “Interband cascade photonic integrated circuits on native III-V chip,” Sensors 21(2), 599 (2021). [CrossRef]
2. R. Q. Yang, “Infrared laser based on intersubband transitions in quantum wells,” Superlatt. Microstruct. 17(1), 77–83 (1995). [CrossRef]
3. I. Vurgaftman, R. Weih, M. Kamp, et al., “Interband cascade lasers,” J. Phys. D: Appl. Phys. 48(12), 123001 (2015). [CrossRef]
4. J. R. Meyer, W. W. Bewley, C. L. Canedy, et al., “The interband cascade laser,” Photonics 7(3), 75 (2020). [CrossRef]
5. R. Q. Yang and R. T. Hinkey, “Ultimate detectivity of multiple-stage interband cascade infrared photodetectors,” Appl. Phys. Lett. 118(24), 241101 (2021). [CrossRef]
6. C. L. Canedy, C. S. Kim, M. Kim, et al., “High-power, narrow-ridge, mid-infrared interband cascade lasers,” J. Vac. Sci. Technol. B 26(3), 1160–1162 (2008). [CrossRef]
7. M. Kim, C. S. Kim, C. L. Canedy, et al., “Interband cascade lasers with sidewall corrugations for enhanced brightness,” Proc. SPIE 10939, 32 (2019). [CrossRef]
8. B. A. Higgins, D. L. Simonson, E. J. Houser, et al., “Synthesis and characterization of a hyperbranched hydrogen bond acidic carbosilane sorbent polymer,” J. Polym. Sci. A Polym. Chem. 48, 3000–3009 (2010). [CrossRef]
9. C. A. Roberts and R. A. McGill, “Bisphenol hypersorbents for enhanced detection of, or protection from, hazardous chemicals,” US Patent 11,779,901 B2 (25 April 2023).
10. E.J. Houser, R.A. McGill, T.E. Mlsna, et al., “Sorbent coatings for detection of explosives vapor: applications with chemical sensors,” Proc. SPIE 3710, 394–401 (1999). [CrossRef]
11. E. J. Houser, T. E. Mlsna, V. K. Nguyen, et al., “Rational materials design of sorbent coatings for explosives: applications with chemical sensors,” Talanta 54(3), 469–485 (2001). [CrossRef]
12. T. G. Grissom, C. A. Roberts, R. Rodrigues, et al., “Sorbent interactions with hazardous chemicals and spectral detection strategies,” Proc. of SPIE 11416, 163 (2020). [CrossRef]
13. R. A. McGill, V. K. Nguyen, R. Chung, et al., “The ‘‘NRL-SAWRHINO’’: a nose for toxic gases,” Sens. Actuators 65(1-3), 10–13 (2000). [CrossRef]
14. I. Voiculescu, R. A. McGill, M. E. Zaghloul, et al., “Micropreconcentrator for enhanced trace detection of explosives and chemical agents,” IEEE Sens. J. 6(5), 1094–1104 (2006). [CrossRef]
15. R. A. McGill, S. V. Stepnowski, E. J. Houser, et al., “Caspar, a microfabricated preconcentrator for enhanced detection of chemical agents and explosives,” Proc. of Eurosensors XX, Göteborg, Sweden, T1A-03, (2006).
16. C. D. Merritt, W. W. Bewley, C. S. Kim, et al., “Gain and loss vs. current density and temperature in interband cascade lasers,” Appl. Opt. 54(31), F1 (2015). [CrossRef]
17. Y. Lin, L. Li, W. Huang, et al., “Quasi-Fermi level pinning in interband cascade lasers,” IEEE J. Quantum Electron. 56(4), 1–10 (2020). [CrossRef]
18. M. C. Phillips and M. S. Taubman, “Intracavity sending via compliance voltage in an external cavity quantum cascade laser,” Opt. Lett. 37(13), 2664 (2012). [CrossRef]
19. M. C. Phillips and M. S. Taubman, “Use of external cavity quantum cascade laser compliance voltage inreal-time trace gas sensing of multiple chemicals,” Proc. SPIE 9370, 93700Z (2015). [CrossRef]
