1. Introduction

The development of multi-analyte sensor systems on a chip has attracted growing interest in recent years since they can provide low-cost sensing solutions along with the ability to perform rapid parallel chemical analysis. Multi-analyte sensors have been realized on different sensing platforms, such as electrochemical sensors [1], fluorescence sensors [2], and label-free refractometric sensors based on integrated optics or surface plasmon resonance [3,4]. Among these technologies, silicon photonic refractometric sensors offer the advantages of low cost and compactness, making them suitable for implementing large-scale sensor arrays on a chip. In particular, silicon microring resonators functionalized with appropriate sensing layers have been shown to be promising for developing label-free sensors with high sensitivities and low limits of detection (LODs) [3,5–8]. Arrays of microring resonators have also been employed for multi-analyte detection using either a spatial multiplexing scheme [3] or a wavelength division multiplexing (WDM) scheme [9]. However, most of the reported sensors have been developed for biosensing in a fluid environment, which is easier to implement than gas sensing since no complicated sensing apparatus is required. One of the few works on integrated photonic multi-gas sensors was reported in [10], which used an array of microring resonators functionalized with four different polymers for detecting four different volatile organic compound (VOC) vapors. However, the sensor performance was severely impacted by the poor selectivities and significant cross-sensitivities to the different target gases, since not only did each functional polymer show significant response to all four VOCs, but different polymers also exhibited similar responses when exposed to the same VOC vapor.

In this paper we report the development of a silicon photonic dual-gas sensor for the simultaneous detection of CO₂ and H₂ gases. Both of these gases are closely monitored in the energy production industry for environmental and safety compliance. In particular, hydrogen is a highly explosive gas which is released in the production of fossil fuels as well as the generation of clean energy such as in hydrogen fuel cells. Carbon dioxide is a greenhouse gas responsible for climate change, so its emission from industrial processes is also closely monitored for regulatory compliance.

We previously showed that the polymer Polyhexamethylene Biguanide (PHMB) is an effective functional material for refractometric sensing of CO₂ gas [11]. In particular, by functionalizing a silicon microring resonator with a thin PHMB layer, we showed that the sensor can detect CO₂ gas concentrations as low as 20 ppm. Moreover, the sensing layer can be regenerated with N₂ gas and shows no cross-sensitivity to H₂ gas. In this work, we further develop this CO₂ sensor platform by demonstrating the possibility of integrating a PHMB-functionalized microring with a H₂ gas sensor on the same chip for dual-gas detection. For refractometric H₂ gas sensing, a commonly used functional material is the palladium metal [12–15]. Using a WDM microring array, we developed a procedure for integrating the two different types of functional materials - a polymer and a metal - on the same chip and demonstrated the simultaneous detection of CO₂ and H₂ in a gas mixture environment. An important finding which emerged from our experiments is that the Pd material was found to exhibit significant cross-sensitivity to CO₂ gas, which has not been reported before for Pd-functionalized H₂ sensors. We showed how this cross-sensitivity issue could be resolved with the dual-gas sensing scheme, allowing for accurate measurement of H₂ concentration in the presence of CO₂ gas. The results of our study could help further the development of silicon photonic multi-gas sensor arrays for environmental greenhouse gas and industrial emission monitoring.

The paper is organized as follows. In Section 2, we present the design, fabrication and functionalization of a silicon photonic dual-gas sensor based on a WDM microring resonator array. Section 3 describes the sensing experiments and reports the sensor performance in terms of sensitivity, resolution and cross-sensitivity. The paper is summarized in Section 4.

2. Design and fabrication of a silicon photonic dual-gas sensor

Figure 1(a) shows a schematic of the dual-gas sensor design, which consists of an array of three microring resonators evanescently coupled to the same straight waveguide. The sensor was fabricated on a Silicon-on-Insulator (SOI) substrate with a 220 nm-thick silicon layer on a 3 μm-thick layer of SiO₂. The waveguide width was designed to be 450 nm, which supports a single TE mode with effective index n_eff = 2.29 at the 1.55 μm wavelength. Microring 1 is left air-cladded and serves as the reference sensor to track temperature and laser power variations. Microrings 2 and 3 serve as sensing elements for H₂ and CO₂ gases and are functionalized with Pd and PHMB, respectively, as shown by the cross-sectional diagrams in Figs. 1(b) and 1(c). Since the PHMB polymer has a refractive index of 1.5 and negligible absorption at the 1.55 μm wavelength, it can be used directly as a cladding for microring 3. On the other hand, Pd is a highly-absorbing noble metal with a complex refractive index n_Pd = 3.16 – 8.12j at 1.55 μm wavelength [16]. To minimize the optical loss while ensuring that the microring could sense the refractive index change in the Pd layer upon exposure to H₂ gas, we deposited a 320 nm-thick HSQ (hydrogen silsesquioxane, index 1.4) cladding layer on top of microring 2 and fabricated a Pd disk concentric to the microring, with its edge offset by a lateral distance d = 275 nm from the edge of the silicon waveguide core (Fig. 1(b)). Simulation of the waveguide mode indicated that the presence of a 13 nm-thick Pd film causes the real part of the effective index of the TE mode to increase by about 1 × 10⁻⁴, while the imaginary part increases to 1.8× 10⁻⁵j, which corresponds to a loss of 1.6 dB/cm. The change in the real part of the effective index is large enough to cause a measurable shift in the microring resonance, while the extra waveguide loss is kept low so as not to significantly degrade the resonator’s Q-factor.

 

Fig. 1 (a) Schematic of the dual-gas sensor consisting of an array of three microring resonators: Ref-MR for reference, Pd-MR for H₂ sensing and PHMB-MR for CO₂ sensing. Cross-sectional diagrams showing (b) the Pd functional layer and (c) the PHMB coating. (d) SEM image of the Pd-MR.

Download Full Size | PDF

We designed the microring resonators to have the same radius R = 10 μm, which gives a free spectral range (FSR) of about 9 nm. Within each FSR, we offset the resonant wavelengths of adjacent microrings by about 2 nm by varying the widths of the microring waveguides by 50 nm around the nominal value of 450 nm. Each microring was also designed to operate near critical coupling condition with a deep resonance notch in order to obtain sensor performance with good signal-to-noise ratio (SNR) and low resolution [17]. This was achieved by adjusting the coupling gap between the common straight waveguide and each microring depending on the optical absorbance of the functional coating on the microring. In particular, the reference microring (Ref-MR) and the PHMB-coated microring (PHMB-MR) have a coupling gap of 170 nm, while the coupling gap of the Pd-coated microring (Pd-MR) is reduced to 140 nm to compensate for the higher loss of the metal film. In the chip layout, the Ref-MR and Pd-MR are separated by a distance of 100 μm, while the PHMB-MR is placed 2 mm away from the other two microrings to facilitate the deposition of the PHMB layer, as described below. Grating couplers are placed at both ends of the straight waveguide for coupling TE-polarized light to and from fibers.

We fabricated the sensor chip using a process consisting of electron beam lithography (EBL) and ICP RIE dry etching. After the chip was fabricated, we functionalized microring 2 with a Pd layer using a metal lift-off process, followed by the functionalization of microring 3 by PHMB spin-coating. To deposit the Pd layer, we first spin-coated a 320 nm-thick HSQ cladding layer on the chip and patterned a 50 × 50 μm² area of HSQ covering the Pd-MR. Next, we spin-coated the chip with a 90 nm-thick layer of PMMA and used EBL to define an 18.5 μm-diameter disk concentric to the microring. The exposed PMMA disk was then removed with a developer and a 13 nm-thick Pd layer was deposited on the chip using a planar magnetron sputtering system. Metal lift-off was performed to leave a Pd disk lying concentric on the microring, as shown by the SEM image in Fig. 1(d). Next, we functionalized microring 3 by spin-coating a 620 nm-thick layer of PHMB on the entire chip. The chip was baked at 110°C for 5 minutes, then dipped partly in de-ionized water to dissolve the PHMB from the region where microrings 1 and 2 were located, leaving only microring 3 covered by PHMB. Finally, the sensor chip was bonded to a fiber array for testing.

3. Sensing experiments and results

The experimental setup used to perform sensing experiments on the dual-gas sensor is shown in Fig. 2. The sensor chip was placed in a sealed test chamber made of Teflon with a volume of about 1 dm³. Gas inlets and outlets were placed at the top of the chamber to allow analyte gases (H₂ and CO₂) and carrier gas (N₂) to enter and exit the chamber. The analyte gases were supplied from two separate gas tanks containing 1% H₂ in N₂ and 1% CO₂ in N₂. The flow of each gas mixture was controlled and regulated by a small-volume mass flow controller (MFC). Pure N₂ gas from a separate tank was also used to flush the chamber and tubes to precondition the system before each test.

 

Fig. 2 Schematic of the setup used to perform sensing experiments to measure the responses of the dual-gas sensor to CO₂ and H₂ gases.

Download Full Size | PDF

Optical measurements of the sensor response were performed using a CW tunable laser, a polarization controller to set the input light to TE polarization, a photodetector and a power meter (or oscilloscope) to record the transmitted power. The tunable laser was a Santec TSL-510A model with a 160 MHz linewidth, wavelength resolution of 1 pm and stability of 3 pm, and power stability of 0.003 dB. Figure 3(a) shows the spectral response of the sensor chip obtained after the test chamber was preconditioned by flooding it with N₂ gas for 10 min. We observe three distinct resonances corresponding to each of the three microrings, as labeled in the plot. The FSR of each microring is around 9 nm. To determine the parameters of each microring, we fitted its resonance spectrum to the power transmission function of an all-pass microring resonator,

 

Fig. 3 (a) Spectral response of the dual-gas sensor measured under N₂ gas flow. (b) Resonant wavelength shifts of the Ref-MR at different H₂ and CO₂ gas concentrations.

Download Full Size | PDF

To test the dual-gas sensor, we first performed a series of sensing experiments to measure the response of each microring to separate H₂ and CO₂ gas flows. In each sensing experiment, we tuned the laser to a fixed wavelength located on the shoulder of the resonance spectrum of the microring under test and monitored the transmitted power. From the transmitted power, we computed the corresponding resonant wavelength shift using Eq. (1) and the extracted microring parameters (κ and α) to obtain the sensor response to the analyte gas. These measurements allowed us to determine the sensitivity, LOD and cross-sensitivity of the Pd-MR and PHMB-MR sensors to each analyte gas. Finally, we performed sensing experiment with a mixture of H₂ and CO₂ gases to demonstrate the simultaneous detection of these two gases. The results of each experiment are reported in the following sub-sections.

3.1 Response of the reference microring

We began by measuring the response of the Ref-MR to H₂ and CO₂ gas flows of various concentrations to determine the noise floor and thermal stability of the sensing system. For each measurement, we preconditioned the chamber for 3 min to achieve a stable sensor baseline response. H₂ or CO₂ gas of a fixed concentration was then switched on at a constant flow rate of 200 sccm and the transmitted power of the Ref-MR was monitored and recorded after it became stabilized. The test chamber was then purged with N₂ gas flow for 3 min to reestablish the baseline before performing the next measurement.

Figure 3(b) shows the resonant wavelength shifts of the Ref-MR versus H₂ and CO₂ gas concentrations. We observe that the microring has a similar response to both CO₂ and H₂ gases, showing a small red shift in the resonant wavelength which is nearly independent of the gas concentration. Over the 1000 - 5000 ppm concentration range of each gas, the measured resonant wavelength shift is 0.41 ± 0.02 pm for CO₂ gas and 0.45 ± 0.1 pm for H₂ gas, which corresponds to an effective index change in the microring waveguide of 1.3$\times {10}^{-6}$ for CO₂ gas and 1.4$\times {10}^{-6}$ for H₂ gas. We can attribute these index changes to two possible causes: differences in the refractive indices of the analyte gas mixtures and carrier gas N₂, and the thermo-optic response of the Si microring waveguide to differences in the temperatures of the gas flows. The former effect, however, is likely to be minor since the refractive index differences between the analyte gas mixtures and N₂ gas are very small. For example, at 1550 nm wavelength, the refractive indices of N₂ and CO₂ are 1.00044 and 1.00028, respectively [18,19], from which we can calculate the index difference between N₂ gas and a gas mixture of 5000 ppm CO₂ in N₂ to be around 8 × 10⁻⁷. This index change is too small to cause the observed resonant wavelength shifts of the microring. Thus the resonance shifts of the Ref-MR are more likely to be due to the temperature changes which occurred when different gases were introduced into the test chamber. Assuming that the resonance shift is due mainly to the index change in the Si waveguide core caused by the difference in the temperatures of the analyte gas (H₂/N₂ or CO₂/N₂) and pure N₂ gas, we estimate that a wavelength shift of 0.45 pm of the Ref-MR would correspond to a temperature change of approximately 0.01°C between the analyte gas and N₂. This level of temperature variations is reasonable for our sensing system and can be taken as the background noise of our measurements.

3.2 Responses of PHMB-MR to CO₂ and H₂ gases

Next we measured the responses of the PHMB-MR to separate CO₂ and H₂ gas flows of different concentrations. Figs. 4(a) and 4(b) show the plots of the resonant wavelength shift of the PHMB-MR versus CO₂ and H₂ gas concentration, respectively. We observe in Fig. 4(a) that the resonant wavelength of the microring is blue shifted upon exposure to CO₂ gas, with the amount of wavelength shift linearly correlated to the CO₂ concentration. This behavior is similar to the response of a standalone PHMB-coated microring sensor to CO₂ gas which we reported in [11]. Over the 0 – 1250 ppm concentration range, the sensitivity of the PHMB-MR on the dual-gas sensor is calculated to be S = 4.83 × 10⁻⁴ pm/ppm, which is smaller than the value of 2.4 × 10⁻³ pm/ppm reported for the standalone sensor in [11]. The difference is due to the much thicker PHMB polymer used in the dual-gas sensor (620 nm vs. 240 nm in the standalone sensor), which might have prevented CO₂ molecules from penetrating deep enough into the functional layer, resulting in weaker perturbation of the waveguide mode and hence lower sensitivity. We expect that the sensitivity of the PHMB-MR can be improved by using a thinner PHMB layer. The error bars in Fig. 4(a) represent the standard deviations of three repeated measurements of the wavelength shift at each gas concentration. The uncertainty of the measurements in terms of wavelength is σ_λ = ± 3 × 10⁻² pm (σ = ± 5.3%), which represents the experimental uncertainty due to laser fluctuations and system noise of the PHMB-MR sensor for detecting CO₂ gas in the 0 − 1250 ppm concentration range. The resolution (in terms of concentration) of the sensor can be calculated from the sensitivity S and uncertainty σ_λ to be R = σ_λ/S = 40 ppm, which is comparable to the reported resolutions of other types of CO₂ sensors. For example, a surface plasmon resonance CO₂ sensor using carbon nanotubes as the functional material was reported to have a resolution of 150 ppm, though with a higher sensitivity of 4 × 10⁻³ pm/ppm [20].

 

Fig. 4 Responses of the PHMB-MR to CO₂ and H₂ gases: (a) plot of the resonant wavelength shift vs. CO₂ gas concentration. The straight line is the linear curve fit of the data (slope S = −4.83 × 10⁻⁴ pm/ppm, y-intercept B = −0.17 pm). (b) Plot of resonant wavelength shift vs. H₂ gas concentration.

Download Full Size | PDF

From Fig. 4(b) we observe that the PHMB-MR exhibits a small blue shift of −0.28 pm in the resonant wavelength upon exposure to H₂ gas. This shift, however, is nearly independent of the H₂ concentration up to 3000 ppm, indicating that PHMB has negligible cross-sensitivity to H₂ gas. More rigorous tests performed on a standalone CO₂ microring sensor in [11] also confirmed that PHMB does not respond to H₂ gas. The small resonant wavelength shifts of the PHMB-MR in Fig. 4(b) were most likely due to the thermo-optic response of the microring to slight temperature variations caused by the H₂ gas flow. We note that the wavelength shifts are within the noise floor caused by temperature fluctuations as measured by the Ref-MR in Fig. 3(b). This noise floor represents a source of systematic uncertainty and places a lower limit on the CO₂ gas concentration that can be reliably measured by the PHMB-MR with the current experimental setup. We call this lower concentration limit the detection threshold of the sensor. Specifically, since the sensitivity of the PHMB-MR to CO₂ is 7.62 × 10⁻⁴ pm/ppm, the dual-gas sensor cannot measure CO₂ gas concentrations less than about 370 ppm, which is ten times larger than the resolution of the sensor. These results underscore the importance of maintaining the gas mixture in a stable thermal equilibrium with the sensor so as to minimize temperature effects.

3.3 Responses of Pd-MR to CO₂ and H₂ gases

Figures 5(a) and 5(b) show the plots of the resonant wavelength shift of the Pd-MR vs. H₂ and CO₂ gas concentration, respectively. In Fig. 5(a) we observe that over the 0 – 5000 ppm concentration range, the Pd-MR exhibits a blue shift in the resonant wavelength upon exposure to H₂, with the magnitude of the wavelength shift linearly correlated to the gas concentration. Similar behaviors have been reported for Pd-functionalized photonic sensors [15]. The blue resonance shifts of the Pd-MR are due to decreases in the refractive index of the Pd layer upon exposure to H₂ gas. This index change can be explained in terms of the lattice expansion of the metal upon hydrogen absorption, which results in the formation of palladium hydride (PdH_x) with reduced conductivity and hence smaller refractive index than Pd [12]. From the linear curve fit of the data in Fig. 5(a), we obtain the sensitivity of the Pd-MR to H₂ gas to be S = 9.15 × 10⁻⁴ pm/ppm. The error bars in the plot represent the standard deviations of three repeated measurements at each gas concentration, yielding an uncertainty of σ_λ = ± 0.2 pm (σ = ± 7.8%) in the measured wavelength shifts and a sensor resolution of 235 ppm for H₂ detection in the 0 – 5000 ppm concentration range. Fig. 5(a) also indicates that the Pd-MR has a lower detection threshold of about 1000 ppm. Below this concentration, the change in the transmitted power was within the noise level and could not be used to accurately determine the resonant wavelength shift. We note that the performance of the Pd-MR on the dual-gas sensor is comparable to those reported for standalone Pd-based H₂ sensors in the literature. For example, a Pd-coated SU-8 microresonator sensor in [15] was reported to have a sensitivity of 3.2 × 10⁻³ pm/ppm and a resolution of 700 ppm, but with a much higher detection threshold of 3000 ppm.

 

Fig. 5 Responses of the Pd-MR to H₂ and CO₂ gases: (a) plot of resonant wavelength shift vs. H₂ gas concentration (straight line is the linear curve fit with slope S = −9.15 × 10⁻⁴ pm/ppm, y-intercept B = 0.71 pm); (b) plot of resonant wavelength shift vs. CO₂ gas concentration (straight line is the linear curve fit with slope S = 1.44 × 10⁻³ pm/ppm, y-intercept B = −0.58 pm).

Download Full Size | PDF

In Fig. 5(b) we observe that upon exposure to CO₂ gas, the Pd-MR exhibits a red shift in the resonant wavelength which increases with increasing CO₂ gas concentration in an approximately linear fashion. The magnitudes of the wavelength shifts are also comparable to the response of the Pd-MR to H₂ gas in Fig. 5(a), indicating that there is a significant response of the Pd layer to CO₂ gas, which to our knowledge has not been reported before for Pd-based H₂ sensors. From the linear fit of the data in Fig. 5(b), we determined the cross-sensitivity of the Pd-MR to CO₂ to be $S_{{\text{CO}}_2}$ = 1.44 × 10⁻³ pm/ppm. To confirm this cross-sensitivity behavior of Pd thin film to CO₂ gas, we also fabricated a standalone Pd-coated microring and performed CO₂ and H₂ sensing experiments, with the results showing similar cross-sensitivity behavior to CO₂. A possible cause of this cross-sensitivity may be the adsorption of CO₂ molecules on the sensor surface by palladium oxide [21], which might have been formed by partial oxidation of the Pd film. This adsorption forms a monolayer of CO₂ molecules on the Pd-MR surface, which perturbs the effective index of the microring waveguide resulting in the measured resonance shifts.

Since the Pd-MR responds to both H₂ and CO₂ gases, it cannot be used to measure H₂ concentration in a gas mixture containing both gas species. However, with the dual-gas sensor, we can independently determine the CO₂ concentration in the mixture from the PHMB-MR. Knowledge of the CO₂ concentration ($C_{{\text{CO}}_2}$) along with the cross-sensitivity of the Pd-MR to CO₂ allows us to determine the H₂ concentration ($C_{{\text{H}}_2}$) from the Pd-MR response using the formula

3.4 Response of the dual-gas sensor to a mixture of CO₂ and H₂ gases

We employed the experimental setup in Fig. 2 to measure the response of the dual-gas sensor to a mixture of 2% H₂ and 0.5% CO₂ in N₂. The mixture was supplied from a certified standard N₂ gas tank containing both H₂ and CO₂ gases with the specified concentration ratios. After preconditioning the test chamber with 200-sccm flow of N₂ for 10 min, we measured the baseline spectral responses of the Pd-MR and PHMB-MR, which are shown by the black curves in Figs. 6(a) and 6(b). Next we introduced 200-sccm flow of the gas mixture into the chamber and measured the transmission spectra of the Pd-MR and PHMB-MR again after the sensor response had stabilized. The results are shown by the red curves in Figs. 6(a) and 6(b), from which we found that the resonance spectra of the Pd-MR and PHMB-MR were blue-shifted by −11 ± 0.3 pm and −2.4 ± 0.3 pm, respectively, upon exposure to the gas mixture. From the resonant wavelength shift of the PHMB-MR, we calculated the CO₂ gas concentration to be 0.46% ( ± 0.06%). Using this value in Eq. (2), we calculated the H₂ gas concentration to be 1.9% ( ± 0.1%), which agrees with the specified H₂ content of the gas mixture. Although the concentrations of H₂ and CO₂ in the gas mixture are higher than the calibrated ranges performed on the individual microrings, the fact that the measured concentrations agree with the expected values within experimental uncertainties implies that the responses of the microring sensors did not depart significantly from the linear relationships, at least up to the concentration levels of the gas mixture.

 

Fig. 6 Measured resonance spectra of (a) the Pd-MR and (b) the PHMB-MR on the dual-gas sensor. Black curves are the initial spectra in N₂ gas; red curves are the spectra in the presence of a mixture of 2% H₂ and 0.5% CO₂ gases in N₂. (c) Time response of the power transmission of the Pd-MR and PHMB-MR to an 8-min flow of the gas mixture. The operating wavelengths of the two microrings are indicated by points A and B in (a) and (b). Blue trace is the response of the Pd-MR; red trace is the response of the PHMB-MR. The gas mixture flow is indicated by the black dashed line.

Download Full Size | PDF

Figure 6(c) shows the time response of the sensor to an 8-min flow of the gas mixture. The plot shows the transmitted powers of the Pd-MR and PHMB-MR monitored at 1560.74 nm (wavelength A in Fig. 6(a)) and 1562.43 nm (wavelength B in Fig. 6(b)), respectively. We observe that the transmitted powers of both microrings decreased upon the introduction of the gas mixture. After the gas mixture was turned off, the transmitted powers rose back again as CO₂ and H₂ were released from the PHMB and Pd functional layers during the regeneration period. Since the operating wavelength of each microring is located on the left shoulder of the respective resonance spectrum, the drops in the transmitted powers upon exposure to the gas mixture correspond to blue shifts in the resonant wavelengths, in agreement with the results in Figs. 6(a) and 6(b). We note, however, that the time trace of the transmitted power of the Pd-MR did not return fully to the baseline after the gas mixture was turned off, indicating that the Pd thin film did not fully return to its initial state. This irreversible change of the Pd metal layer typically occurs at H₂ concentrations over 1% due to the irreversible phase transition of the PdH_x system and has been reported for other Pd-based hydrogen sensors [12,14]. Heating the Pd film to above 300°C could reverse this phase transition but may cause damage to the PHMB functional layer. Another approach to avoid the irreversible response is to use Pd-Ni alloy as the functional material, but at the cost of reduced sensitivity [22].

4. Conclusion

In this paper we reported the design, implementation and operation of a silicon photonic dual-gas sensor for H₂ and CO₂ gases. Simultaneous detection of the two gases was achieved using a WDM microring resonator array functionalized with two different types of materials: Pd metal for H₂ gas and PHMB polymer for CO₂ gas. Results of the gas sensing experiments showed that the PHMB microring had a high sensitivity to CO₂ gas and good selectivity against H₂ gas. On the other hand, the Pd microring was found to exhibit sensitivity to both H₂ and CO₂ gases, making it ineffective for detecting H₂ in a gas mixture containing CO₂. We showed that the dual-gas sensor can allow for accurate measurement of H₂ concentration in the gas mixture by accounting for the cross-sensitivity of Pd to CO₂ gas. Our work can be extended to incorporate sensors for other industrial or greenhouse gases on the same chip for rapid multi-gas analysis for safety and environmental monitoring applications.