Enhancing methane sensing with NDIR technology: Current trends and future prospects

2.1 IR light source

The IR light source, often an incandescent lamp or light-emitting diode (LED), generates the initial IR light beam. It produces a broad spectrum of IR light to include the wavelengths absorbed by the target gas. For example, Ye et al. [30] selected an IRL715 lamp, widely used in NDIR sensors. Gibson and MacGregor [5] used an LED. LEDs offer benefits but early devices had challenges concerning power, stability, lifetime, and cost. The light source produces the initial IR beam sent into the sample cell.

2.2 Gas chamber

The sample cell contains the sample gas and allows IR light to pass through. The cell design ensures maximum light passes through the gas, enhancing sensitivity. The cell size and shape varies by application and required sensitivity. There are two cell designs: maximizing resolution or signal-to-noise ratio (SNR). High resolution is used for low gas concentrations. Cell length plays a crucial role. A longer path enhances sensitivity but decreases SNR [31]. New research proposes sensors without cells. Vafaei et al. [32,33] reported a method using a light source with reflector, measuring CO₂ without a cell. They created 4 structures incorporating reflectors, light source and detector.

2.3 Optical filter

An optical filter selects the specific absorbed wavelengths. The filter enhances selectivity by passing only these wavelengths to the detector. This reduces the impact of other gases that absorb IR at different wavelengths. Filter design is often a research focus, with different designs improving performance. Fleming et al. [34] reduced N₂O cross-talk using thin film filters. Zhou et al. [35] theoretically demonstrated multichannel mid-IR filters. The design uses Ge₂Sb₂Te₅ grating defect layers in a photonic crystal, exciting defect and resonance modes.

2.4 Detectors

The detectors are typically pyroelectric or thermopile, converting IR light into an electrical signal. There are usually two detectors: one for the sample gas and another for reference. The reference detector measures the light source intensity, compensating for changes over time.

2.5 Amplifier and signal processing unit

The detector signals are small and require amplification. An amplifier amplifies the signals. A signal processing unit compares the measurement and reference signals. It calculates the gas concentration based on the difference in the signals.

3.1 Factor compensation

NDIR sensors are relatively resistant to environmental conditions such as temperature and humidity changes, making them suitable for use in a wide range of environments. While these factors can affect the sensor’s readings, NDIR sensors can be designed with built-in compensations for these effects, ensuring accurate measurements under varying conditions. For example, the compensation of spectral impact from water vapor in air is necessary because methane and water vapor have overlapping absorption spectra in the IR region. When measuring the methane concentrations using a NDIR sensor, the presence of water vapor in the air can interfere with the accuracy of the methane measurements. The overlapping absorption lines of methane and water vapor can lead to an increase in the reported methane concentration, causing inaccuracies in the measurements. The compensation approach proposed by Gaynullin et al. [40] involves using a multispectral NDIR sensor, which measures the optic transmission in two specific spectral bands: the methane absorption spectra at 3.375 μm and the H₂O absorption spectra at 2.7 μm. Additionally, a third channel at 3.95 μm is used as a reference for “zero-level” calibration. This channel does not contain absorption lines for any gas present in the atmosphere. To compensate for the spectral impact of water vapor on the methane measurements, the researchers perform a calibration procedure in an environment with variable humidity. They measure the absorptions in both the methane and H₂O channels under different humidity levels and establish a relationship between the two. By simultaneously sensing both channels in the same environment, they can calculate the absorption of water vapor specifically at the methane spectral region (αH₂O-_at-methane) based on the known absorption in the separate H₂O channel (αH₂O). With this information, they can then correct the real methane transmission (τ _methane) by eliminating the impact of water vapor, using the following formula:

where τ _methane is the compensated transmission value for methane; τ _repmethane is the transmission reported from the methane channel in a humid environment, containing the combined absorption of both methane and water vapor; and αH₂O_-at-methane is the absorption of water vapor specifically at the methane spectral region, derived from the calibration in the humid environment.

By applying this compensation algorithm, the researchers were able to reduce the uncertainty in methane concentration measurements to only 15–25 ppm relative to the reference concentrations, compared to the much higher uncertainty of 180–200 ppm for the non-corrected methane concentration.

3.2 Ethane interference

However, it should be noted that while the NDIR sensor is highly specific for methane, it is not perfect. Other gases, such as ethane (C₂H₆), also absorb light at similar wavelengths, although typically not as strongly as methane. Therefore, in environments with significant concentrations of such gases, some cross-sensitivity might occur. To solve this interference problem, Tian et al. [41] used a wavelength modulation spectroscopy (WMS) technique combined with a compact home-made dense-pattern multipass cell (Figure 1). They used two tunable semiconductor lasers operating at different wavelengths: 1.653 μm for methane and 1.684 μm for ethane. Using lasers with specific wavelengths for each gas, they were able to target the specific absorption lines of methane and C₂H₆ separately, reducing the interference between the two gases during detection. The use of a dense-pattern multipass cell also increased the optical path length, enhancing the sensitivity of the sensor to low concentrations of both gases.

Figure 1

Schematic diagram of the dual-gas sensor. Reproduced with permission from ref. [41].

The problem of ethane interfering with the detection of methane was addressed in the design and algorithm of the hand-held unit based on NDIR sensor in the article published by Hennig et al. [42]. The hand-held unit is specifically designed to simultaneously detect both methane and ethane using a single tunable distributed feedback (DFB) laser diode. By reading both gases’ absorption lines simultaneously, the sensor can differentiate their concentrations accurately. The software running the sensor incorporates a sophisticated algorithm to analyze the acquired transmission signals. The acquired signals are compared with a stored background signal, normalized, and further processed to extract the concentrations of methane and ethane. The algorithm takes into account background effects, optical resonances, and variations in pressure and temperature. During the initialization process, the sensor records absorption reference lines for both methane and ethane at known concentrations, temperatures, and pressures. These data are then used for future monitoring and corrections during the operation. Although there is some cross-sensitivity between methane and ethane due to their overlapping absorption lines, the sensor applies correction methods to compensate for this interference. The algorithm is designed to differentiate and accurately quantify the individual contributions of methane and ethane to the overall signal.

This problem can also be solved by a dual-gas sensor system that uses two different detection techniques, DFB interband cascade lasers [43]. The two methods used for gas detection are laser direct absorption spectroscopy for methane and second-harmonic WMS (2f-WMS) for C₂H₆. Using two distinct detection methods, each optimized for the specific gas, the dual-gas sensor system can effectively differentiate between C₂H₆ and methane even in the presence of interference. This allows for simultaneous and accurate monitoring of both gases at ppbv concentration levels.

To address ethane interference, researchers have also proposed solutions: using lasers with specific wavelengths for methane and ethane, targeting their separate absorption lines. This reduces the overlap and interference; using gas filter correlation (GFC) techniques and sophisticated algorithms that can differentiate and quantify the contributions of methane and ethane to the sensor signal; and developing dual-gas sensor systems that use distinct detection methods optimized for methane and ethane. The selectivity and reliability of NDIR sensors depend on factors such as the design of the optical filter, signal processing techniques, and compensation for environmental factors. With proper calibration and validation, NDIR sensors can achieve good reproducibility and system suitability. However, ethane and propane interference still remains a challenge, especially at high concentrations of these gases. NDIR sensors show the highest selectivity for methane when ethane and propane levels are relatively low.

3.3 Light source

NDIR sensors, particularly those that use incandescent light sources, can have relatively high power consumption. While this may not be a significant concern for some applications, it can limit the use of NDIR sensors in battery-powered or remote devices. Using an MEMS-miniaturized IR light source enhances the sensitivity and precision of the system, concurrently minimizing power consumption and instrument size [32]. Introducing a black body radiating coating to the IR light source improves the emission of IR wavelengths [44]. Additionally, by creating micro-sized coating structures tailored to specific wavelength characteristics, narrow-band spectroscopy can be further optimized for enhanced efficiency [45].

Fanchenko et al. [46] proposed a new LED light source for methane detection emitting at a wavelength of 2.3 µm. They highlighted that recent advancements in laser optics have enabled the development of semiconductor heterostructures that emit light in the IR region, making it possible to create LEDs operating at specific wavelengths within the IR spectrum. The performance of the LED-based methane detector is discussed in the context of a NDIR gas detection scheme. They used a dual-beam optical system, with one LED emitting at 2.3 µm for the measurement channel and another LED emitting at 1.7 µm for the reference channel. The use of a reference channel helps to eliminate the influence of methane absorption bands close to the reference emitter, ensuring accurate measurements. The analytical model suggested that the LED-based NDIR detector with a 2-channel optical scheme is promising for methane detection near the lower explosive limit. The calibration curve for the LEDs is derived to determine gas concentration from NDIR measurements. The first prototype of the LED-based methane detector was constructed and tested, and the initial measurements show promising results. The methane concentration can be measured on the level of tenth percentage, indicating good sensitivity and accuracy for detecting methane gas.

3.4 Design of optical cavity

The field of NDIR methane detection continues to be an active area of research, with scientists and engineers working on several promising innovations that could further improve the performance of these sensors. One significant focus of current research is enhancing the sensitivity and selectivity of NDIR sensors. For instance, new designs for optical cavities are being explored that could increase the effective path length of the light, thereby improving the sensitivity of the sensor. For example, Ye et al. [30] proposed a compact pentahedron gas cell (Figure 2). The gas cell is designed to enhance the performance of the NDIR sensor for methane detection by increasing the optical path length while maintaining a compact and small-sized structure. The design of the gas cell incorporates a paraboloid concentrator, two biconvex lenses, and five planar mirrors. This configuration allows the IR light emitted by the incandescent lamp (IRL715) to undergo multiple reflections and significantly increase the optical path length. The extended optical path enhances the sensitivity of the sensor and improves the accuracy of methane detection. Despite the increased optical path length, the gas cell maintains a compact and small-sized structure. This is crucial for portable and handheld sensors that require a balance between performance and portability. The design allows for efficient gas absorption and detection in a reduced volume. The performance of the sensor was evaluated experimentally, and it demonstrated a 1σ detection limit of 2.96 ppm with a 43 s averaging time.

Figure 2

(a) Mathematical model of cross-section and (b) three-dimensional spiral structure of the pentahedron gas cell. Reproduced with permission from ref. [30].

3.5 New materials

In parallel, researchers are investigating new materials for the IR source and detector. These materials could potentially offer higher efficiency, longer lifespan, and better stability, leading to more reliable and longer-lasting sensors. Ahmed et al. [47] used lead zirconate titanate (PZT) to serve as the sensing material in the NDIR sensor (Figure 3). PZT is a piezoelectric material, which means that it generates an electrical signal in response to mechanical pressure or thermal changes. In this case, PZT is used as a pyroelectric material, where it generates an electric signal when exposed to IR radiation and experiences temperature changes due to the absorbed IR energy. The LOD of a methane gas sensing system using NDIR spectroscopy was improved by an order of magnitude with the plasmonic IR detector. The LOD before adding the plasmonic crystal structure was 261 ppm, and after adding the plasmonic crystal structure, it was reduced to 20 ppm.

Figure 3

Schematic diagram of the pyroelectric IR detector with a plasmonic crystal structure. Reproduced with permission from ref. [47].

Macroporous silicon filters were used as narrow-band optical filters for gas sensing in NDIR sensors [48]. The main purpose of these filters is to selectively isolate the absorption bands of specific gases from the broadband light source used in the detection system. This selective filtering allows the sensor to target and measure the absorption lines of the gas of interest accurately, which is crucial for achieving high sensitivity and selectivity in gas sensing. Macroporous silicon filters can be designed to have narrow bandwidths that correspond to specific absorption lines of the target gas. By filtering out unwanted frequencies, the filters significantly reduce interferences from other gases present in the measurement environment. This selective absorption allows for accurate identification and quantification of the target gas even in the presence of other gases with overlapping absorption spectra.

Wang et al. [49] deposited single layers of Nb₂O₅ and Ge onto optical substrates, which were subsequently used in the design of the bandpass filters. These materials were then sputter-deposited onto MBE-grown heterostructural diodes with the peak detectivity tuned to 3,300 nm using microwave plasma-assisted pulsed DC magnetron sputtering. Nb₂O₅ is a dielectric material with tunable optical properties, making it suitable for designing optical interference filters. Dielectric materials are used in multilayer thin film coatings to create interference effects that selectively transmit or reflect specific wavelengths of light. By carefully controlling the thickness and refractive index of the Nb₂O₅ layer, it becomes possible to create interference effects that allow certain wavelengths of light (in this case, those corresponding to the methane absorption band) to pass through while reflecting or blocking unwanted wavelengths. Germanium is also a dielectric material and has specific optical properties that can be useful for optical filter applications. In this work, Ge was likely used as another layer in the bandpass filters to further shape the spectral response of the filter and improve its performance in the target methane detection application. Similar to Nb₂O₅, the Ge layer’s thickness and refractive index would be carefully engineered to create additional interference effects that align with the desired spectral characteristics of the bandpass filter. Using multiple layers with different refractive indices and thicknesses, the overall transmission and blocking properties of the filter can be fine-tuned to achieve the required spectral response.

3.6 Advanced signal processing

Advanced signal processing techniques are being explored to improve the performance of NDIR sensors. These techniques could help to eliminate noise, compensate for environmental factors, and accurately interpret the sensor’s output, leading to more precise and reliable measurements. For example, a single-frequency filter denoising algorithm proposed by Zhu et al. [50] is a digital signal processing technique used to reduce the noise level in the output signal of the NDIR methane gas sensor. The algorithm is based on Fourier transform and aims to extract the signal component corresponding to the lamp modulation frequency while filtering out unwanted noise and interference from other frequencies. The algorithm applies the DFT to the input signal obtained from the IR detector. After performing the Fourier transform, the algorithm selects the amplitude of the signal component at the lamp modulation frequency as the filtered output signal. Since the output signal of the gas sensor is a sinusoidal or triangular-like wave with the dominating lamp modulation frequency, this filtering process allows only the signal component of interest to pass through. The algorithm effectively reduces noise and interference in the output signal because most of the unwanted frequency components are filtered out during the single-frequency filtering process. This leads to an increase in the SNR of the system. The main advantage of this algorithm is its ability to significantly enhance the SNR of the methane gas sensor. By extracting the signal component at the lamp modulation frequency and filtering out noise, the algorithm improves the sensitivity and accuracy of the sensor. The algorithm allows for adjusting the bandwidth of the filter by changing the sampling period. This flexibility enables optimizing the filtering process for the specific application and achieving the best SNR. The algorithm is implemented in software and does not require additional complex hardware components, making it a cost-effective solution for enhancing the performance of the methane gas sensor.

Wavelet filtering algorithm has been used to remove noise from the methane gas sensor’s signal, which is crucial for improving detection limit and SNR of the sensor [51]. The algorithm decomposes the signal of the sensor into different frequency components using the wavelet transform. The wavelet transform allows the algorithm to analyze the signal at various scales and resolutions. After decomposition, the algorithm applies a thresholding process to the wavelet coefficients. Thresholding involves setting small coefficients to zero, effectively removing noise from the signal. The choice of the thresholding method can be crucial in achieving an optimal denoising effect. Once the noise coefficients have been removed through thresholding, the algorithm reconstructs the denoised signal using the modified wavelet coefficients. The effectiveness of the wavelet filtering algorithm is demonstrated using comparisons with traditional low-pass filtering methods. The study showed that the wavelet filter outperformed the low-pass filter in reducing noise and improving the SNR. The SNR was improved by approximately 50 dB using the wavelet filter, which indicates a significant enhancement in sensor’s performance.

3.7 Multi-gas detection

In the context of detecting multiple gases, the presence of overlapping absorption in the IR spectrum leads to cross-talk issues, significantly impacting detection accuracy [52]. Consequently, for multi-gas NDIR analyzers, the elimination of interference between gases is of paramount importance. To address this challenge, previous studies have put forth several solutions, including the use of narrow-band optical bandpass filters, GFC techniques, machine learning algorithms, and the integration of multiple detection units.

The NDIR system in the report published by Tan et al. [53] achieves the detection of three gases (methane, carbon dioxide, and carbon monoxide). The gas chamber is designed with two crossed elliptical surfaces coated with a reflective gold film. This design allows the IR light to be reflected twice before reaching the detectors, increasing the optical path length and enhancing sensitivity. Four single-channel pyroelectric sensors are integrated into the miniature gas chamber. Each sensor is equipped with a specific filter piece to detect one of the gases or act as a reference channel (Figure 4). The system incorporates a compensation method for environmental parameters such as ambient temperature, humidity, and pressure. This compensation method helps overcome the influence of environmental variations on the detectors, thereby improving the accuracy of gas concentration measurements. By combining these features, the NDIR system can effectively detect and quantify the concentrations of methane, carbon dioxide, and carbon monoxide gases, making it a valuable tool for multi-gas monitoring applications in various industries and environments.

Figure 4

Schematic diagram of NDIR system for methane, carbon dioxide, and carbon monoxide detection. Reproduced with permission from ref. [53].

Genner et al. [54] developed a NDIR gas sensor based on tunable Fabry–Pérot filter technology for online monitoring of methanol and methyl formate (MF) concentrations in the exhaust gas of an industrial formaldehyde production plant. They designed and implemented a prototype gas sensor that integrates a modulated thermal IR source, a temperature-controlled gas cell for absorption measurements, and a Fabry–Pérot interferometer–detector device. The use of tunable Fabry–Pérot filters allows for the selective measurement of methanol and MF in the gas phase. The gas sensor was calibrated in the laboratory to measure the methanol and MF concentrations in the range of 660 to 4,390 ppmV and 747 to 4,610 ppmV, respectively. The achieved limits of quantification were 184 ppmV for methanol and 165 ppmV for MF.

The NDIR sensor proposed by Fonseca et al. [55] uses a non-specific filter microarray and a thermopile microarray. The filter microarray contains multiple non-specific filters with broad transmittance bands in the mid-IR region. When exposed to a gas sample, each filter will capture various absorption features of different gases, including CO₂ and methane, as the gas absorbs specific IR wavelengths. For the quantitative determination of gas mixtures of CO₂ and methane, the system showed prediction errors in the range of tens to hundreds of ppm. The reported results indicated that the sensor could separate and quantify different mixtures of CO₂ and methane, as well as detect them individually. It is important to note that the reported detection limits for CO₂ and methane in this study are in the range of tens of ppm. While this sensitivity level may be suitable for certain applications, it might not be sufficient for highly sensitive applications requiring detection at lower concentrations, such as environmental monitoring or greenhouse gas studies. In another work [31], the optical E-nose uses NDIR sensors to detect CO₂ and methane gases. The optical E-nose was equipped with four emitter–detector pairs (Figure 5), each encapsulated in individual heated sensor chambers. These emitter–detector pairs are tuned to specific IR wavelengths corresponding to the absorption frequencies of the gases of interest. When a gas sample containing CO₂ or methane is passed through the sensor chamber, the gas molecules absorb the IR light emitted by the emitter, and the detector measures the amount of light that reaches it after passing through the gas sample. The performance of the optical E-nose in detecting CO₂ and methane is evaluated using the gas rig test. The sensor responses to different concentrations of CO₂ and methane gases are recorded, and the results show that the response is linear to polynomial (second degree) for changes in concentration. The detection limits for CO₂ and methane using this technology are found to be below 25 ppm and below 1 ppm, respectively.

Figure 5

Optical electronic nose 2 × 2 gas sensor chamber formation: (a) 3D model and (b) manufactured prototype. Reproduced with permission from ref. [31].

Jiang et al. [56] proposed that a system uses four individually controlled DFB lasers, each emitting light at one of the selected wavelengths. These lasers are driven and controlled independently using thermoelectric controller modules and field programmable gate array boards. A specialized Herriott gas cell is used to provide a long optical path for the laser beam to interact with the gas sample. The gas sample, containing the dissolved hydrocarbon gases, is pumped into the gas cell. The transmitted light is detected by a photodetector, and a lock-in amplifier is used to extract the second harmonic signal (2f) from the modulated light intensity. This 2f signal is related to the gas concentration and is used for further analysis. To achieve the multi-gas detection, the system uses an optical switch to time-share the laser channels. This allows each laser to sequentially measure the absorption at different wavelengths, effectively enabling the detection of multiple gases in the same gas cell. The combination of specific wavelength selection, controlled lasers, long optical path gas cell, signal processing, and optical switching enables the system to perform simultaneous detection of methane, ethyne, ethene, and ethane dissolved in the transformer oil.

Liu et al. [57] reported a six-gas NDIR gas sensor for CO, CO₂, methane, H₂CO, NH₃, and NO detection by six dual-channel IR detectors, each equipped with a specific filter that corresponds to the IR absorption peak of a particular gas.

5.1 Industrial applications

One of the key applications of NDIR sensors in methane detection is particularly in coal oil industries. Tian et al. [65] developed a new monitoring system based on NDIR spectroscopy. This system uses NDIR sensors and specialized algorithms to measure gas concentrations, particularly CO and methane, which are key indicators of coal fire hazards. The NDIR-based system meets industrial standards for monitoring these gases and can be installed close to underground coal fire risk areas, offering real-time and in situ detection. Field applications have demonstrated the high accuracy, quick analysis, and excellent safety features of the system, leading to its successful implementation in various coal mines.

NDIR camera has been used to detect and quantify methane emissions from various sources in the anaerobic digester (AD) system at the municipal wastewater treatment plant [66]. The researchers used the NDIR camera to visually identify methane-emitting point sources within the AD system. These point sources included locations such as the sludge outlet at the digester’s head, leaking manhole sealing, and cracks in the concrete structure. Additionally, the NDIR camera was used to perform quantitative measurements of methane emissions. It allowed the researchers to determine the emission rates of methane from the identified point sources. This information was crucial for calculating the total methane loss from the AD system. The NDIR camera successfully detected methane-emitting point sources within the AD system, enabling the researchers to identify locations with potential methane emissions. The NDIR camera provided data on the emission rates of methane from the identified point sources. This information was used to calculate the total methane loss from the AD system. The measurements from the NDIR camera indicated that methane emissions from the AD system were relatively low compared to literature values for other plant components. The total methane loss from the AD was calculated to be approximately 0.4% of the produced biogas. Load-dependent behavior: The NDIR camera measurements revealed that methane emissions from the AD system exhibited a strong load-dependent behavior. The emission rates correlated with the sludge retention time in the sludge shaft and the amount of displaced digested sludge.

The successful application of the NDIR gas sensor based on tunable Fabry–Pérot filter technology in an industrial setting involved its installation and integration into the formaldehyde production plant operated by Metadynea Austria GmbH [54]. The gas sensor was linked to the plant’s process control system via a dedicated microcontroller, allowing for automated and real-time monitoring of the process off-gas. The gas sensor was installed at the production site, where it could access the exhaust gas streams from the formaldehyde production process. The compact design and robust construction of the sensor made it suitable for the industrial environment, with consideration given to factors such as humidity, vibration, and exposure to chemical substances in the air. The gas sensor was capable of sequentially monitoring up to five process streams in an automated manner. This feature allowed for continuous monitoring of different production lines or reactors within the plant, providing comprehensive information on the efficiency and performance of each individual process. The gas sensor demonstrated reliable performance in the challenging industrial environment of a formaldehyde production plant. Its ability to withstand temperature variations, humidity, and other environmental factors ensured consistent and accurate measurements over extended periods of operation.

5.2 Landfill applications

Landfills produce a significant amount of methane, a potent greenhouse gas. Monitoring these emissions is essential both for environmental protection and for potential energy recovery. Mahbub et al. [67] successfully designed and developed a low-cost NDIR-based methane gas sensor using a rapidly pulsed NIR LED with flexible and on-the-fly real-time data handling capabilities. A landfill site was chosen as the location for field testing (Figure 6). They selected a specific area within the landfill where methane emissions were expected to be present. The methane gas sensor was taken to the landfill site and deployed at various sampling points within the selected area. The sensor was operated in a handheld mode, allowing the researchers to move around and take measurements at different locations. The sensor was used to measure methane concentrations at each sampling point, and the data were used to create a 3D spatial map of methane concentrations in the area. This spatial mapping provided valuable information about the distribution and variation of methane emissions at the landfill. The sensor continuously sampled methane concentrations at each location for a specific duration. The collected data included methane concentration levels and corresponding spatial coordinates. They interpreted the results to understand the methane emission patterns at the landfill. They identified areas with increased methane concentrations, which were likely associated with active decomposition of organic waste.

Figure 6

(a) 3D spatial map of the methane concentrations at ten sampling points and (b) the image of the landfill site and the sampling path’s position in it.

Hahn and Grande [68] designed the methane-monitoring system, using the MQ-4 methane gas sensor, proved to be effective and flexible for detecting methane emissions in various environments, including landfills at different depths, ducts, and other places with methane emissions. The system demonstrated reliable performance in measuring methane concentrations, and the timing of measurements was optimized to achieve accurate results. The study also found that the solenoid coil temperature remained within safe limits during operation, making the system suitable for use in potentially explosive environments.

5.3 Agricultural applications

Agricultural activities, particularly livestock farming, also produce notable quantities of methane. NDIR was one of the two methods used by Rey et al. [69] compared for measuring methane emissions in cattle. The NDIR method is a sniffer method that measures methane concentration in exhaled breath or eructated air from the cows. It draws air through a gas sampling tube from the front of a cow’s head to the analyzer, continuously measuring methane concentration in the cow’s breath. The study found that the repeatability of methane concentration measurements with the NDIR method was relatively high (0.42). Repeatability measures how consistently the same method provides similar results when repeated measurements are taken. A higher repeatability indicates that the method is more consistent and less affected by random variations. The study observed a moderately high and positive correlation (0.73) between the NDIR measurements and the hand-held laser methane detector (LMD) measurements for methane concentration. Correlation measures the degree of association between the results obtained by the two methods. A higher correlation indicates a stronger association between the measurements taken by the two methods. The agreement between NDIR and LMD was assessed using concordance correlation coefficient (CCC) and the coefficient of individual agreement (CIA). The CCC was found to be moderate (0.62), indicating that the agreement between the two methods is not perfect but still reasonable. The CIA values were moderately high (0.83), suggesting that the individual agreement between the two methods is relatively good. Overall, the NDIR method demonstrated a reasonably good performance in measuring methane concentration in cattle, showing relatively high repeatability and a moderate level of agreement with the hand-held LMD. However, the study also highlighted that the NDIR method cannot be used interchangeably with the LMD, and the sources of disagreement between the two methods need to be identified and corrected for in order to use them jointly for research purposes or mitigation strategies.

In a similar work [70], NDIR was used in two different variations: NDIR peaks and NDIR CO₂t1. NDIR peaks have been used for measuring the methane concentration in eructation peaks (burping events) of the cows during milking or feeding. The purpose of NDIR peaks was to capture the highest methane concentration during the eructation events and to assess its correlation with respiration chambers, which are considered the gold standard for measuring methane emissions. NDIR CO₂t1 has been used for detecting CO₂ as a tracer gas in combination with NDIR technology to estimate the daily methane output. The daily CO₂ output was calculated based on milk yield, live weight, and days pregnant for each cow. The purpose of NDIR CO₂t1 was to evaluate its correlation with respiration chambers and other methods. The performance of NDIR peaks and NDIR CO₂t1 was evaluated in terms of their correlation with respiration chambers, their accuracy, and precision in measuring methane emissions from individual cows. The study found that both NDIR peaks and NDIR CO₂t1 showed high correlations with respiration chambers and other methods. Specifically, NDIR peaks had a correlation of 0.89 ± 0.07, and NDIR CO₂t1 had a correlation of 0.97 with the respiration chambers. Overall, NDIR technology performed well in capturing methane concentrations during eructation peaks and estimating daily methane output using CO₂ as a tracer gas. It demonstrated good agreement with the respiration chambers, which is essential for its potential application in large-scale genetic evaluation studies of methane emissions in dairy cattle. Other evaluation studies of methane emissions in dairy cattle using NDIR can be found in previous studies [71–74].

6.1 Increased miniaturization

As technology continues to evolve, we expect to see further miniaturization of NDIR sensors. This could open up new possibilities for applications in which size and weight are crucial factors, such as wearable devices for personal safety or drones for environmental monitoring.

6.2 Integration with artificial intelligence (AI)

The integration of AI and machine learning techniques with NDIR sensors is a promising trend. This could allow for more sophisticated analysis of the sensor data, leading to more accurate detection and quantification of methane, better compensation for environmental factors, and potentially even the identification of different methane sources based on their unique “signatures.”

6.3 Improvements in power efficiency

Future developments in materials science and electronics are expected to result in NDIR sensors that are even more power efficient. This could extend the battery life of portable devices or enable the use of energy-harvesting technologies, such as solar power, to provide a virtually unlimited operating life for stationary sensors.

6.4 New materials and designs

Future research and development are likely to uncover new materials and designs for NDIR sensors that offer even better performance. This could include materials that provide higher reflectivity for the optical cavity, more efficient IR sources, and more sensitive detectors, as well as innovative designs that enhance the sensitivity and selectivity of the sensor.