Yanfang Li， Yuejin Zhao， Tingting Zhang， Jie Hu， Yubin Wei and Tongyu Liu

(1.School of Optoelectronics, Beijing Institute of Technology, Beijing 100081, China;2.Shandong Key Laboratory of Optical Fiber Sensing Technologies, Jinan 250102, China)

Abstract: A methane sensor system was designed based on tunable diode laser absorption spectroscopy (TDLAS) technology and the feature of vertical cavity surface emitting laser (VCSEL) with wavelengths up to several nanometers. We studied the gas present in the methane sensor’s common operation environment. Through absorption spectrum stimulation and experiments,we analyzed the cross interference of the gas of H2O, CO and CO2 to methane detection. The results prove that this laser methane sensor has the characteristics of high stability and anti-cross interference.

Key words: tunable diode laser absorption spectroscopy(TDLAS); vertical cavity surface emitting laser (VCSEL)； methane sensor;cross interference

Methane (CH4) is a colorless, odorless gas, which is lighter than air. It is the principal component of natural gas.It widely exists in our life and it is one of the main components of greenhouse gases. Methane is an explosive and acts as asphyxiate, it can therefore cause serious harm to human’s health. Methane detection is required in the areas like coal mines, power plant, waste water treatment and petroleum chemical industry[1]. With the development of optical fiber and semiconductor laser technology, tunable diode laser absorption spectroscopy (TDLAS) has become an important technology in the field of gas detection.

The TDLAS technology was first developed in the 1970s by Hinkley and Reid[2]. With the continuous development and improvement of the technology, more and more gases can be detected, such as O2and H2O whose contents are high in the atmosphere, and atmospheric pollution gases of CO, CO2, SO2, NOxand CH4[3-6]. And the detection sensitivity is becoming more and more high. With the rapid development of optoelectronic technology and optical communication technology in the past decade, the tunable semiconductor laser has small size, long life and high power, which further promote the TDLAS technology use in industrial, environmental, and medical science fields. The gas detection technologies are developing towards multi-component, system miniaturization, and open optical path directions[7-10].Generally speaking，TDLAS includes direct absorption method and modulation method. To reduce the system’s complexity and costs, we choose the direct absorption method and used a saw tooth current to scan the absorption peak in our system.

In this paper, we designed a laser methane sensor based on the Lambert-Beer law and by detecting the optical power change of a VCSEL laser source before and after the gas absorption.By introducing a reference cell, we realized the output wavelength of the laser to track the gas absorption line automatically. This senor is commonly used in coal mines, chemical plants and other places whereother gases, such as H2O, CO and CO2, also exist. If there are other interfering gases in the measurement environment, a traditional sensor might have large errors, whichcan cause great inconvenience to production, wrong operation and economic loss. A gas sensor based on TDLAS could identify different gases, and achieve a trace detection of a specific gas. We did cross interference experiments to check the anti-cross interference of our laser methane sensor. And the results show that our sensors can resist cross interference.

1 Measurement Principle and System Design

1.1 Measurement principle

Based on the Beer-Lambert law, when a monochromatic light with wavelengthλgoes through the sample gas, the intensity of the transmitted lightI(λ) and original lightI0(λ) meets the following relationship as

I(λ)=I0(λ)exp [-α(λ)CL]=
I0(λ)exp [-PS(T)φ(λ)CL]

(1)

whereα(λ) is the gas absorption coefficient;Cis the volume concentration;Lis the length of the absorption path (cm);S(T) is the intensity of gas characteristic spectral line that indicates the intensity of the absorption and is only a function of temperature;Pis the pressure of the sample gas;φ(λ) is the line profile function related to the temperature, the pressure and the contents of the gas.

By doing logarithm calculation on both sides of Eq.(1) followed by integration, we have

(2)

Then we can obtain the concentration calculation formula as

(3)

According to Eq.(3), ifP,S(T) andLare all known, we can obtain the concentration if we put the integral values in the frequency domain of -ln (I/I0) into Eq.(3). Instead of directly integrating the spectrum absorption signal, we usually find a proper fitting curve to obtain the value of -ln(I/I0). In the actual design of the sensors, firstly we assumePandS(T) are all fixed constants. After the manufacture of the sensor, the length of optic pathLis fixed, so we can easily get the gas concentration by detect the power change before and after gas absorption.

1.2 System design

1.2.1System structure

In TDLAS system design, the first and most important step is to determine the absorption spectrum of the system according to the HITRANdatabase. At room temperature, the absorption spectrum of methane is shown in Fig.1. After the analyses of the spectrum factors such as the absorption intensity of methane and the deviation from center wavelength, we choose 1 650.5 nm as the absorption line. Then we began to design our methane sensor system.

Fig.1 Absorption line of methane

The system structure was shown in Fig.2a. In our system, we drive the light source and the temperature/pressure detection with CPU. Then the light comes to photoelectric conversion and signals processing after optical path. The electric signals from detector are transmitted to AD conversion, then to CPU. After data processing, we could get the methane concentration, and then the frequency signal corresponding to concentration, RS485, alarm signal and LCD display were output. In the data processing, we introduce consumption of temperature and pressure to improve the veracity and reliability of measurements.

Optical part of the system mainly includes laser, coupler, detectors and probe. To reduce the cost of this sensor, we select detectors in communication band—InGaAs detector to detect the light signal before and after the absorption, the response wavelength range of this detector is 600 nm to 1 700 nm, and the responsiveness in the 1 550 nm is 0.85 mA/mW. After the conversion voltage value and stability measurement of these PD,we confirm that the detector can be used in 1 650 nm band.The sensor has a gas cell, in macro analysis. It’s effective optical pathlength is about 6 cm, in trace analysis, it’s effective optical path length is about 20 m. In addition, we introduce a wavelength reference probe, through which the absorption wavelength could be easily find. Depending on this, we could make sure that the absorption line is in the wavelength scanning range by modulating the drive current.

Combining our process design, a methane sensor was manufactured as Fig.2b. The operating temperature of this sensor is 0-40 ℃. It outputs the methane concentration signal with standard serial port (RS485).The physical diagram of the sensor is as follows. Its machine dimension is about 90 mm×70 mm×20 mm.

Fig.2 Structure of the system and the sensor photo

1.2.2Sensor calibration and verification

Fig.3 Display value

Per the calibration method given in the coal mine safety standard, we use standard sample gas to calibrate our sensors. In this process, we make sure that the response time T90 is not greater than 10 s. After the calibration, to check the performance at 0 ℃ and 40 ℃， we inlet standard gases with volume fraction of methane to N2being 1.49%,3.47%,8.48%,20% and 59.4% respectively. Fig.3 shows the display value results of one sensor at 0 ℃ and 40 ℃ respectively. Compared with the standard gas concentration, the relative errors are all less than 3%. This error is far below the requirements of coal mine safety standards (less than 6%). In all our experiments, the relative expanded uncertainty of standard gas is about 2.0%.

2 Experiments of Cross Interference

To further verify the anti-cross interference of the laser methane under normal temperature and pressure, we select H2O, CO and CO2as the interference gases which generally exist in coal mine environment.

2.1 Analysis of cross spectrum

(4)

Fig.5 Absorption line simulation

Based on HITRAN database, there are cross interference gases nearby 1 650.9 nm. According to Eq.(4), we assume the absorption length of the gas cell is 100 cm, the volume fraction of CH4, H2O, CO and CO2to N2are 0.1%,15%,0.1%, and 10% respectively. The simulation results is shown in Fig.4.

Fig.4 Absorption line simulation

From Fig.4, in the macro measurement (the methane measurement error is greater than 0.06%)， the absorption intensity of cross gas and the target gas is more than 5 orders of magnitude. We can ignore the effects of cross gases directly. In experiment I, to verify anti-cross interference of this laser methane sensors, we selected 3 methane sensors (1#,2#and 3#) randomly as our sample prototype. Then we inlet H2O, CO and CO2to the gas cell and record value data.The result are as follows.

In the trace gas measurement, we couldn’t ignore the influence of crossed gas. Taking CO2as an example, we studied cross interference data processing techniques of this sensor. Fig.5 shows the simulation results with the absorption length of the gas cell being 20 m, the volume fraction of CH4and CO2to N2being 0.000 2% and 0.1% respectively.

From Fig.5, we couldn’t ignore the influence of CO2for trace methane detection. In our data progress, we used correlation function method to reduce the impact of CO2. If we assume the real concentration of CH4and CO2areC1andC2respectively, and the absorption peak of CH4and CO2areP1andP2. At these two peak, the CH4absorption intensity isα1(self-intensity) andα12(interference intensity). CO2absorption intensity isα21(interference intensity) andα2(self-intensity). Then the total absorption intensity of these two peaks satisfied

which can be written in matrix form as

C=A-1Ap

(5)

Through the simulation or standard gas calibration, we could get the value self-absorption intensity ofα1andα2and interference intensity ofα12andα21. Once we know the value of every element of matrix, we could get the inverse matrix of A. Then according to Eq.(5), we could get CH4real concentrationC1.

2.2 Experiment I—macro measurement cross interference

2.2.1Cross interference of H2O

Because of the effect of groundwater and spray water on the working surface in coal mine field, the air humidity is high and can be up to 90%-100%, and the air intake lane is very wet even in winter. Combined with the above investigation and laboratory testing conditions, in the preliminary test experiment, the water vapor that is 37 ℃ and has relative humidity of 95%, is set as the initial test sample gas.We began our experiment by resetting the sensor value as 0. After the instrument is stable, the sample water vapor was introduced and 3 instrument display values were recorded at 30 s, 60 s and 120 s. The 3 values were averaged as the final test concentration of the instrument to calculate the interference errors of this sample instrument.The results are shown in Tab.1.

Tab.1 Interference test results of RH 95% H2O on methane

2.2.2Cross interference of CO

The source of coal mine carbon monoxide mainly include: the gas and coal dustexplosion, coal oxidation, direct fire suppression with water and blasting operation of working face. “Coal mine safety regulations” in the provisions of the maximum allowable volume fraction of CO in the air is not more than 0.002 4%. And the range of CO volume fraction alarm used in coal mine environment is generally 0-0.1%. Based on the above investigation and existing standard gas in our lab, the volume fraction of carbon monoxide to N2is selected to be 0.1% in our experiment.We began our experiment at sensor value of 0, after the instrument is stable, 0.1% CO is introduced, and 3 instruments display values were recorded at 30 s, 60 s and 120 s. The values were averaged as the final test concentration of the instrument to calculate the interference errors of this sample instrument.The result is shown in Tab.2.

2.2.3Cross interference of CO2

“Coal mine safety regulations” in the provisions of the maximum allowable concentration of CO2in the air is 0.5%. And the range of CO2concentration alarm used in coal mine is generally 0-5%. In this experiment, 10% CO2is selected as interference gas. The results are as follows.

Tab.2 Interference test results of 1×10-3 CO on methane

The main source of CO2in coal mine mainly include: pit prop become bad, slowly oxidation of coal and carbon, and so on. In addition, human breath, explosion of underground gas and coal dust, fire accidents also produce large amounts of CO2. “Coal mine safety regulations” in the provisions of the maximum allowable concentration of CO2in the air is 0.5%. And the range of CO2concentration alarm used in coal mine environment is generally 0-5%. Combined with the above investigation and existing standard gas in our lab, the initial concentration of CO2is selected as 10%. We began our experiment at the sensor shows the value of 0, after the instrument stable, pump the 10% CO2to the gas cell, and then in the 30 s, 60 s and 120 s, we recorded 3 instrument display values as the stable display value and then averaged them as the final test concentration of the instrument to calculate the interference errors of this sample instrument. The result is given in Tab.3.

2.3 Experiment II—trace measurement cross interference

In this experiment, the effective absorption length is 20 m and the measuring volume fraction range of the sensor is 0-0.003 0. We choose 0.04%， 0.1% and 1% CO2as the interference gas. Before pumping the distributed gas for measurement, we blew the gas cell with high purity nitrogen gas (99.999%) for 3 min. The volume fraction of CH4to N2are 0.000 2%, 0.001%, and 0.003% respectively. After 30 s of stabilization, we recorded the display value. Tab.4 gives the test result. This result show that our data processing method is effective in eliminating cross interference.

Tab.3 Interference test results of 10%CO2on methane

Tab.4 Display value before and after compensation 10-6 vol/vol

3 Conclusion

The experimental results show that our laser methane sensor had high stability and good anti-interference performance to H2O, CO or CO2. To improve the performance further, we would improve the concentration of interference gas in future studies. For some coal mine, C2H2, H2S and other cross interference gases also need to be further verified. In addition, in our experiment, we found that measurement value of the sensor in the hydrogen environment had a large variation, so algorithm optimization to improve the sensor performance should also be studied in the future.