Integrated near-infrared QEPAS sensor based on a 28 kHz quartz tuning fork for online monitoring of CO2 in the greenhouse.

In this paper, a highly sensitive and integrated near-infrared CO2 sensor was developed based on quartz-enhanced photoacoustic spectroscopy (QEPAS). Unlike traditional QEPAS, a novel pilot line manufactured quartz tuning fork (QTF) with a resonance frequency f 0 of 28 kHz was employed as an acoustic wave transducer. A near-infrared DFB laser diode emitting at 2004 nm was employed as the excitation light source for CO2 detection. An integrated near-infrared QEPAS module was designed and manufactured. The QTF, acoustic micro resonator (AmR), gas cell, and laser fiber are integrated, resulting in a super compact acoustic detection module (ADM). Compared to a traditional 32 kHz QTF, the QEPAS signal amplitude increased by > 2 times by the integrated QEPAS module based on a 28 kHz QTF. At atmospheric pressure, a 5.4 ppm detection limit at a CO2 absorption line of 4991.25 cm-1 was achieved with an integration time of 1 s. The long-term performance and stability of the CO2 sensor system were investigated using Allan variance analysis. Finally, the minimum detection limit (MDL) was improved to 0.7 ppm when the integration time was 125 s. A portable CO2 sensor system based on QEPAS was developed for 24 h continuous monitoring of CO2 in the greenhouse located in Guangzhou city. The CO2 concentration variations were clearly observed during day and night. Photosynthesis and respiration plants can be further researched by the portable CO2 sensor system.

Introduction

Carbon dioxide (CO2) gas is an important component of the ambient air and greenhouse gas in the atmosphere. The emission sources of CO2 include industrial waste gas, photosynthesis of plants, and automobile exhaust [1], [2], [3], [4]. Highly sensitive detection of CO2 concentration is of great significance in public health care, agricultural greenhouse monitoring, chemical plants hazardous detection, air pollutant monitoring, industrial process control, medical diagnosis, etc [5]. In the field of public health care, the detection of CO2 can effectively evaluate the air quality level in public places [6]. Real-time monitoring of CO2 concentration in agricultural greenhouses can improve the photosynthesis of fruits and vegetables by increasing the CO2 content in time [7]. Monitoring the concentration of CO2 in the atmosphere is an important way to analyze the greenhouse effect and haze phenomenon [8]. Therefore, it is necessary to develop highly sensitive CO2 gas sensors.

In recent years, several CO2 detection technologies have been extensively developed. The electrochemical sensor uses CO2 gas to generate an oxidation-reduction reaction at the electrode of the electrolyte to obtain the CO2 concentrations by measuring current [9]. These electrochemical CO2 sensors have poor gas selectivity and short service life [10]. The solid electrolyte CO2 sensor has attracted the attention of researchers because of its good sensitivity and the characteristics of being less affected by temperature [11], [12], [13]. The CO2 sensors mentioned above have the shortcomings of relatively poor detection sensitivity, susceptibility to environmental noise, and slow response time, compared with the laser sensing technologies. Laser absorption spectroscopy (LAS) has the advantages of fast response, online monitoring, non-invasive, highly sensitive and selective detection, etc [14], [15]. Tunable diode laser absorption spectroscopy (TDLAS) and photoacoustic spectroscopy (PAS) have been used for CO2 gas detection. These sensors usually reach the detection limit of ppm CO2 concentration level [16], [17]. Photoacoustic spectroscopy (PAS) technology converts the absorption of light energy into sound energy and performs measurement [18]. The intensity of sound waves can be obtained through a microphone to invert the gas concentration. However, most microphone-based PAS cells usually have a low resonance frequency (<5 kHz), which makes such cells sensitive to environmental noises and sample gas flow noise.

An innovation of microphone-based PAS is quartz-enhanced photoacoustic spectroscopy (QEPAS) which was firstly reported in 2002 by Kosterev [19]. The novelty of this technology lies in the use of industrial mass-produced quartz tuning fork (QTF) as the acoustic wave transducer, replacing the miniature microphone in the traditional PAS, and realizing highly sensitive detection of weak photoacoustic signals [20], [21], [22]. QEPAS has become a research hotspot in recent years due to its immunity to environmental noise, high detection sensitivity, small size, wide dynamic range, and QTF operation insensitive to the excitation wavelength [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35]. In recent years, researchers have used QEPAS technology to detect and analyze CO2 gas concentration. In 2014, Pietro Patimisco et al. proposed a gas detection technology called intra-cavity QEPAS (I-QEPAS), which can obtain a detection limit of 300 ppt with an integration time of 20 s for CO2 gas [36]. Zheng et al. developed a double pass QEPAS sensor for atmospheric CO2 detection in 2015 [8]. A reflective concave mirror was used to enhance the absorption optical path to improve detection sensitivity, and consequently achieved a detection limit of 1.74 ppm with an integration time of 247 s. Recently, Maxime Duquesnoy et al. designed and fabricated a low-frequency QTF to adapt to gas molecules with slow relaxation rates and obtained a minimum detection limit of CO2 gas of 44 ppm with an integration time of 1140 s [37]. Therefore, the gas sensor based on QEPAS is an excellent alternative for trace gases detection due to its merits of immunity to environmental noise, fast response, highly sensitive [38].

A typical PAS acoustic detection module (ADM) usually consists of a microphone and a gas cell. The acoustic resonators operating on longitudinal, azimuthal, and radial resonances are designed and manufactured in the internal structure of a gas cell, as shown in the Fig. 2 of ref [39]. However, in QEPAS acoustic detection module (ADM), the gas cell and acoustic micro resonator (AmR) are separated, as shown in Fig. 1(a). The separated AmR and gas cell results in a more complicated and unstable structure than traditional PAS ADM.

Fig. 2

(a) Schematic diagram of CO2 gas sensor system based on QEPAS (b) Top view of the ADM (c) Side view of the ADM.

Fig. 1

(a) Diagram of traditional QEPAS ADM. (b) Diagram of Integrated QEPAS ADM in this work, only optical part. (c) Front view of integrated ADM. (d) Back view of integrated ADM.

In this paper, we designed and demonstrated an integrated near-infrared QEPAS ADM based on a 28 kHz QTF. A pilot line manufactured custom 28 kHz QTF was employed as the photoacoustic transducer. The details of the 28 kHz QTF have been reported in the reference [35]. An integrated QEPAS ADM was designed and developed first. The laser fiber, QTF, resonator, gas cell, and preamplifier were integrated to form a compact and robust detection module, as shown in Fig. 1(b). A series of coaxial holes were drilled on the aluminum gas cell to integrate the fiber focuser, acoustic resonator, and buffer volume. The real pictures of the integrated ADM were shown in Fig. 1(c) and (d). The integrated ADM has an optimum on-beam AmR configuration in the inner architecture. No separated AmR was required to be assembled with the QTF. A compact QEPAS instrument based on the integrated ADM was developed for trace CO2 gas detection. A near-infrared distributed feedback (DFB) laser diode emitting at 2.004 µm was employed as the excitation source. The instrument sensitivity, linear response and the long-term stability were evaluated in detail. Reliability and stability of the CO2 QEPAS instrument were proved by the 24 h continuous monitoring of CO2 concentrations inside the greenhouse located in Guangzhou city, China.

Experiment setup

The experimental setup of the CO2 gas sensor system based on QEPAS is shown in Fig. 2(a). A near-infrared DFB laser diode (NTT Electronics) wavelength tuning range measured from 2.002 µm to 2.006 µm was employed as the excitation source. The output laser beam from the DFB laser was divided into two parts, 99% of which was used as the excitation light source for detection, and the remaining 1% was used for the reference cell which was filled with pure CO2 sample. 99% of the laser was collimated and focused by an optical fiber focuser (OZ Optics). The focal length was ~11 mm, and the focused spot diameter was about 100 µm. The laser beam was focused through the AmR and QTF prong spacing in the photoacoustic detection module (ADM). Two cylindrical “holes” were drilled on both sides of the ADM wall perpendicular to the QTF prongs, thus resulting in acoustic resonance inside the “holes” as AmR. According to the one-dimensional pipe resonator theory, the optimum hole length L should be in the range of λ/4 40], [41], for on-beam QEPAS configuration, the optimal length L and inner diameter (ID) were 5.4 mm and 0.6 mm respectively. The 3D model of the integrated QEPAS ADM by Solidworks software was shown in Fig. 2(b) and (c). Modulation and tuning of the laser current were controlled by applying a sinusoidal dither to the direct current ramp. The sinusoidal dither was set at the half of the QTF resonance frequency. The drive and temperature control of the laser were accomplished by the custom control electronic unit (CEU). The piezoelectric signal generated by the QTF was detected by a low noise transimpedance amplifier (TA) with a 10 MΩ feedback resistor and converted into a voltage signal. The piezoelectric signals amplified by the preamplifier were transmitted to the lock-in amplifier (Stanford Research System, SR830 DSP) for second harmonic demodulation. The laser from the reference cell was detected by a photodiode and then introduced to CEU with the third harmonic setting. The 3 f signal was used as an error signal of proportion integration differentiation (PID) module to lock the laser wavelength to the selected CO2 absorption line. The demodulated signals were recorded and analyzed by a PC equipped with a data acquisition card (DAQ). The whole experiment was controlled by the automatic program written by LabView. The gas stream from two cylinders was fed to the mass flow controller (MFC) system to dilute 1000 ppm CO2 in nitrogen (N2). The diluted CO2: N2 mixture gas was controlled at the flow rate of 100 ml/min to flush the QEPAS ADM.

Experimental results

Frequency response of QTF

The frequency response of the bare QTF and the integrated QEPAS ADM is shown in Fig. 3. A function generator (Tektronix AFG3102) was used to provide a sinusoidal signal. The frequency of the sinusoidal signal was scanned from 27970 Hz to 28010 Hz with the step of 0.01 Hz. The peak-to-peak amplitude of the sinusoidal signal was set to 400 mV. The QTF output signal was demodulated by a lock-in amplifier (Stanford Research System, SR830 DSP) in 1 f mode. Fig. 3 shows that the resonance frequency and Q factor of bare QTF were 27987.4 Hz and 6802 respectively. This is due to the influence of air damping. The obtained resonance frequency and the Q factor of the integrated QEPAS ADM were 27986.5 Hz and 2809 respectively. The frequency slightly shift < 1 Hz, however the Q factor decrease by 3993. The huge decrease in Q factor can be attributed to the coupling energy transfer between the QTF and the integrated ADM. The decrease of Q factor indicates that more energy is stored in the AmR of the integrated ADM.

Fig. 3

Frequency response curve of custom 28 kHz QTF.

Influence of humidifier on the QEPAS

For gas molecules with slow relaxation rates such as CO2, the amplitude and modulation frequency of the QEPAS signal largely depends on the vibration-translation (V-T) relaxation rate. Adding water vapor to the gas mixture can effectively promote the energy transfer of the excited CO2 molecules in the V-T state [24]. A silicone hollow fiber membrane humidifier was added to the experimental gas circuit to humidify the CO2 gas. The absolute humidity at room temperature was increased from ~2700 ppm to ~9200 ppm, which was determined by the performance of the hollow fiber membrane. According to the HITRAN database [42], the H2O interference-free CO2 absorption line located at 4991.25 cm−1 with the intensity of 1.292 × 10−21 cm/molecule was selected as the target absorption line for detection. The temperature of the laser was controlled at 17.3 ℃, the injection current was scanned from 120 mA to 140 mA, and the integration time of the lock-in amplifier was 1 s. The whole experiment was carried out under atmospheric pressure. Fig. 4 depicts the second harmonic QEPAS signal obtained by the CO2 sensor system under dry and wet conditions respectively. The absolute humidity of the wet gas was limited by the performance of the used hollow fiber membrane. The QEPAS signal amplitude of the CO2 sensor system was 9.31 × 10−5 V under dry condition, and the QEPAS signal amplitude increased to 1.14 × 10−4 V after humidification. As a result, the humidifier increases the QEPAS signal amplitude by 22.8% by promoting the relaxation rate of CO2 molecules.

Fig. 4

The 2 f QEPAS signals under dry and wet conditions.

Sensor evaluation

The amplitude of the QEPAS signal S is expressed by the following formula [43]:where K is the sensor constant, P is the laser power, Q is the quality factor of QTF, α is the peak intensity of the 2 f absorption spectrum and ε is the radiation-to-sound conversion efficiency, given by [44]:where f is the modulation frequency and τ is the relaxation time of the target gas. According to formulas (1) and (2), the amplitude of the QEPAS signal S is inversely proportional to the frequency f. Therefore, a QTF with a low resonance frequency f is conducive to obtaining large QEPAS signal. In this research, a QTF with a f0 of 28 kHz was used. The advantages of low-frequency QTF were verified by comparing the QEPAS signal amplitude of a commercially available standard QTF with a frequency of 32.768 kHz and a custom pilot line manufactured QTF with a frequency of 28 kHz. Fig. 5 shows the results of measuring the QEPAS signal amplitude of 1000 ppm CO2 with two QTFs with different frequencies under the same experimental conditions. It is found that the QEPAS signal amplitude was 3.81 × 10−5 V when using a commercial QTF with a frequency of 32.768 kHz, and the QEPAS signal amplitude was 1.13 × 10−4 V when a custom QTF with a f0 of 28 kHz was used. It obvious that compared to a QTF with a f0 of 32.768 kHz, the QEPAS signal amplitude increased by > 2 times when a QTF with a f0 of 28 kHz was used. As a result, custom 28 kHz QTF is proved to show a better sensitivity for CO2 detection than commercial 32.768 kHz QTF. The advantage of lower modulation frequency is obvious.

Fig. 5

The QEPAS 2 f signals for 32 kHz QTF and 28 kHz QTF.

According to the gas absorption theory of photoacoustic spectroscopy, there is a good linear relationship between photoacoustic spectroscopy signal and concentration [45]. In order to verify the linear concentration response of the CO2 sensor system based on integrated QEPAS module, QEPAS signals of different CO2 concentrations were obtained. A 1000 ppm CO2: N2 mixture gas was diluted with 99.999% pure N2 to obtained different CO2 concentrations. At a standard atmospheric pressure, the gas stream was controlled at a flow rate of 100 ml/min. The CO2 sensor system based on QEPAS ran in wavelength locking mode. The laser wavelength was locked the 4991.255 cm−1 CO2 absorption line by PID. Fig. 6(a) shows the QEPAS signal amplitude of different concentrations of CO2. Each concentration of CO2 was measured by 60 data points. According to Fig. 6, QEPAS signal increased with the increment of CO2 concentrations. The noise of the sensor system was measured by filling the gas cell with pure N2. The obtained noise level was 4.55 × 10−6 V with a 1σ standard deviation of 6.31 × 10−7 V. This non-zero signal may come from the bottleneck at the QTF and the scattered light induced noise. Considering that the average QEPAS signal amplitude of 1000 ppm was 1.17 × 10−4 V, the detection SNR was calculated to be ~185.1. Therefore, the detection limit of CO2 is 5.4 ppm with an integration time of 1 s. The average value of QEPAS signal amplitudes at different concentration levels as a function of CO2 concentration is shown in Fig. 6(b). The R-Square value obtained by linear fitting was 0.999, which proves that the CO2 sensor system based on QEPAS has an excellent linear response to gas concentration.

Fig. 6

(a) QEPAS signal amplitude of different CO2concentrations. (b) Linear fitting for different CO2 concentrations.

To assess the long-term stability of the CO2 sensor system based on the integrated QEPAS module, the amplitude of the 1000 ppm CO2 signal was continuously measured in the wavelength locking mode, and the Allan variance analysis was performed. The 1000 ppm CO2 standard gas was flushed into the gas cell with a flow rate of 100 ml/min and measured continuously for 30 min. Allan variance was performed on these data, and the results were shown in Fig. 7, which implies that a minimum detection limit (MDL) of 0.7 ppm was obtained with an optimum integration time of 125 s. Before 125 s, Johnson (thermal) noise was the dominant noise source, and after that, the system drifts started to dominate.

Fig. 7

Allan variance analysis for the CO2 sensor system based on QEPAS.

The basic elements of plant growth are CO2 and H2O, which photosynthesize carbohydrates and release oxygen [46]. Greenhouse system is regarded as an innovation of modern agriculture, which provides good environmental conditions for crops. Greenhouse is a relatively closed environment that the concentration of CO2 in the greenhouse and the outside environment will have a certain difference. The CO2 concentrations in the greenhouse will directly affect the growth of crops. The enrichment of CO2 reduces the inhibition of photosynthesis by oxygen and increases the net photosynthesis of plants, which is the basis for the increase in the growth rate of crops caused by carbon dioxide under low and high light conditions [47]. At the same time, the increased carbon dioxide concentration will also increase the optimal temperature for growth [48]. CO2 enrichment approaches in the greenhouse usually include composting [49], chemical reaction [50] and greenhouse ventilation [51], etc. In addition, in the intelligent organic greenhouse, CO2 concentration should be accurately controlled to optimize the fertilization and photosynthesis. Traditional non-dispersive infrared analyzers became inefficient in many greenhouses, this is due to the cross-talking effect of H2O and CO2 absorption caused by high humidity.

A CO2 sensor system based on the integrated QEPAS module was used to continuously measure CO2 concentration in a semi-enclosed greenhouse located in Guangzhou city. The CO2 sensor was located 268 m away from S4 Huanan Express and 336 m away from Huangpu Avenue, which are two main streets of Guangzhou city, shown in Fig. 8(a). For measurements, an inlet Teflon tube was placed next to crops and connected to a drying tube, shown in Fig. 8(b). The air in the greenhouse was pumped into the CO2 sensor system and fed to the integrated QEPAS module by a 6 m Teflon tube. Considering a maximum pump flow of 320 ml/min, the time delay for the sensor system was estimated to be ~ 14 s. A needle valve and a mass flow controller were used to control a 100 ml/min flow into the ADM to avoid possible gas flow noise. The CO2 sensor system with a dimension of 45 × 34 × 18 cm3, and a weight of ~10 kg, shown in Fig. 8(c). The results of continuous measurements of CO2 concentration for 24 h are shown in Fig. 8(d). The humidity in the greenhouse changed from 63.1%RH to 97.6%RH between day and night. The CO2 concentration dropped from 592 ppm to 269 ppm at 05:39–12:02. This is due to the vigorous photosynthesis of crops in the daytime. After 14:50, the CO2 starts to increase. Photosynthesis stops at night, crops respiration releases CO2, and the concentration of CO2 in the greenhouse gradually increases. According to our measurements, the CO2 in the greenhouse changed between 256 ppm and 608 ppm in 24 h.

Fig. 8

(a) Location of the sensor. (b) The photograph of the greenhouse. (c) The picture of the CO2 sensor system (d) The real-time concentration of CO2 in the greenhouse in 24 h.

Conclusions

An integrated near-infrared QEPAS sensor based on a 28 kHz QTF was demonstrated in this paper. A DFB laser diode with a center wavelength of 2.004 µm was employed as the excitation light source for CO2 detection. A pilot line manufactured QTF with a low resonance frequency f0 of 28 kHz was used as an acoustic wave transducer to improve the photoacoustic signal. Two cylindrical “holes” were drilled on both sides of the ADM wall perpendicularly to the QTF prongs, thus resulting in on-beam AmR QEPAS configuration. A super compact and integrated QEPAS module was developed consisting of QTF, acoustic micro resonator (AmR), gas cell, and laser fiber. The frequency of the integrated ADM was 27,986.5 Hz and the Q factor was 2809, which means that there was a good coupling resonance effect between the QTF and the ADM. A humidifier was added to the gas stream to promote the molecular relaxation rate of CO2 to increase the QEPAS signal amplitude. By comparing the QEPAS signal amplitude of 32 kHz QTF with that of 28 kHz QTF, it was obtained that the QEPAS signal amplitude of 28 kHz QTF was 160% higher than that of 32 kHz QTF. It is verified that 28 kHz QTF was more beneficial to improve the QEPAS signal. With the wavelength locking mode, the QEPAS signal amplitude of different concentrations of CO2 gas was continuously measured to calibrate the QEPAS sensor system. The functional relationship between the QEPAS signal amplitude and CO2 concentration was obtained, corresponding to a linear correlation coefficient of 0.999. A minimum detection limit (MDL) for CO2 is 5.4 ppm with an integration time of 1 s. With an integration time of 125 s, the MDL can be improved to be 0.7 ppm, indicating that the CO2 sensor system is capable of good stability. The developed integrated QEPAS sensor realized the continuous and accurate measurement of CO2 concentration in the greenhouse located in Guangzhou city. The high sensitivity of the integrated QEPAS sensor can meet the requirements for CO2 gas detection in applications such as atmospheric monitoring, industrial production, and medical diagnosis.

Declaration of Competing Interest

The authors declare that there are no conflicts of interest.