Trace Hydrogen Sulphide Gas Sensor Based on Cu/rGO Membrane-Coated Photonic Crystal Fibre Michelson Interferometer

3.1 Sensor Fabrication

The standard SMF and EPCF fibres were bought from Yangtze Optical Fibre and Cable Joint Stock Limited Company (YOFC, Wuhan, China). The cladding diameter of EPCF is 125 μm, with multilayer air holes of 9.5 μm in diameter, arranged in a hexagonal structure with a length of 31.5 mm. The optical fibre fusion machine is S178C type, Furukawa, Japan. The light source is ASE broadband light source (Kangguan, Beijing, China). The annulus is PIOC315P210 type (Xianlian Co., Ltd., Sichuan, China), and the spectrum analyser is used by AQ6370D spectrometer (Yokogawa, Tokyo, Japan). In the preparation process of sensing structure, the manual splicing procedure is adopted, and the parameters are set in Table 1. At this time, the ends between SMF and EPCF are spliced to form a shrunk taper, and the other end of EPCF is spliced with another SMF and connected with an FRM. Thus, the Michelson interferometer is obtained and illustrated in the inset of Figure 1.

Figure 2:

The SEM morphology of sensing film: (a) SEM image of the thickness of cross section and (b) SEM image of the side surface.

3.2 Preparation and Characterisation of Sensing Membrane

The 3.2 g/L nano-copper with diameter of 5–10 nm and GO aqueous dispersion solutions were bought from Nanjing Xianfeng Nano Material Technology Co., Ltd., Nanjing, China. The Cu/rGO-sensing film is prepared by the dip-coating method. First, 500 μL graphene oxide (GO) aqueous dispersion solution is mixed in 300 μL isopropanol and 200 μL 30 % hydrogen peroxide solution and then dispersed uniformly by ultrasound for 20 min to obtain GO mixed solution. Second, 700 μL nano-copper aqueous dispersion is added to 300 μL isopropanol for ultrasonic dispersion. The EPCF is, respectively immersed in GO and nano-copper solution for 10 min, and then it is dried by the vacuum freeze dryer, and the above dip-coating process is repeated alternately. Lastly, the coated fibre is placed in a tubular furnace, which is accessed by nitrogen and calcined at 300 °C for 2 h, and then the Cu/rGO membrane–coated EPCF is obtained. The scanning electron microscopy (SEM) results (Fig. 2) show that the film thickness is about 24.56 nm, and the film surface is dense and smooth. To further determine the element composition of the sensing film, X-ray photoelectron spectroscopy (XPS) and Raman spectroscopy are used.

Figure 3 is the XPS of Cu/rGO composite membrane. Figure 3a is the full spectrum scanning of elements in the composite. The composite mainly contains three elements: C, O, and Cu. Figure 3b is a narrowband scan image of C1s after peak splitting. The binding energies corresponding to C–C/C=C, C–O, C=O, and O–C=O are 284.3, 285.7, 287.1, and 288.4 eV, respectively. Figure 3c is a narrowband scanning image of Cu2p after peak splitting. The binding energies of copper Cu⁺ 2P3/2, Cu²⁺ 2P3/2, Cu⁺ 2P1/2, and CuO 2P1/2 are 932.6, 934.6, 952.4/954.1, and 963 eV, respectively. These are in good agreement with the XPS results of the known Cu/rGO composites [21], [28], [29], [30]. It can be seen that nano-coppers are partially oxidised to CuO. In addition, the photoionisation cross section of H element has a low content, and the signal is too weak. Thus, the energy spectrum of H element cannot be displayed in XPS. Figure 4 is the Raman spectrum of the membrane. The intensity ratio of D to G peak is 1.053, which is consistent with the Raman peaks reported in literature [26], proving that GO is reduced to rGO.

Figure 3:

(a) XPS image of CuO/rGO membrane and the corresponding high-resolution XPS scans of (b) C 1s and (c) Cu 2p.

Figure 4:

Raman spectrum of Cu/rGO membrane.

3.3 Gas-Sensing Properties

Figure 1 is a schematic diagram of a single tapered Michelson interferometric hydrogen sulphide sensor based on the Cu/rGO-sensing film. H₂S gases 0–70 ppm are introduced into the gas chamber. Ambient air is used here in this experiment for gas dilution. The interference spectra obtained are shown in Figure 5. In the range of 0–70 ppm H₂S gas concentration, the interference spectra of the sensor show a significant red shift with the increase of H₂S gas concentration. The reason is that when the H₂S gas molecules contact with Cu/rGO-sensing film, the refractive index of the film will increase because of the adsorption of H₂S gas. It can be seen from formula (3) that the interference spectra of the sensor will be red-shifted with the increase of H₂S gas concentration. The experimental results are in good agreement with the theoretical analysis. By fitting the spectral offset, the relationship between the spectral offset and the measured gas concentration is obtained. The linearity of the sensor is 0.97662, and the sensitivity is 13.23 pm/ppm. The limit of detection (LOD) of the gas sensor can be obtained according to the formula in literature [31], [32], that is, LOD = 3σ/K, in which σ (0.000723 nm) is the standard deviation for the wavelength of the sensing system in the absence of H₂S, and K (0.01323 nm/ppm) denotes the slope of the fitting curve; thus, the low LOD (0.164 ppm) of this sensor is achieved. It is worth noting that the film with different thicknesses will all cause the interference spectra to shift to various extents. For a determined film thickness, the sensor’s gas detection ability will not be affected because the reference valley wavelength is a determined and fixed value, which will not change during the measurements.

Figure 5:

The wavelength shift upon the concentration of hydrogen sulphide. The inset shows the reflectance spectra of the sensor in various concentrations of hydrogen sulphide.

Figure 6:

Dynamic response of the hydrogen sulphide sensor.

Figure 6 shows the response–recovery time curve of the sensor. The response time t_r and recovery time t_f of the sensor are approximately 70 and 88 s, respectively. Figure 7 shows the gas selectivity of the sensor. Compared with the same concentration of carbon monoxide, carbon dioxide, hydrogen, nitrogen, ammonia, and oxygen, the sensor has a high selectivity for hydrogen sulphide. The main reason is because hydrogen sulphide gas molecule is polar molecule. Compared to nonpolar molecules, such as carbon dioxide, nitrogen, and oxygen, this Cu/rGO-sensing film results in a better adsorption on H₂S molecules. In addition, hydrogen sulphide is a reducing gas, which is easy to lose electron. There is a very weak chemisorption between the hydrogen sulphide gas molecule and nano-copper. Electrons obtained by copper ions (Cu²⁺ or Cu⁺) correspond to the response process of the gas sensor. On the contrary, it corresponds to the recovery process of the gas sensor. The relevant process formula is as follows: Cu⁺ Cu or Cu²⁺ Cu⁺. At the same time, H₂S has the maximum molecular radius among the above gas molecules; the binding capacity of the outermost electrons on the 3P orbital is weak and easy to give electrons to the sensing film. Thus, the sensor has a high selectivity for hydrogen sulphide gas. Among these molecules, except hydrogen sulphide, the selectivity of H₂ and NH₃ is relatively slightly higher than those of other gases, mainly because the rGO has a certain hydrogen adsorption capacity [33]. Ammonia is a polar molecule. It can be attracted/absorbed to the polar functional groups that are located on the surface of rGO. Although these adsorptions are weaker than those of other chemical bondings, they are still much stronger than physical adsorption. It is worth noting that CO’s polarity is weak because of the π backbonding. Therefore, the selectivity of CO is poor.

Figure 7:

Selectivities for hydrogen sulphide, carbon monoxide, carbon dioxide, hydrogen, nitrogen, ammonia, and oxygen.

Figure 8:

The interference valley wavelength shift dependence with various relative humidity (monitoring at 300 K).

The sensor’s response to moisture has been performed by the experiment. Figure 8 is the interference valley wavelength shift dependence on various relative humidity levels (monitoring at 300 K). With the increase of relative humidity, the spectral interference valley shows a blue shift. When the relative humidity is 80 %, the interference valley wavelength moves 0.32 nm, compared to that of 30 % humidity. When the relative humidity increases from 30 % to 50 %, the interference valley wavelength moves 0.06 nm. However, the relative humidity increased from 50 % to 80 %; the interference valley wavelength is shifted 0.26 nm. Therefore, in the relative humidity range of 30 %–50 %, the sensor is less affected by the moisture. Therefore, the sensor has certain application value when the relative humidity is lower than 50 %.