Michelson Interferometric Hydrogen Sulfide Gas Sensor Based on NH₂-rGO Sensitive Film

3.1 Preparation of NH₂-rGO

Graphene oxide dispersion (GO; 2 mg/mL) was obtained from Nanjing Xianfeng Nanomaterials Technology Co., Ltd. (Nanjing, China. Ethylenediamine (C₂H₈N₂, EDA; 99 %), ammonia (NH₃ ⋅ H₂O; 25–28 %), sodium borohydride (NaBH₄; 98 %), hydrogen peroxide (H₂O₂; 30 %), sulfuric acid (H₂SO₄; 98 %), N,N-dimethylformamide (DMF; AR), and anhydrous ethanol (C₂H₆O; AR) were obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China) and were used without further purification.

GO dispersion (30 mL) and absolute ethanol (50 mL) were fully mixed together, and ammonia was slowly added to the mixed solution dropwise until pH = 10. Then EDA (15 mL) was added to the mixed solution and the sample was heated at 80 °C for 3 h with stirring. It was filtered and washed with de-ionised water and ethanol, respectively, thereby removing unreacted ammonia and EDA. The obtained sample was ground for 1 h after vacuum-drying at 60 °C for 12 h, dispersed in 50 mL DMF using ultrasound, and then 0.02 g sodium borohydride was added to the mixture with stirring at 60 °C for 3 h. After that, the product was filtered and washed by de-ionised water, and finally dried it at 60 °C for 12 h. Then NH₂-rGO was obtained.

3.2 Instruments and Methods

The morphology of the sensing-film-coated TCF was characterised using a Gemini SEM 300 field emission scanning electron microscope (FE-SEM; Zeiss, Germany). Raman spectra were obtained using a Raman spectrometer (LabRAM HR Evolution, Horiba Scientific, France). X-ray photoelectron spectroscopy (XPS) of NH₂-rGO was performed using an XPS-Theta probe instrument (Escalab250Xi, Thermo Fisher, MA, USA).

3.3 Fabrication of Sensor

Two TCFs and several standard SMFs were prepared. The splicing machine (Furukawa, S178, Japan) parameters were as follows: first discharge starting strength +90, the second discharge intensity +40, the pre-welding time +160 ms, automatic discharge time +1300 ms, the second discharge intensity +100, the section interval +20, and other parameters all zero. As shown in Figure 1, one end of the optical fibre is connected to the spectrometer and the other end to the ASE light source to form the Mach–Zehnder interferometer (MZI). In addition, one end of the optical fibre is connected to the fibre circulator and the other end to the FRM to form the Michelson interferometer. Figure 2 shows the related structural parameter optimisation. Shown in Figure 2a are the spectra of MZI without FRM and MI with FRM. By magnifying the waveband around 1575 nm, it can be clearly seen that MI with FRM has stronger anti-interference ability than MZI. The result shows that FRM can improve the signal-to-noise ratio and enhance the stability of the interferometer. The reason is that when the input light passes through the FRM, the polarisation angle of the input light is rotated by 45° and the light is deflected by 45° again after being reflected back by the highly reflective material. The FRM makes the input light and the output light to have mutually perpendicular polarisation directions, which can eliminate the polarisation interference of the spectrum. It should be pointed out that when only one TCF is used, the interference peaks are fewer, the depth of the spectral peaks is not enough, and the interference phenomenon may be not obvious for the sensitivity test.

Figure 2:

Structural parameter optimisation. (a) Spectra of MZI without FRM and MI with FRM. (b) Interference spectra with different TCF-SMF-TCF lengths. (c) FFT spectrum of STSTS-F structure after coating. The inset is the interference spectrum after coating.

In order to investigate the influence of length of the TCF and SMF on the interference spectra, the optimum optical fibre length and high-stability reflectance spectrogram were obtained. As shown in Figure 2b, when the lengths of TCF-L and TCF-R are 3 cm and the length of the SMF is 5 cm, there are fewer interference peaks with only two obvious troughs. When the lengths of TCF and SMF are 2 and 6.5 cm, respectively, the output spectrum shows more obvious interference. However, the obvious trough noise makes it difficult to observe the trough movement. When the lengths of TCF-L, TCF-R, and SMF are 2.5, 2, and 6.5 cm, respectively, the maximum intensity attenuation of the corresponding mth-order mode is about 30 dBm, and the interference pattern is most stable. So, 2.5 cm TCF-L, 2 cm TCF-R, and 6.5 cm SMF were used to fabricate the STSTS-F MI hydrogen sulfide gas sensor.

Two TCFs with lengths 2.5 and 2 cm were spiced between the SMFs. Piranha solution was prepared by adding 30 % hydrogen peroxide to 98 % sulfuric acid at a volume ratio of 3:7 under ultrasonication for 5 min. First, the fused TCFs were cleaned with ethanol, then dried, and further soaked in 70 °C sulfuric acid and hydrogen peroxide solution for 30 min. Surface-modified TCFs with hydroxylation were obtained after drying in nitrogen atmosphere. Second, a suitable amount of NH₂-rGO was dispersed in 60 mL ethanol. After magnetic stirring and cryogenic ultrasonication, the solution was fully mixed. Finally, the hydroxylated TCF segments were immersed in NH₂-rGO for 20 s and vacuum-dried for 1 h at 60 °C. The above steps were repeated several times until a light black film was formed on the surface of the optical fibres. The coated optical fibres were placed in the vacuum drying chamber at 60 °C for 6 h.

The interference spectra and the fast Fourier transform (FFT) spectra of the MI sensor with STSTS-F structure after coating are shown in Figure 2c. Most of the light is transmitted in the core, and the core fundamental mode (LP₀₁) is coupled with the stimulated high-order cladding mode (LP_0m) to form the interference.

4.1 Surface Morphology

The NH₂-rGO-coated TCF was characterised by SEM, and the surface morphologies are shown in Figure 3. From Figure 3a, a uniform and compact sensing film is formed on the surface of TCF. From the end-face diagram of TCF, as shown in Figure 3b, the thickness of the sensing film is about 500 nm.

Figure 3:

Surface morphology. (a) Outside surface of TCF coated with NH₂-RGO. (b) Cross-section of NH₂-rGO-coated TCF.

4.2 Raman Spectra and XPS Analysis

Raman spectra of GO and NH₂-rGO are shown in Figure 4a. The D band represents the defect of the C atom lattice and the G band represents the in-plane stretching vibration of sp²-hybridised C atom [32]. The D and G bands of GO are located at 1339 and 1585 cm⁻¹, respectively, while the D band, G band, and 2D band of NH₂-rGO are located at 1336, 1588, and 2927 cm⁻¹, respectively. These results are consistent with those in the literature [33]. The intensity ratio (I_D/I_G) of the D and G bands of GO and NH₂-rGO are 0.990 and 1.184, respectively. The larger the ratio, the more the number of defects or vacancies. That is to say, NH₂-rGO has more defects than GO.

Figure 4:

Raman and XPS spectra. (a) Raman spectra of GO and NH₂-rGO. (b) XPS image of NH₂-rGO. (c) High-resolution XPS scan of C 1s. (d) High-resolution XPS scan of N 1s.

Figure 4b–d shows the results of XPS of NH₂-rGO. From Figure 4b, the sample contains C, N, and O elements, which are consistent with the expected compositions. H 1s does not exist because the ionisation interface of H is too small, and the 1s electron of H is easily transferred and bound to other atoms [34]. The binding energies of C 1s, N 1s, and O 1s correspond to 284.7, 398.7, and 532 eV, respectively, and C and O Auger energies correspond to 1221.4 and 979.7 eV, respectively. As shown in Figure 4c, high-resolution peak splitting fitting results of C 1s binding energies of C–C, C–N, C–O, and C=O correspond to 284.5, 285.5, 286.0, and 287.2 eV, respectively. From Figure 4d, the splitting peak fitting results of N 1s show those of –CO–NH and –CH₂–NH₂, corresponding to 398.2 and 399.2 eV, respectively. The formation of –CO–NH– is due to the dehydration and condensation reaction of –COOH and –NH₂ when they graft with the amino groups. The existence of the –CH₂–NH₂ peak indicates that –NH₂ is successfully grafted onto rGO.

4.3 Gas Sensing Properties

Hydrogen sulfide gas was injected into the chamber, and the gas concentration and the corresponding monitoring wavelength were recorded. As shown in Figure 5a, with the increase in hydrogen sulfide concentration, the wave trough shows a blue shift. The sensitivity is 21.3 pm/ppm and the linear fitting value is R² = 0.98096. The limit of detection (LOD) of the sensor is given as LOD = 3σ/k [35], where σ (∼0.0112 nm) is the standard deviation of the slope and k (∼21.3 pm/ppm) is the sensitivity; thus the LOD of the sensor is 1.577 ppm. The response of the sensor to some typical gases, such as CO, H₂S (the concentration of both gases is 30 ppm), as well as pure N₂, O₂, and CO₂ at room temperature are shown in Figure 5b. It can be concluded that the NH₂-rGO film is most sensitive to hydrogen sulfide, which indicates that NH₂-rGO film has good selectivity to hydrogen sulfide gas. When 60 ppm hydrogen sulfide gas was injected into the chamber, and the data were collected every 10 s (see Fig. 5c), the response and recovery time of the sensor were 72 and 90 s, respectively. A detailed comparison of the designed sensor with other fibre gas sensors is shown in Table 1 [34], [36], [37]. The results indicate that this sensor has some good characteristics such as sensitivity and selectivity for H₂S.

Figure 5:

Gas sensing properties. (a) Wavelength versus concentration of hydrogen sulfide. The inset shows the spectral response of the sensor under various concentrations of hydrogen sulfide. (b) Selectivity for hydrogen sulfide, helium, nitrogen, carbon dioxide, and oxygen. (c) Dynamic response of the hydrogen sulfide gas sensor.

The following is the error analysis of the experimental data. From Table 2, the arithmetic mean of wavelength deviation can be obtained as

The variance can be written as

The standard deviation of wavelength is

The small variance denotes small fluctuation. That is, the difference Δλi between the theoretical value and the actual measured value is relatively stable. The small fluctuation indicates that the standard deviation is also small (σ = 0.0290). In addition, the arithmetic mean Δλ¯=0.0476 (nm) of the difference between the theoretical value and the actual measured wavelength value is obtained. Because the resolution of the spectrum analyzer is 0.02 nm, and the difference values between the theoretical and experimental data are so tiny and can be ignored.

4.4 Temperature Response of the Sensor

As shown in Figure 6, the effect of temperature on the sensing performance was evaluated. When the temperature is between 20 °C and 45 °C, the effect of temperature on the sensor is very small. However, when the temperature is higher than 45 °C, the effect on the sensor becomes obvious. Therefore, the sensor is suitable for H₂S detection in a small temperature range (20–45 °C).

Figure 6:

Influence of temperature change on sensing performance.

4.5 Sensing Mechanism

Since the amino group (–NH₂) can donate an electron on the rGO surface, amino functionalisation enhances the negative carrier concentration of rGO [38]. Because of the enhancement in charge density in NH₂-rGO by the presence of amine groups on the surface, NH₂-rGO behaves like a p-type semiconductor, which can act as an acceptor. As a reduced gas, H₂S easily loses electron to the sensing film. After H₂S is injected into the gas chamber and makes contact with the film surface, the film surface is considered to effect the migration of a small amount of electronic charge transfer at room temperature (Fig. 7), resulting in a change in thickness of the sensing film and hence a change in the RI of the film. Thus, the resonant wavelength will be shifted. When fresh air is introduced into the gas chamber, desorption occurs, and the negative carrier is transferred from the film to H₂S.

Figure 7:

Sensing mechanism of the sensing film.