1. Introduction

Because humidity is important for human's normal production and daily life, humidity sensors have been widely used in many fields, such as in aerospace, power generation, textile, food, medicine, agriculture and other industries [1–4]. Humidity sensors can be divided into resistive humidity sensors, capacitive humidity sensors, and photoelectric humidity sensors according to the measurement principle [5–8]. At present, varied kind of humidity sensors have been developed and can be operated in a common environment well [9–10]. However, their measurement range, accuracy and component life are not satisfactory for humidity measurement in a harsh environments, such as high temperature, strong electromagnetic radiation and industrial gases with severe pollution. To improve the humidity sensing performance, fiber-optic (FO) humidity sensors have been rapidly developed due to its advantages of compact size, remote sensing, corrosion resistance, and small temperature coefficient [11–14]. In general, FO humidity sensors require moisture sensitive materials and transducer mechanisms to achieve humidity measurement. The reason is that the light propagation in fiber follows the principle of total reflection and this leads to an immunity from the surrounding environment for the propagating wave. Therefore, in order to realize the humidity sensing with optical fiber, a necessary step is to modify the fiber structure to allow light leakage. However, the processes are often inseparable from specialized equipment and are destructive to the fiber structure making it more fragile, including fused tapering, misalignment, hydrofluoric acid etching and side polishing and fiber grating [15–19].

Moisture sensitive layer is also an important part for the FO humidity sensor. At present, many moisture sensing materials have been investigated and used as a functional coatings, such as polyvinyl alcohol, agarose, gelatin, polyethylene oxide, carbon nanotubes, graphene oxide (GO), and graphene [20–26]. As a moisture sensitive coating for FO sensors, GO has distinct advantage for humidity detection. For example, GO can be prepared massively and efficiently by the process of graphite oxide material [27–28]. Besides, there are many chemical groups distributed on the surface of GO sheet, which can increase the hydrophilicity of GO and enhance the humidity sensitivity of the sensors [29–30]. Herein, we demonstrated an evanescent field FO sensor based on a GO coated capillary in fiber line for humidity sensing. We noticed that the capillary wall allows the presence of a number of higher-order modes propagation with GO and air as the external medium. The propagation path of the light covers the entire capillary wall and its evanescent field can permeate through the capillary wall and interact with the GO coating. Because the evanescent field is extremely sensitive to permittivity change of GO coating, this sensor can achieve a high sensitivity relative humidity measurement.

2. Fabrication and operation principle

The sensor was manufactured by splicing two single mode fibers symmetrically at both ends of the capillary. Both of the single mode fibers and capillary were treated by an optical fiber cleaver to obtain flat end faces before splicing. The structure of the sensor consists of three parts as shown in Fig. 1(a). The first part and the third part are standard single-mode fibers, which are used for light input and output. The middle part is a capillary with outer diameter and wall thickness of 350um and 50um. When light is injected into the capillary wall through the lead-in fiber, the light gradually spreads and fills the entire wall of the capillary. Various high-order modes can be excited in the capillary wall which continuously reflect between the inner and outer surfaces of the capillary wall and gradually transmits to the other end of the capillary. Some of these modes can enter into the core of the lead-out fiber while others are dissipated in the air. Because these high-order modes entering in the lead-out fiber were in different optical path through the capillary, interference occurred when they met each other in the lead-out fiber. To describe and verify the light propagation behavior in the capillary wall, we modeled the propagation process by using a simulation software based on the finite-difference time-domain method. The simulation results show the electromagnetic field distribution of the capillary cross-section at three different distances in Fig. 1(b), which presents the beam entering the capillary rapidly spreads along the curved capillary wall and travels forward in a spiral-like path.

 

Fig. 1. (a) Schematic of beam propagation in the capillary wall; (b) Electromagnetic field distribution in cross sections of capillary in different distance; (c) Photograph of sensor with capillary of 1.5mm length; (d) Experimental setup for relative humidity measurement.

Download Full Size | PDF

The sensor and humidity measurement setup we used for testing sensor is shown in Fig. 1(c) and 1(d), which consists of a custom-made fiber-optic sensing interrogator and a humidity chamber. We put the sensor in a humidity chamber with relative humidity range of 25%RH-80%RH and temperature range of 10-50°C. Because the bending loss can influence the transmission of the sensor, the sensor was fixed on a glass plate to keep the sensing region straight. We connected the sensor to a custom-made fiber-optic sensing interrogator with a built-in light source to detect output light. The interrogator has a resolution of 4pm and a detection range of 1515nm-1590nm. A computer equipped with a data acquisition card was connected to the interrogator for spectral analysis. The detected spectrum was recorded and processed through a program developed based on the Labview software.

3. Graphene oxide coating

In this work, we have coated the capillary part with GO sheets which has many hydrophilic groups and can be used as a moisture sensitive material. Before coating the GO sheets, we ultrasonically cleaned the capillary surface with deionized water, acetone and isopropanol. Then, the sensor was dried with nitrogen and soaked in piranha solution for 2 hours. A 1 mg/mL GO aqueous solution was subjected to alkaline treatment with a 0.1 mol/L sodium hydroxide solution to improve the dispersibility of GO in water. Next, the capillary wall was immersed in GO aqueous solution to be coated by a dip-coating method. In the experiment, we treated the sensor with GO coating of 1 cycle, 3 cycles, and 5 cycles. The sensors were examined by scanning electron microscopy as shown in Fig. 2. Without GO coating, the surface of capillary wall was very smooth [Fig. 2(a)]. As the number of cycle increases, more and more wrinkled GO film appeared on the capillary surface [Figs. 2(b)–2(d)], which means that the distribution of GO sheets on the surface of the capillary became denser with the coating cycle. In order to verify the quality of the GO coating, we also collected the Surface-enhanced Raman spectroscopy (SERS) spectra of the GO coated sensor. As shown in the Fig. 2(e), we can see that the intensities of the SERS peaks increase with the cycles of GO coating, indicating an increasing amount of GO sheets attached to the capillary surface.

 

Fig. 2. Scanning electron micrograph of the capillary wall of sensor: (a) an uncoated capillary surface; (b)–(d) Capillary surface coated by GO with 1, 3, 5 cycles coating; (e) Decoration characterization of GO film on the surface of capillary wall.

Download Full Size | PDF

As a moisture sensitive material, GO film can adsorb water molecules by through hydrogen bond because GO has numerous hydrophilic groups (Fig. 3). With the relative humidity increases, water molecules will occupy the available active sites of GO film. When the humidity is further increased, the layer of water molecules on the surface of the GO film gradually exhibit liquid water properties and the adsorbed water molecules will also penetrate into the intermediate layer of the GO film at a high relative humidity. Besides, the absorption of water molecules causes a decline in permittivity of the GO film. Moreover, the permittivity change of the GO film will influences the effective refractive index of the high-order modes propagating in capillary wall through evanescent field, which will lead to a wavelength shift of the interference fringe in the transmission.

 

Fig. 3. Illustration of the proposed sensor for relative humidity sensing based on GO and evanescent tunneling.

Download Full Size | PDF

We have investigated the interaction between GO layer and the evanescent field by simulation. The GO layer has a complex refractive index of 2.25 + 0.18i and a thickness of 5 nm. As the simulation result shown in Fig. 3, the power of optical field increases significantly in the capillary-air interface with GO layer. This indicates that the energy is coupled from the low-order mode to the higher-order mode and the evanescent field is enhanced. This effect can be explained as following: The amplitude enhancement factor of evanescent wave excited on the capillary surface can be expressed as ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}} \over E} _0}\textrm{exp}\left[ { - \frac{{2\pi }}{\lambda }z\sqrt {{{\left( {\frac{{\sin i}}{{{n_{21}}}}} \right)}^2} - 1} } \right]$.

where, ${\mathord{\buildrel{\lower3pt\hbox{$\scriptscriptstyle\rightharpoonup$}} \over E} _0}$ presents the incident light wave. λ and i are the incident wavelength and the incident angle. ${n_{21}}$ is the ratio of the refractive index of the capillary to the external environment. z is the vertical distance from the surface of the capillary. Compared to air, the GO film covers the surface of the capillary wall as a high refractive index medium, which means the total reflection critical incident angle decreases and the relative refractive index ${n_{21}}$ increases. It can be found that these changes cause an increase in the amplitude factor of the evanescent wave. In addition, the refractive index of GO film will decrease after absorption of water molecules, which also make a more significant RI change than that of moist air enveloping capillary surface.

4. Results and discussion

To verify the capillary element and GO coating configuration and optimize sensor performance, we examined the sensor characteristics by changing the length of capillary and compactness of GO coating. In the experiment, we applied 1, 3 and 5 cycles GO coating on capillaries with length of 0.8 mm, 1.5 mm, and 2.2 mm. The influence of GO coating and evanescent field strength on sensor spectra and sensitivity were investigated by comparing the humidity responses of these sensors. Because each GO coating process was prepared by the same solution and operation, for the same sensor, any difference in its spectral response and sensitivity should be attributed to the difference in the compactness of the GO coating on capillary surface.

The working principle of the sensor was based on the interaction between the evanescent field excited on capillary surface and the GO layer adsorbed with water vapor molecules. The appearance of GO layer can lead to the refractive index increase of the external medium, the extinction of the total internal reflected light and the adsorption of water molecules. In addition, because the GO layer can adsorb more water vapor molecules with the increase of the coating cycle, the high-order modes in the capillary wall has a strong interaction with the water vapor through the evanescent field on capillary surface and provides sensitivity to the relative humidity. The sensitivity of the sensor is related to the penetration depth of the evanescent wave and the light reflections in the sensing area. The number of reflections of total reflected light in the capillary wall is closely related to the length of the capillary. However, because cladding absence of capillary wall results in a transmission loss and GO layer has a certain extinction on total reflection, it means that longer capillary isn't always better in experiment and the GO coated capillary has an optimal length.

We examined the effect of GO coating on the performance of sensor with a short capillary (0.8mm). We performed GO coating operations on the capillary wall and tested the sensor’s relative humidity response. The experimental results are shown in the Fig. 4. We can find the dips in transmission [Fig. 4(a)] had slightly sensitivities to relative humidity after 1 cycle GO coating. Then we the coated the sensor with 3 cycles and tested its response. In Fig. 4(b), we can find more interference fringes appear in the spectrum while the spectral response to relative humidity vibration is still weak. The sensor had a further spectral degradation after 5 rounds of GO coating operation [Fig. 4(c)] but the spectrum had a more obvious change than before. The GO-induced spectral degradation can be explained as: The appearance of GO layer changed the total internal reflection condition and affected the energy distribution of modes in capillary wall. As a result, for the sensor with 0.8 mm capillary, the GO coating made the sensor more sensitive to the relative humidity although without a regular and significant response.

 

Fig. 4. (a), (b) and (c): Spectral response of sensors with 0.8mm capillary to different relative humidity after 1, 3 and 5 cycles of GO coating;

Download Full Size | PDF

Thereafter, we selected a GO coated sensor with a capillary length of 1.5mm for the relative humidity measurement. The spectral intensity of the sensor was one magnitude lower than the spectrum of the sensor with 0.8 mm capillary. As shown in Fig. 5(a), there are two distinct dips in the spectrum of the sensor, but their responses to relative humidity are weak [Fig. 5(b)]. The reason may be that the GO layer coated with 1 cycle coated was sparse and the adsorption capacity for water molecules was limited, which resulted in an inadequate interaction between the evanescent field and water molecules. As the GO coating increased to 3 cycles, the sensor showed a response to relative humidity change. In Fig. 5(c) and 5(d), the wavelengths of dip 1 and dip 2 had a red-shift and their sensitivities are evaluated to be 0.006 nm/% RH and 0.037 nm/% RH. When the sensor coated with 5 cycle of GO coating, the sensitivities of dip 1 and dip 2 in the spectrum [Fig. 5(e)] were further improved. The sensitivities of dip 1 and dip 2 were evaluated to be 0.147 nm/% RH and 0.211 nm/% RH by using the slopes of fitting curves in Fig. 5(f), which increased more than 20-fold and 6-fold compared to the sensitivities of the sensor with 3 cycle coating. The sensitivity improvement of the sensor can be explained as: Because the compactness of GO layer increased with coating cycle and the thickness of the GO sheet is much smaller than the penetration depth of the evanescent wave, there are more water vapor molecules adsorbed by the GO layer interacted with evanescent field on capillary surface. Besides, from the above experimental results, we can also find that the GO coating has a significant improvement on the humidity sensitivity of the sensor and the influence on the two dips in spectrum is different. This indicates that we can tune the wavelength sensitivity of dips by controlling the amount of GO sheets on the capillary wall.

 

Fig. 5. (a), (c) and (e): Spectral response of sensor to different relative humidity after 1, 3 and 5 cycles of GO coating; (b), (d) and (f): Relative humidity sensitivities of dips in the spectra.

Download Full Size | PDF

We also studied the response of a sensor with 2.2 mm capillary to relative humidity. As shown in Fig. 6(a) and 6(b), after a single cycle of GO coating, the intensity of the spectrum had dropped drastically and did not show a sensitive relative humidity response. Thereafter, the sensor was coated by GO sheets with 3 cycles and tested. Its spectral response are shown in Fig. 6(c) and 6(d). It can be seen that the relative humidity sensitivity of the sensor was enhanced and several dips have significantly wavelength shifts. We calculated the sensitivities of dip 1 and dip 2 which were 0.043nm/%RH and 0.057nm/%RH, respectively. After the sensor was treated with 5 cycles GO coatings, the spectrum was greatly degraded as Fig. 6(e) shown. We can find the minimum of several dips on the spectrum is already lower than the detection limit of the interrogator [Fig. 6(f)]. This can be explained as: the capillary of the sensor is relative long, which results in largely transmission loss and light absorption by the GO coating. Although the wavelength shifts of the dips are unable to be measured, it is still apparent that the interference fringes on the sensor spectrum shows sensitive response to relative humidity. For example, the wavelengths of peak 1 and peak 2 shifted by 1.5 nm and 2 nm in the range of relative humidity from 30%RH-70%RH. Overall, more coating operations increase the amount of GO sheets attached to the capillary surface, which not only allows the sensor to absorb more water molecules and enhance sensitivity, but also results in a more significant extinction. In addition, the longer capillary brings more transmission losses, but also allows more total reflections of light in the sensing region to enhance the interaction of the evanescent field with water molecules.

 

Fig. 6. (a), (c) and (e): Spectral response of sensor to different relative humidity after 1, 3 and 5 cycles of GO coating; (b), (d) and (f): Relative humidity sensitivities of dips or peaks in the spectra.

Download Full Size | PDF

The GO coating is beneficial to enhance the interaction between evanescent field and water molecules on surface of capillary wall, which also brings a transmission loss and changes in total reflection conditions. The energy mainly distributed in the low-order mode for the sensor with 0.8mm capillary. Moreover, with the appearance of GO coating, part of the high-order mode pass through the capillary wall into the GO layer and dissipate in air due to the changes in total reflection condition. Because the proportion of high-order mode is much smaller than the low-order mode, we can see from the Fig. 4 that the transmitted spectrum of sensor was not very sensitive to relative humidity vibration. It is most unfavorable for high-order mode propagation in the 2.2mm capillary among the three lengths (0.8mm, 1.5mm and 2.2mm) because the capillary has no cladding. We can find the sensitivity of the sensor with 2.2mm capillary increases when GO layer increase from 1 cycle to 3 cycles because the interaction between evanescent field and water molecules is enhanced. However, due to the long length of capillary, the GO coating also brings a large transmission loss and is not conducive to the propagation of high-order modes in capillary, which can be proved by the experimental results in Fig. 6(e). When the coating cycle increased from 3 to 5, the transmitted light intensity also decreased significantly. Compared with the former two sensors, the presence of GO coating on the 1.5mm capillary can enhance the strength of interaction between the evanescent field and water molecules while the resulting transmission loss has not caused a severe recession of the high-order mode, which ensures the sensor has an increasing sensitivity with the GO coating from 1 cycle to 5 cycles.

We also tested the sensors with 1.5mm and 2.2mm capillary when relative humidity decreasing from 70%-30%RH. The 1 cycle coated sensors with 1.5mm and 2.2mm capillary did not show a sensitive response to relative humidity (Fig. 7 and Fig. 8), but the wavelengths of dips for sensors with 3 cycles and 5 cycles have blue shifts when relative humidity decreases. In the Fig. 7(c) and (d), the wavelength shift of sensor with 1.5mm capillary and 3 cycles coating are Δλ_dip1=0.25nm and Δλ_dip2=1.35nm from 70%RH to 30%RH. The wavelength shift of sensor with 1.5mm capillary and 5 cycles coating are Δλ_dip1=6nm and Δλ_dip2=9nm [Fig. 7(e) and (f)], respectively. Δλ_dip1 and Δλ_dip2 are similar with the wavelength vibrations in Fig. 5(d) and (f) when relative humidity increases from 30%RH to 70%RH. Moreover, the response of sensor with 3 cycles coating changes at a faster rate from 70%RH to 30%RH than that of 30%-70%RH, which results in the fitting curve in Fig. 7(d) to be quadratic rather than linear.

 

Fig. 7. Relative humidity response of sensors with 1.5mm and equations of the fitting curves from decreasing humidity from 70% to 30% RH. (a), (c) and (e): Spectral response of sensor to different relative humidity after 1, 3 and 5 cycles of GO coating; (b), (d) and (f): Relative humidity sensitivities of dips in the spectra.

Download Full Size | PDF

 

Fig. 8. Relative humidity response of sensors with 2.2mm equations of the fitting curves from decreasing humidity from 70% to 30% RH. (a), (c) and (e): Spectral response of sensor to different relative humidity after 1, 3 and 5 cycles of GO coating; (b), (d) and (f): Relative humidity sensitivities of dips or peaks in the spectra.

Download Full Size | PDF

As shown in Fig. 8(c) and (d), the spectrum of sensor with 2.2mm capillary and 3 cycles coating also has a blue wavelength shift and a faster change rate from 70%RH to 30%RH than that of 30%-70%RH. The wavelength variations of two dips are 2nm and 2.7nm, which are similar with the wavelength vibrations in Fig. 6(d) when relative humidity increases from 30%RH to 70%RH. However, in the Fig. 8(e) and (f), sensor with 2.2mm capillary and 5 cycles does not show a regular response to the relative humidity change.

5. Conclusion

We have presented a fiber-optic evanescent field humidity sensor based on a capillary element and GO sheet coating. Because the large surface-to-volume ratio and the rich chemical groups are very helpful for water molecules absorption, the GO film significantly enhances the sensitivity of the evanescent field to humidity. Through theoretical analysis and numerical simulation, we described the interaction between the GO film and the evanescent field. In the experiment, we applied different cycles of GO coating on the outside surface of capillary wall. With an increase in coating cycle, the sensor gradually became sensitive to the relative humidity, and the dip wavelength linearly shifted with the increase of relative humidity. The GO-coated sensor with different capillary length were tested to investigate the influences caused by the GO sheet and the transmission loss and the number of reflection of capillary. The experimental results show that the humidity sensitivity of the sensor increases with the increase of compactness of the GO coating on the capillary surface and the capillary of the sensor has an optimal length. As a result, the cycles of GO coating determined the humidity sensitivity of the sensor and the coating process was controllable, which means that the sensor was capable to perform tunable sensitivity. This evanescent field humidity sensor may have potential applications in environmental sensing and health monitoring.