1. Introduction

The ratio of the partial vapor pressure of water to the saturation vapor pressure at a specific temperature, i.e., relative humidity (RH), is a significant parameter for various industrial applications, such as meteorology, medicine, food, and agriculture etc. The conventional methods for measuring RH include mechanical hygrometer, chilled mirror hygrometer, wet and dry bulb psychrometer, infrared optical absorption hygrometer, electronic element and electrochemistry [1,2]. However, these sensors have drawbacks, such as long response time, low sensitivity, electromagnetic inference etc.

In recent years, fiber-optic humidity sensors have attracted a lot of research interests due to their unique advantages, including low weight, small size, immunity to electromagnetic interference, and remote sensing capability [3–24]. Optical and mechanical properties of the sensing material, including optical absorption, scattering, fluorescence and mechanical expansion, can be utilized to sense the RH of the surrounding environment. For instance, cobalt chloride [5], phenol red [6] and crystal violet [7] can be used to make RH sensors, since their optical absorptions depend on humidity. Evanescent-wave scattering is another principle suitable for humidity sensing, in which porous material is used to act as a cladding to scatter the transmitted light and the RH can be measured through the intensity changes of transmitted light [8,9]. However, these intensity-based RH sensors will be influenced by fluctuations in light source and temperature. Although fluorescence is a well-established method to overcome these issues [10–13], fluorescence based RH sensors lack long-term stability due to the bleaching and leaching issues of the dyes in host materials.

Hygroscopic expansive polymers provide a promising approach to make humidity sensors because the polymer’s volume will expanse after absorbing water. The volume expansion induced elongation or refractive index (RI) change can be used for humidity sensing if the polymer is coated on the side surface of a fiber Bragg grating (FBG) [14–16], a long-period grating [17,18], a U-shape polymer fiber [19,20], a taped fiber [21,22] or used to form a Fabry-Perot cavity [23,24].

In this paper, we propose a new type of fiber-optic RH sensor based on thin-core fiber modal interferometer (TCFMI) and FBG hybrid structure in which an electrostatic self-assembled polymer nanocoating is coated on the side-surface of the sensor for humidity sensing. The TCFMI is used to measure the humidity-induced RI change of the nanocoating, whereas the FBG inscribed in the middle of TCFMI is used for the compensation of temperature sensitivity. A fast, linear and reversible response with high resolution will be experimentally demonstrated in a large RH measurement range.

2. FBG incorporated thin-core fiber modal interferometer sensor

The schematic diagram of the RH sensor is shown in Fig. 1(a) . A 20 mm long thin-core fiber is inserted into standard single-mode fiber to form an in-fiber modal interferometer. It was demonstrated that the spectrum dip of thin-core fiber modal interferometer is an ideal sensing signal for RI measurement [25,26]. If one further inscribes an FBG in the middle of thin-core fiber modal interferometer, one can eliminate the cross sensitivity to the temperature.

 

Fig. 1 Schematic configuration of the fiber-optic RH sensor based on an FBG incorporated TCFMI (a) and its reflection spectrum (b) and transmission spectra (c) before (solid curve) and after (dashed curve) self-assembling of sensing nanocoating.

Download Full Size | PDF

The TCFMI was fabricated with a commercial thin-core fiber (Nufern 460-HP) whose core diameter is ~3.0 μm and cutoff wavelength is ~450 nm. The TCFMI was then hydrogen loaded at 110 °C under pressure of 10 MPa for four days to enhance the photosensitivity. An FBG was fabricated on the thin-core fiber by using a KrF excimer laser (TuiLaser Ltd., Germany) with a phase-mask grating-writing technique. A phase mask with a pitch of 1070.6 nm is employed to make FBGs with period of 535.3 nm. The grating length is 10 mm. Figure 1(b) and (c) show the reflection and transmission spectra of the fabricated sensor, respectively, which were recorded by using an optical spectrum analyzer (ANDO, AQ6317).

Figure 2 shows the spectral response of the sensor to the change of external RI and temperature. As one can see in Fig. 2 (a), when the concentration of glycerol/deionized-water solution increases, the transmission dip shifts to longer wavelength with the sensitivity of 140 nm/R.I.U., whereas the reflection peak almost keeps unchanged. Figure 2(b) shows the temperature responses of the sensor, which is measured by using a temperature-controlled oven. One can see that both the transmission dip and the reflection peak shift to longer wavelengths when the temperature increases. The measured temperature sensitivities of the transmission dip and reflection peak are 15.3 pm/°C and 15 pm/°C, respectively. It means that the FBG is a good reference element with very close temperature response to TCFMI, but totally insensitive to external RI.

 

Fig. 2 The spectral response of the sensor to the change of external refractive index (a) and temperature (b).

Download Full Size | PDF

3. Self-assembly of the humidity sensing nanocoating

The sensing nanocoating was fabricated on the side surface of the fiber sensor by using a layer-by-layer (LbL) electrostatic self-assembly technique [27,28]. The method is based on the construction of molecular multilayer through the electrostatic attraction between oppositely charged polyelectrolytes. In our experiments, poly (N-ethyl-4-vinylpyridinium chloride) (P4VP·HCl) and poly (vinylsulfonic acid, sodium salt) (PVS), whose molecular structures are shown in Fig. 3(a) , were used as cationic and anionic materials. P4VP·HCl was synthesized as described in our previous work [27], and PVS (25% aqueous solution) was purchased from Sigma-Aldrich. All reagents are of analytical grade without further purification. Deionized water with a resistance of 18 MΩ·cm was used in all experiments.

 

Fig. 3 Chemical structures of the materials (a) and schematic diagrams of electrostatic self-assembly technique (b).

Download Full Size | PDF

The LbL electrostatic self-assembly was carried out on the substrates (quartz slides, AT-cut quartz crystals with gold electrodes or the side surface of the optical fiber) as shown in Fig. 3(b). The concentrations of the P4VP·HCl and PVS are 2.0 g/L. After cleaned with piranha solution (7:3 of concentrated H₂SO₄ and H₂O₂), the substrate was flushed by a large volume of deionized water and dried with nitrogen. The substrate was then immersed into the P4VP·HCl (positively charged) and PVS (negatively charged) alternatively for 10 min at room temperature. After deposition of each layer, the substrate was rinsed by deionized water three times (1 min every time) to remove the excess of adsorbed materials and dried with nitrogen. Each polycation/polyanion layer (P4VP·HCl /PVS) is called a bilayer. The same cycle was repeated until the desired amount of bilayers had been reached. After the deposition of 10 bilayers, the substrate was baked in a temperature-controlled oven at 60 °C for 10 hours. As presented in Fig. 1(c), the transmission dip of the TCFMI shifted to 1585 nm after the deposition of nanocoating.

The fabrication procedure was tested by UV-vis absorption spectroscopy (Cary 100Bio). Since P4VP·HCl has an absorption peak at 256 nm, we measure the absorption spectra around 256 nm to monitor the growth of sensing nanocoating. Figure 4(a) shows the measured absorption of the self-assembled nanocoating at 256 nm with increment of bilayers, and the inset shows the measured absorption spectra. The regular increment of the absorption indicates that the fabrication of nanocoating is well performed in the experiment. The thickness of the self-assembled nanocoating was measured by using a quartz crystal microbalance (QCM, Resonance Probe GmbH, Goslar, Germany). AT-cut quartz crystals (Maxtek) with a fundamental frequency f ₀ of 5 MHz and gold electrodes were used in the experiments. The frequency shift Δf of the quartz crystal after deposition of each bilayer was measured at the third overtone order n = 3 (i.e., 15 MHz). If we assume that the density is 1.0 g·cm⁻³ [27], the thickness of the nanocoating can then be deduced by the Sauerbrey equation [29]:

 

Fig. 4 Growth of absorption (a) and thickness (b) of (P4VP·HCl /PVS)₁₀ bilayers as a function of bilayer number.

Download Full Size | PDF

Figure 5 shows the scanning electron microscopy (SEM) photos of optical fiber without (a) and with (b) assembled nanocoating. One can see that the surface of the bare optical fiber is very smooth, but it becomes quite rough after the deposition of nanocoating. As shown in Fig. 5(b), the nanocoating shows a bumpy surface with many protuberant crests. The existence of those crests provides channels for water molecules to diffuse into or out of the nanocoating which thus can decrease the response time of the sensor.

 

Fig. 5 Scanning electron microscopy images of the surface of optical fiber without (a) and with (b) self-assembled nanocoating. Both scale bars in the above two images are 1 μm.

Download Full Size | PDF

4. Results and discussion

The sensor was tested by using a setup with two glass chambers. A small chamber is used to fix the sensor, and a large one is used to install the heating and steam controlling system. With a commercial hygrometer, the temperature and humidity of the large chamber can be adjusted to specific values. A separation door between the two chambers can be used to separate or equilibrate the environment between the two chambers. Figure 6 (a) shows the spectral response of the sensor to the RH at different temperatures. One can see that the transmission dip shifts to longer wavelengths with increment of RH. This is because that the refractive index of the nanocoating increases after absorbing water [18]. Meanwhile, the reflection peak keeps constant with the change of humidity, and shifts only when the temperature is changed. Moreover, the slope of the RH response is somewhat increased when the temperature increases. The sensitivities of the transmission dip to RH are 84.3 pm/1%RH, 87.9 pm/1%RH, and 97.2 pm/1%RH at 20 °C, 40 °C and 60 °C, respectively. If using one hundredth part of 3-dB bandwidth of the sensor transmission dip, i.e., ~0.065 nm, one can estimate the detection resolution of the sensor as 0.78%RH. Figure 6 (b) and (c) present the measured reflection spectra at different temperatures and the evolution of transmission spectra with the increment of RH when the temperature is 40 °C, respectively.

 

Fig. 6 The measured responses of the sensor to RH at different temperatures (a), the shift of reflection spectra with the increment of temperature (b) and the evolution of transmission spectra with the increment of humidity when the temperature is 40 °C (c).

Download Full Size | PDF

The dynamic response of the RH sensor was also tested in the experiments. In order to shorten the sweep time of the optical spectrum analyzer, the bandwidth of the measured spectrum is set to be around 2 nm during the dynamic test of the sensor. Figure 7 shows the measured dynamic response of the sensor. The red line is not the direct reading of the hygrometer, but the presumed humidity values around the sensor since the small chamber is not big enough for installing commercial hygrometer. The dashed red lines mean that separation door is closed, and the abrupt changes of red line mean that the door is abruptly opened. Since the volume of the small chamber is much smaller than that of the large chamber, the relaxation time of humidity between the two chambers has been ignored. It shows that the rise time (t_r) is about 2 s and the fall time (t_f) is about 10 s for a change of 20%RH. The response of the sensor is quite fast because the nanocoating is ultra-thin and very rough, as shown in Fig. 5. After several rounds test of swelling/deswelling processes, the transmission dip of the sensor for a given RH almost keeps constant and no significant degradation of sensitivity is observed, which means that the sensor has good repeatability and reusability.

 

Fig. 7 The dynamic response of the fabricated RH sensor to the change of humidity.

Download Full Size | PDF

5. Conclusion

We have presented a new fiber-optic RH sensor based on an FBG incorporated TCFMI architecture. The sensing nanocoating was made with an electrostatic self-assembly technology. The fabrication procedure of the sensing nanocoating was characterized by UV-vis absorption spectroscopy and QCM and its morphology was measured by using SEM. The response of the sensor has also been tested, which showed that the sensor has a high sensitivity to RH in a large range (RH range: 20% ~90%) and the FBG can be utilized to indicate the environment temperature for compensation of the cross sensitivity to temperature. Such a miniature and reusable fiber-optic RH sensor is very promising technology for e.g., breath analysis applications.