1. Introduction

Nowadays, relative humidity (RH) measurements are becoming more and more important in certain fields including meteorological services, chemical, biomedical and electronic processing, and sophisticated construction monitoring etc [1–4]. Fiber-optic based RH sensors have attracted considerable attentions in the past decades because of its unique advantages including electromagnetic immunity, high sensitivity, passivity, low crosstalk, and in particular, excellent large scale multiplexing capability [5–7].

Fiber-optic based RH sensors can be realized by means of fiber Bragg grating sensing technology and polymer coatings as sensitive elements. When the moisture sensitive polymer is coated on the fiber to form a moisture sensitive layer, this sensing layer will undergo a volume variation as the change of the number of water molecules adsorbing, which directly generates a strain imposed on the fiber. Based on this correlation, the humidity measurements can be derived from strain measurements using fiber sensing. Xie etc. proposed a RH sensor structure by forming an extrinsic Fabry-Perot cavity, composed of sensitive multilayers of Ti₃O₅ and SiO₂, on the distal end multimode fiber [8]. The humidity measurement was carried out by analyzing the shift of interference fringe. Alwis etc. used a long-period grating (LPG) coated with tailored polyimide layer to allow the wavelength shift responding to the humidity change [9]. All the achievements promote the development of fiber-optic based RH sensors. However, there is less studies about building fiber-optic based RH sensors network to increase the capacity.

Meanwhile, huge capacity is one of the characters of developing trends towards the next generation of optical fiber sensing network. Hu etc. combined two semiconductor optical amplifiers (SOAs) and time-division multiplexing (TDM) technique on low-reflectance FBGs, and demonstrated a single optical fiber network with over 1000 sensors [10]. Owing to the simplicity of structure and fabrication, FBG-based sensors have been accepted widely. However, to the most of our knowledge, there is few reports about the study of FBG-based RH sensor network.

In this paper, a serial TDM polyimide coated RH sensing network based on identical low-reflectance FBGs is proposed and demonstrated with experimental results. It can be expected that thousands of FBG-based RH sensors multiplexed in series. In addition, the identity of the RH sensors simplifies the sensing link and makes the mass production of the sensing array affordable. Furthermore, this method is not limited in RH sensing application and in wide range of FBG-based chemical sensing applications.

2. Principle

2.1. Sensing principle

A uniform FBG sensor contains a series of grating along the fiber axis, which is irradiated by photon induced periodic refractive index modulation. When the light propagates along the optical fiber, if the Bragg condition is met, the specific wavelength signal reflected at each of the grating planes will add constructively to form a back reflected signal with central length called Bragg wavelength, expressed as Eq. (1), according the mode coupling theory.

where$n_{eff}$is the effective refractive index,$\Lambda$is the spatial period of grating plane.

The$n_{eff}$and$\Lambda$of FBG will be changed by any strain or ambient temperature. Then the relative magnitude of the shift of Bragg wavelength is given by

where$P_e$is the effective photoelastic constant, $\epsilon$is the induced stress,${\alpha }_T$and$\xi$are the thermal expansion coefficient and the thermo-optic coefficient of the single mode fiber, respectively and$\Delta T$is the temperature change.

Since the optical fiber is immune to the humidity, the FBG-based RH sensor is fabricated by coating moisture sensitive polymer on the surface of fiber cladding covering the Bragg grating section. When the ambient RH changes, the volume expansion of the polymer layer convert the change of RH into mechanical strain and imposes on the FBG to induce a shift on Bragg wavelength. As a result, applying Eq. (2) on the RH measurement,$\epsilon$is the combination of thermal and volume expansion of polymer layer.

The strain imposed on the FBG due to thermal expansion of polymer layer is given by

where${\alpha }_{RH}$is the thermal expansion coefficient of polyimide film.

Assuming the humidity-induced strain due to volume expansion of polymer layer is a linear function of the change of RH,${\epsilon }_{RH}$can be expressed as [11]

where$\beta$is an average moisture expansion coefficient, can be back-calculated through experimental measurements.

Therefore, the total shift of Bragg wavelength of the RH sensor can be obtained accordingly, expressed as

It is nature to use the solutions of strain measurement by FBG to reduce the humidity-temperature cross sensitivity, such as taking another normal FBG to detect the surrounding temperature as a differential reference.

2.2. Serial TDM sensing network

Assuming$n$identical FBGs are written on a fiber serially, in response to each input laser pulse, successive pulses return from the serially cascaded FBGs. The delay time between two neighboring sensors is [12]

Where${\tau }_i$is the arriving delay of FBG_i reflected pulses, $c$is the speed of light in vacuum, $n$is the effective refractive index of optical fiber and$L_i$is the distance of FBG_i from the light entrance.

If the multiple reflections among the gratings are neglected, the returned power from the FBGi can be approximated from literature

where$P_{ref}{{}_{,}}_i\left({\lambda }_B\right)$ is the reflected power with all other reflectivity neglected,$P_0\left({\lambda }_B\right)$is the input power of fiber,$\alpha \left({\lambda }_B\right)$is the spectral reflectivity of the FBG, which is approximately a Gaussian function.

In order to avoid the sensing signal overlapping from multi FBG, the input pulse width is chosen to be less than the time delay. Thus, in time field, different FBGs can be identified by different time delays. By sweeping the input light wavelength, the reflection spectrum of every FBG along the fiber can be resolved by

where$A$,$B$,$c$are the adjusted parameters, amplitude, center and deviation,$f\left({\lambda }_i\right)$is the calculated spectra of${\lambda }_i$. The peak value is$A$when${\lambda }_i=B$.

3. Experiments

3.1. Sensor fabrication

The large-scale FBG fabrication system for on-line writing FBG arrays during drawing fiber has been successfully developed [13]. The piece of sensing fiber with five ultra-weak FBGs used in this work was chopped from a fiber array with 1124 FBGs written by this technology. The RH measurement is to determine the shift of central wavelength. Before experiment, the initial central wavelength of all five sensors was recorded. The peak reflectivity of these FBGs was ~0.04% and 1 meter equidistance between FBG neighbors. First of all, the original coating layer protecting FBGs should be removed. Prior to the polymer deposition, the surface of clapping sections was cleaned by anhydrous alcohol and then the ultrasonic cleaner for 10 minutes under 30°C. It was heat-treated at 80°C for 20 minutes. Then the FBGs were dip-coating the silane coupling agent (silane coupling agent: alcohol: deionized water = 20%:72%:8%) for 10 minutes to enhance the adhesion at polymer interface, which can ensure the effective transfer of strain generated by polymer expansion to the gratings. Finally the FBG array were backed by drying cabinet at 80°C for 1 hour, then it was ready for the polymer layer coating.

Polyimide (ZKPI-305IIE, solid content: 12%~13%, viscosity: 5000-6000cp) was chosen as the polymeric layer material because of its linear and reversible response with humidity change. The fiber grating was dipped into the polyimide solution for 5-10minutes and dried in a drying cabinet for a short thermal treatment at 150°C.This process will be repeated several times in order to obtain the desired film thickness. The polyimide coating thickness is proportional to the repeating times, and can be measured by electron microscope. The rate of dip coating has strong effect on the smoothness and homogeneity of the polyimide film. In order to get a uniform polyimide film on the surface of the fiber. We set the rising and dropping rate at 10 μm/s. The configuration system of the dip-coating process is shown in Fig. 1.

 

Fig. 1 Dip-coating method for polyimide film on FBG fiber chain.

Download Full Size | PDF

Finally the coated FBG sensors were put into an oven for curing from 140°C to 240°C for about 1 hour at every 20°Cand then 250°Cfor 7 hours.

The coating profile of the five sensors were measurement by 500 × optical microscope (VHX-100), and the measured thickness were about 12.7μm, 36.9μm, 39.3μm, 51.8μm, 55.3μm, respectively, as shown in Fig. 2. Without loss of generality, the cross section of FBG1 was shown in the subplot. It witnessed the uniformity of Polyimide coating adhered on the fiber.

 

Fig. 2 Profile of five FBG sensors coated with polyimide film of different thickness.

Download Full Size | PDF

3.2. Experimental set-up

As shown in Fig. 3, the experimental set-up includes the following units: light source, semiconductor optical amplifier(SOA), optical circulator, FBG wavelength interrogation unit, polyimide-coated FBG sensors array, saturated salt solutions and terminal for data acquisition, processing and display.

 

Fig. 3 Experimental set-up of the RH measuring system.

Download Full Size | PDF

The light from amplified spontaneous emission (ASE) is filtered to nanosecond pulses. The SOA functions as a gating device as well as the first stage optical amplifier driven by a fixed phased pulse generator. The light out of SOA feeds to port 1 of the optical circulator which directs the signals to the FBG-based RH sensing array through port 2. The signal reflected by the RH sensors return to the input port of FBG wavelength interrogation unit via port 3 of the circulator. During the whole signal acquiring process, interrogation unit transmits the reflected signals of sensors to real-time programs on computer, which carry out the wavelength fitting and display the experimental results. The Bragg wavelength has a measurement uncertainty of ± 2pm. The ASE light source has average intensity of 13.5dBm in broadband from 1525nm to 1565nm. The position of RH sensors along the sensing array is positioned via reconstructing the reflection spectra of each FBG during the entire scanning period in time domain by ultra-high speed sampling (40Gs/s). The demodulation system reaps the benefits of multiplexing large capacity of RH sensors in the sensing network, and the cost for the demodulation of each RH sensor is relatively low.

The FBG wavelength interrogation unit generates nanosecond electrical pulse to control SOA. The light pulse outputs when SOA is open. In order to demodulate the central wavelength of every FBG sensor along the fiber easily, only one nanosecond light pulse is injected into the sensors array during one time demodulation. All the FBGs reflections by the light pulse are interval distributed in time domain. The time interval of two FBG sensors is determined by their optical distance. Then the spectrum of every FBG reflection can be reconstructed and the central wavelength can be calculated by Eq. (9). Two conditions are required in the interrogation system: firstly, the period of the scan light pulse injected into FBG array is longer than the double optical distance of the length of fiber; secondly, the light pulse width is narrower than the time interval of optical distance of neighbor FBG sensors.

Different saturated salt solution exhibits different RH stably in room environment, according to the international standard of salt solution saturated humidity value issued by the International Organization of Legal Metrology (IOML). Table 1. shows four kinds of typical saturated salt solutions with humidity values above the air of each saturated salt solution at 28.5°C. The RH sensor was suspended near above the surface of an enclosed in a sealed container. In order to eliminate the interference of strain and temperature, the circulator, the fiber with RH sensors and the saturated salt solutions container were all fixed on the test-bed, and the room temperature is fixed at 28.5°C. The actual RH value was measured by electric-chemical type (Center313, Temperature accuracy: ± 0.7°C, RH accuracy: ± 2.5%).

Table 1. Humidity Values Above the Air of Each Saturated Salt Solution at 28.5°C

View Table

3.3. Experimental results

In order to overview the RH characteristics of the five sensors TDM network, the sensors were placed into the humidity chamber and set to 28.5°C and 40.3%RH, 61.1%RH and 83.4% RH, respectively. After passed through optical circulator again, the light reflected from the sensors array was then fed into optical spectrum analyzer (YOKOGAWA AQ6370B). The spectrum of the sensor array as the RH changed was shown in Fig. 4. As expected, the superposition of reflected signals of five sensors witnessed the shift of wavelength according to the change of RH. Since all the FBGs were written at the same condition and their optical characteristics, including central wavelength were nearly identical, all the reflected spectrum of the five RH sensors were mostly overlapping in frequency domain, but separated in time domain as expressed in Eq. (7). Only the optical spectrum analyzer could not show the details of the status of each RH sensor along the fiber, not to mention the shift of wavelength. Then TDM technology was used in this wavelength demodulation system.

 

Fig. 4 Spectra overlap of five RH sensors with different humidity.

Download Full Size | PDF

Wavelength shifts of each FBG based RH sensor under different humidity levels were experimentally investigated. The reflected spectra of the five FBG-based RH sensors of the array were measured by wavelength demodulation system, as shown in Fig. 5, by placing all the sensors in 83.4% RH container. Based on Eq. (9), the peak wavelengths were 1552.9924nm, 1552.8426nm, 1552.8128nm, 1552.894nm, and 1552.8128nm, respectively. The average value of their 3-dB bandwidths was ~0.12nm. In case of the used SMF-28 fiber, 1 ns time delay corresponds to about 0.1m distance along the fiber based on Eq. (1). The time of achieving the peak wavelength was 13ns, 23ns, 33ns, 42ns, 52ns in time field, which were close to the theoretical value of time delay between two adjacent FBGs was 10ns.

 

Fig. 5 Spectra of five RH sensors at the RH level of 83.4%.

Download Full Size | PDF

Since the balance of RH in the containers would be damaged each time while changing different saturated salt solutions, 10 minutes was given for the new balance reached in the replaced container before recording and measuring. Figure 6 show the spectrum shift of FBG2 when changed to 23.8%RH, while the other four FBGs were kept in the container of RH equal to 83.4%RH. Wavelength shift of each FBG was calculated with a Gaussian fitting algorithm in Eq. (9).

 

Fig. 6 Spectra of FBG2 at 23.8%RH and the others at 83.4% RH.

Download Full Size | PDF

Figure 7 presented the experimental results of the five sensors at different RH. The measurement data of each sensor was fitted to obtain its RH sensitivity, and the fitted coefficients were 1.134, 1.633, 1.639, 1.770 and 1.832pm/%RH, respectively. Since the experiments were conducted under room environment with constant temperature, the last terms of Eq. (6), related to the temperature effect of sensor, could be ignored. The moisture expansion coefficient of the optical fiber was tiny and could be neglected comparing to that of polyimide. The experimental results witnessed that the shift of wavelength exhibited the same linear trend in terms of change with RH, which was consistent with the theoretical analysis.

 

Fig. 7 Measurement results of the five sensors at different RH.

Download Full Size | PDF

The RH sensitivity values of sensors with different coating thickness were shown in the last subplot of Fig. 8.

 

Fig. 8 The RH sensitivity values of sensors vs. different Coating thickness.

Download Full Size | PDF

It can be concluded that the RH sensitivity exhibited nearly linear correlation to the coating thickness. Since the experiments were conducted at a fixed temperature, based on Eq. (6), ignoring the effect of temperature and the defect of the FBGs etc. and assuming all FBGs have the same fitting curve ratio, then β and P_e can be extracted with value of 1.999 and 0.22 respectively.

4. Conclusions

In this work, polyimide coated RH sensors based on ultra-weak FBGs was proposed and experimentally demonstrated by combined an SOA and TDM method for interrogation of a large number of serial RH sensing network. Limited by the fabrication capability, a large number of sensors, such as thousands of sensors, have not yet been confirmed. Based on the report of the multiplexing capacity of ultra-weak FBGs, the number of multiplexing can reach to several thousand. The proposed method was demonstrated with 5 RH sensors with the reflectivity of FBG ~0.04%. Thus, it is expected that this scheme of ultra-weak FBG-based RH sensor network can potentially multiplex thousands of RH sensors in single fiber.