1. Introduction

Hydrogen (H₂) plays a pivotal role in renewable alternatives and in reducing undesirable emissions [1,2]. As clean energy, it is commonly used as a power source for rockets, life support systems, and computers in the space environment. However, there is a special need to ensure the safety concentrations of hydrogen because it is explosive flammable, odorless, colorless, and volatile. However, there is a special need to ensure safe concentrations of hydrogen because of a low explosive limit of 4% at room temperature, combined with its low ignition energy and high flame propagation speed. Also, it is flammable, odorless, colorless, easily volatile, and quickly spread, so it can eventually leak from the container owing to its reduced molecular size. Consequently, there has been an increased demand for the development of low-cost, compact H₂ sensors with high performance, including a low limit of detection, and fast response, at room temperature.

Currently, electrical and optical technologies are widely used in commercially available hydrogen sensors. As an accurate technique, electrical sensors with a reasonable service life are popularly used in the industry. However, such electrical gas sensors suffer from increased sensitivity at high operating temperatures, which causes safety concerns [3,4]. Alternatively, the optical sensor has been based on various optics configurations, e.g. a glass prism [5] or a micro-mirror [6]. Sensors based on optical fiber have appeared, which offer many advantages, including corrosion resistance, safety, or suitability in harsh environments [7–9]. To date, various principles of optical fiber sensors, e.g. interference-based [10], hollow core fiber [11], exposed cores [12], or U-bent fiber [13], have been demonstrated. However, such fiber-optic sensors must disrupt their structure to achieve a special architecture for interacting with the external environment. Such destructed structures of the optical fiber may cause the unstable performance of sensors with less service life.

The use of fiber gratings is advantageous because it allows us to preserve the integrity of the fiber and provide temperature-insensitive measurements [14,15]. Especially, as short-period gratings, tilted fiber Bragg gratings (TFBGs) can excite a large of cladding resonance modes since its periodic and constant modulation of the refractive index of the optical fiber core. Some of these cladding resonance modes can interact with the out medium of the fiber and thus sensitivities to the surrounding refractive index. In other words, TFBGs would be sensitive to the surrounding refractive index because of the excitation of the cladding resonance modes by their structure.

Here, we propose a hydrogen optical fiber sensor by combining TFBGs and palladium (Pd). Unlike other hydrogen-sensing materials such as WO₃ [16] and ZnO [17], palladium is the most widely used for hydrogen sensing since its reversible phase transition can only be influenced by the concentrations of hydrogen change, so palladium is a particularly suitable functional material for specific hydrogen detection at room temperature [18]. Hydrogenation occurs when the H atoms occupy sites of the interstitial lattice (α–phase) and even leads to saturation of the Pd lattice (β–phase) at high hydrogen concentration [19]. This phase transition of palladium cause a change in the dielectric function (refractive index) and thus makes it suitable to deposit on the TFBGs surface as a hydrogen-sensitive material. The proper palladium thickness on the TFBGs surface was optimized using a complex vectorial finite-difference algorithm. Finally, the optical response of Pd-coated TFBGs to low hydrogen levels with dry and wet atmospheres is studied and a limit of detection (LOD) of ∼348 ppm was achieved.

2. Methods

2.1 Fabrication of the Pd-coated TFBGs

A 10 mm long TFBGs of tilt angle 10-degrees was inscribed in the commercial single-mode optical fiber based on the phase mask method, using a fiber grating manufacturing system (Noria, NorthLab Photonics Ltd.) and a custom phase-mask with a tilted interferometric holographic pattern [20,21]. Then, the deposition of palladium on the surface of the TFBGs was achieved through sputtering using a radio-frequency power source with the condition of pressure 1E-3 Pa and of radio-frequency 125W.

2.2 Sensing method and simulation

Figure 1 shows cladding modes of excitation with Pd-coated TFBGs. It can couple the incident core mode into a number of cladding modes, e.g. guided cladding modes, cut-off mode, and leaky cladding. Some of these special cladding modes provide a mechanism of interaction between the light and surrounding medium for sensing the surrounding environment. Especially, the cut-off cladding mode exhibits a maximum sensitivity of the surrounding refractive index since its evanescent field extends furthest on the surface of the fiber.

 

Fig. 1. Cladding modes of excitation with Pd-coated TFBGs.

Download Full Size | PDF

On the contrary, the guided cladding modes don’t sensitive to the out medium of the fiber since its fully reflected inside the fiber resulting from its higher effective refractive index than the refractive index of the out medium of the fiber. The leaky cladding modes, on the other hand, are hard to be tracked because they are totally coupled to the outside of the fiber caused by the less effective refractive index of the leaky cladding mode compare to the refractive index of the out medium of the fiber [22,23].

In this work, we monitor a cut-off cladding mode using an optical spectrum analyzer to interrogate the refractive change of the palladium coating when hydrogenation happens.

The transmission spectra of TFBGs with different thickness palladium depositions were simulated in order to optimize the thickness of the palladium film to improve the hydrogen sensing performance. To account for the theory of coupled modes to adapt the results to the specific case of transmission of the TFBGs, resonance modes of the TFBGs structure were calculated using a complex vector finite difference algorithm. The modeling includes the core, cladding, palladium deposition, and the surrounding medium [25]. Modeling parameters of the TFBGs can be seen in Table 1. The refractive index and radius of the fiber, as well as various thickness values of the palladium with the refractive index, have been set in the modeling. It is worth to be mentioned that the palladium thickness values are ranging from 5 nm to 50 nm.

Table 1. Parameters of the TFBGs in the simulation

View Table

Figure 2(a) shows the simulated results of the cut-off mode coupled relationship with the evanescent field. As can be seen in the gray-shaded region, the cut-off mode appears for several values of thickness, which is materialized by the attenuation of the cladding mode resonances. The amplitude of the cut-off mode resonance intensity should be as high as possible to increase the sensor's effective interrogation [26]. Thus, a 10 nm thick palladium deposition on the sensor is the best-optimized result. As can be seen, TFBGs with 10 nm thick palladium led to a cut-off location of around 1336.7 nm in the air atmosphere. This is where the sharply reduced amplitude of the resonance intensity, while the maximum optical power is coupled to the evanescent field. Figure 2(b) presents the measured transmission spectrum of the as-fabricated sensor with the wavelength ranges of 1330 ∼ 1360 nm which is the same as the simulations. As can be seen, the cut-off mode in the expected wavelength range is consistent with the simulation results shadowed in grey. As the boundary between of the leaky cladding and the guided cladding mode, the cut-off area showed clearly by the envelope of the transmission of the Pd-coated TFBGs.

 

Fig. 2. (a) Simulation of the mode attenuation near the cut-off mode of TFBGs with different Pd thicknesses shadowed in grey, (b) measured transmission spectrum of a 10 nm thick Pd-coated TFBGs, and (c) transmission spectra of the palladium coated TFBGs with the cut-off mode marked by the blue asterisk in air. The inlays highlight the detail of the cut-off region and the core mode (λ_Bragg), respectively.

Download Full Size | PDF

The full transmission of the TFBGs coated with different thicknesses of palladium layer are shown in Fig. 2(c) where it is possible to compare the amplitude of the mode intensity of the cut-off cladding mode. The inlay illustrates the amplitude change of the intensity of the cut-off mode when the palladium-coated TFBGs with a thickness of 10 nm and 100 nm are in the presence of air. As can be seen, the cut-off areas are characterized by a suddenly increased mode loss, which induces a sharply reduced amplitude of cladding mode. The result shows the biggest reduced amplitude of the resonance when the TFBGs is coated with a 10 nm thickness of palladium. Finally, the location of the core mode (λ_Bragg) is at 1580.82 nm. Commonly, the core mode is a reference for the stability of environmental temperature due to its sensitivity to the temperance whereas insensitivity to the surrounding refractive index [27]. Thus, any undesirable temperature and intensity level fluctuation effects can therefore be subtracted from the sensor response.

Figure 3 shows the process of the reversible phase transition of palladium from a metal to a metal hydride state. This process was caused by the insertion of atomic hydrogen into the palladium crystal lattice and consequently induced the effective refractive index (ERI) modulation of the palladium film covering the surface of TFBGs. Finally, the cut-off resonance of the palladium-coated TFBGs would change in relative intensity shift or wavelength shift because of the hydrogen-induced ERI modulation.

 

Fig. 3. Phase transition of palladium when exposed to the hydrogen atmosphere.

Download Full Size | PDF

3. Results and discussion

Figure 4 presents the setup to evaluate the performance and suitability of the sensor which was introduced into a gas chamber. As the figure illustrate, two cylinders of hydrogen and synthetic air were connected to the chamber via two gas inputs, respectively. Then, the changes in hydrogen concentrations can be reached by controlling the flow of hydrogen and air. In order to select a radial polarization, a polarization controller was connected to a source (ASLD-CWDM-5-B-FA, 1250∼1650 nm, Amonics Ltd.). Finally, an optical spectrum analyzer with a resolution of 0.02 nm and a wavelength range of 600–1700nm (AQ6370D, Yokogawa Ltd.), as an interrogator was carried out to monitor the change of the transmission when the sensor was exposed to the presence of hydrogen.

 

Fig. 4. Experimental setup for evaluating sensor performance in a hydrogen atmosphere.

Download Full Size | PDF

Figure 5(a) shows the simulation results of the cut-off response as varied refractive of the surrounding medium. As the figure shows, the relative intensity was changed when the surrounding refractive index of the sensor changed from 1.00027 to 1.00032. In addition, the result shows that the cut-off would change even though the surrounding refractive index (SRI) is as small as 1E-5. Actually, both wavelength and intensity can be used to interrogate the cut-off cladding mode [28] when the difference of the SRI was obvious, e.g., 1E-2 [29], but the second method would be used when the change of SRI is too small. Figure 5(b) shows the change in the intensity of the sensor’s cut-off mode when it is in the air to the one in the atmosphere of a 1% hydrogen concentration (by volume). The intensity change because of the transformation of palladium into palladium hydride. The method of intensity change was used here since it is the predominant alteration in the spectrum in the measured results owing to the refractive change of the concentration of hydrogen from 0.1% to 0.4% being too small.

 

Fig. 5. (a) Simulation of intensity change of the cut-off cladding mode when 10 nm thick Pd-coated TFBGs exposure to varied surrounding refractive index, (b) experiment of the cut-off mode intensity changes when the same thickness Pd-coated TFBGs exposure to hydrogen atmosphere with concentrations between 0% and 1%, (c) the intensity and stabilization time values of the hydrogen response for association phase when the sensor was exposed in wet and dry air, and (d) the linear response of the sensor with the concentration range from 0.05%∼0.4% in volume.

Download Full Size | PDF

Figure 5(c) presents the sensor’s performance for hydrogen detection. The wet and dry air atmospheres were used to simulate different kinds of leaks by performing the changes from the air environment to others with some hydrogen content. All of the simulated hydrogen concentrations are lower than the explosive limit of hydrogen. As the figure illustrates, the sensor reacts to hydrogen with different concentrations from 0.1% to 0.4% (by volume). The relative intensity changes would be towards previous values after introducing airflow to flush hydrogen out. As shown in the figure, a maximum intensity change of 0.62 dB can be achieved when the sensor is exposed to 0.4% hydrogen in wet air. The relative change in intensity is higher when the sensor was in a wet atmosphere because of the increased absolute concentration of hydrogen caused by the humidity. Additionally, during the association phase, the sensor’s average stabilization time is 4 min. During the dissociation phase, on the other hand, the sensor’s stabilization time is longer owing to the kinetics of the palladium deposition, corresponding to a range of 3∼20 min. The higher concentration of hydrogen, the longer the stabilization time the sensor need. According to Fig. 5(c), the intensity fluctuation value of ± 0.01 dB, which was the average standard deviation ($\sigma$) of the cut-off mode amplitude from all constant concentrations (including baseline), was calculated. And it was considered to estimate the sensor’s theoretical limit of detection (LOD), using the following Eq. [8]:

Combining the black line from Fig. 5(d), $f\left( x \right) = - 0.029 + 1.692x$, thus, the limit of detection (${x_{LOD}}$) was calculated to be 348 ppm. Meanwhile, the sensitivity of the sensor in wet air can be obtained via the slope of the black line, corresponding to 1.692 dB/%. Although the sensor shows a better performance in a wet atmosphere, the linear fitting is lower than the one in a dry atmosphere. As Fig. 5(d) shows, the sensor’s linear fitting is 98% when it works in dry air with a concentration range of 0.05% ∼ 0.4%.

4. Conclusion

The work reported here descript the procedure to achieve cut-off cladding mode excitation with palladium-coated TFBGs. After coating this metal, the surrounding refractive index around TFBGs is changed thus influencing the excitation of the cut-off mode in the air. Simulations to optimize the thickness of palladium were carried out, and results show that a maximum evanescent field can be excited by the cut-off mode resonance when the TFBGs is coated with a 10 nm thickness of palladium. Then, a 10 nm thick palladium-coated TFBGs hydrogen sensor, whose transmission spectrum is good to agree with previous simulations, was fabricated. And the hydrogen responsivity of the as-fabricated sensor was tested. A fast response time of 4 min to hydrogen and a low theoretical LOD of 0.0348% or 348 ppm are achieved, respectively.

Together with the inherent properties of optical fibers, the palladium-coated TFBGs opens the possibility for hydrogen detection in hard-to-reach areas. Moreover, these sensors provide promising platform to develop various smart sensors after combination of other advanced functional materials for artificial intelligence technology.