Using hydrogen as fuel presents a potential risk of explosion and requires low cost and efficient leak sensors. We present here a hybrid sensor configuration consisting of a long period fiber grating (LPFG) and a superimposed uniform fiber Bragg grating (FBG). Both gratings are covered with a sensitive layer made of WO3 doped with Pt on which H2 undergoes an exothermic reaction. The released heat increases the temperature around the gratings. In this configuration, the LPFG favors the exothermic reaction thanks to a light coupling to the sensitive layer while the FBG reflects the temperature change linked to the hydrogen concentration. Our sensor is very fast and suitable to detect low hydrogen concentrations in air whatever the relative humidity level and for temperatures down to -50 °C, which is without equivalent for other hydrogen optical sensors reported so far.
© 2008 Optical Society of America
Nowadays gas detection is a key issue in numerous fields such as environment, automotive, process control and chemical industries. Besides cumbersome and expensive classical detection systems, there exists a huge need for miniaturized systems, allowing to multiply the measurement points. Optical fiber sensors are particularly interesting, especially because of their explosion proof nature and their ability to provide numerous sensing points.
Among the gases concerned by detection systems, hydrogen is essential since it participates to a wide range of chemical processes and appears during energy production and transport. It is also an attractive fuel for clean engines. However, with its high diffusivity, H2 is extremely reactive. In air, it can burn at concentrations from 4 % with a flame velocity almost ten times higher than that of natural gas. Therefore, in order to reduce the risk of explosion due to potential leaks, H2 sensors are required. Owing to their numerous advantages, H2 optical fiber sensors have been the subject of recent researches.
In the past, Palladium (Pd) has been widely used with optical sensors as it exhibits a high and selective affinity for H2 . Among the different optical fiber sensor configurations, a great deal of the researches has been devoted to fiber Bragg gratings (FBGs) since they offer quasi-distributed sensing characteristics with a wavelength-encoded response. For Pd-coated FBGs [1–5], the sensing mechanism results from the Pd-coating swelling and induces a stress on the grating, shifting its central wavelength. Although efficient, these sensors suffer from two major limitations. First, except for the solution proposed in  that exploits the in-fiber laser light to increase the temperature around the grating, the operating time is long (several minutes) and leads to a hysteresis effect between the responses obtained for increasing and decreasing H2 concentrations. Second and more importantly, these sensors have only been tested in nitrogen and are not able to properly operate in air. It is an issue since practical applications such as the monitoring of storehouses and pipe lines require H2 sensors in air.
In this paper, supported by a recent work , we report a novel sensor prototype able to detect H2 leaks in air. The sensor is composed of hybrid fiber gratings covered with a catalytic sensitive layer on which H2 undergoes an exothermic reaction. The resulting release of heat increases the temperature around the gratings and shifts their resonance bands. Our sensor presents a very good sensitivity and is capable to detect H2 concentrations below the explosion limit whatever the relative humidity level of the environment. Our sensor also operates at temperatures down to -50 °C, which has no equivalent for other types of optical sensors.
2.1 Hydrogen sensor composition
Uniform FBGs and long period fiber gratings (LPFGs) were inscribed into standard single mode fiber by means of a frequency-doubled Argon-ion laser emitting at 244 nm. Prior to the UV exposure, the optical fiber was hydrogen-loaded at 70 °C and 200 atm during 48 hours. After the inscription, the gratings were annealed at 100 °C during 24 hours in order to stabilize their properties. As further explained, LPFGs were used for their radiative properties since they are transmissive gratings that couple light from the fiber core to the cladding .
To obtain the sensitive layer, nano-sized tungsten oxide powder was prepared using the solgel method . Aqueous sol-gel of tungstic acid (H2WO4) was obtained from Na2WO4 with protonated cation-exchange resin. In a first stage, a gel consisting of WO3. H2O was formed. The gel was washed, centrifuged several times with demineralized water and dried in air at 60 °C for 6h. Appropriate amounts of hexachloroplatinic acid (H2PtCl6) solution were added to the obtained powder. The mixture was finally annealed at 500 °C for 1h in order to obtain WO3 doped with Pt on its surface. At the end of the process, the active layer consists of WO3 nano-lamellae (squares of about 1 µm×1 µm×50 nm) with Pt dispersed on their surface. It was finally dispersed in a solvent in order to deposit a uniform layer of the sensitive material (several microns) on the stripped optical fiber using the dip-coating technique, ensuring in any case the same experimental conditions. The molar ratio Pt/W was about 1/14. Let us also add that the used layer does not react with other pollutants such as methane or carbon monoxide.
2.2 Experimental set-up
The set-up used to test the H2 sensors in different air environments of various relative humidity levels is depicted in Fig. 1. It was composed of two gas bottles, one of pure H2 and the other one of dry air. A bubbler filled with distilled water was used to modify the relative humidity level of air between 0 and 90 %. Three mass flow controllers provided a mixture of air and hydrogen with variable H2 concentrations from 0 to 4 % (the accuracy on the determination of the H2 concentration being equal to 1 %). The gas chamber was made of a 20 cm long and 1.5 cm wide glass cylinder with an inlet and an outlet to allow flowing in an out of the gas mixtures. Temperature and gas flow were continuously monitored. The amplitude spectrum of the tested gratings was measured by means of an ASE source covering the C+L bands (1520–1620 nm) and an optical spectrum analyzer (OSA) with an accuracy of 15 pm.
When the sensitive layer is in contact with a gas mixture of air and hydrogen, the oxidation of H2 molecules by O2 molecules contained in air occurs on its surface. This reaction is exothermic so that the temperature locally increases around the gratings. H2 sensing is therefore based on the monitoring of the resonance wavelength shift induced by the temperature change. To properly operate, this reaction requires an activation energy equal to 0.15 eV . This leads to a minimum detectable H2 concentration called detection threshold. In our previous work , we have shown that the detection threshold was equal to 3 % in dry air for a 1 cm long uniform FBG. We have also demonstrated the possibility to decrease the threshold value to 1 % in dry air with strongly reflective uniform FBGs. However, in that case, the FBGs are so strong that the accurate measurement of their Bragg wavelength is impossible. Moreover, a drastic increase of the threshold was obtained in wet air and for temperatures below 0 °C, which strongly limits the use of uniform FBGs for H2 sensing.
The presence of the threshold and the modification of its value due to the experimental conditions can be understood by examination of the heat flows exchanged on the sensitive layer surface in the presence of H2. The temperature experienced by the grating results from the equilibrium between the heat flow delivered by the exothermic reaction ϕr and the heat flow lost by exchange with the surrounding medium ϕth. In a very good approximation, we may consider that ϕth is dominated by radiation so that it can be expressed by:
where ε is the emissivity of the WO3 layer (close to 0.9 as it was estimated from infrared measurement of a surface of WO3 heated to a known temperature), σ is the Stefan-Boltzmann constant (5.673 10-8 W/(m2.K4)), T is the absolute temperature measured by the grating and Te is the absolute ambient temperature. ϕr is defined as:
where v is the reaction rate and ΔHr is the reaction heat and is equal to -57.8 kcal/mol .
The reaction rate v is proportional to the concentration of hydrogen CH2 and depends on the temperature through the usual Arrhenius formula. Ea is the activation energy (0.15 eV), R is the gas constant equal to 8.31 J/(mol.K) . k0 depends on the amount of Pt per surface unit and reflects the sensitive layer efficiency. Its value is not precisely known in practice. Using Eq. (1) and (2), the sensor behavior can be semi-quantitatively simulated from the condition φth=φr for different operating conditions. Figure 2 presents simulation results obtained for different ambient temperatures and for different sensitive layer efficiencies. Conformingly to our experiments, one can see on Fig. 2 that the detection threshold increases when both the ambient temperature and the sensitive layer efficiency decrease. Decreasing k0 gives the same qualitative effect as a humidity level increase since water molecules tend to inhibit the exothermic reaction occurring on the sensitive layer. Hence, it may explain the increase of the threshold level. However, for the global evolution, as the temperature increases while hydrogen reacts, the adsorbed water molecules disappear from the surface and the occupied reaction sites are released and can participate to the reaction. Consequently, as soon as the temperature exceeds 60 °C, the effect of water almost vanishes and the equilibrium response remains the same as in absence of water. This effect was not simulated so far in our analysis.
The results presented in Fig. 2 correspond to the situation where no external energy is provided to the sensitive layer. To avoid the limitations of uniform FBGs, we use in this work a hybrid configuration consisting of the superimposition of a uniform FBG within an LPFG. This solution takes profit of the light coupling to the cladding modes induced by the LPFG. Indeed, as the refractive index of the sensitive layer, consisting essentially of WO3, is slightly higher than that of pure silica, the light outcoupled in the cladding penetrates the sensitive layer and its energy (photon energy around 1500 nm ~0.7 eV) favors the exothermic reaction. The FBG is used as a probe to reflect the temperature change. The sensing mechanism is therefore based on the monitoring of the wavelength shift due to a temperature change in the reflected spectrum of the uniform FBG.
Both kinds of gratings were designed to minimize the detection threshold. The periodicity of the LPFG was chosen so as to obtain a strong resonance band (light coupled to the cladding) inside the spectral range of the optical source. A periodicity of 475 µm allowed us to obtain such a feature. A 1 cm long 545 nm period uniform FBG was then superimposed on the 3 cm long LPFG so as to obtain a point sensor. As shown in Fig. 3, it was written at the extremity of the LPFG to take a maximum profit of the light coupling provided by the LPFG. Fig. 3 also presents the transmitted and reflected spectra of a typical hybrid configuration.
Our first experiments consisted in testing the behavior of hybrid configurations composed of 3 cm long LPFGs characterized by different refractive index modulations, yielding various transmission losses in the C+L bands. The LPFGs were inscribed with the same optical power (55 mW) but different exposure times (translation velocity of the UV beam along the exposed fiber length set to 0.5 cm/s) so that the laser fluence was modified between the different inscriptions. In any case, the FBGs were characterized by a transmission loss of about 12 dB.
The radiating efficiency of every coated LPFG was evaluated by computing the ratio between the injected and transmitted optical powers integrated over the LPFG resonance bandwidth. The difference between these two quantities indeed reveals the total amount of radiated optical power. In addition, an examination of the fiber surface with an IR camera revealed that the radiation extends on a few centimeters beyond the LPFG end and is not uniform along the grating length. Hence, only a part of the total radiated power can be collected by the sensitive layer and is useful to favor the exothermic reaction. From our experiments, this part was roughly estimated to the third of the total radiated power.
Figure 4 shows that the LPFG radiating efficiency increases with respect to the exposure time. In particular, with the LPFG characterized by a 15 dB transmission loss, more than 60 % of the injected light is coupled out of the fiber core. For this grating and with the ASE source used here, we approximated to 0.2 mW the optical power collected by the sensitive layer. This quantity corresponds to about 1.5 1020 photons/(s.m2 of fiber), which is sufficient to favor the reaction with respect to the total number of H2 adsorption sites per surface unit of the sensitive layer (estimated to 1019/m2 of fiber). The exact quantum efficiency is currently unknown and the involved mechanisms are thus being investigated.
Figure 5 confirms the threshold reduction obtained provided by LPFGs. With the LPFG characterized by a 15 dB loss, the threshold has been measured equal to 0.6 % of H2 concentration, which is 2.4 % less than a single uniform FBG. Error bars (15 pm in wavelength and 1% in H2 concentration) display the measurement uncertainty. It must be noticed that for all graphs, the response (wavelength shift versus concentration) is approximately the same, yielding the same sensitivity. This is due to the fact that the response is directly linked to the temperature reached thank to the exothermic reaction. For all sensors, the response is linear and reversible, with a mean sensitivity equal to 198 pm per 0.1 % of H2 concentration, which is easy to detect with a standard instrumentation. The sensor response time was measured of about one second for H2 concentrations above the detection threshold.
The sensor behavior was also tested in wet air environments and at various temperatures. Figure 6 presents the Bragg wavelength shift of a hybrid sensor (configuration with the LPFG characterized by the 15 dB transmission loss) due to the H2 concentration in different wet air environments. The H2 detection threshold increases as the relative humidity level increases. In 90 % wet air, the threshold has been measured equal to 0.9 % H2 concentration. This is a real improvement in comparison to a single 1 cm long uniform FBG sensor for which the threshold value has been measured higher than 3 % in 90 % wet air.
Finally, Fig. 7 confirms that a surrounding temperature decrease limits the sensor performances since more energy is required to initiate the exothermic reaction for a given H2 concentration. However, while a single FBG is not sensitive to H2 concentrations up to 4.0 % at -50 °C, the hybrid configuration presents a detection threshold equal to 1.5 %.
Hence, in comparison to the use of single FBGs, the hybrid configuration presents the important advantage to decrease the detection threshold value for any experimental conditions. This feature is linked to the use of LPFGs and is possible without increasing the optical source power density. It thus allows to work with a standard optical source (total power of 15 mW in our case), which keeps as low as possible the sensor price. Let us also mention that using uniform FBGs as probes instead of LPFGs presents two assets. First, the sensor response is encoded in the Bragg wavelength and is therefore not influenced by bending effects, which is not the case for LPFGs . Second, while the resonance band of an LPFG extends on several tens of nanometers (complicating the realization of quasi-distributed sensors with such gratings), a wavelength window of several nanometers is sufficient to record the reflected spectrum evolution of uniform FBGs. Consequently, owing to the presence of uniform FBGs, the hybrid configuration can be used in frequency multiplexed systems. In particular, with LPFGs characterized by a transmission loss of 3 dB, up to 15 hybrid sensors can be cascaded along an optical fiber in the range of the used ASE source.
Superimposed hybrid fiber gratings were coated by a catalytic sensitive layer that heats the gratings in the presence of hydrogen in air. In this hybrid configuration, the LPFG provided a light energy coupling to the sensitive layer to decrease the H2 detection threshold while the FBG was used to track the temperature increase. Very good sensing performances have been reported: fast response, high sensitivity, reversibility, frequency multiplexing capability and H2 concentrations detection well below the explosion limit of 4 %, whatever the relative humidity level and for temperatures down to -50 °C.
C. Caucheteur is supported by the Fonds National de la Recherche Scientifique. Authors acknowledge the Attraction Pole Program of the Belgian Science Policy IAP 6/10.
References and links
1. R. Maier, B. Jones, J. Barton, S. McCulloch, T. Allsop, J. Jones, and I. Bennion, “Fibre optics in palladium-based hydrogen sensing,” J. Opt. A: Pure Appl. Opt. 9, S45–S59 (2007). [CrossRef]
2. B. Sutapun, M. Tabib-Azar, and A. Kazemi, “Pd-coated elastooptic fiber optic Bragg grating sensors for multiplexed hydrogen sensing,” Sens. Actuators B 60, 27–34 (1999). [CrossRef]
3. T. Peng, Y. Tang, and J. Sirkis, “Hydrogen sensors based on palladium electroplated fiber Bragg gratings,” SPIE Proc. 13th International Conference on Optical Fiber Sensors and Workshop on Device and System Technology toward Future Optical Fiber Communication and Sensing, 171–179 (1999).
4. D. Zalvidea, A. Diez, J. L. Cruz, and M. V. Andres, “Wavelength multiplexed hydrogen sensor based on palladium-coated fibre-taper and Bragg grating,” Electron. Lett. 40, 301–302 (2004). [CrossRef]
5. R. Maier, J. Barton, J. Jones, S. McCulloch, B. Jones, and G. Burnell, “Palladium-based hydrogen sensing for monitoring of ageing materials,” Meas. Sci. Technol. 17, 1118–1123 (2006). [CrossRef]
6. M. Buric, K. Chen, M. Bhattarai, P. Swinehart, and M. Maklad, “Active fiber Bragg grating hydrogen sensors for all-temperature operation,” IEEE Photon. Technol. Lett. 19, 255–257 (2007). [CrossRef]
7. C. Caucheteur, M. Debliquy, D. Lahem, and P. Mégret, “Catalytic fiber Bragg grating sensor for hydrogen leak detection in air,” IEEE Photon. Technol. Lett. 20, 96–98 (2008). [CrossRef]
8. S. James and R. Tatam, “Optical fibre long-period grating sensors: characteristics and application,” Meas. Sci. Technol. 14, R49–R61 (2003). [CrossRef]
9. S. Okazaki, H. Nakagawa, S. Azakura, Y. Tomiuchi, N. Tsuji, H. Murayama, and M. Washiya, “Sensing characteristics of an optical fiber sensor for hydrogen leak,” Sens. Actuators B 93, 142–147 (2003). [CrossRef]
10. C. Appel, J. Mantzaras, R. Schaeren, R. Bombach, and A. Inaen, “Catalytic combustion of hydrogen-air mixtures over platine: validation of hetero/homogeneous chemical reaction schemes,” Clean Air 5, 21–44 (2004).