Three modifications have been proposed for hydrogen sensors in which hydrogen in the air causes changes in the reflection spectra of interferometric film stacks that are interrogated via optical fibers. One of the films in the stack in each such sensor contains palladium, which reacts with hydrogen to form palladium hydride. The differences between the optical absorbances and indices of refraction of palladium and palladium hydride give rise to observable changes in spectra.
In an unmodified sensor of this type, the film that contains the palladium is reflective and is the outermost film in the stack, which is deposited on a thin, transparent glass substrate. The substrate, in turn, is adhesively bonded to the polished tip of a ferrule in which the interrogating optical fiber is terminated. Unfortunately, it is difficult to obtain a robust change in the reflectance spectrum of this or any similar interferometric film stack when the outermost film is the one that changes. In this case, the differences between the indices of refraction and optical absorbances of palladium and palladium hydride in the outermost layer are not large enough to elicit the desired robust change.
The situation would be different if the film containing the palladium were the innermost layer in the stack (the layer closest to the interrogating optical fiber), because in such a case, the film would spectrally modulate all of the light entering and leaving the stack. Because the other films in the stack are made of non-porous materials with low permeability by hydrogen, it would not be practical to attempt to exploit this effect by placing the film containing the palladium on the glass substrate at the bottom of the stack. However, if the stack from the unmodified design were simply inverted and positioned with a small airgap between the palladium-containing layer and the optical fiber, the air containing the hydrogen could diffuse through the gap to the palladium-containing layer: this is the point of departure for one of the three proposed modifications.
Figure 1 illustrates a reflection-based sensor head as thus modified. The interferometric film would be deposited on a narrow sensor bar with the palladium-containing film on the outside. The bar would then be mounted in a plane perpendicular to the axis of the optical fiber, with the palladium-containing film facing the polished end of the fiber across an airgap about 0.1 mm wide. The gap would be maintained by use of a glass or ceramic ring bonded adhesively to the tip of a capillary tube, which would serve as a pedestal and as a ferrule to hold the optical fiber. A thin-film electrical heater could be deposited on the sensor bar along with the interferometric films to obtain intimate thermal contact and thus highly energy-efficient heating of the films to maintain the required operating temperature. Electrical contact with external circuitry that supplies the heating power could be made via metal pads evaporated onto the glass or ceramic ring.
The response delay of the sensor would be governed by the diffusion of hydrogen. Given the known diffusivity of hydrogen in air, it has been estimated that at standard atmospheric pressure and a temperature of 20 °C, the characteristic diffusion time for a sensor bar 0.2 mm wide would be about 0.11 ms; this would be an acceptable response delay in a typical application. The delay could be reduced by use of an air-sampling unit that provides sufficient flow in contact with the sensor bar and its environs.
Another proposed modification is based on the concept of measuring the transmission spectrum instead of the reflection spectrum. The impetus for this modification is the fact that the change in the transmission of a palladium-containing film upon exposure to hydrogen exceeds the change in its reflection; consequently, at a given hydrogen concentration, the output of a transmission-based sensor should exceed that of the reflection-based sensor described above.
Figure 2 shows two alternative configurations for the proposed transmission-based sensor. In the first configuration, light from a small source would be collimated by a lens. The light would pass through the etalon, then through another lens that would focus the light onto a photodetector. Hydrogen would diffuse to the etalon through a small airgap between the etalon and the focusing GRIN lens.
The source of light could be two light-emitting diodes (LEDs) of different colors mounted in proximity on the same header. One LED would be chosen to have a wavelength (about 450 nm) in the vicinity of maximum change in transmission upon exposure to hydrogen. The other LED would be chosen to have a wavelength (typically in the infrared) for which there is little change in optical properties on exposure to hydrogen; the signal from this LED would serve as an amplitude reference.
Alternatively, the source of light could comprise multiple LEDs that would be turned on and off in sequence to obtain the transmission responses at the various LED wavelengths. Yet another alternative would be to illuminate the etalon with white light, in which case it would be necessary to analyze the transmitted light by use of a spectrometer. A suitable spectrometer could be constructed from a diffraction grating and a linear array of photodiodes. This alternative would provide more information than would the two-LED design described above, albeit at increased complexity and cost.
Although lenses shown in the figure are of the gradient-index-of-refraction (GRIN) type, other types could be used. The advantage of GRIN lenses in this application is that their ends can be made flat; this makes it possible to fabricate the etalon directly on the end of the collimating lens, reducing the size and complexity of the sensor.
The second configuration would involve the same principle of operation, but light would be coupled into and out of the sensor by different means. In this configuration, light would be delivered to the etalon from a remote source via a first optical fiber, and light transmitted through the etalon would be delivered to the photodetector(s) or spectrometer via a second optical fiber. In this case, the etalon could be fabricated on the polished output end of the first optical fiber, and the airgap would lie between the etalon and the input end of the second optical fiber. (Alternatively, the etalon could be fabricated on the end of the second optical fiber.) The core of the second fiber would be made wider than the core of the first fiber so that the second fiber could capture all of the transmitted light, without need for a collimating and a focusing lens.
The use of optical fibers would make it possible to mount the sensor away from the associated electronic circuitry. This could be advantageous in situations in which the circuitry would be vulnerable to electromagnetic interference or to damage if it were placed at the sensor location.
The third modification would be the introduction of pulsed optical heating to decrease response times. This modification would exploit the observations that (1) the response time in desorption of hydrogen is typically about 12 times that in absorption, (2) the response time decreases with temperature, and (3) optical modulation goes to zero as the temperature is increased to 80 °C, suggesting that all of the hydrogen becomes desorbed from the palladium-containing film at this temperature. These observations suggest that the overall response time could be greatly shortened if the desorption time could be shortened, and, in particular, if the film containing Pd could be heated quickly to 80 °C to effect complete desorption. In a typical case, the thermal-response (and thus desorption) time would be of the order of a millisecond. Thus, for practical purposes, the operating cycle could be shortened to the absorption time, which would typically be of the order of a minute.
One way to accomplish the rapid heating to 80 °C would be to pulse the interrogating source of light with enough power to obtain the required temperature rise. Optionally, if the pulsed light were infrared, then one of the films in the etalon (not necessarily the one containing Pd) could be made of InSb to enhance absorption of infrared. In the case of a transmission-based etalon sensor, the fraction of light transmitted would change during each illuminating pulse; specifically, the film containing Pd would become less transparent as the hydrogen was desorbed. Thus, the hydrogen content of the air could be inferred from the decrease, during each pulse, in the fraction of light transmitted. As an important side benefit of this approach, there would be no need to provide additional means to generate a reference or baseline signal because the optical signal would be returned to its baseline level as the hydrogen was expelled with each pulse.
This work was done by Elric W. Saaski and Charles Young of Research International, Inc., for Kennedy Space Center.