A5. Hydrogen sensor system

The work on the experimental program for development of hydrogen sensor systems was performed simultaneously at INERIS and UNIMAN.

INERIS conducted a test program based on the international standards IEC 61779-1&4 and was aimed at assessing the performance of some commercial available hydrogen sensors/detectors. These devices were both of the electrochemical and catalytic types i.e. the two types most often used in the industry.
These tests consisted of:

• determining the calibration curve for each of these devices;
• measuring the response and recovery time following a sudden variation of concentration; and
• studying the signal response to variation in environmental temperature.

From the results of the study, the following conclusions were drawn:

1. Continuous or discontinuous calibration leads to the same calibration curve. Since it leads to a faster calibration, the continuous approach might be a better alternative to the standard test defined in IEC 61779.
2. Because of its operating principle, the catalytic sensor needs a gas flow rate higher than 200 ml/min to give a reliable reading.
3. 2The catalytic sensor is five times faster than the electrochemical one to respond to a sudden exposure of hydrogen. To be more precise, the response time is approximately 10 s for the catalytic sensor and 50 s for the electrochemical sensor. These figures also apply for the recovery time.
4. According to JRC, the time required by the electrochemical sensor to respond to a hydrogen exposure of known concentration becomes higher when the gas flow rate is reduced (it could be twice as much by solely reducing the flow rate from 100 to 30 ml/min). This finding is particularly important when a sensor is intended to control the formation of an explosive atmosphere within a fuel cell cabinet. Indeed, the dynamic response measured in a laboratory could be significantly different from that in the cabinet. This important behaviour was not observed by INERIS. Further tests might be necessary to clarify this point.
5. Catalytic detectors were also prone to loss of sensitivity and drift of zero after a prolonged exposure of hydrogen. This emphasises the need to regularly calibrate these devices, perhaps weekly at the beginning and then less often when their use has increased.
6. Continuous measurement of the effect of temperature on sensor sensitivity does not appear reliable. It is better to let the signal stabilise for a moment before recording the output, while controlling very carefully parameters such as the sensor position, the gas velocity and its composition. Though more tests might be necessary to fully understand the effect of temperature, preliminary results seem to indicate that it has only minor impact on the signal deviation of the catalytic detectors.
7. Discontinuous measurement of the effect of humidity seems to be preferable too. Using this method, a higher humidity tends to increase the reading of a catalytic detector.
8. A catalytic detector is very sensitive to the presence of CO but the interference is only temporary. When the CO exposure ceases, the detector behaves in a normal way.

The thermal-conductivity sensor was not found to be very sensitive to temperature variation. But an increase of relative humidity could lead to a significant change in sensor reading (e.g. from 10 to 10.4% v/v when the relative humidity rises from 0 to 40% according to KI). This sensor is also prone to drift with time.

The part of the work performed by UNIMAN was intended to evaluate hydrogen sensors for leak detection and generate Deliverable D5.6 Report on Hydrogen Sensor Systems. It was agreed that the scope of the work in the sub-task focused on:

• Where to put the sensors in a room or enclosure;
• What sensors are available and their specification (response time, concentration range and sensitivity, lifetime of the sensor, ability to detect the LEL);
• Gaps and limitations in sensors commercially available;
• Comparative specification;
• Practical aspects of what to recommend in the IPG.

For the primary evaluation of hydrogen sensors, a literature review of optical fibre- based hydrogen sensors was completed. This literature survey considered the advantage of optical fibre sensors for measuring hydrogen, such as intrinsic safety of an all-optical probe within a flammable atmosphere. The range, sensitivity and application of various optical fibre hydrogen sensors developed to date was considered, together with disadvantages of palladium as a hydrogen sensor, including sensitivity to humidity, cross-sensitivity to other gases, problems with long-term stability and reversibility of sensors.

As a result of the analysis, the following conclusions were formulated:

• Optical fibre sensors are attractive for measuring hydrogen, due to the intrinsic safety of an all-optical probe for sensing a gas forming an explosive mixture;
• Palladium interacts with hydrogen to form palladium hydride, affecting both its mechanical and optical properties, creating strain and refractive index changes in the palladium coating. These effects can be monitored optically by using the hydrogen-induced strain in palladium to perturb a fibre Bragg gratings (FBGs), or by using the optical changes induced in palladium by exposure to hydrogen to affect light transmission in optical coatings interrogated using long period gratings, micro-mirrors, evanescent field and surface plasmon resonance effects.

For further evaluation of hydrogen sensors including the ways of their improvements, an examination of the optical effect of volumetric expansion and wavelength-specific transmission losses of palladium/polymer coatings on polymer optical fibres and other coated substrates was performed by UNIMAN Partner.

The concentrations of hydrogen used in the experiments ranged from 1 to 20%, with the upper limit set at 20% for safety reasons.

This work concluded the following:

• the application of palladium to the surface of a polymer optical fibre is unlikely to be the basis of an optical hydrogen gas sensor. Previous publications have used dry air or nitrogen in their experiments but for a field sensor a wide range of humidity levels will be encountered. The comparison between dry air and dry nitrogen highlights the problem of concluding results by comparing dry nitrogen and dry hydrogen.
• The colorimetric option offers a pigment-based sensor that is not affected by humidity and temperature, even though the overall transmission may be affected by these factors, there are reference and signal wavelengths available to compensate for this effect.