Schottky based hydrogen sensors
Researchers at QUB have developed novel Schottky junction-based detectors showing a hydrogen detection capability of 1 ppm (part per million).
Hydrogen sensors, an essential component of the future hydrogen energy industry, have drawn extensive research interest. Various types of hydrogen sensors, based on catalytic metals, Pd and Pt – for example, have been extensively studied over the past decade. In contrast to existing devices the QUB sensors use a simple, yet effective, method of detecting changes to the work function of the semiconductor material caused by the diffusion of the hydrogen. In spite of the relatively complicated physical mechanism the hydrogen concentration in such sensor is measured via simply monitoring the resistance of the thin catalytic metallic film. The substantial change of the resistance of the metal electrode alone can then be used to calculate the hydrogen concentration
The Schottky device is fabricated by depositing a thin Pd strip (length ~ 160 μm, width ~ 4μm and thickness < 10 nm) over a n-InP substrate of doping density ~ 1 x 1017 cm-3. Metal contacts are deposited on the metal layer (for four-point contact resistance measurements) along with an Ohmic back contact on InP (for complementary IV measurements). The devices are currently fabricated at the EPSRC National III-V Laboratory, Sheffield, UK. The devices display a drop in resistance with increasing temperature across a limited temperature range (typically 50-100 K wide). Fig. 1 illustrates the case of Pd/p-Si devices where the resistance spans over an order of magnitude for the most sensitive device (uppermost curve).Exposure to hydrogen causes the regime of the resistance drop to recede to substantially lower temperatures. If the devices are on the edge of the resistance drop at ambient temperature, then the devices act as very sensitive hydrogen sensors. Figure 2 illustrates the Schottky diode sensitivity to the presence of hydrogen with concentration ranging from 20 ppm to 100 ppmHowever, the time scale of the response is currently slow – the initial response when the detector is ‘charging’ with hydrogen is on the order of 5000 s, thereafter the response time is reduced to a few 100s.
The devices can be tailored, however, for applications where there is a need to detect hydrogen at the ~1% level on a time scale of ~10s. These requirements can be met by monitoring the resistance of the devices – for example, there will be an initial glitch above 1500W in the resistance of figure 2 - or by operating a conventional Pd strip-on-insulator type of hydrogen detector in parallel to the Pd-on-semiconductor (Schottky) device. Both can be fabricated on the same chip at the same time, one on an exposed semiconductor region and the other on an oxide covered region.
The devices are inherently low cost. They are made using standard semiconductor fabrication techniques in standard semiconductor processing facilities. Even with very modest photolithography resolution (say 4 μm line width) very effective devices can be made. Allowing for an undemanding device size of say 200 x 200 μm (including provision for wire bonding), device density becomes 2,500 cm-2 which translates to ~ 175,000 devices per 4-inch wafer. The metal that is required for Schottky diode formation can be expensive (e.g. Pt or Pd) but only a ~10 nm layer is required which keeps costs down.
There is no need to integrate sophisticated signal processing electronics with the sensor - rather a simple sub-dollar microcontroller solution based on, for example, the Microchip PIC16F family will suffice. This can be programmed to use the basic temperature sensor data to create a digital output linearly proportional to the hydrogen concentration. As usual the final cost of the electronics will be dominated by the cable, the type of connector used, and, maybe, an RS485 physical level interface. The most expensive part of the final sensor will be the device packaging especially if the sensors are used in liquids.
Suitable packaging might involve either a semi-permeable coating applied directly to the semiconductor surface or a semi-permeable membrane integrated into the sensor package including a PCB with built-in electronics.
Hydrogen has the potential to be an important source of clean fuel and the use of accurate hydrogen monitoring will be a key component of several of its emerging areas.
Hydrogen monitoring will be important in hydrogen refuelling stations, service pump areas, and in the engine compartments of hydrogen-powered fuel cell vehicles. Any future moves towards the ‘hydrogen economy’ will thus demand high volumes of sensors as, for example, fast leak detectors
There is a large existing industrial market for slow but sensitive ‘hydrogen dipole’ type sensors. At the moment this market is largely served by technologies other than catalytic that are significantly more expensive.
Within the energy sector, Power transformers are expensive constituents of the electricity grid; they are also owned and used by some industrial customers which use substantial amounts of electricity.
Estimates of the total installed population of transformers worldwide depend on definition but there are probably at least 700,000 very large transformers in use. Perhaps unsurprisingly, some 2% of this installed base fails each year. Some of these are catastrophic failures with associated loss of life. Some 80% of transformer failure modes are predictable. Predictable failure modes generally develop slowly and are accompanied by detectable phenomenon such as electrical partial discharge signals during the early stage of insulation breakdown; gases (especially hydrogen) generated during oil or insulation breakdown