The original work on the diffusion of hydrogen through palladium was carried out by Thomas Graham in 1866, when he was Master of the Royal Mint, in London (1). However, for many years the commercial use of palladium as a hydrogen diffusion membrane was seriously inhibited by the fact that at relatively low temperatures the adsorption of hydrogen into palladium induces an α → β phase transformation, which consequently changes the atom spacing in the metal lattice. This dimensional change is large enough to distort the palladium membrane to such an extent that after only 30 hydrogenation-dehydrogenation cycles very little of the original structure remains. As a consequence of this, purification of hydrogen using a palladium membrane was only possible if the membrane was kept above a critical temperature of about 300°C throughout its working life, a requirement that did not favour its commercial use.

It was nearly a century after Graham’s discovery that Dr. J. B. Hunter, while working on the diffusion of hydrogen through palladium alloys, found that the maximum rate of hydrogen diffusion through silver-palladium alloys occurred at a composition close to the maximum solubility of silver in palladium (2). Unexpectedly the alloy was found to be stable when thermally cycled through the α − β phase transition temperature in the presence of hydrogen. Furthermore the addition of silver made metal fabrication easier as well as reducing the intrinsic value of the alloy. This was a major step forward and led to the development of a range of commercial diffusion cells.

Early hydrogen diffusion equipment was of a simple design. Fabricators of the platinum group metals rarely make a complete product. Generally they are involved in supplying components which are often intricate and complex in design, but which comprise only a part of the complete product. Thus when fabrication of hydrogen diffusion equipment commenced only the diffusion cells were made; market indications suggesting that potential purchasers would want to construct diffusion plants to their own design, or to incorporate diffusion cells into existing plant. The diffusion cells were of robust construction and had an output of 1 to 500 standard cubic feet per hour (approx. 0.03 m³/h to 14m³/h). They were designed to operate at temperatures of 500°C and pressures of 500 psi (3); years later some are still in everyday use. Three early models are shown in Figure 1. The actual diffusion membrane consists of a number of silver-palladium tubes, which are contained within the cylindrical cell.

Although the supply of diffusion cell components met the needs of a certain market it soon became apparent that this approach did not satisfy all demands and that there was a requirement for complete purification units. In response to this a series of small laboratory diffusion units were designed and manufactured. Since modified in design, such units still provide the purest form of hydrogen in a great many laboratories.

To provide a considerably higher rate of hydrogen output (286 cubic metres of hydrogen per hour) a very few much larger diffusion units were built up from standard diffusion cells but the safeguards against contaminating impurities were less than adequate and these early units did not have satisfactory lifetimes.

The expansion of the electronics industry, which uses hydrogen during the manufacture of silicon chips, resulted in a rapid increase in the demand for hydrogen of a suitable quality. The hydrogen is used to carry doping elements onto the surface of the silicon wafer, and the gas must be of the highest purity. In this high technology industry a reliable supply of high purity gas is regarded as an essential everyday service, and new designs of palladium alloy diffusion units were required to satisfy this demand for hydrogen of near absolute purity. In practice it was found that a moisture meter installed in the gas stream served to monitor the purity of the hydrogen, and once in service dew points below −70°C are steadily recorded in diffusion units. Two typical purification units which use silver-palladium diffusion membranes are shown in Figure 2.

Many electronics factories are located long distances from traditional sources of commercial cylinder hydrogen, making them vulnerable to delivery delays caused by unusual weather or traffic conditions. A need for hydrogen self-sufficiency was recognised and led to the development of simple generators for on-site hydrogen production (4). These are fuelled by a methanol/water mixture which is cracked over a catalyst, the resulting hydrogen being extracted by diffusion through a palladium alloy membrane (5). The methanol/water mixture and its constituents are cheap and internationally available.

The use of this fuel effectively allows hydrogen to be stored as a liquid, moreover as a liquid which can be contained in thin walled vessels at atmospheric pressure, so providing a safer method of hydrogen storage than when the gas is kept under compression in high pressure cylinders.

The first methanol/water hydrogen generator built by Johnson Matthey was successfully used in the Antarctic by the British Antarctic Survey, who in 1975 installed it in a simple shelter built out of its own packing case. This shelter is now buried deep in the Antarctic ice, under many years’ accumulated snowfalls, and a third replacement unit has been supplied. The new G2 generator, capable of producing up to 2 cubic metres of hydrogen per hour at atmospheric pressure, is installed in one section of a caboose (Figure 3) where the electrical controls are also housed. However, the low pressure hydrogen store is located in an adjacent section. The whole caboose is sledge mounted and it is hoped that by frequently moving this around burial in the snow will be prevented.