Products extracted from air currently account for more than 60% 0f the current US$45 billion world industrial gas business and more than 65% if associated services are included. In addition, the equivalent to a further US$10 billion of air gases is produced by end users.

The uses for these gases are endless, but basically depend on the strong reactive properties of oxygen and generally on the relative inertness of nitrogen and the nearly complete inertness of argon.

The other interesting feature of this business is that the raw material, air, is free and industrial gas companies have grown on their ability to develop economic technologies to separate and deliver its various components. Another unusual feature of the air gas business has been the ability of the industrial gas sector to develop \\$quot;market pull\\$quot; by applications research and development on its potential customers businesses. Typically, they will have shown just how much better it is to use oxygen in place of air in a particular process and generated market for themselves from early adopters and for the gas industry in general as the rest of the potential users \\$quot;catch up\\$quot;.

The separation of air into its components has origins in the early 17th century chemists who realised that air had components that supported life and a \\$quot;denatured\\$quot; air that didn\\$quot;t. In the 19th century oxygen was being produced by chemical means but it was the advent of the ability to liquefy and distil air that made all air gases available in reasonable quantities. Most people know that when a gas expands through a nozzle from a high pressure to a lower pressure cooling occurs. Linde of Germany and Hampson of the UK realised that if the expanded gas was used to cool down the unexpanded gas in a heat exchanger, the whole system would get colder and colder until the air liquefied. Figure 1 shows a picture of a Linde air liquefier in 1900.

Claude of France made some efficiency improvements by using an expansion engine instead of a pressure let-down valve and by the beginning of the 20th century the scene was set for the \\$quot;cryogenic\\$quot; distillation of air. By the second decade of the century Linde, BOC and Air Liquide were exploiting the inventions of their engineers and offering a range of plant sizes that are still appropriate at the smaller end of today\\$quot;s market.

The interesting developments that followed were driven by commercial ambition, but relied on the development of both production and user technologies. The development of oxygen use in steel-making was initially constrained by the availability of oxygen in sufficient quantities. As technology improved most steel was made using oxygen and oxygen price dropped. The innovative decision by Pool, the founder of Air Products, to site production plants at the customer site and sell the molecules, made the by product nitrogen readily available for blanketing and conveying duty. As plants got bigger and the cost of oxygen fell, chemical and other industries started to use oxygen, or enriched air in place of air in combustion and reaction systems. They too found use for nitrogen to ensure the safety of their plants and storage. With larger plants it became technically and economically possible to recover the nearly 1 % of argon in the air. This enabled the development of near inert shielding of cutting and welding and various spin-offs into other industries.

In simple money terms oxygen is the same price now that it was in 1920 and in real terms less than half the price. This is more than simple economies of scale; it represents continuous value improvement of the product for more than 80 years.

The industrial gas industry has also been subject to disruptive technologies in its own production systems. For many years, small users bought cylinder gases, medium users bought liquid gases and large users were served by a dedicated plant or a pipeline system linking several plants. The pricing regimes reflected both the cost of supply and the next best choice of the end customer. Onsite customers usually signed a 15 year contract with a high degree of \\$quot;take-or-pay\\$quot;. Bulk liquid customers signed 3-5 year contracts depending on jurisdiction and cylinder customers generally annual contracts. The last two modes were only challenged at the margin by very large users, but usually economic factors kept them in their place. Then a number of things happened in the 1970\\$quot;s that challenged the system; the German Coal Research Institute developed a \\$quot;coke\\$quot; molecular sieve that could separate air economically by pressure swing adsorption (PSA) to produce nitrogen of reasonable purity and a few major chemical companies decided to develop membranes to separate air. The former wished only to license its technology and provide the separation medium, the latter, Monsanto and Dow, were interested in the industrial gas market. Most of the industrial gas companies realised that these technologies were a threat to their liquid nitrogen business and that many customers did not really need the purity (of a few ppm of oxygen) that was the standard for liquid nitrogen (LIN). In most cases for blanketing and purging duties purities as low as 98% were quite adequate. Fortunately the users were very conservative on \\$quot;safety\\$quot; issues and market penetration was likely to be slow. The industrial gas industry responded in a variety of ways. A few companies took out licenses from Bergbau Forschung, others ignored the problem for a while and then undertook their own research to develop alternative molecular sieve material. Forecast done at the time showed that PSA based nitrogen could take 30% of the market and even with growth this would lead to unloading of existing liquid production assets. It was also realised that what could be done non-cryogenically for nitrogen could probably be done for oxygen and this was placed under development. Membrane systems were initially ignored because they were at a very early stage of process development and suffered from manufacturing and headering problems. They were only viewed a suitable for the smallest duties.

The principal problem for a variety of companies trying to break into the industrial gas market using PSA was that although the economics were good, there were so many moving valves, switching every few minutes, that reliability was an issue. The answer was simple; provide a liquid back-up system. The problem was that only the major industrial gas companies had the capacity to provide the liquid. In the end, the industrial gas industry converted its own most vulnerable liquid customers to small onsite customers served by PSA. The rights to the membrane systems were sold by their developers to industrial gas companies for the same reason.

The use of air gases continues to grow at around 5% per year. Oxygen finds new applications in combustion and gasification process and in the rapidly growing \\$quot;Homecare\\$quot; sector; nitrogen is used for tertiary oil recovery and natural gas calorific value correction. The markets for the two are a similar size. The noble gases find applications in numerous high technology fields from krypton lighting to xenon anaesthesia.

The growth opportunities continue to mount up. The gasification of oil residues offers the potential to double oxygen demand in the next 20 years. Twenty years ago a thousand tonne per day oxygen plant was a very large plant. Today the gas to liquid projects in the Middle East demand trains of 12000 \\$quot;“ 14000 tpd in 3600 tpd modules. Future projects will see the module size grow to 5000 tpd and even 7000 tpd. Serious recovery of oil and gas will demand both onshore and offshore nitrogen supplies of similar sizes. At the other extreme small membrane and PSA units have improved enough to provide the purity and economics required for many of the smaller customers.