In the late 1970s John Cadotte of America's Midwest Research Institute and the FilmTec Corporation created a much-improved membrane by using a special cross-linking reaction between two chemicals atop a porous backing material. His composite membrane consisted of a very thin layer of polyamide, to perform the separation, and a sturdy support beneath it. Thanks to the membrane's improved water flux, and its ability to tolerate pH and temperature variations, it went on to dominate the industry. At around the same time, the first reverse-osmosis plants for seawater began to appear. These early plants needed a lot of energy. The first big municipal seawater plant, which began operating in Jeddah, Saudi Arabia, in 1980, required more than 8 kilowatt hours (kWh) to produce one cubic metre of drinking water.

The energy consumption of such plants has since fallen dramatically, thanks in large part to energy-recovery devices. High-pressure pumps force seawater against a membrane, which is typically arranged in a spiral inside a tube, to increase the surface area exposed to the incoming water and optimise the flux through the membrane. About half of the water emerges as freshwater on the other side. The remaining liquid, which contains the leftover salts, shoots out of the system at high pressure. If that high-pressure waste stream is run through a turbine or rotor, energy can be recovered and used to pressurise the incoming seawater.

The energy-recovery devices in the 1980s were only about 75% efficient, but newer ones can recover about 96% of the energy from the waste stream. As a result, the energy use for reverse-osmosis seawater desalination has fallen. The Perth plant, which uses technology from Energy Recovery, a firm based in California, consumes only 3.7kWh to produce one cubic metre of drinking water, according to Gary Crisp, who helped to oversee the plant's design for the Water Corporation, a local utility. Thermal plants suck up nearly as much electricity, but also need large amounts of steam. ?A thermal plant only is practical if you can build it in such a way that it can take advantage of very low-cost or waste heat,? says Tom Pankratz, a water consultant based in Texas, who is also a board member of the International Desalination Association.

Economies of scale, better membranes and improved energy-recovery have helped to bring down the cost of reverse-osmosis seawater-desalination. Although the cost of desalination plants and their water depends on where they are, as well as the local costs of capital and operations, prices decreased from roughly $1.50 a cubic metre in the early 1990s to around 50 cents in 2003, says Mr Pankratz. As a result, reverse osmosis is preferred for most modern seawater-desalination (though rising energy and commodity prices mean the cost per cubic metre has now risen to around 75 cents). Experts reckon that further gains in energy efficiency, and hence cost reductions, will be increasingly difficult, however. According to a recent report on desalination from America's National Research Council, energy use is unlikely to be reduced by much more than 15% below today's levels?though that would still be worthwhile, it concludes.

Sometimes, using desalination within water management may be the only way to ensure supply.

To achieve these reductions, researchers want to find better membranes that allow water to pass through more easily and are less likely to get clogged up. Eric Hoek and his colleagues from UCLA, for example, have developed a membrane embedded with tiny particles containing narrow flow channels, producing a significant increase in water flux. The membrane's smooth surface is also expected to make it harder for bacteria to latch onto. Depending on a plant's design, the new membranes could reduce total energy consumption by as much as 20%, reckons Dr Hoek. The technology is being commercialised by NanoH2O, a company on UCLA's campus.

Meanwhile, the possibility of making membranes out of carbon nanotubes, which consist of sheets of carbon atoms rolled up into tubes, has also garnered attention. A study published in the journal Science in 2006 demonstrated unexpectedly high water-flow rates. But insiders think it will be a decade before the idea is ready for commercialisation.

As desalination becomes more widespread, its environmental impacts, including the design of intake and discharge structures, are coming under increased scrutiny. Some of the damage can be mitigated fairly easily. Reducing the intake velocity enables most fish species and other mobile marine life to swim away from the intake system, though small animals, such as plankton or fish larvae, may still get caught in the intake screens or sucked into the plant.