Making Reverse Osmosis Membranes

Su Lv shows off a reverse osmosis membrane cartridge.

Making ultra-thin materials with holes the size of water molecules: While visiting GE's China Technology Center, we got to take a look at reverse osmosis membranes. Reverse osmosis is the most energy-efficient means of removing dissolved substances from water. It's what's used commercially for desalination, the process of producing drinking water from seawater.

The term "membrane" is typically used to mean a thin sheet of some material (in fact, the word "sheet" appears in the definition of the term). But for some of the things GE is using it for, the membranes were thin yet robust tubes, each one capable of supporting the weight of a bowling ball. Despite that toughness, features on the tubes are so fine that they can allow water molecules to pass through but reject many things that are roughly the same size, such as the salt ions found in seawater.

This all raises an obvious question: how do you actually produce anything like that? We decided to look into the process of making reverse osmosis membranes. It quickly became clear that the toughness of the membranes is a key feature. Water purification systems need to survive repeated cleaning cycles and go right back to use. We talked to Sijing Wang in Shanghai, who said that some membranes in the systems GE makes can be used for up to eight years.

That toughness, however, is provided by polymers that are microporous, in that they have features a thousand times larger than what is needed for reverse osmosis. These would do little to help remove salts from water, but they provide structural support for membranes that can. The large pores also ensure that water can easily flow through the system once it has passed through the membrane that acts as a filter.

 A test setup that allows Sijing Wang to see how membranes (held in the rectangular cases, lower right) respond to different types of waste material.

The interactions between that membrane and the water it's purifying help dictate the efficiency of the system. Since the membrane doesn't interact well with water, more water will flow through it when it's thinner. For reverse osmosis membranes, this layer is kept to a micrometer or less in thickness. The pores within it have to be kept small so that the ions of the salts in the water can't pass through a pore without interacting with the membrane, which will repel them. (Technically, the pores aren't small enough to physically block the ions from passing through, but the interactions between ions and the membrane keep them from getting too close to the pore opening.)

So you need to both layer a thin membrane across your support membrane and control the size of the pores that form within it, typically limiting them to less than 10 nanometers.

There are two methods of creating thin membranes. One involves forming a polymer but keeping it dissolved in a solvent that also mixes with water (often an alcohol of some sort). As you increase the fraction of water present, the polymer will eventually precipitate out. There are several ways of doing this. The simplest is to just heat the solution so that the solvent evaporates, which increases the fraction of water until the polymer precipitates. Alternatively, you can place the solution in a humid environment until the fraction of water goes up.

The most common method, however, is to create a viscous, 20 percent polymer solution and dunk it directly in water. The solution is so viscous that it won't mix into the water; instead, water infiltrates it and causes the polymer to drop out of solution. This process is often done on a continuous roll of material that's sent through a vat of water.

In all these instances, the action takes place at the interface between the polymer surface and the environment. As a result, the membrane primarily forms at this interface, creating the very thin barrier needed for reverse osmosis. Polymer deeper in the solution tends to form a larger, more open structure, which allows water to flow freely away from the membrane.

An alternative approach that functions in a similar manner is to use a building block for the polymer that dissolves in some solvents and a chemical activator that dissolves in an immiscible one. The two solvents will form two different layers (much like oil and water), and the building block and activator will only meet each other at the interface. As a result, polymerisation only takes place at this interface, resulting in a very thin layer.

How do you put holes in it? To a certain extent, the process takes care of that itself. As water begins to enter a solvent it's not fully compatible with, it will form tiny droplets that are held together by surface tension. The polymer will form around those droplets, leaving small holes behind. The size of these holes is determined by the speed of the process; the quicker it takes place, the smaller the water droplets will be and the smaller the resulting pores. By varying the solutions being used and the speed of the process, it's possible to have fine control over the pore formation process.

There are also additional layers of control possible. It's possible to include molecules that act as "pore generators" in the solutions, which are then removed when the membrane is rinsed later. Wang said most of the polymers GE uses are made of aromatic polyamine—which means a carbon ring that nitrogens are attached to. These chemicals do allow a certain degree of flexibility, in that they can be different sizes (one or more rings) and have slightly different chemical properties. (They're also carcinogenic before they're polymerised, but they're inert afterward. While water purification systems can be said to "contain a carcinogen," they pose absolutely no threat to human health.)

By adjusting the chemistry of the polymer, as well as the process by which it's formed, it's possible to have very fine control over the membrane that ultimately forms. This allows manufacturers to customise membranes for different tasks and to provide the durability that's needed for multiple years of use.

Sidney Loeb (left) with first RO membrane.

First Demonstration Of Reverse Osmosis: In the late 1940s, researchers began examining ways in which pure water could be extracted from salty water. During the Kennedy administration, saline water conversion was a high priority technology goal-"go to the moon and make the desert bloom" was the slogan. Supported by federal and state funding, a number of researchers quickly advanced the science and technology of sea water conversion, but UCLA made a significant breakthrough in 1959 and became the first to demonstrate a practical process known as reverse osmosis (RO).

At that time, Samuel Yuster and two of his students, Sidney Loeb and Srinivasa Sourirajan, produced a functional synthetic RO membrane from cellulose acetate polymer. The new membrane was capable of rejecting salt and passing fresh water at reasonable flow rates and realistic pressures. The membrane was also durable, and could be cast in a variety of geometric configurations. The impact of this discovery has been felt worldwide, ranging from applications in home demineralizers to "rivers of fresh water" in the Middle East and North Africa, where desalination facilities produce trillions of gallons of pure water every day. About 60 percent of the world's desalination capacity is located on the Arabian peninsula.

The process of osmosis through semipermeable membranes was first observed in 1748 by Jean-Antoine Nollet (pictured). For the following 200 years, osmosis was only a phenomenon observed in the laboratory. In 1949, the University of California at Los Angeles first investigated desalination of seawater using semipermeable membranes. Researchers from both University of California at Los Angeles and the University of Florida successfully produced fresh water from seawater in the mid-1950s, but the flux was too low to be commercially viable until the discovery at University of California at Los Angeles by Sidney Loeb and Srinivasa Sourirajan at the National Research Council of Canada, Ottawa, of techniques for making asymmetric membranes characterised by an effectively thin "skin" layer supported atop a highly porous and much thicker substrate region of the membrane. John Cadotte, of FilmTec Corporation, discovered that membranes with particularly high flux and low salt passage could be made by interfacial polymerisation of m-phenylene diamine and trimesoyl chloride. Cadotte's patent on this process[4] was the subject of litigation and has since expired. Almost all commercial reverse osmosis membrane are now made by this method. By the end of 2001, about 15,200 desalination plants were in operation or in the planning stages worldwide.