Pioneering the research in reverse osmosis desalination were Dr. Srinivasa Sourirajan, Sidney Loeb, and associates at UCLA. For their initial testing of various osmotic membrane properties, cellophane and cellulose ester films were used. Cellulosic structure allows for hydrogen bonding, thus making it appropriate for the preferential sorption of water from aqueous salt solutions (Sourirajan, 1970). Cellulose acetate also has a relatively high impact strength and a low softening point, which make it easy to work with (Billmeyer, 1971). Basing their choice on these known qualities of cellulosic membranes, Sourirajan and colleagues conducted extensive testing with the membranes in various high-pressure osmosis chambers. Through these initial studies they were able to hypothesize on a mechanism for the osmotic process. A microporous interfacial surface area over a macroporous interior were determined to be the membrane qualities that, along with compatible chemical nature, would produce the best solvent product flow. Assuming their mechanism and structure hypothesis to be correct, they sought methods of preparing membranes that best fit their model.
Their starting material was cellulose acetate,
which is prepared by esterification in the following manner:
Seven parts (by weight) of purified cotton
linters or wood pulp are combined with a solution of 0.36 parts concentrated
sulfuric acid in 38 parts acetic acid. This mixture is agitated to insure
uniform wetting of the cellulose. The reaction mixture is then closed and
stored at room temperature for an hour. After this pretreatment, a mixture
of 39 parts 95% acetic anhydride and 15.2 parts acetic acid is added and
the reaction mixture is maintained for 15 minutes in a water bath at 50
degrees Celsius (Braun, et. al, 1972). High temperatures are avoided to
prevent molecular weight degradation (Billmeyer, 1971). From this primary
solution, the triacetate and the 2-1/2 acetate can be prepared (Braun,
et. al, 1972).
For the cellulose triacetate, 38 parts of 80 percent acetic acid at 60 degrees Celsius is added to the primary solution to destroy the excess acetic anhydride, taking care that no precipitation of cellulose acetate occurs. After maintaining the solution for another 15 minutes at 60 degrees Celcius, 36.2 parts of water is cautiously added while stirring. 289 parts of water are added to this mixture, causing the cellulose triacetate to precipitate as a white, friable powder. The product is recovered by suction filtration, slurried in 434 parts of water, which is decanted after 15 minutes. This process is repeated until the wash water shows a neutral reaction. The polymer is filtered under aspiration and is then dried at 105 degree Celsius. Films are made from this product by melting and then casting in forms or sheets.
The cellulose 2-1/2 acetate is prepared by only partial saponification of the cellulose units. 72 parts of 70 percent acetic acid at 60 degrees Celcius and 0.3 parts concentrated sulfuric acid are added slowly with stirring to the primary solution. This reaction mixture is maintained at 80 degrees Celsius in a closed system for 3 hours. After the setting time has elapsed, an initial 36.2 parts of water is carefully added while stirring. An additional 280 parts of water are added to the mixture. The cellulose 2-1/2 acetate precipitates, is collected on a filter under aspiration, and is slurried in 434 parts of water. After letting the suspension settle for 15 to 20 minutes, the polymer is collected on a filter under suction and is dried at 105 degrees Celsius. Films are made from this product (40 percent acetyl group content) by melting and then casting in forms or sheets (Braun, et. al., 1972).
Loeb and Sourirajan (1970) obtained their cellulose acetate from commercial suppliers, thus saving themselves from having to prepare their basic raw material. Prepared as in the above mentioned procedures, the cellulose acetate that was preferred had a 54 to 56 percent by weight content of acetyl groups.
22.2 parts of this cellulose acetate is mixed with 11 parts aqueous magnesium perchlorate (a pore-producing agent which produces a structure that allows high fluid flow rate--1 part magnesium perchlorate and 10 parts water) and 66.7 parts acetone (a solvent that prevents the solution from becoming too viscous) (McDermott, 1970).
This sticky mixture is cast cold on a glass plate at a temperature between minus 7.5 to minus 16 degrees Celsius (optimum temperature-- minus 11 degrees Celsius). The solvent is allowed to slowly evaporate in the low temperature. The reduced temperature permits the initiation of the water-cellulose structural organization. Complete drying is not permitted. The glass plate and film are immersed in ice water after three minutes of drying time. After a few minutes of immersion in the water bath, the membrane will be ready to peel off. The membrane is heated in a water bath between glass plates at an optimum temperature of 82 degrees Celsius. The heating (for at least an hour) causes membrane shrinkage, reducing the interfacial pore size as well as causing the completion of the membrane's structural organization. The film is mounted wet on the reverse osmosis pressure chamber's porous membrane support. This membrane-support assembly is sealed against evaporation in an airtight container until installment in the reverse osmosis unit (McDermott, 1970).
The resulting film is opalescent, yet transparent, with a porosity (ratio of open volume to total volume) of 25 to 40 percent. The side of the film away from the glass plate during casting must be used against the salt solution. The other side must be against the porous membrane support. Reverse osmosis will not occur if this membrane is facing the other direction. The membrane is originally approximately 0.004 inch thick before heating and pressure stressing. After stressing on the porous membrane support the membrane thickness may be as little as 0.0025 inch thick (Sourirajan, 1970). The membrane consists of two layers--one, which comprises 99.8 percent of the volume, is highly porous, and is permeable to water and salt; the other layer, in which the salt-water separation occurs is only 0.0000001 inch (2.54 Angstroms) thick (the H-O-H distance of water is 1.98 Angstrom) (Friedlander, 1965).
In tests and actual desalinating apparatus, the membranes of this process have been able to reduce the concentration of a 5.25 percent sodium chloride solution to 500 parts per million in a single pass at a flow rate of 8 gal/ft2 /day under an applied pressure of 1,500 lb-ft/in2 (McDermott, 1970).
The modifed cellulose acetate membranes were the first semipermeable membranes to be tested for reverse osmosis desalination and in subsequent tests have proven to be the most efficient in salt-water separation. Yet, there is much research and development taking place with a great variety of other membranes.
Commercially available reverse osmosis desalination units have the standard cellulose acetate membrane or have an aromatic polyamide membrane. The aromatic polyamide (nylon) membranes are of an undisclosed structure.
Other nylons have promising characteristics that are being studied for possible use in reverse osmosis desalination. Nylon membranes seem to be more stable to hydrolysis, attack by microorganisms, and elevated temperature operation than the cellulosic membranes. Unfortunately, they have a considerably lower semipermeability (or salt rejection) to sodium chloride than the cellulosic membranes (Podall, 1971).
Nylon type 8, nylon type 6, and nylon tyhpe 6-6 are raw materials for preparing membranes (McDermott, 1970). One example of a method of preparation is as follows:
A bath of 600 grams paraformaldehyde dissolved in 1,942 grams tetraethylene glycol and 2 cc. or 50 percent sodium hydroxide solution at 90 degrees Celsius is prepared. To this formaldehyde solution is added 600 grams glacial acetic acid and 126 grams oxalic acid dihydrate. This solution is maintained at 90 degrees Celsius for 48 hours. A piece of 2 mil type 6 nylon film is placed in the bath and allowed to soak for 1 hour.
The N-beta-hydroxyethoxyethoxyethoxyethoxymethyl (from McDermott, 1970--Not named according to IUPAC rules) modified nylon film is then removed from the bath and is placed in distilled water for 2 hours after which it is dried for at least 8 hours. This film is then ready to be placed on the porous reverse osmosis unit membrane support (McDermott, 1970).
The modified nylon film additives (-OCH2CH3, -OH) allow for more hydrogen bonding (intermolecular and intramolecular), thus permitting substantial water sorption for efficient solvent flow. These nylon films can be treated in the formaldehyde bath in the sheet form to produce a flat membrane sheet, or can be treated in the formaldehyde bath after being spun into hollow fibers. The hollow tubular membranes produced by the latter are used in the hollow fiber spiral-wound reverse osmosis units (Channabasappa, 1971).
An inorganic membrane that has come under investigation is graphitic oxide (Sourirajan, 1970). Made by the oxidation of graphite flakes with strong oxidizing agents such as nascent chlorine dioxide (from potassium chlorate) in concentrated sulphuric acid or sulphuric acid-nitric acid mixtures, the graphite particles retain their shapes and readily form a loose membrane when carefully deposited on a porous supporting base of glass, fiber paper, or polymer.
Semipermeable membranes for the reverse osmosis process must meet two basic requirements: 1) They must have a chemical nature such that the film material has a preferential sorption and/or preferential repulsion for one or more constituents of the fluid mixture; and 2) They must have a large number of pores of the required size on the area of film (the thinnest layer of the film) at the film-solution interface, with comparatively big interconnecting pores in the interior of the film material (Sourirajan, 1970).
With cellulose acetate, the cellulosic scructure imparts a high hydrogen-bonding ability which allows for the preferential sorption of water from the solution to be separated. In deuterium replacement studies researchers were able to distinguish between hydrogen-bonded and non-hydrogen-bonded hydroxyl functional groups. It was indicated that there is intramolecular hydrogen bonding involving the oxygen atoms attached to the carbon atoms 6 and 2 of adjacent glucose residues. X-ray investigations confirmed this theory (Davidson, 1967). In addition to the cellulosic hydrogen bonding, the acetate groups' double-bonded oxygens are also involved with hydrogen bonding.
Cellulosic acetate film cast straight from an acetone solvent is not porous enough for the reverse osmosis process. Pores most be built into the film's organization. A water-soluble additive to the film casting solution which can be leached out of the film structure after it has been stabilized is needed. Magnesium perchlorate provides the porosity required and can be leached out of the structure by water once the membrane has set (Sourirajan, 1970).
Nylon membranes have varying structural
formulas but possess a common interunit linkage of 
(Ravve, 1967).
The N-hydrogen of the amide group readily participates in hydrogen bonding with the double-bonded oxygen of other amide groups. This hydrogen bonding can be intramolecular (if the molecule is folded over upon itself) or between separate molecules. Although the hydrogen bonding that occurs in the nylon molecular structure allows for solvent (water) sorption, many of the nylons used in reverse osmosis must be treated by the addition of more functional groups capable of hydrogen bonding. This treatment not only increases the solvent flow characteristics, but through the hydrogen bonding a gridlike structure is formed. The grid structure permits only small-size molecules to pass through the membrane. At the present time, chemical technology has not yet found a way to make nylon membranes more than a grid-filter system. Ways may be found soon, to produce nylone membranes with a thin, uniformly-porous (gridded?) film-solution interfacial surface above a macroporous, spongy interior layer.
Graphitic oxide membranes consist of loose layers of the oxidized graphite flakes (which retain their shapes) deposited on a porous support. The other bridges and tertiary hydroxy groups in the oxidized crystals provide an intracrystalline space of approximately 12 Angstroms with many available hydrogen bonding sites. Graphitic oxide is capable of a high-degree of intercrystalline swelling on exposure to water or water vapor. These two characteristics provide graphitic oxide with a high water (solvent) sorption capability (Sourirajan, 1970).
R. F. Probstein (1973) expressed the idea that all saline water conversion processes may be thought of as involving a semipermeable barrier, which is permeable to either the water or salt but rejects the other constituent. Distillation and freezing involve the nonisothermal transfer of heat and the barrier they utilize is a phase boundary. In electrodialysis and reverse osmosis, a real membrane is utilized and the process is essentially isothermal.
Where in the other separation processes the semipermeable barrier is only a minor location of the operation involved, the osmotic process depends entirely on the semipermeable membrane. The separation of the solution takes place entirely on or in the membrane. There is no general scientific concensus of how the pure solvent is passed through the membrane. There are a number of hypotheses in contention, but no one theory on which all can agree (Ocean Industry, 1976).
The term "osmosis" is familiarly used to describe the spontaneous flow of pure water into an aqueous solution, or from a less to a more concentrated aqueous solution, when separated by a suitable membrane. In order to obtain potable drinking water from saline water in a similar process, the direction of pure water flow must be reversed; for instance, pure water must flow from a more to a less concentrated solution--thus, the latter process has been conveniently termed "Reverse Osmosis".
The osmosis phenomenon is not restricted to the passage of water from aqueous solutions and it is not restricted to 100 percent solute separation. Neither "osmosis" nor "reverse osmosis" explain the mechanics involved with the phenomenon, thus it is misleading to explain "reverse osmosis" as the reverse of "osmosis" (Sourirajan, 1970).
Osmosis is a common process in living organisms, and is used to transport cellular fluids. In most of these processes pure fluid moves through cell walls (membranes) toward less pure or more concentrated solutions (Ocean Industry, 1976). Although Scholander (1972) reported that mangrove trees, which obtain their transpiration water from the sea actually use a physical ultrafiltration process similar to "reverse osmosis", most natural osmotic phenomenon have a net movement of diffusion of solvent into the more concentrated solution.
Under isothermal conditions, in both osmosis and reverse osmosis, the preferential transport of material through the membrane is always in the direction of lower chemical potential. This is a thermodynamic requirement, and it cannot specify which component of a solution will be preferentially transported through a given membrane, and the mechanism by which such transport takes place (Sourirajan, 1970).
The mechanism of membrane semipermeability might be explained by supposing the solvent to be soluble in the membrane whereas the solute is insoluble. Another theory isthat the solvent, for example, water, molecules are attracted by the membrane surface--they are absorbed resulting in a continuous connection between the solvent molecules on both sides of the membrane, thus the molecules pass through without difficulty. Solute molecules are not able to penetrate the complex network of pores that consitute the semipermeable membrane, since there is no direct connection from one side to the other.
An interesting theory that is losing foothold with scientists is the vapor transfer theory. This theory holds that there are fine capillaries in the semipermeable membrane that are not wetted by the liquid. It states that molecules of solvent vapor are able to diffuse into these pores, so that when pure solvent is placed on one side of the membrane and solution is placed on the other side, distillation occurs through the capillaries from the region of high vapor pressure (solvent) to the region of lower vapor pressure (solution), resulting in osmosis (Glasstone, 1960).
The main theory that has come into being has been the work of Loeb and Sourirajan (1970). This theory holds that at semipermeable membrane film-solution interfaces, a chemical organization of the solvent and solute takes place, whereby the smaller molecules pack closer and less randomly at the membrane interfacial area. These small molecules can readily pass through any pores in the membrane large enough for them to fit through. If there are no pores, there will be no osmosis. This theory, when applied to reverse osmosis, holds well.
Loeb and Sourirajan incorporated this theory into their hypothesized membrane structure and proceeded to test the reverse osmosis membranes with regard to these qualities. Their tests and subsequent preparations of membranes made their theory on the mechanics of osmosis more acceptable to the scientific community.
Thus, the readily approved mechanism is illustrated in the following figure (Sourirajan, 1970).
Billmeyer, Fred W. Jr., Textbook of Polymer Science. 2nd ed. New York: Wiley-Interscience. 1971.
Braun, Dietrich., Harald Cherdron, and Werner Kern. Techniques of Polymer Synthesis and Characterization. Translated by R.F. Bender from Praktikum der Makromolekularen Organischen Chemie. (Wiley-Interscience, New York, 1972.)
Channabasappa, Kendere C., "Reverse Osmosis." In Saline Water Conversion Report 1970-1971. Membrane Division. Office of Saline Water. U.S. Department of Interior. Washington: n.p., 1971.
Davidson, Eugene Abraham. Carbohydrate Chemistry. New York: Holt, Rinehart, and Winston, Inc., 1967.
Friedlander, Gordon D., "Science and the Salty Sea." IEEE Spectrum. August, 1965. in Desalination and Ocean Technology. ed. Sumner N. Levine. New York: Dover Publications, Inc., 1968.
Glasstone, Samuel, and David Lewis. Elements of Physical Chemistry. 2nd ed. Princetown, New Jersey: D. Van Nostrand Co., Inc., 1960.
McDermott, John. Desalination by Reverse Osmosis 1970. Park Ridge, New Jersey: Noyes Data Corporation, 1970.
Podall, H.E., "Reverse Osmosis." In Saline Water Conversion Report 1970 -- 1971. Polymer and Biophysics Division. Office of Saline Water. U.S. Department of Interior. Washington: n.p., 1971.
Probstein, Ronald F. "Desalination." American Scientist. May-June, 1973.
Ravve, A., Organic Chemistry of Macromolecules: an Introductory Textbook. New York: Marcel Dekker Inc., 1967.
"Reverse Osmosis Units Produce Fresh Water from the Sea." Ocean Industry. July 1976.
Scholander, P.F., "Tensile Water." American Scientist. Sept--October, 1972.
Simson, Hugh C., and R.S. Silver. "Technology of Sea Water Desalination". In Desalination and Ocean Technology. ed. Sumner Levine. New York: Dover Publications, Inc., 1968.
Sourirajan, Srinivasa. Reverse Osmosis.
New York: Academic Press, 1970.
Culligan International Co.
Northbrook, IL
Millipore Corporation
Bedford, MS