Electrodeionization is latest configuration for UPW ion exchange
08/01/2003
Method is quickly gaining ground in pharmaceutical cleanroom operations for continuous, on-line regeneration capability free of regeneration chemicals
By Robert Donovan
Ion exchange (IE) columns and reverse osmosis (RO) stages are probably the most important steps in the treatment chain (filtration, degasification and sterilization are other typical operations) used to produce the ultrapure water (UPW) required by all semiconductor manufacturers.
Electrodeionization (EDI) is a reasonably recent, commercially available design configuration that simplifies IE operation and reduces the hardware required for it; as a matter of fact, most UPW users now choose EDI as a preferred approach for carrying out at least some traditional IE functions.
Over its years of development, EDI has exhibited significant advantages over conventional IE. These advantages are being exploited as the use of EDI continues to grow, not only in semiconductor UPW water systems but in other applications as well, including power generation and pharmaceutical production.
EDI may not always be the best answer for a UPW water system at a given site, but it certainly shouldn't be dismissed or overlooked. Its performance record of successful installation and operation at semiconductor manufacturing facilities and power generation plants already makes for impressively convincing reading.
Conventional ion exchange
The chemistry of ion exchange reactions is straightforward and simple as summarized in the following reversible reactions:
 |
null
The left hand side of Equation 1 represents a fresh IE column consisting of cationic resin beads saturated with attached hydrogen ions (RCH) immersed in an aqueous solution containing cationic impurities (M+).
The ion exchange reaction is a simple exchange of resin attachment between the impurity ion and the hydrogen ion in which the cation replaces the hydrogen ion on the resin bead and the hydrogen ion goes into solution—or, the right hand side of Equation 1.
This direction of the Equation 1 chemical reaction dominates until the concentrations of the reaction products reach the values defined by an equilibrium constant, K:
 |
null
Once Equation 3 is satisfied, an equilibrium exists among the four species of Equation 1; their concentrations no longer change, as equal ion exchange takes place in both directions and many impurity cations are released as are being captured by the cationic resin beads. The contribution of the reversible reaction proceeding at the left of the equation balances that proceeding to the right.
The cation resin is now saturated with impurity cations; and before this cationic IE column can remove more cations from the solution, it must be taken off line and regenerated.
Regeneration consists of flooding the IE column with H+ supplied by hydrochloric acid—a nasty solution to handle in the large quantities necessary for column regeneration in a typical semiconductor UPW system.
Increasing the concentration of H+ in the column shifts the dominating direction of reaction on the left of Equation 1 in order for the concentration of the [M+] _ [RCH] product in the denominator of Equation 3 to increase and maintain a constant K value as [H+] increases in the numerator.
Captured cations are replaced by the excess H+ and released as impurity ions, restoring the initial conditions on the left of Equation 1. The contaminated liquid created during this regeneration step becomes discarded wastewater.
A similar action takes place in the anionic column, as the regeneration solution introduced into the off-line anionic column develops a high [OH-] concentration, such as NaOH—another hazardous solution to handle.
This sequence of on-line operation, followed by off-line regeneration, has been the industry standard practice for more than 50 years. The frequency of regeneration varies from several days to several weeks or longer, depending on the impurity concentration in the feed water entering the IE column.
IE costs include: Taking a column off-line for regeneration requires that extra IE columns be available to maintain rated capacity during the regeneration period; regeneration also requires the storage of hazardous chemicals and safeguards for using them during the regeneration cycle; and the wastewater discharged from the columns during regeneration degrades the quality of the site effluent water being fed to the municipal sewer system.
Nonetheless, IE performance justifies these costs and no UPW system operates without IE at one or more stages of its treatment chain.
The electrodeionization cell
The good news now is that carrying out IE with the electrodeionization (EDI) configuration eliminates all three of the above-cited costs. EDI operates in a mode featuring continuous, on-line regeneration without chemicals so that the only impurities added to the sewer discharge are those removed from the incoming feed water.
At the same time, EDI avoids the personnel hazards associated with the handling of the regeneration chemicals and their associated capital and operating costs. As the name implies, EDI achieves these benefits by substituting dc electrical power-driven regeneration for the chemically based regeneration of conventional IE operation.
 |
null
The basic EDI cell, illustrated in Figure 1, consists of three types of compartments: the concentrate, the dilute and the electrode isolated from each other by membranes. Multiple concentrate and dilute compartments can be arranged in series between one anode and one cathode compartment.
Figure 1 shows just one dilute compartment between two concentrate compartments. In full-scale installations, EDI is typically carried out as a stack of EDI cells—an alternating series of dilute and concentrate compartments extending between the membranes isolating the two electrode compartments through which feed water flows in parallel.
 |
null
The total number of cells in a stack depends on the available voltage supply and the cell voltage requirements. The total flow capacity of an EDI system then depends on the total number of stacks operated in parallel. The modular stack design means that stacks can be added in parallel to meet virtually any required flow capacity.
The electrode compartments simply isolate the electrodes from the adjacent membranes and are filled with relatively conducting water to minimize their contribution to the electrical resistance between the electrodes. In some designs, the metallic electrodes are actually located in the end concentrate compartments so that no isolated electrode compartments are present.
 |
null
A crucial feature of EDI, distinguishing it from conventional IE, is the membranes used to define the walls separating the compartments. These membranes are solid sheets made of resin material similar to that of the resin beads, and are either anionic or cationic membranes.
They act as semipermeable membranes in the sense that an anionic membrane passes just anions, rejecting all cations. A cationic membrane does the reverse; it allows cations to pass through and rejects all anions. Ideally, both types of membranes leak zero water as well as zero ions of the opposite polarity.
Water enters all compartments of the cell—entry ports are seen at the top of the sketch in Figure 1. The dilute compartment is filled with conventional resin beads. These resin beads can be all anionic or all cationic or, as is most common in EDI, a mixed bed of both (Figure 1), making the dilute compartment similar to a mixed bed IE column of conventional resins.
And, indeed, the initial interaction between the impurities in the feed water and the resin beads is identical to that in the conventional IE column. The difference is that in conventional IE, once captured, impurity ions remain attached to the resin beads until the regeneration step. In EDI, the dc electric field created by the voltage difference between the external anode and cathode causes the captured impurity ions to drift through the bed of beads and the isolating membranes, toward one electrode or the other.
No chemically captured impurity ion remains stationary under the action of the electric field. Impurities attached to anionic resin beads in the dilute compartment drift toward the anode, migrating through a chain of contiguous anionic resin beads, then through the anionic membrane and into the adjacent concentrate compartment.
The cationic membrane separating this concentrate compartment and the anode compartment blocks further ion motion toward the anode electrode. The impurity anions remain confined to the concentrate channel. Similarly, cations, captured by the cationic resins in the dilute compartment, drift through a chain of contiguous cationic resin beads, through the cation permeable membrane and into the concentrate compartment on the opposite side of the dilute compartment. An anionic membrane now blocks their further motion toward the cathode.
This ion motion, induced by the applied dc field, removes ions captured by the resins from the dilute compartment and deposits them in the adjacent concentrate compartments. No resin motion is required for this action; only the impurity ions move. The ion removal process is continuous with no interruption and constitutes an on-line regeneration process.
Ionic conduction mechanisms
Ion drift is primarily through the resin beads and not the surrounding water. The enhanced ionic conductivity of the resin beads is the reason that resin beads fill the dilute compartment.
Figure 2 plots the electrical resistivity of water as a function of ionic impurity concentration, and also that of representative resin beads in that same water. At low ionic concentration, ion conductivity through the beads exceeds that through the water by several orders of magnitude.
This property makes resin beads a necessary ingredient for achieving practical EDI in UPW. Of course, under an applied dc field, ion drift will take place in water without the presence of any resin beads.
The configuration of Figure 1 without the resin beads is an electrodialysis cell, which is widely used for removing objectionable ions from heavily contaminated liquids, such as human blood. Electrodialysis alone, however, does not match the needs of UPW purification—resins are an essential part of an EDI used in UPW systems because of the differences in ionic resistivity between aqueous solutions alone and aqueous solutions containing resin beads (as depicted in Figure 2).
Ionic impurity transport through the dilute compartment packed with resin beads is most efficient when direct bead-to-bead ion transfer occurs. Such direct ion transfers between resins occur only between resins of the same type. Cations transported along the surface sites of a cationic resin bead must contact another cationic resin bead in order to directly transfer to that neighboring resin bead and continue an uninterrupted journey toward the cathode.
Without contact with a bead of the same type with like sites for ion transfer, the impurity ion must first be displaced into the surrounding water and then recaptured from the water by a remote (non-contacting) resin bead of the same type.
This indirect process costs electrical power and reduces cell efficiency. Thus, a rule of thumb is that the practical width of a mixed bead, dilute compartment should be no greater than approximately three bead widths. Thicker compartments of mixed beads result in significantly reduced probability of having a contiguous chain of same type resin beads extending across the dilute compartment from one membrane to the other.
To avoid this complication, some EDI designs consist of two dilute compartments in series, each compartment having just one type of bead—usually, a cation resin compartment followed by an anionic resin compartment. These separate bed EDI designs typically do not perform as efficiently as the mixed-bed designs. Cations leaking into a downstream separate bed anionic column, for example, see no cationic capture sites and are not removed. This observation is similar to that noted for conventional IE.
Common practice in conventional IE is to locate separate anion and cation columns upstream of a mixed bed column. Having a mixed bed as a final column usually produces higher overall water quality in both IE and EDI.
A better solution for overcoming the dilute compartment width limitation of mixed-bed EDI designs is to deliberately avoid having a well-mixed distribution of the two resin bead types. Loading the resins into the dilute compartment as alternating bands of single type beads guarantees that the desired uninterrupted chains of each type of resin will extend across the dilute compartment from membrane to membrane. In addition, the compartment so filled will still be a mixed bed and presumably retain the advantages of a mixed-bed compartment.
Resin beads can also reduce the series resistance of the concentrate compartment. Some cell designs now include resin beads in the concentrate compartment as well—unlike the configuration sketched in Figure 1. This feature serves to lower the transverse voltage gradient across the concentrate compartment, where ion drift is unimportant, and thus maintain higher voltage drops across the dilute channels and the isolating membranes where the electric field is the major driving force for moving impurity ions out of the dilute compartment.
 |
null
Figure 4 illustrates a configuration having resin beads in both the diluting and concentrating compartments. This schematic shows one concentrating compartment between two diluting compartments, unlike Figure 1, which showed one diluting compartment between two concentrating compartments. Again, an EDI stack of either design consists of multiple, alternating compartments between the external electrodes.
Two functions of the electric field
Drifting impurity ions out of the dilute compartment is just one of the important roles of the applied dc electric field. This action dominates in the upstream, entry region of the dilute compartment where the concentration of impurity ions in the feed water is the greatest. Toward the exit end of the dilute compartment, where relatively few impurity ions are chemically attached to the resin beads, the applied electric field has few ions to drift, so that much of the field ends up dissociating water molecules in the electric fields between adjacent anionic and cationic resins.
This creation of H+ and OH- ions provides the ions required for regeneration, and insures the availability of suitable resin sites for capturing impurity ions. Maintaining highly regenerated resin beads within the bed is particularly important for capturing the weakly ionized impurities, such as silica and various boron compounds, that are not efficiently captured by the resins at the entry of the column where capture sites are preferentially occupied by the strongly ionized impurities in the feed water.
In a separate bed EDI configuration, water dissociation takes place at the electrodes, generating OH- and hydrogen gas at the cathode, and H+ and oxygen gas at the anode (as indicated in Figure 3).
The increase of H+ in the cation compartment means that the water pH decreases in the direction of water flow through the compartment. Thus, the water exiting the cation dilute compartment and feeding the anionic dilute compartment has very low OH- concentrations.
According to Equation 2, this pH state enhances the capture of anions—it favors the dominance of the reversible reaction proceeding to the right. The lowered pH of the water leaving the cation compartment makes anionic impurity removal more favorable in a separate bed than in a mixed bed where the regeneration reactions contribute more or less equal concentrations of OH- and H+, and thus have little effect on compartment pH.
In some applications, pH variations could be important and desirable, although the inability of a separate bed design to capture trace cations in the downstream anion compartment generally remains the more important effect.
EDI advantages
The main advantage of EDI is, of course, its continuous, on-line regeneration capability free of regeneration chemicals. Users report other gains as well, including improved performance over conventional IE, enhanced mechanical and thermal properties, bacteria destruction, reduced maintenance costs and reduced total costs.5
The most common use in UPW systems is to replace the primary conventional IE with EDI but retain polishing mixed-bed IEs of conventional design. Under this arrangement, with the high-quality water output of the EDI unit feeding the conventional mixed-bed IE, the conventional mixed-bed resins typically last for more than a year and are usually never chemically regenerated. It is more economical to simply replace them.
EDI maintenance requirements also prove to be less onerous than those associated with IE. According to Deb Mukhopadayay, "The [chemically] regenerable mixed bed ion exchange system represents, by far, the most expensive (and complicated) single unit operation/process in the entire UPW treatment system."6
Regeneration of a conventional mixed bed IE column requires a means of separating the mixed resins so that the appropriate regeneration chemicals for each type of resin can be delivered to the appropriate resin—and just that resin type.
Techniques for achieving such separation in a mixed bed are not perfect, and regeneration is often less complete than in an IE bed composed of just one type resin. As described earlier, EDI encounters no such problem. The applied dc field moves the ionic impurities and not the resin beads; it separates anionic and cationic impurities by moving them in opposite directions.
Thus, in EDI, mixed resin beds are as easily regenerated as separate resin beds—the electric field separating the ionic impurities without requiring the separation of the mixed anionic/cationic resin beads.
EDI limitations
EDI resins can be fouled, especially by hard water. Thus, the feed water to the dilute compartment needs to be essentially free of calcium and magnesium. R/O product water produces near ideal feed water for EDI and, indeed, as noted above, the earliest and still most common use of EDI in a UPW system is as a replacement for a conventional EI in the primary loop, typically located downstream of an R/O unit.
Robert P. Donovan is a process engineer assigned to the Sandia National Laboratories and a monthly columnist for CleanRooms magazine. He can be reached at: [email protected]
References
- Ganzi, G. C., "Electrodeionization for High Purity Water Production", AIChE Symposium Series 261, vol. 84, New Membrane Materials and Processes for Separation, K. K. Sirkar and D. R. Lloyd, editors, 1988, pp. 73-83
- Harada, K., T. Karibe, H. Uchino and M. Umeda, "Water Reclamation Using Continuous Electro-Deionization", 1996 UPW Reclaim Proceedings, pp. 69-84 (Balazs Analytical Laboratory, 46409 Landing Parkway, Fremont, CA 94538)
- Verbeek, H. M., L. Fuerst and H. Neumeister, "Digital Simulation of an Electrodeionization Process", Computers Chem Engng, Vol 22, Suppl., 1998, pp. S913-S916
- Ganzi, G. C., A. D. Jha, F. DiMascio and J. H. Wood, "Theory and Practice of Continuous Electodeionization", Ultrapure Water 14 (6), July/August 1997, pp. 64-69
- Liang, L. S. and Wang, L., "Continuous Electrodeionization Process for Production of Ultrapure Water", Proceedings of the 20th Annual Semiconductor Pure Water and Chemicals Conference, Feb 27 - March 1, 2001, Monterey, CA, pp. 1- 4 (Balazs Analytical Laboratory, 46409 Landing Parkway, Fremont, CA 94538)
- Mukhopadhyay, D. and S. Whipple, "High Efficiency Reverse Osmosis System", Proceedings of the 16th Annual Semiconductor and Pure Water and Chemicals Conference, March 3 - 7, 1997, Santa Clara, CA, vol 1, pp. 1-20 (Balazs Analytical Laboratory, 46409 Landing Parkway, Fremont, CA 94538)