Most reef aquarists use artificial seawater mixes, and these always need to be made with freshwater. Likewise, all reef aquaria need their evaporated water to be replaced with freshwater in some fashion. Consequently, all aquarists need access to suitably pure freshwater. Unfortunately, tap water often contains contaminants, including chlorine, chloramine, copper, and a variety of other undesirable compounds, that make it unsuitable for aquarium use. There are several ways to purify tap water to make it suitable for reef aquaria, and the best method for achieving this is likely the combination of reverse osmosis and deionization (RO/DI) purification.

This article describes what these multistage systems are comprised of, what each stage accomplishes, and how to make the most of an RO/DI system. Specifically, the contents are:

• What Contaminants are Present in Tap Water?
• What is an RO/DI System?
  • Sediment Filters
  • Carbon Filters
  • Reverse Osmosis Membranes
  • Deionizing Resins
  • Deionizing Resin Depletion Issues
• Final Effluent pH
• Recharging DI Resins
• Tips on Using RO/DI
• Conclusions
• References

What Contaminants are Present in Tap Water?

Tap water can contain a variety of undesirable impurities. Some are present in the freshwater source before it is collected by a municipal water supply company. These can include nitrate, phosphate, silica, certain potentially toxic metals (such as chromium), and a variety of organics (Tables 1-4).

Some "contaminants" are intentionally added to the water to make it suitable for human consumption; these include chlorine and chloramine, as well as silica (added to some water supplies to raise the pH and reduce corrosion and release of copper and lead into drinking water). From the aspect of aquarists, chloramine can be among the worst of these offenders, with many water supplies targeting levels of 2-4 ppm chlorine equivalent (ppm-Cl). Some organisms are sensitive to chloramine at levels far below this concentration. In its assessment of chloramine toxicity to marine invertebrates, Environment Canada (the Canadian equivalent of the United States Environmental Protection Agency, EPA) determined the Estimated No-Effects Value (ENEV) based on this type of data to be 0.002 ppm-Cl for marine and estuarine environments. Consequently, chloramine must be removed before using tap water that contains it. Unfortunately for aquarists who want to use tap water, it is long-lived, and will not rapidly dissipate the way that chlorine will.

Lastly, some contaminants are more likely to come from the pipes in the aquarist's home than from the water supply itself. Consequently, these contaminants do not depend on the quality of the source water as much as the pH of the water and the nature of the pipes within the home. Many aquarists are fooled into thinking that their town has very clean water, so they need not worry about anything in their tap water. Unfortunately, that can be untrue. Chief among these that are of concern to aquarists is copper. Copper is allowed by the U.S. EPA to be present in drinking water at levels exceeding 1 ppm. Some homes in recent surveys have been found to exceed 1.3 ppm (Table 5).

Figure 1. The RO/DI system that I have been using for my reef aquarium needs for the past ten years.

That level is well above the threshold for toxicity to marine organisms, and is about a hundred times higher than I found in my aquarium the last time I tested it for copper (about 10-15 ppb copper).

What is an RO/DI System?

At a minimum, an RO/DI system consists of a reverse osmosis membrane through which the water flows (and is purified), followed by a deionizing resin that removes any residual charged compounds. Typically, other necessary parts ensure that these two main parts function properly. These other parts may include sediment filters, activated carbon filters, pressure gauges and conductivity or total dissolved solids (TDS) monitors. Figure 1 shows the RO/DI system that I have been using for nearly 10 years. It has each of these components and, other than changing filters when needed, hasneeded no maintenance. Not all aquarists are so lucky, however, and sometimes the RO membranes become clogged or broken and need to be replaced. Maintenance issues are discussed later in this article. The function of each of these components is described below.

Sediment Filters

Sediment filters are the first thing that incoming tap water encounters, and they do just what their name implies: filter out sediment. If this sediment were to get past the filter, the sediment would then rapidly clog the activated carbon filter, making it much less useful and potentially risking damage to the RO membrane (as described below). The sediment filter also prevents sediments from getting through that might coat the inside of the reverse osmosis membrane. Typical sediment filters used for this application have pore sizes in the 0.5 to 1 mm (micrometer, or millionth of a meter) range. If the water contains a lot of sediment, it is sometimes useful to have a series of sediment filters with steadily decreasing pore size, increasing the lifetime of the filters.

Carbon Filters

The next filter in line typically contains activated carbon. The primary purpose of this activated carbon is to break down the chlorine and chloramine in the tap water (Figure 2). If not removed, these compounds may damage the RO membrane. They will also pass through the remaining filters (RO and DI) fairly readily, and so can end up in the final filtered water.

The reactions within the activated carbon that break down these compounds rely on having enough active surface area for these catalytic reactions to take place. If the sediment filter is faulty, sediment may clog the activated carbon and reduce its ability to break the compounds down. Chlorine and its hydration product in water, hypochlorite ion (ClO^-), are broken down into chloride ion (Cl^-) and oxygen as shown in reactions 1 and 2 below. Equation 1 occurs at the water supply site where the water is chlorinated, and equation 2 takes place on the activated carbon in the RO/DI filter (where C* stands for the activated carbon and CO* stands for the activated carbon with an attached oxygen atom).

(1)        Cl₂ + H₂O  à OCl^- + 2H⁺ + Cl^- 

(2)        ClO^-  + C*  à  Cl^-  +  CO*

Some of the oxidized activated carbon remains, and some breaks down to produce oxygen (O2):

(3)        2CO*  à  2C*  +  O₂

None of the products of reactions 2 or 3 is of concern to reef aquarists.

The reaction of chloramine (NH2Cl) on activated carbon is a bit more complicated, and produces ammonia (NH3), chloride (Cl^-), and nitrogen gas (N2). Equations 4 and 5 show the processes.

(4)        NH₂Cl + H₂O + C* à NH₃ + Cl^- + H⁺ + CO*

 (5)       2NH₂Cl + CO* à N₂ + H₂O+ 2H⁺ + 2Cl^- + C*

Unlike the breakdown of chlorine, one of the byproducts of chloramine breakdown, specifically ammonia, is of significant concern to reef aquarists. Ammonia will pass fairly readily through an RO membrane (discussed later in this article), but if there is a DI system, it should effectively remove ammonia.

The carbon filters used in an RO/DI system will also remove certain organic chemicals from the water. Most of these are large enough not to pass through the RO membrane, but a few may be small enough to get through, yet still be reasonably well bound by the carbon. An example of something bound by the carbon that might otherwise get through the RO membrane is trichloromethane (CHCl3). In general, however, I do not believe that this binding is important for reef aquarium applications using normal potable water as the source.

Often the carbon filter also acts as a secondary sediment filter, and can therefore become clogged over time. A pressure drop at the RO membrane that is not solved by replacing the sediment filter may be due to a clogged carbon filter. In some cases, clogged filters may be cleared somewhat by backwashing, if the system permits reverse flow. Both carbon and sediment filters can become clogged with bacteria as well. In such a situation, some aquarists dry the filters out, thereby opening the pores, and then reusing them. I have never backwashed or dried any of my filters.

Reverse Osmosis Membranes

Reverse osmosis membranes consist essentially of a sheet of porous organic polymer. A variety of different materials is used commercially, including cellulose acetate/triacetate blends (sometimes called CTA), thin film/thin layer composites (sometimes called TLC or TFC), and modified polysulfone (sometimes called SPSF). The relative advantages of each of these materials are detailed in other articles, but some important points are:

• CTA membranes are inexpensive and resistant to oxidation by chlorine.

• TFC membranes are more costly, but have high impurity rejection. They must be protected from chlorine and chloramine.

• SPSF membranes are generally optimal only in special situations, such as very soft source water.

If the membrane's pore sizes are made just a bit larger than water molecules, then water can pass through them, but larger compounds cannot. Size in this case is a somewhat simplified idea. Many ions are smaller than a water molecule (Figure 3), but it turns out that charged ions (such as sodium, Na⁺) in solution contain several very tightly bound water molecules. Removing all of these attached water molecules requires a lot of energy, so when passing through a porous membrane, they act as if they are as large as the whole hydrated assembly (Figure 4). These larger assemblies cannot pass through an RO membrane as readily as they could without the tightly bound water molecules (Figure 5).

The more charges an ion has, the more water molecules are attached and the harder they are to remove. It has recently been suggested that the ratio of the hydrated volumes of two ions approximates the ratio of the square of the charges of the same two ions. So, for any simple inorganic X, Y, and Z, X⁺ is one-quarter the size of Y⁺⁺, and X⁺ is one-ninth the size of Z⁺⁺⁺. The same holds true for negatively charged ions.¹ For these reasons, the relative order of rejection by RO membranes is typically trivalent > divalent > monovalent, as shown below.

Since RO membranes purify based on size, they are subject to some obvious limitations. Certainly, anything that is very large cannot pass though them. In this category would be bacteria (although they may colonize both sides of the filter, they cannot pass through it), viruses, large organic molecules such as proteins, and inorganic mineral particulates that were small enough to pass through the sediment and carbon filters (often called colloids).

Also, in order to get a sufficiently fast flow of water through the membrane, membrane pores are actually significantly larger than a water molecule. For this reason, some of the molecules of compounds that are somewhat larger than a water molecule can still get through (sodium ion, for example, is not perfectly rejected).

However, at the small end of the spectrum a number of compounds can pass through a reverse osmosis membrane to some extent and are, therefore, of concern to reef aquarists. These include carbon dioxide (CO2), ammonia (NH3), hydrogen sulfide (H2S, especially a concern with well water) and silicic acid (Si(OH)4, which is the uncharged and predominate form of silicate at pH values below 9.5). All of these should be trapped by a functioning DI resin (discussed below), but can still be a concern.

In the case of CO2, for example, there can be a lot of it in certain well waters, and DI resins may become rapidly depleted because the CO2 so readily passes through RO membranes (how to deal with this is discussed later in this article). As another example, ammonia that comes from chloramine in the water can be significant, and is one reason that RO/DI is greatly preferred to RO alone in those situations where chloramine is added to the tap water.

In the case of silicic acid, some types of RO membranes can be better than others at excluding it, even before it gets to the DI resins. For example, a thin-film polyamide membrane might let only 0.3% of the silicic acid pass, while a similar cellulose acetate membrane might let 12.7% of it pass.

In order to function properly, the RO membrane is coupled with a flow restrictor that allows pressure to build on the upstream side of the membrane, rather than letting the water simply run out of the unit and down the drain. This pressure helps force the water molecules (and other small molecules) through the membrane. After passing through the pores, the water then continues on to the DI resin.

Many systems will include a pressure gauge that measures the line pressure ahead of the RO membrane. It is the pressure across the RO membrane that forces water through. At low pressure, the water may simply run past the restrictor and down the drain. Most membranes need at least 40 PSI or so to get reasonable flow and purification. In my system, the pressure drops over time as sediment clogs the sediment and carbon filters. I use this gauge as an indicator that the filters before the membrane need to be replaced. Some RO/DI manufacturers (e.g., Spectrapure and Kent) sell kits that allow the membrane to be flushed with water, permitting loose sediments and calcium/magnesium carbonate that can clog it to be washed away.

Various factors, such as temperature and pressure, impact not only the flow rate through the membrane but also the purity of the resulting water. Lower temperatures make the water more viscous and less likely to flow through the small pores, reducing the production of purified water. The effect of temperature on purity is much smaller, with purity decreasing slightly at higher temperatures. Higher line pressure across the RO membrane results in higher rates of production and quality, although a pressure that is too high can damage the membrane. Any backpressure on the effluent will degrade performance. Very high TDS (total dissolved solids) in the source water also leads to higher osmotic backpressure, reducing the membrane's effectiveness. As a rough guide, every 100 ppm of TDS produces 1 psi of osmotic backpressure.

For those interested in a great many more details on reverse osmosis membranes and their engineering applications, a big library of articles is available online at General Electric's website: "What is Reverse Osmosis."

Deionizing Resins

The final filter in an RO/DI system is the deionizing resin. A DI resin traps all charged molecules passing through it, and leaves uncharged (neutral) molecules free to pass through. Water, for example, passes through it, as would other uncharged inorganic molecules such as oxygen (O2), nitrogen (N2) and chloramine (NH2Cl, if any remained from the previous filters). Uncharged organic molecules also pass through a DI resin, including ethanol (CH3CH2OH), methanol (CH3OH), methane (CH4), propane (CH3CH2CH3), carbon tetrachloride (CCl4), and methylene chloride (CH2Cl2). Ions such as sodium (Na⁺), copper (Cu⁺⁺ or Cu⁺), ammonium (NH4⁺), phosphate (PO4^---), silicate (Si(OH)3O^-), and acetate (CH3CO2^-) all get caught.

All atoms or molecules that are in rapid and significant equilibrium with their charged forms will be caught and removed just as their charged forms are. These include ammonia (NH3) caught as ammonium, silicic acid (Si(OH)4) caught as silicate, carbon dioxide caught at least partially as bicarbonate (HCO3^-) or carbonate (CO3^--), and acetic acid (CH3CO2H) caught as acetate, etc.

To catch these ions, the resin consists of porous beads that have fixed charges attached to them. The counterions to these fixed charges start off as H⁺ and OH^- in a fresh resin. Normally, there are different beads intended to bind cations and anions. In a mixed bed DI resin, the beads are mixed together in a single filter. In a separate bed system, each bead type will be in a different filter, thereby potentially allowing the DI to be recharged (a process that is discussed later in this article).

Figure 6 shows a cation-binding resin bead ready to bind sodium (the fixed charges on the resin are not shown, only the replaceable H⁺ ions). As the sodium ions enter the bead, they bind to the fixed negative charges, and H⁺ is released as they swap places. The chloride ions pass through this bead unchanged since they are not attracted to the negatively charged sites in the bead. After all of the sodium ions that entered the bead are bound, the product is a hydrogen chloride solution that passes on to the next bead in the bed (Figure 7).

Figure 8 shows an anion-binding resin bead ready to bind chloride (the fixed charges on the resin are not shown, only the replaceable OH^- ions). As the chloride ions enter the bead, they become bound to the fixed positive charges, and OH^- is released as they swap places. As the chloride ions that entered the bead are bound and OH^- is released, the H⁺ and OH^- ions combine to form water molecules (H2O). In this way, none of the original sodium and chloride ions remain in solution, and only pure water passes out of the DI resin chamber (Figure 9).

Deionizing Resin Depletion Issues

Eventually, all of the H⁺ and OH^- originally installed in a DI resin will become depleted, and ions will pass through unchanged (Figure 10). When charged ions begin to pass through the DI resin, the effluent's conductivity rises. Many RO/DI systems use an inline conductivity meter to alert users when ions are starting to appear, indicating that the resin needs to be replaced. Without such an inline meter, users need to periodically monitor the effluent’s conductivity (in mS/cm or ppm TDS; details are given in the tips section on what conductivity to target for resin replacement).

Some DI resins incorporate color changes to indicate when the DI is depleted. Such indicators are typically pH indicating dyes that change color when the pH in the interior of the beads shifts from the very high or low pH values when OH^- or H⁺ are the dominant counterions, to more neutral values when other ions dominate (such as Na⁺ or Cl^-). Such color changes may be less effective than measuring the effluent’s conductivity for indicating early breakthrough of ions. The color change may not indicate that some beads or parts of beads may become depleted before others due to channeling of the ion flow. Consequently, I would not rely exclusively on such color changes unless they have proven to accurately predict the rise in conductivity of the effluent for a given brand of DI filter and bead.

Several issues arise relating to the depletion of the DI resins that aquarists need to be aware of. Primary among these is that when a DI resin becomes depleted, that does not simply mean that the water passes through just as it came from the RO effluent. It may actually be much worse from an aquarist’s perspective. The reason for this is that while the DI resin is functioning properly, all ions will be caught. But when it is depleted, not only the new ions are coming through and might show up in the product water, but so are all the ions that ever got into the DI resin in the first place. The total concentration of ions coming out of the exhausted DI resin will not be raised as compared to the RO's effluent, but which ions are released may be very different.

In the DI descriptions above, I did not address the fact that some ions will show a greater preference for attachment to the resin than will others. When the resins are not depleted, it does not matter what the ions’ affinity is, as all are bound. But in a depleted scenario, when there are more ions than ion binding sites, those with a higher affinity for the resin will be retained, and those with a lower affinity will be released. It turns out that silicate is found at the lower end of affinity for anion resins. Consequently, if the DI resin has been collecting silicate for a long period and is then depleted, a large burst of silicate may be released.

Perhaps even more of a concern is ammonia. In a system with chloramine in the tap water, the DI resin will serve the important function of removing much of the ammonia produced by the chloramine breakdown. Ammonia has a poorer affinity for many cation-binding resins than do many other cations (e.g., calcium or magnesium). Consequently, when the DI resin first becomes depleted, a big release of ammonia from and through the DI resin is likely. I recently had a DI resin become depleted, and the effluent contained so much ammonia that I could easily smell it.

Other complications can also impact resin depletion. One potentially important issue is that the anion and cation-binding sites may not become depleted at the same time. Figure 10 shows this scenario when both types become depleted together, with sodium and chloride in the effluent. But, it is possible for one to become depleted first, and in that case, the pH of the effluent can swing far from neutral. Figures 11 and 12 show what happens when a lot of carbon dioxide is present, as is the case with some well waters. Initially, it is mostly bound as bicarbonate, and the effluent is essentially pure water. Note, however, that as the bicarbonate is removed, the anion binding resin is being taken up with bicarbonate, while the cation-binding resin is unchanged and is therefore not being depleted.

Eventually, the anion-binding sites become fully occupied (Figure 13). At that point, additional ions coming through (such as sodium and chloride) are no longer equally swapped out to produce pure water. The sodium is swapped for H⁺, but the chloride does nothing, potentially leaving the effluent water with a very low pH.

A similar effect can be hypothesized for silicic acid in the RO permeate:

Si(OH)₄  à  H⁺  +  Si(OH)₃O^-

The effect on pH of the DI resin’s initial depletion would be similar here to the effect of carbon dioxide in the tap water.

The same can happen in the opposite sense with ammonia. If a lot of ammonia gets through the RO membrane (as is the case when chloramine is present), the ammonia will be bound in the DI resin as ammonium:

NH₃  +  H₂O  -->  NH₄⁺  +  OH^-

The ammonium depletes the cation-binding resin, while the OH^- does not impact the anion-binding resin. Eventually, then, the cation-binding capacity can become depleted before the anion binding is depleted, and Na⁺/Cl^- passing through is converted into Na⁺ and OH^-, with a potentially high pH.

Final Effluent pH

Aside from the issues discussed above concerning the effluent’s pH when the DI resin becomes depleted, the final pH coming out of an RO/DI system should not significantly concern reef aquarists. Many aquarists with low pH problems have asked, for example, if their aquarium’s low pH may be caused by their replacing evaporated water with RO/DI water that they measure to have a pH below 7. In short, the answer is no, this is not a cause of low pH nor is it something to be generally concerned about, for the following reasons:

1. The pH of totally pure water is around 7 (with the exact value depending on temperature). As carbon dioxide from the atmosphere enters the water, the pH drops into the 6’s and even into the 5’s, depending on the amount of CO2. At saturation with the level of CO2 in normal (outside) air, the pH would be about 5.66. Indoor air often has even more CO2, and the pH can drop a bit lower, into the 5’s. Consequently, the pH of highly purified water coming from an RO/DI unit is expected to be in the pH 5-7 range.

2. The pH of highly purified water is not accurately measured by test kits, or by pH meters. There are several different reasons for this, including the fact that highly purified water has very little buffering capacity, so its pH is easily changed. Even the acidity or basicity of a pH test kit’s indicator dye is enough to alter pure water’s measured pH. As for pH meters, the probes themselves do not function well in the very low ionic strength of pure freshwater, and trace impurities on them can swing the pH around quite a bit.

3. The pH of the combination of two solutions does not necessarily reflect the average (not even a weighted average) of their two pH values. The final pH of a mixture may actually not even be between the pH’s of the two solutions when combined. Consequently, adding pH 7 pure water to pH 8.2 seawater may not even result in a pH below 8.2, but rather might be higher than 8.2 (for complex reasons relating to the acidity of bicarbonate in seawater vs. freshwater).

Recharging DI Resins

When DI resins are present as mixed bed filters, they are essentially one-time use devices that must be thrown away when depleted. These are the types supplied with many commercial RO/DI systems, and are what I use. Separate bed DI resins have certain advantages, however. In particular, if they are kept in different cartridges, they can be recharged. Trendy deals of the Makro Specials and exclusive price range.

The recharging process is essentially the reverse of the deionizing process (Figures 6-8). A strong acid (usually hydrochloric acid, HCl) is used to swap H⁺ for all of the positively charged ions on the cation-binding resin (Na⁺, K⁺, Ca⁺⁺, Mg⁺⁺, etc.). Likewise, a strong base (usually sodium hydroxide, NaOH) is used to swap OH^- for all of the negatively charged ions on the anion binding resin. Both hydrochloric acid and sodium hydroxide are readily available and inexpensive. They are, however, potentially dangerous to work with. The detailed procedures and safety precautions are beyond the scope of this article, but can be found elsewhere online.

Tips on Using RO/DI

1. If you have a pressure gauge on the RO membrane, change the sediment filter when the pressure drops to about 40 PSI. If the pressure does not rise adequately after replacing the sediment filter, change the carbon filter.

2. If you do not have a pressure gauge, change the sediment and carbon filters when the output of purified water becomes significantly reduced.

3. Avoid running the RO/DI if your town’s water system is flushing fire hydrants in your area, as the sediment and carbon filters will clog very fast.

4. If the sediment and carbon filters do not seem to get clogged, at least replace the carbon filter periodically to ensure that chlorine and chloramine are being broken down appropriately. Both of these can be measured with an inexpensive chlorine test kit if you want to be sure they are being removed.

5. If you have especially hard water, you might consider running it through a water softener before the RO/DI filter. That process removes calcium and magnesium by swapping sodium for them. While aquarists are not generally worried about calcium and magnesium in their source water, they can foul RO membranes by forming calcium and magnesium carbonate precipitates inside it. That precipitation will reduce flow and eventually make the membrane unusable.

6. Make sure that your waterline pressure is appropriately high (at least 40-60 PSI). In some situations where the pressure is low, an inline pump may be desirable before the RO/DI system to boost the pressure.

7. If there is excessive carbon dioxide in the source water (such as well water), you might consider degassing the water first to remove some of the carbon dioxide, and thereby reduce the rate of depletion of the DI resins. Note, however, that this option is not inexpensive, and typically involves repressurizing the water with a pump.

8. If you are evaluating an existing RO membrane and can collect water from the tap and after the RO membrane, the conductivity (in mS/cm or ppm TDS) should drop by a factor of more than 10 across it (to as much as 100), relative to the tap’s water. If the drop is less than a factor of 10, it is not working properly, and may have holes in it.

9. Monitor the DI resins by measuring the effluent’s conductivity, either with an inline meter (set to its most sensitive level), or by measuring the effluent manually. If you are using a TDS or conductivity meter, then the measured value should drop to near zero, or maybe 0-1 ppm TDS or 0-1 mS/cm. Higher values indicate that something is not functioning properly, or that the DI resin is becoming saturated and needs replacement. That does not necessarily mean, however, that 2 ppm TDS water is not OK to use. But beware that the flow of impurities and the conductivity may begin to rise fairly sharply when the resin becomes saturated. Do not agonize over 1 ppm versus zero ppm. While pure water has a TDS well below 1 ppm, uncertainties from carbon dioxide in the air (which gets into the water and ionizes to provide some conductivity; about 0.7 mS/cm for saturation with normal levels of CO2, possibly higher indoors) and the conductivity/TDS meter itself may yield results of 1 or 2 ppm even from totally pure water by not being exactly zeroed properly. Also note that the first impurities to leave the DI resin as it becomes saturated may be things that you are particularly concerned with (such as ammonia if your water supply uses chloramine or silica if there is a lot in the source water).

10. If you recharge your DI resins yourself, be very careful with the acid and base used, as they can be dangerous.

Conclusions

RO/DI is likely the most effective way for reef aquarists to adequately purify all tap waters they are likely to encounter. Most aquarists will find that a standard system consisting of a sediment filter, a carbon filter, a reverse osmosis membrane, and a deionizing resin cartridge is perfectly suitable. It is also useful to have the ability to measure the conductivity of the effluent inline, and to measure the pressure across the RO membrane.

In some cases, it may be desirable to degas the incoming tap water if it contains excessive carbon dioxide. It may also be desirable to have two DI cartridges in series if there is a lot of silica or other easily penetrating problem ions in the RO permeate. If you have well water that smells bad or has other peculiarities, it may be worth discussing the issue with a top-of-the-line RO/DI system manufacturer to make sure that you are treating the water appropriately.

Whatever brand or setup you use, be sure to change the filters at appropriate intervals.

Happy Reefing!

References:

1. Dynamic hydration numbers for biologically important ions. Kiriukhin, Michael Y.; Collins, Kim D. Department of Biochemistry and Molecular Biology, University of Maryland Medical School, Baltimore, MD, USA. Biophysical Chemistry (2002), 99(2), 155-168.