Raw Water Quality

In addition to potential pathogens, raw water may contain contaminants that may interfere with the disinfection process or may be undesirable in the finished product. These contaminants include inorganic and organic molecules, particulates, and other organisms, e.g., invertebrates. Variations among these contaminants arise from differences in regional geochemistry and between ground- and surface-water sources.

Particulates and Aggregates

To inactivate organisms in water, the active chemical species must be able to reach the reactive site within the organism or on its surface. Inactivation will not result if this cannot occur. Microorganisms may acquire physical protection in water as a result of their being adsorbed to the enormous surfaces provided by clays, silt, and organic matter or to the surfaces of solids created during water treatment, e.g., aluminum or ferric hydrated oxides, calcium carbonate, and magnesium hydroxide. Viruses, bacteria, and protozoan cysts may be adsorbed to these surfaces. Such particles, with the adsorbed microorganisms, may aggregate to form clumps, affording additional protection. Organisms themselves may also aggregate or clump together so that organisms that are on the interior of the clump are shielded from the disinfectant and are not inactivated. Organisms may also be physically embedded within particles of fecal material, within larger organisms such as nematodes, or, in the case of viruses, within human body cells that have been discharged in fecal material.

To disinfect water adequately, the water must have been pretreated, when necessary, to reduce the concentration of solid materials to an acceptably low level. The primary drinking water turbidity standard of 1 nephelometric turbidity unit (NTU) is an attempt to assure that the concentration of particulates is compatible with current disinfection techniques. Where it is possible to obtain lower turbidities, this is desirable.

Disinfection studies in which the complications of adsorbed organisms, aggregation, or embedment were thought to occur were excluded from this study. The conclusions in this report should not be extrapolated to such situations as the disinfection of turbid or colored waters.

The Importance of Residuals

Water supplies are disinfected through the addition or dosage of a chemical or physical agent. With a chemical agent, such as a halogen, a given dosage should theoretically impart a predetermined concentration (residual) of the active agent in the water. From a practical point of view, most natural waters exert a "demand" for the disinfectant, as discussed above, so that the residual in the water is less than the calculated amount based on the dosage. The decrease in residual, which is caused by the demand, is rapid in most cases, but it may be prolonged until the residual eventually disappears. In addition, the chemical agent may decompose spontaneously, thereby yielding substances having little or no disinfection ability and exerting no measurable residual. For example, ozone not only reacts with substances in water that exert a demand, but it also decomposes rapidly. To achieve microbial inactivation with a chemical agent, a residual must be present for a specific time. Thus, the nature and level of the residual, together with time of exposure, are important in achieving disinfection or microbial inactivation. Because the nature of the dosage-residual relationship for natural waters has not been and possibly cannot be reliably defined, the efficacy of disinfection with a chemical agent must be based on a residual concentration/time of exposure relationship.

Residual measurements are important and useful in controlling the disinfection process. By knowing the residual-time relationship that is required to inactivate pathogenic or infectious agents, one can adjust the dosage of the disinfecting agent to achieve the residual that is required for effective disinfection with a given contact time. Thus, the effectiveness of the disinfection process can be controlled and/or judged by monitoring or measuring the residual.

Following disinfection of a water supply at a treatment plant, the water is distributed to the consumers. A persistent residual is important for continued protection of the water supply against subsequent contamination in the distribution system. Accidental or mechanical failures in the distribution system may result in the introduction of infectious agents into the water supply. In the presence of a residual, disinfection will continue and, as a result, offer continued protection to the users. Physical agents such as radiation may provide effective disinfection during application, but they do not impart any persistent residual to the water.

The dosage of a chemical agent that is used to effect microbial inactivation should not be so great that it imparts a health hazard to the water consumer. From another point of view, the aesthetic quality of the finished water should not be impaired by the dosage of the chemical agent or the residual that is required for effective disinfection. These qualities might include discoloration of water from potassium permanganate (KMnO₄) or iodine or problems of taste and odor from excessive chlorine.

Application of the Disinfectant

Optimum inactivation occurs when the disinfectant is distributed uniformly throughout the water. To disperse the chemical disinfectant when it is added to the water, it must be mixed effectively to assure that all of the water, however small the volume, receives its proportionate share of the chemical. Additions of a disinfectant at points in a flowing water stream, e.g., from submerged pipes, is seldom adequate to assure uniform concentration. In such cases, mechanical mixing devices are needed to disperse the disinfectant throughout the water. Disinfection by radiation treatment also requires good mixing to bring all of the water within the effective radiation distance.

Microbiological Considerations¹

Comparison of the biocidal efficacy of disinfectants is complicated by the need to control many variables, a need not realized in some early studies. Halogens in particular are significantly affected by the composition of the test menstruum and its pH, temperature, and halogen demand. For very low concentrations of halogen to be present over a testing period, halogen demand must be carefully eliminated. Different disinfectants may have different biocidal potential. In earlier work, analytical difficulties may have precluded defining exactly the species present, but new techniques allow the species to be defined for most disinfectants. Information on the species of disinfectant actually in the test system should be included in future reports on disinfection studies.

Investigators studying efficacy have usually adopted one of two extremes. Some have conducted carefully designed laboratory experiments with controls for as many variables as possible. Certain of these investigators have reduced the temperature to slow the inactivation reactions. Although these experiments yield good basic information and can be used to determine which variables are important, they often have little quantitative relationship to field situations. The other extreme, a field study or reconstruction of field conditions, is difficult to control. Moreover, their results are often not repeatable.

In addition to the variables noted above, prereaction of chemicals in the test system, the culture history of the organism being used, and the ''cleanup" procedures applied to it may also affect the observed results. Despite these problems, there have been some attempts to standardize efficacy testing.

Chemistry of Chlorine in Water

Chlorine has an atomic number of 17, a melting point of -102°C, a boiling point of -35°C, and an oxidation potential of -1.36 V at 25°C (i.e., 2 Cl^- Cl₂ + 2 e^-). It is a green-yellow gas at room temperature.

When chlorine is added to water, the following chemical reactions occur:

(1)

(2)

Extremely little molecular chlorine (Cl₂) is present at pH values greater than pH 3.0 and total chlorine concentrations of less than ~1,000 mg/liter (White, 1972). The hypochlorous acid (HOCl) that is produced further ionizes to form hypochlorite ion (OCl^-) and hydrogen ion (H⁺) (Reaction 2). The dissociation of hypochlorous acid is dependent chiefly upon pH and, to a much lesser extent, temperature, with almost 100% hypochlorous acid present at pH 5 and almost 100% hypochlorite ion present at pH 10 (Figure II-1). Free available chlorine refers to the concentration of hypochlorous acid and hypochlorite ion, as well as any molecular chlorine existing in a chlorinated water.

Figure II-1

Effect of pH on quantities of hypochlorous acid (HOCl) and hypochlorite ion (OCl^-) that are present in water. Data from Fair et al., 1948.

Chloramine Formation: Inorganic Chloramines

During chlorination of a water supply for disinfection, chlorine will react with any ammonia (NH₃) in the water to form inorganic chloramines.

Furthermore, ammonia is sometimes deliberately added to chlorinated public water supplies to provide a combined available chlorine residual, i.e., inorganic chloramines. Chlorine will also react with organic amines. The organic chloramines that are produced (see below) are considered encompassed in the term "combined available chlorine."

Although inorganic chloramines are less effective oxidizing and disinfecting agents than hypochlorous acid and hypochlorite ion, they are more stable. Consequently, they will produce a residual in water that will persist for a longer time (Symons et al., 1977).

Inorganic chloramines are formed when hypochlorous acid reacts with ammonia:

(3)

(4)

The chloramine that is formed in the reaction depends upon the ratio of ammonia to hypochlorous acid and the pH of the system. Dichloramine (NHCl₂) is the predominant form of chloramine at a 1:1 molar ratio of ammonia to chlorine at pH values of 5 and below, whereas at pH values of 9 and above, monochloramine (NH₂Cl) predominates. Figure II-2 shows the proportions of monochloramine and dichloramine formed for pH values of 4 to 9 and temperatures of 0°C, 10°C, and 25°C (Morris, 1978, personal communication).

Figure II-2

Proportions of mono- and dichloramine (NH₂Cl and NHCl₂) in water chlorination with equimolar concentrations of chlorine and ammonia. Data from J.C. Morris, personal communication.

Organic Chloramines

Chlorine is also known to combine slowly with organic or albumenoid nitrogen (amines) to form organic chloramines (Taras, 1953):

(5)

Although the reaction between organic amines and chlorine is generally considered to be slow, organic chloramines may be formed, thereby producing a stable combined available chlorine residual after many hours of contact. It is generally accepted that most organic chloramines have little disinfecting capability, i.e., less than the inorganic chloramines (Feng, 1966; Nusbaum, 1952).

Breakpoint Chlorination

Hypochlorous acid and other chlorine compounds having disinfecting ability by virtue of their being oxidizing agents will oxidize sulfites (SO₃^2-), sulfides (S^-), and ferrous (Fe²⁺) or manganous (Mn²⁺) ions. The disinfecting species are reduced, and the products have no disinfecting activity. All of the interfering compounds that destroy the disinfecting ability of the added chlorine exert a "chlorine demand," which may be defined as the difference between the amount of chlorine applied and the quantity of free or combined available chlorine residual measured in the water at the end of a specified contact period. When chlorine is added to water with no chlorine demand, a linear relationship is established between the chlorine dosage and the free chlorine residual (Figure II-3).

Figure II-3

Diagrammatic representation of completed breakpoint reaction. From Morris, 1978, personal communication.

However, when increasing amounts of chlorine are added to water containing reducing agents and ammonia, the so-called breakpoint phenomenon occurs. The breakpoint is that dosage of chlorine that produces the first detectable amount of free available chlorine residual.

When chlorine is added to water, it reacts with any reducing agents and ammonia that are present. It is believed that chlorine reacts first with the reducing agents. Since the chlorine is destroyed, no measurable residual is produced. Following the oxidation of these reducing agents, e.g., sulfides, sulfites, nitrites (NO₂^-), and ferrous ions, the chlorine will react with ammonia to form inorganic chloramines. The quantity of monochloramine and dichloroamine that is formed is determined primarily by the pH of the water and the ratio of chlorine to ammonia. When the ratio by weight is less than 5:1, or the molar ratio is less than 1:1, and the pH is in the range of 6.5 to 8.5, the combined available chlorine residual is probably due primarily to monochloramine (Reaction 3). With additional chlorine, the ratio of chlorine to ammonia changes with the result that the monochloramines are converted to dichloramines (Reaction 4). When all of the ammonia has been reacted, a free available chlorine residual begins to develop. As the concentration increases, the previously formed chloramines are oxidized to nitrous oxide (N₂O), nitrogen trichloride (NCl₃), and nitrogen (N₂). The reactions leading to the formation of these oxidized forms of nitrogen destroy the combined available chlorine residual so that the measurable residual in the water actually decreases. Upon completion of the oxidation of all the chloramines, the addition of more chlorine creates the breakpoint phenomenon. At the breakpoint dosage, some resistant chloramines may still be present, but at such small concentrations that they are unimportant.

As pointed out by Morris (1970), the occurrence of reactions giving rise to the ''breakpoint" is most rapid in the pH range 7.0 to 7.5. At greater and lesser pH values, it becomes slower and less distinct, e.g., at pH's < 6 or > 9 the concept of "breakpoint" is not significant. In the pH range 7.0 to 7.5 the "breakpoint" is about half developed within 10 min at 15°C to 20°C and is then substantially completed within about 2 hr.

Analytical Methods and Their Evaluation

Standard Methods (1976) lists six acceptable methods for the determination of chlorine residuals in natural and treated waters: iodometric methods, amperometric titration, the stabilized neutral orthotolidine (SNORT) method, the ferrous diethyl-p-phenylenediamine (DPD) method, the DPD colorimetric method, and the leuco crystal violet (LCV) method.

The amperometric, LCV, DPD, and SNORT methods are unaffected by dichloramine concentrations in the range of 0 to 9 mg/liter (as Cl₂) in the determination of free chlorine. If nitrogen trichloride is present, it reacts partially as free available chlorine in the amperometric, DPD, and SNORT methods. Nitrogen trichloride does not interfere with the LCV procedure for free chlorine. The sample color and turbidity may interfere with all colorimetric procedures. Thus, a compensation must be made. Also, organic contaminants in the sample may produce a false-free chlorine reading in most colorimetric methods.

Standard Methods (1976) contains data on the precision and accuracy of the methods used in the measurement of chlorine. These data were obtained from participating laboratories by the Analytical Reference Service (1969, 1971), which then operated in an agency that preceded the Environmental Protection Agency. However, as noted in Standard Methods (1976), these results are valuable only for comparison of the methods tested, and many factors, such as analytical skill, recognition of known interferences, and inherent limitations, determine the reliability of any given method. Moreover, some oxidizing agents, including free halogens other than chlorine, will appear quantitatively as free chlorine. This is also true of chlorine dioxide. Also, some nitrogen trichloride may be measured as free chlorine. The actions of interfering substances should be familiar to the analyst because they affect a particular method.

Although orthotolidine (i.e., orthotolidine and orthotolidine arsenite) methods have been widely used in many disinfection studies, they are omitted from the 14th edition of Standard Methods (1976) primarily because of their inaccuracy and high overall (average) total error in comparison with other available methods.

Research studies on disinfection are restricted by the limitations that are inherent in the methods themselves or by poor selection of methods by the investigator. The chemical conditions of the test water have not always been well-defined. The types of titratable chlorine, i.e., free (hypochlorous acid or hypochlorite ion) or combined (mono- or dichloramine) in the chlorinated water, have not always been differentiated, and the rates of microbial destruction or inactivation have not always been studied in experimental systems with little or no chlorine demand. In fact, reports prior to the 1940's have been especially difficult to interpret, because reliable test methods for distinguishing between free and combined chlorine, between hypochlorous acid and hypochlorite ion, and between mono- and dichloramine in solution were not developed until the 1950's. For example, many earlier researchers claimed to have tested mono- and dichloramine by controlling the pH and the ratio of chlorine to nitrogen. They used methods such as the orthotolidine or thiosulfate titrations to determine total chlorine residual. Much of this early work is now questionable, since it was not possible to detect free chlorine contamination in their chloramine solutions or the quantitative ratios between the mono- and dichloramine tested. in addition, these earlier studies had high chlorine demand in the test systems.

Some more contemporary studies have lacked quantitated information on chlorine residual and/or types of chlorine present in the test systems.

Biocidal Activity

In the absence of reducing agents, inorganic ammonia, and organic amines, the addition of chlorine to municipal water supplies will result in free available residual chlorine, represented by the hypochlorous acid or hypochlorite ion. The pH determines the relative amounts of each species. However, inorganic chloramines will be formed if the background level of ammonia in the water supply is significant or if ammonia is intentionally added during treatment. If such is the case, monochloramine would predominate due to the alkaline pH of most finished water (see Figure II-2).

In 1966, Feng proposed that the active forms of chlorine would exhibit disinfection properties in the following descending order:

Butterfield et al. (1943) published the first treatise on the use of chlorine demand-free water for studies of water disinfection. They proposed that to study the disinfectant capacity of any chlorine species, the test medium must meet certain exacting criteria. It must be nontoxic to bacteria except for the variables under study such as chlorine and pH, well buffered at the desired pH, free of all ammonia and organic matter capable of forming chlorine-addition products, free of background chlorine, and of such a nature that calculated additions of chlorine are recoverable after 5 min without a loss in residual and that free chlorine must still be present several hours after contact.

Most studies of combined chlorine have dealt with poorly defined mixtures of mono- and dichloramine. Also, test conditions have often been inadequately defined, poorly controlled, or both.

Efficacy against Bacteria

Free Chlorine (HOCl and OCl^-)

Butterfield et al. (1943) studied percentages of inactivation as functions of time for E. coli, Enterobacter aerogenes, Pseudomonas aeruginosa, Salmonella typhi, and Shigella dysenteriae. They used different levels of free chlorine at pH values ranging from 7.0 to 10.7 and two temperature ranges—2°C to 5°C and 20°C to 25°C. Their work is of great importance, since very few other studies have been conducted that dealt with the action of disinfectants on pathogens. Generally, they found that the primary factors governing the bactericidal efficacy of free available chlorine and combined available chlorine were:

• the time of contact between the bacteria and the bactericidal agent, i.e., the longer the time, the more effective the chlorine disinfection process;
• the temperature of the water in which contact is made, i.e., the lower the temperature, the less effective the chlorine disinfecting activity; and
• the pH of the water in which contact is made, i.e., the higher the pH, the less effective chlorination.

Thus, the test bacteria will be killed more rapidly at lower pH values and at higher temperatures. Since hypochlorous acid would predominate at lower pH's (Figure II-1), the data of Butterfield et al. show that it is a better bactericide than the hypochlorite ion. For example, to produce a 100% inactivation of an initial inoculum of 8 × 10⁵ E. coli in 400 ml of sterile chlorine demand-free water (2,000/ml) at 20°C-25°C with a chlorine level of 0.046 to 0.055 mg/liter, 1.0 min was required at pH 7.0, but at pH 8.5, 9.8, and 10.7, between 20 and 60 min of exposure were needed. At higher concentrations of chlorine, i.e., from 0.1 to 0.29 mg/liter, exposure of 1.0 min was required at pH 7.0, 10 min at pH 8.5, 20 min at 9.8, and 60 min at 10.7. A similar pH effect was noted for S. typhi.

Unfortunately, Butterfield et al. (1943) lifted their cells from agar slants but failed to wash them in demand-free water. The cells probably carried trace amounts of albumenoid nitrogen from the slants to the test flasks, thereby creating the small chlorine demand that the investigators had tried so carefully to avoid. The effect of such a trace amount of chlorine demand would be most apparent in test solutions with very low chlorine levels. In studies using approximately 0.1 mg/liter or less free chlorine at decreasing pH values, Butterfield et al. (1943) observed that the disinfection of the organisms required a very long time. This might indicate interference at the low levels due to the formation of combined chlorine.

Under very exact controlled test conditions of pH and temperature and using chlorine demand-free buffer systems, Scarpino et al. (1972) observed that at 5°C E. coli was 99% destroyed by 1.0 mg/liter hypochlorous acid at pH 6 in less than 10 s and at pH 10 by 1.0 mg/liter hypochlorite ion in about 50 s. Their studies, which totally eliminated any form of combined chlorine from the test solutions, indicated that hypochlorous acid was approximately 50 times more effective than the hypochlorite ion as a bactericide. Fair et al. (1948) and Berg (1966) analyzed the data of Butterfield et al. (1943) on the destruction of E. coli. They reported that hypochlorous acid was 70-80 times as bactericidal as hypochlorite ion.

Engelbrecht et al. (1975) investigated new microbial indicators of disinfection efficiency. In their chlorination studies, they noted the following decreasing order of resistance to free chlorine at pH values of 6, 7, and 10 and at 5°C and 20°C: acid-fast bacteria > yeasts > poliovirus > Salmonella typhimurium > E. coli.

Monochloramine (NH₂Cl)

In 1948, Butterfield summarized previous results on the bactericidal properties of chloramines (and free chlorine) in water at pH values ranging from 6.5 to 10.7 and in two temperature ranges—2°C to 5°C and 20°C to 25°C. The test bacteria included strains of Escherichia coli, Enterobacter aerogenes, Pseudomonas aeruginosa, Salmonella typhi, and Shigella dysenteriae. Although he admitted that adequate tests for separate determination of free and combined chlorine forms were not used in these studies, the solutions were vigorously prepared to ensure the exclusions of free chlorine. Chloramines were determined using orthotolidine; readings made after 10 to 30 s at 20°C gave free chlorine levels, and those after standing for 10 min at 20°C were recorded as total residual chlorine. Since no free chlorine was reported (and should not have been found, according to the authors), the 10-min readings of total residual chlorine were also those of total chloramine levels. No distinction could be made between monochloramine and dichloramine. However, he estimated that at pH 6.5, 7.0, 7.8, 8.5, 9.5, and 10.5 the chloramines were present as monochloramine at 35%, 51%, 84%, 98%, 100%, and 100%, respectively. The balance was believed to be dichloramine.

Butterfield and his associates (Butterfield, 1948; Butterfield et al., 1946) found that chloramine disinfection was always slower than that for free chlorine. For example, in order to achieve a 100% inactivation of the initial number of bacteria tested after 60 min of contact time at 20°C, 0.6 mg/liter chloramine was required at pH 7.0 and 1.2 mg/liter at pH 8.5. At 4°C, a 100% inactivation required 1.5 mg/liter chloramine at pH 7.0 and 1.8 mg/liter at pH 8.5. However, only 0.03 to 0.06 mg/liter free chlorine was needed at pH ranges of 7.0 and 8.5 at either 4°C or 22°C to achieve 100% inactivation in 20 min.

The bactericidal effects of monochloramine alone were confirmed by Siders et al. in 1973. They found that at 15°C E. coli was 99% destroyed in approximately 20 min using 1.0 ppm monochloramine in pH 9 borate buffer. Since E. coli was less resistant to monochloramine than were the animal viruses tested, Siders et al. questioned the validity of using E. coli as an indicator organism for measuring the viral quality of a chlorinated water supply. Chang (1971) had previously calculated that 4 mg/liter monochloramine would be needed to give 99.999% reduction of E. coli bacterium in 10 min at 25°C.

Dichloramine (NHCl₂)

Chang (1971) calculated that 1.2 mg/liter dichloramine would be needed to give 99.999% reduction of enteric bacteria in 10 min at 25°C.

In carefully conceived studies, Esposito et al. (1974) examined the destruction rates of test organisms in contact with dichloramine in demand-free phthalate buffer at pH 4.5 and 15°C. Figure II-4 shows the comparisons that were made among enteroviruses (poliovirus 1 and coxsackievirus A9), the bacteriophage ΦX-174, and E. coli (ATCC 11229).

Figure II-4

Inactivation of various microorganisms with dichloramine (NHCl₂) at pH 4.5 and 15°C. From Esposito, 1974.

From a review of the literature and an analysis of the data, Chang (1971) calculated that the relative bactericidal efficiency of dichloramine to monochloramine was 3.3:1. However, Esposito (1974) and Esposito et al. (1974) showed experimentally that Chang's estimate was conservative. They found that dichloramine was 35 times more bactericidal than was monochloramine, not 3.3. At 15°C, poliovirus 1 was 17 times more resistant than coxsackievirus A9, 83 times more resistant than ΦX-174, and 1,700 times more resistant than E. coli to dichloramine (Figure II-4). They observed that dichloramine was a better bactericide than monochloramine.

Organic Chloramines

These chlorine derivatives exhibit some bactericidal activity, but markedly less than either free chlorine or the inorganic chloramines (Feng, 1966; Nusbaum, 1952).

In summary, the bactericidal efficiency of hypochlorous acid, the hypochlorite ion, monochloramine, and dichloramine have been accurately defined in recent years by investigators using rigidly controlled test conditions. The order of disinfection efficiency presented by Feng (1966) has been confirmed. Comparative c · t values are shown in Table II-3 at the end of the section on chlorine.

TABLE II-3

Dosages of Various Chlorine Species Required for 99% Inactivation of Escherichia Coli and Poliovirus 1.

Efficacy against Viruses

In reviewing disinfection of enteroviruses in water, Clarke and Chang (1959) excluded all studies on the inactivation of viruses by chlorine that were conducted before 1946. Their justification for this exclusion was the failure of these studies to differentiate between free and combined chlorine. Furthermore, they attributed the irregular virucidal results of some studies to the use of animal inoculation methods for assaying virus concentrations. For these same reasons, those studies have been omitted in this report. The advent of viral propogation techniques using tissue cultures (Enders et al., 1949) enabled research of a more exacting nature to be performed, resulting in more precise virus inactivation data.

Free Chlorine (HOCl and OCl^-)

Generally, enteroviruses are more resistant to free chlorine than are the enteric bacteria (Chang, 1971; Clarke and Kabler, 1954; Scarpino et al., 1972). For example, in what was probably the first well-defined study, Clarke and Kabler (1954) used purified coxsackievirus A2 to investigate viral inactivation in water by free chlorine. They carefully controlled their free chlorine residuals with a modified form of the orthotolidine test to determine total chlorine and an orthotolidine-arsenite method for free chlorine. (Combined chlorine was then calculated as the difference between "total" and "free" chlorine readings.) They measured virus recoveries by using suckling mice and the LD₅₀ quantitation procedure. Their results indicated that inactivation times for the virus increased with increasing pH (6.9 to 9.0), decreasing temperatures (27°C-29°C to 3°C-6°C), and decreasing total chlorine concentration. They estimated that approximately 7 to 46 times as much free chlorine was required to obtain comparable inactivation of coxsackievirus A2 as was required for a suspension of E. coli cells (Butterfield et al., 1943). For instance, Butterfield et al. (1943) found that at pH 7.0 and at 2°C to 5°C, 99.9% of E. coli cells were inactivated in 5 min with 0.03 mg/liter of free chlorine. At approximately the same pH and temperature ranges, Clarke and Kabler (1954) observed 99.6% inactivation of coxsackievirus A2 in 5 min with 1.4 mg/liter of free chlorine, i.e., 46 times as much free chlorine as that required to inactivate E. coli cells. At a pH of 8.5 at 25°C, 99.9% of E. coli cells were inactivated in 3 min with 0.14 mg/liter of free chlorine (Butterfield et al., 1943), while at a pH of 9.0 at 27°C to 29°C, 99.6% of the virus was inactivated in 3 min by 1.0 mg/liter of free chlorine. Thus, Clarke and Kabler's work showed that 7 times as much free chlorine was required to inactivate the test coxsackievirus compared to the time necessary to kill the bacterium E. coli. In a subsequent study, Clarke et al. (1956) found that adenovirus type 3, E. coli, and Salmonella typhi were all inactivated or destroyed at approximately the same concentration of free chlorine.

In 1958, Weidenkopf reported his studies on the rate of inactivation of poliovirus 1 as a function of free available chlorine (HOCl and OCl^-) and pH at 0°C. His results showed that 99% inactivation of poliovirus 1 was obtained with 0.10 mg/liter free chlorine in 10 min at pH 6.0 and of 0°C. At pH 6, most of the free chlorine should have been present as hypochlorous acid. Increasing the pH to 7.0 increased the time required for the same degree of inactivation by approximately 50%. At that pH, the free chlorine should have been a mixture containing predominantly hypochlorous acid and significant levels of hypochlorite ion. Both Weidenkopf (1958) and Clarke et al. (1956) indicated that an increase in pH (from 7.0 to 8.5, Weidenkopf; from 8.8 to 9.0, Clarke et al.) increased the inactivation time about sixfold. At these pH's, the free chlorine should have been present as mixtures of both hypochlorous acid and hypochlorite ion, but predominantly as the ion.

After comparing these studies, Clarke et al. (1964) concluded that at 0°C to 6°C poliovirus 1 and coxsackievirus A2 were considerably more resistant to hypochlorous acid than was E. coli, while adenovirus type 3 was more sensitive (Figure II-5). For 99% destruction of E. coli, a 99-s contact time was required when the system was dosed with 0.1 mg/liter free chlorine as hypochlorous acid. The same percentage of the adenovirus was inactivated in approximately one-third of that time by the same concentration of hypochlorous acid. Under the same conditions, 8.5 min was required to inactivate 99% of the poliovirus, i.e., approximately 5 times the contact time required for E. coli. Coxsackievirus required a contact time for 99% inactivation in excess of 40 min, more than 24 times that required for E. coli.

Figure II-5

Concentration-time relationship for 99% destruction of E. coli and several viruses by hypochlorous acid at 0°-6°C. From Clarke et al., 1964.

Clarke and Kabler (1954) and Clarke et al. (1956) reported that the time required for inactivation of coxsackievirus A2 and adenovirus increased with increasing pH, decreasing total chlorine concentrations, and decreasing temperatures. Clarke and Chang (1959) concluded from data in the literature that a 10°C increase in temperature increased the rate of virus inactivation by a factor of 2 to 3.

Kelly and Sanderson (1958) reported that each of six enteric viruses possessed a different sensitivity to chlorine. Their results suggested that the inactivation of enteric viruses in water at pH 7.0 and 25°C required a minimum free residual chlorine concentration of 0.3 mg/liter with a contact time of at least 30 min. With combined chlorine in water, a concentration of at least 9.0 mg/liter was necessary for a 99.7% inactivation of poliovirus at 25°C and a pH of 7.0. Poliovirus 1 (strain MK 500) was the most resistant strain tested and coxsackievirus B5 the most sensitive. Poliovirus 1 (Mahoney strain), poliovirus 2, coxsackievirus B1, and poliovirus 3 were intermediate in resistance. The virucidal efficiency of hypochlorous acid was more than 50 times greater than that of the chloramines (Kelley and Sanderson, 1958, 1960).

Liu et al. (1971) studied the manner in which 20 strains of human enteric viruses responded to free chlorine. They used Potomac River water that had been partially treated by coagulation with alum and filtration through sand. Chlorine was added to the water at one dosage, 0.5 mg/liter. The final pH was 7.8. They stored the sample at 2°C. There was a wide range of resistance to chlorine by the viruses. The most sensitive virus was reovirus type 1, which required 2.7 min for inactiving 4 logs (99.99%) of the virus with 0.5 mg/liter of free chlorine. The most resistant, as judged by extrapolating the experimental data, was poliovirus 2, which required 40 min for the same degree of inactivation. Using actual experimental data, the most resistant virus was echovirus 12, which required a contact time of greater than 60 min for 99.99% inactivation. Liu et al. (1971) concluded from their extrapolated values that the reoviruses were the least resistant to chlorine treatment, that both adenoviruses and echoviruses were less resistant, and that the polioviruses and coxsackieviruses were the most resistant. However, assuming a 20-min contact time, most of the viruses tested at pH 7.8 and 2°C would have been 99.99% inactivated with a free chlorine residual of 0.5 mg/liter.

Using six of the same virus strains studied by Liu et al. (1971), Engelbrecht et al. (1978) investigated the effect of pH on the kinetics of chlorine inactivation at 5.0 ± 0.2°C. The suspending medium was buffered, chlorine demand-free, distilled-deionized water. Each virus stock was also prepared so as to be chlorine demand-free. Table II-1 summarizes the results, giving the chlorine levels used and the time required for two logs (99%) inactivation of the viruses at pH 6.0 and 7.8 in phosphate buffer and at pH 10.0 in borate buffer. Because of the use of two different buffer systems, i.e., at pH 6.0 and 10.0, virus inactivation was determined at pH 7.8 with each of the two buffer systems. The kinetics of inactivation of poliovirus 1 at pH 7.8, using both the phosphate and borate buffer, were the same. The results shown in Table II-1 indicate that there is a significant difference in the time required for two logs inactivation for the various viruses at pH 6.0 and 10.0. In every case, the rate of inactivation at pH 10.0 was significantly less than at pH 6.0. The rank ordering in Table II-1 shows that there is also a wide range of sensitivity of related viruses to chlorine disinfection. For example, at pH 10.0, coxsackie B5 was 40 times more resistant than coxsackie A9. There are several cases in which the relative sensitivity to chlorine was altered (rank ordering) between pH 6.0 and 10.0, suggesting important effects of pH on the virion as well as on the chlorine species, i.e., hypochlorous acid versus. hypochlorite ion. This observation can be seen more clearly in Table II-2 in which the time required for two logs of inactivation of the various viruses and the ratio of inactivation times at pH 6.0 and 10.0 are compared. Even at pH 7.8, differences in relative sensitivity appear when ranked and compared to results at pH 6.0 or 10.0 (Table II-1).

TABLE II-1

Time Required for 99% Inactivation by Free Residual Chlorine at 5.0°C ± 0.2°C.

TABLE II-2

Comparison of Virus Inactivation by Free Residual Chlorine at pH 6.0 and 10.0, at 5.0°C ± 0.2°C.

Combined Chlorine (NH₂Cl and NHCl₂)

Viral inactivation rates with chloramines have been found to be much slower than with free chlorine. For example, Kelly and Sanderson (1958) studied the effects of chlorine on several enteric viruses. They reported that at pH 7 at 25°C-28°C, 0.2-0.3 mg/liter free chlorine inactivated 99.9% of all test viruses in 8 min. At the same temperature and pH, combined chlorine at 0.7 mg/liter and at least 4 hr of contact time were needed to achieve 99.7% inactivation of the test viruses.

Although most viral inactivation studies with chloramines have not differentiated between mono- and dichloramine (Kelly and Sanderson, 1958; Lothrop and Sproul, 1969), Kelly and Sanderson (1958, 1960) noted that viral inactivation by chloramines proceeds more rapidly at pH 6-7 than at pH 8-10. This tendency indicates that dichloramine may be more virucidal than monochloramine since its proportion increases with increased hydrogen ion concentration.

In 1971, Chang proposed that 5.0 mg/liter dichloramine or 20 mg/liter monochloramine would be needed to inactivate enteroviruses by 99.99% in 10 min at 25°C. Subsequently, Siders et al. (1973) presented evidence of the first comparative studies on viral inactivation due solely to monochloramine. At pH 9 at 15°C, poliovirus 1 (Mahoney) was 10 times more resistant to monochloramine than was E. coli. Similarly, coxsackievirus A9 was approximately 4 times more resistant than E. coli. Siders' data can be compared to Chang's theory on the disinfectant capacity of monochloramine. Assume that poliovirus inactivation has a temperature coefficient for a 10°C rise (Q₁₀) of 3. Based on Chang's (1971) calculations that 20 mg/liter monochloramine at 25°C would be needed to achieve 99.999% reduction of enterovirus in 10 min, one would extrapolate that 3 × 20 or 60 mg/liter monochloramine would be needed at 15°C to achieve an equivalent reduction of poliovirus in 10 min. However, examination of the data of Siders et al. reveals that approximately 18 min were needed to achieve only 99% inactivation using 60 ppm monochloramine. Therefore, it appears that Chang slightly overestimated the virucidal capacity of monochloramine.

Mechanism of Action

One of the earliest references to the mechanism of inactivation of microorganisms by chlorine resulted from the work of Chang (1944a,b). While studying the inactivation of E. histolytica cysts by chlorine, he observed greater uptake of chlorine and less survival at low pH than at high pH. This observation was associated with the increased inactivation efficiency of the undissociated hypochlorous acid. Supportive evidence for the hypothesis that permeability of the uncharged chlorine species is important in determining sensitivity to chlorine has been provided by Skvortsova and Lebedeva (1973), Kaminski et al. (1976), and Dennis (1977). Chang (1944a) also noted that the inactivation of amoebic cysts was accompanied by microscopic damage to the cell nucleus, which was dependent on chlorine penetration.

The importance of penetration and/or damage to the permeability barrier of the cell membrane as a result of exposure to chlorine has been observed by several investigators. In 1945, Rahn suggested that the inactivation of bacteria by chlorine was due to multiple injuries to the cell surface. From their work with bacterial spores, Kulikovsky et al. (1975) implicated permeability damage as a mechanism of chlorine inactivation. Studies with Escherichia coli have shown that chlorine causes leakage of cytoplasmic material, first protein, then RNA and DNA, into the suspending menstruum. It also inhibits the biochemical activities that are associated with the bacterial cell membrane (Venkobachar, 1975; Venkobachar et al., 1977). Friberg (1957) observed that E. coli also loses nondialyzable phosphorus following exposure to chlorine. In a recent study, Haas (1978) demonstrated that chlorine caused certain bacteria and yeast to release organic matter or UV-absorbing material, presumably protein or nucleic acid or their precursors. This investigator also noted that chlorine affected the uptake and retention of potassium by these same microorganisms.

Green and Stumpf (1946) and Knox et al. (1948) indicated that destruction of bacteria by chlorine was caused by an inhibition of the mechanism of glucose oxidation. Specifically, they suggested that chlorine affected the aldolase enzyme of E. coli by oxidizing the sulfhydryl group that is associated with the enzyme. Venkobachar et al. (1977) recently reported that chlorine significantly inhibits both oxygen uptake and oxidative phosphorylation. The latter effect was attributed to inhibition of the respiratory enzyme rather than to a deficiency in phosphate uptake. However, it is unclear whether free or combined chlorine was used in these studies. Haas (1978) also observed chlorine to affect the respiration of bacteria as well as the rate of synthesis of protein and DNA. Others have also noted that chlorine affects the nucleic acids or physically damages DNA (Bocharov, 1970; Bocharov and Kulikovskii, 1971; Fetner, 1962; Rosenkranz, 1973; Shih and Lederberg, 1976a,b).

It appears that chlorine, having penetrated the cell wall, encounters the cell membrane and alters its permeability. Simultaneously or subsequently, the chlorine molecules may enter the cytoplasm and interfere with various enzymatic reactions. It should be noted that permeases and respiratory enzymes are associated with the cytoplasmic membrane of bacteria.

Chang (1971) supported the hypothesis that the rapid destruction of vegetative bacteria by chlorine was due to the extensive destruction of metabolic enzyme systems. He also addressed the subject of virus inactivation, commenting that viruses are generally more resistant to chlorine than bacteria. He associated this observation with the fact that viruses completely lack a metabolic enzyme system. He speculated that inactivation of viruses by chlorine probably result from the denaturation of the capsid protein. Furthermore, since protein denaturation is more difficult to achieve than destruction of enzymatic R—S—H bonds by oxidizing agents, it is understandable why greater levels of chlorine are required to inactivate viruses than bacteria. However, from their experimental work with the bacterial virus f2, Olivieri et al. (1975) concluded that chlorine caused initial lethal damage to the viral genome and that the capsid protein was affected after the virus was inactivated. Dennis (1977) reported that the incorporation of chlorine into the f2 bacterial virus is dependent on pH and that the higher rates of incorporation occur at lower pH values.

There is limited information in the literature on the mechanism of inactivation of microorganisms by chloramines. Nusbaum (1952) proposed that since low levels of inorganic chloramines were effective in inactivating bacteria, the mechanism of action must be essentially the same as that of hypochlorous acid on enzymes. Ingols et al. (1953) showed that monochloramine was not able to immediately and irreversibly oxidize sulfhydryl groups. Such oxidation would have resulted in the rapid inactivation of the bacteria. They hypothesized that since monochloramine required higher concentrations and longer contact times to destroy bacteria completely and could not readily and irreversibly oxidize the sulfhydryl groups of the glucose oxidation enzymes, its ability to inactivate microorganisms should be attributed to changes in enzymes that may not be involved in the inactivation of the organism by hypochlorous acid. Thus, while the sulfhydryl group may be the most vulnerable to a strong oxidant like hypochlorous acid, changes in other groups produced by the weaker oxidant, monochloramine, may lead also to microbial inactivation. More recent information indicates that the destructive effects of chloramine might be associated with the effects of chloramine on nucleic acids or DNA of cells (Fetner, 1962; Shih and Lederberg, 1976a,b).

Nusbaum (1952) suggested that the disinfective activity of dichloramine occurs by a mechanism similar to monochloramine, but there do not appear to be any data to support this contention.

Considering the mechanism of destruction or inactivation of microorganisms by chlorine and associated compounds, it is interesting to note that Fair et al. (1948) speculated that there might be three or four "targets" or points of attack and that perhaps all must be affected before there is death. This "multiple hit" concept supported the observation that monochloramine must alter groups other than the sulfhydryl group to be effective in the destruction of microorganisms.

Thus, the action of chlorine on microbes such as bacteria and amebic cysts may involve some or all of the steps in the following sequence: penetration of the disinfectant through the cell wall followed by attack on the cell membrane (the site of cellular respiration in bacteria) and disruption of permeability of the cell membrane, which leads to a loss of cell constituents, thereby disrupting metabolic functions within the cell including those involving nucleic acids. Changes in viability may result from this process. Experimental studies on virus (Olivieri et al., 1975) demonstrated that chlorine caused initial damage on the viral nucleic acid while leaving the capsid protein unaffected until after the virus was inactivated.

Summary

Chlorine is the most widely used water supply disinfectant in the United States. Depending upon the predominant species of chlorine, hypochlorous acid, and/or hypochlorite ion, disinfection with chlorine can achieve greater than 99.9% destruction of bacteria. For example, a chlorine residual of 0.2 to 1.0 mg/liter and a contact time of 15 to 30 min will inactivate 99.9% of E. coli (Walton, 1969). According to Walton, a properly designed, constructed, and operated water treatment plant, consisting of chemical coagulation, sedimentation, filtration, and disinfection, can remove or destroy more than 99.999% of the coliform bacteria that are present. Although most investigations on the removal or destruction of bacteria have used E. coli, there is evidence that the bacterial pathogens, e.g., Salmonella typhi, respond somewhat similarly to E. coli.

Laboratory studies have demonstrated that that there is limited virus inactivation after the added chlorine has reacted with any ammonia that is in the water. Most inactivation probably occurs in the first few seconds before the chlorine has completed its reaction with ammonia (Olivieri et al., 1971).

Table II-3 displays c · t values for E. coli and poliovirus inactivation for the various species of free and combined chlorine.

Research Recommendations

Recent reports of enhanced chlorine resistance of certain viral and bacterial strains should be investigated and the mechanism of increased resistance elucidated, if the reports are corroborated.

Other recommendations, applicable to other agents as well as to chlorine, are included after the evaluations of the other methods of disinfection.

Chemistry of Ozone in Water

Ozone has the molecular formula O₃, a molecular weight of 48 g/mol, and a density, as a gas, of 2.154 g/liter at 0°C and 1 atm. It is approximately 13 times more soluble in water than is oxygen. At saturation in water at 20°C, a 2% weight mixture of ozone and oxygen contains about 11 mg of ozone and 40 mg of oxygen per liter.

Ozone has a half-life in pure distilled water of approximately 40 min at pH 7.6, but this decreases to 10 min at pH 8.5 (Stumm, 1958) at 14.6°C. Rising temperatures increase the rate of decomposition. Because its half-life is so short, ozone must be generated on the site where it is to be used.

Ozone is a powerful oxidant that reacts rapidly with most organic and many inorganic compounds. It does not convert chloride to chlorine under test conditions (U.S. Environmental Protection Agency, 1976), nor does it react extensively with ammonia (NH₃) (Singer and Zilli, 1975). However, bromide and iodide are oxidized to bromine and iodine. Singer and Zilli (1975) reported that oxidation of ammonia was pH-dependent. At pH 7.0, 9% of a 29 mg/liter ammonia nitrogen (NH₃—N) solution was oxidized in 30 min, but at pH 9.0 70% of a 24.4 mg/liter solution was oxidized in the same time. During disinfection, only minor amounts of ammonia are oxidized when ozone is used. Ozone's limited reaction with ammonia is desirable, but its fast reaction rate with most organic and many inorganic compounds further shortens its persistence in water.

Production and Application of Ozone

Ozone is produced on site from a stream of clean dry air or oxygen by passing an electrical discharge between electrodes that are separated by a dielectric. Approximately twice the percent of ozone by weight is obtained if oxygen, rather than air, is used as the feed stream. Power requirements are about 13 to 22 kWh/kg of ozone that is generated from air and approximately half that when oxygen is used (Rosen, 1972). Compressors and dryers may increase these requirements by 20% to 50%. Other factors affecting efficiency are the rate of gas flow, applied voltage, and the temperature of the gas. The heat that is produced during the process must be removed by cooling with either air or water.

The ozone gas stream must be fed into the water to effect the transfer of ozone. The usual methods are to inject the ozone gas stream through an orifice at the bottom of a co- or countercurrent contact chamber or to aspirate the gas into a contact chamber where it is mixed with the water mechanically. Successful design and operation of the contactor system is necessary to minimize costs of the operation.

Commercial equipment is available in a wide range of capacities—from a few grams of ozone per day to more than 40 kg/day. Larger capacities are obtained by adding additional units. Successful delivery of ozone to the water to be treated requires a dependable power supply and reasonably maintenance-free ozonization equipment.

Ozone has been used in a great number of water treatment plants throughout the world. However, in small institutions and private residences, its use appears limited, because it requires dependable power supplies and, usually, a second disinfectant to furnish a disinfecting residual in the system. The maintenance and repairs that are required for the specialized ozone generation equipment provide further barriers against the use of ozone by small institutions.

Analytical Methods

The disinfection process is usually controlled in one of two ways: by the dosage of a specified amount of ozone or by the maintenance of a specified minimum residual for a given time. Residual measurements in both the gas stream and water are sometimes required. Standard Methods (1976) contains descriptions of the measurement of ozone in water by the iodometric, orthotolidine-manganese sulfate, and orthotolidine-arsenite methods. Of these methods the iodometric method, which is subject to the fewest interferences, is the method of choice. Determinations must be made immediately since ozone decomposes rapidly.

In all three methods, the oxidant compounds that result from the reaction of ozone with contaminants in water may react with the test reagents, thereby indicating a higher concentration of ozone than is actually present. This is particularly true in the presence of organic matter, which results in the formation of organic peroxides. In the iodometric method this interference and others are minimized by stripping the ozone from the sample with nitrogen or air and absorbing it from this gas stream in an iodide solution. These and other methods are described in Standard Methods.

Schecter (1973) developed a UV spectrophotometric method to measure the triiodide that is formed by the oxidation of iodide by ozone. She reported a better sensitivity at low ozone concentrations (0.01 to 0.3 mg/liter) than achieved with the normal titration method. The effects of interferences on the direct measurement of ozone without sparging the ozone to a separate iodide solution were not indicated. These effects are noted in Standard Methods.

Analytical determination of ozone in water in the presence of other oxidants is poor. Considerable work in this area is needed.

Residual

As discussed above, ozone is unstable in water with a half-life of approximately 40 min at pH 7.6 and 14.6°C. Many regard the half-life in water supplies at higher ambient temperatures to be 10 to 20 min.

In a recent review of the literature, Peleg (1976) concluded that the possible species to be found in an ozone solution were ozone, hydroxyl radicals (·OH), hydroperoxyl radicals (HO₂·), oxide radicals (O·), ozonide radicals (O₃·), and, possibly, free oxygen atoms (·O·). Hydrogen peroxide (H₂O₂) may also be present by dimerization of the hydroxyl radicals. There have been no studies on the disinfecting activities of these individual species except for those on hydrogen peroxide, which is a poor biocide (when compared to chlorine). Peleg concluded that evidence indicates that the dissociation species are better disinfectants than ozone.

Biocidal Activity

Efficacy against Viruses

Katzenelson et al. (1974) investigated the inactivation of poliovirus 1 at 5°C and pH 7.2. A 99% inactivation of the virus occurred in less than 8 s with 0.3 mg/liter of initial residual ozone. Their experimental methods were the same as those described above for bacteria. They also observed that inactivation resulted from two distinct stages (rates) of action. The first stage of inactivation, less than 8 s long, produced a virus inactivation of from 99% to 99.5%, depending upon the ozone residual. The second stage, which lasted from 1 to 5 min, still left some viruses infective. Additional work showed that the slower second stage inactivation apparently involved the inactivation of viruses that were clumped together. The single virus particles were inactivated during the first stage. After ultrasonic treatment, 99.5% of the virus was inactivated within 8 s and more than 99.99% within 3 min at an initial ozone residual of 0.1 mg/liter.

Coin et al. (1964) investigated the inactivation of poliovirus 1 (Mahoney) in a batch system in which a measured amount of ozone was introduced into a virus suspension in distilled and filtered river water. In distilled water, nearly 1 log of the virus was inactivated in 4 min at a 4-min ozone residual of approximately 0.23 mg/liter. In river water, the inactivation was in excess of 99.99% in 4 min when the ozone concentration, also measured at 4 min, exceeded 0.3 mg/liter. The ozone was measured by the iodide titration without stripping of the ozone. Rather large ozone demands existed in this system. An initial 5 mg/liter residual ozone concentration in the distilled water decreased to about 0.6 to 0.8 mg/liter in 4 min. The same initial residual in river water decreased to between 0.2 to 0.6 mg/liter. Temperature and pH were not reported.

Keller et al. (1974) studied ozone inactivation of viruses in a water supply by using both batch tests and a pilot plant with a 38 liter/min (10 gal/min) flow rate. Inactivation of poliovirus 2 and coxsackievirus B3 in 5 min was greater than 99.9% in the batch tests when the ozone residual was between 0.8 and 1.7 mg/liter at the end of the 5-min contact period. Initial ozone residuals varied from 1.6 to 2.8 mg/liter. At the pilot plant, greater than 99.999% inactivation of coxsackievirus B3 could be achieved with an ozone dosage of 1.45 mg/liter, which provided an ozone residual of 0.28 mg/liter after 1 min in lake water. Temperature and pH conditions were not reported. Ozone was measured by the iodide method without prior stripping of the ozone.

Burleson et al. (1975) obtained inactivations of > 99.9% for vesicular stomatitis virus, > 99.99% for encephalomyocarditis virus, and 99.99% for GD VII virus in 15 s with an ozone residual of approximately 0.5 mg/liter of ozone. The experiment was conducted in the same manner as that described above for their studies with bacteria. Evison (1977) reported data for the inactivation of a number of viruses in buffered water at pH 7.0 and 25°C (Table II-4).

TABLE II-4

Ozone Required for Inactivation of Viruses in 10 Minutes at pH 7.0 and 25°C.

The reported ozone concentrations were evidently those measured initially and maintained by the addition of ozone during the experiment. Ozone was measured by a colorimetric version of Palin's DPD technique. The Evison data show that more ozone or a longer time are required for inactivation than do the data of other workers. This may have resulted from the virus purification used. Her viruses were purified by low-speed centrifugation and filtration through an 0.2-µm membrane filter. These cleanup procedures are neither as complete nor as thorough as those used by other investigators. The unremoved cell debris and organic matter offer protection to the virus. Either higher ozone residuals or longer contact times would be required to inactivate such preparations to the same extent as clean virus. Other data showed that the inactivation of coliphage 185 was relatively unaffected by pH's ranging from 6.0 to 8.0 at ozone concentrations exceeding 0.05 to 0.10 mg/liter. Evison (1977) also concluded that the rate of inactivation of the coliphage 185 by ozone was much less affected by temperature than was the inactivation by chlorine.

Farooq (1976) observed a 99% inactivation of 10⁵/ml poliovirus 1 in 30 s in distilled water at an initial ozone residual concentration between 0.23 and 0.26 mg/liter at 24°C and pH 7.0. In this study, ozone was measured by the Schecter (1973) method.

Sproul et al. (1978) reported total inactivation in 10 s of poliovirus 1 (Sabin), which had initial concentrations of 1.3 × 10² to 2.4 × 10³ plaque-forming units (PFU)/ml. The initial ozone residual of 0.012 to 0.085 mg/liter decreased by approximately one-third in 40 s. The experiments were conducted with the Sharpe dynamic reactor. Ozone was measured by the Schecter (1973) method.

Mechanism of Action

Investigations of the inactivation of bacteria by ozone have centered on the action of ozone on the cell membrane. Scott and Lesher (1963) concluded from their work with E. coli that the primary attack of ozone occurred on the double bonds of the fatty acids in the cell wall and membrane and that there was a consequent change in the cell permeability. Cell contents then leaked into the water. This was confirmed by Smith (1967). Prat et al. (1968), examining extracts from ozonized E. coli, reported that thymine was more sensitive to ozone attack than were cytosine and uracil.

Riesser et al. (1977) showed that ozone attacked the protein capsid of poliovirus 2 in such a manner that the virus was not taken up into susceptible cells. An electrophoretic study showed complete loss of viral proteins in a poliovirus 2 sample that had showed an inactivation of 7 logs in 20 min.

Summary and Conclusions

Inactivation with ozone at specified ozone residuals is relatively insensitive to pH's between 6.0 and 8.5. Moreover, ozone does not react with ammonia over this same range when short detention times are used. The data on temperature are not sufficiently firm to permit conclusions concerning its effect on disinfection. Ozone must be generated on site, and the process is relatively energy intensive. To make economic comparisons of ozone with other disinfectants, the cost of local power must be ascertained.

Available kinetic data on ozone inactivation are presented in Table II-5. The c · t product for 99% inactivation of E. coli appears to be approximately 0.006 at near-neutral pH values and at temperatures of ≤10°C. There was less consistency for the c · t products of poliovirus 1. These values varied from < 0.005 to 0.42 at pH 7 and at temperatures from 5°C to 25°C.

TABLE II-5

Concentration of Ozone and Contact Time Necessary for 99% Inactivation of Escherichia coli and Polio 1 Virus.

The c · t products vary over a broad range. These variations illustrate the difficulty of doing quantitative experimentation with ozone and microorganisms in water. Among other reasons, these difficulties are caused by undetected ozone demand in the water, poor analytical techniques for residual ozone, and nonuniformity of microorganisms from one laboratory to another. As an example of the latter, different strains of poliovirus with different inactivation rates are used, but the inactivation data are frequently not reported as strain-specific. Furthermore, viruses frequently exist in an undetected clumped state rather than in the presumed single discrete particle state. Higher c · t products are required for the clumped viruses than for the unclumped ones.

Because of ozone's relatively short half-life in water, another disinfectant must be added to maintain a disinfection capability in the distribution system. The most effective disinfectant, its optimum concentration, and method of addition must be determined. The disinfection process with ozone will probably be controlled by specifying the ozone residual at the beginning and end of a given contact time.

Research Recommendations

Future research studies with ozone should be conducted to:

• provide data on the inactivation of enteric pathogens;
• provide more data on the inactivation of protozoan cysts, especially those of Giardia lamblia;
• provide analytical methods that are specific for ozone;
• provide additional definitive data on bacteria and viruses, and eliminate discrepancies; and
• provide operating data from full-scale drinking water plants to demonstrate reliability of operation, operating costs, ozone dosages, residuals, contact times, and disinfection results.

The Chemistry of Chlorine Dioxide in Water

Chlorine dioxide reacts with a wide variety of organic and inorganic chemicals under conditions that are usually found in water treatment systems (Stevens et al., 1978). However, two important reactions do not occur. Chlorine dioxide per se does not react to cause the formation of trihalomethanes (THM's) (Miltner, 1976). However, THM's will be formed if the chlorine dioxide is contaminated with chlorine. Such a situation may occur when chlorine is used in the preparation of chlorine dioxide.

Chlorine dioxide does not react with ammonia, but will react with other amines (Rosenblatt, 1978). The amine structure determines reactivity. Tertiary amines are more reactive with chlorine dioxide than secondary amines, which, in turn, are more reactive than primary amines.

Production and Application

Chlorine dioxide condenses to form an unstable liquid. Both the gas and liquid are sensitive to temperature, pressure, and light. At concentrations above 10% in air, chlorine dioxide may be explosive, and at 4% in air, it can be detonated by sparks (Gall, 1978; Sussman, 1978). As a result, the preparation and distribution of chlorine dioxide in bulk have not been deemed practical. It has been generated and used on site.

Sodium chlorate (NaClO₃) or sodium chlorite (NaClO₂) may be used to generate chlorine dioxide. The method of production will depend upon the amount of chlorine dioxide that is required. The reduction of sodium chlorate is the more efficient process and is generally used when large volumes and high concentrations of chlorine dioxide are needed. Commercial processes that are used in North America for large-scale production of chlorine dioxide are based on the three reactions listed below. To reduce the sodium chlorate, each process uses a different agent: sulfur dioxide (SO₂), methanol (CH₃OH), and the chloride ion (Cl^-).

All of these processes are used in the pulp and paper industry. They can also be used to prepare chlorine dioxide for the large waterworks that might require several metric tons per day. Small quantities of chlorine are formed during the side reactions and intermediate reactions in these processes. A more detailed review of the chemistry that is involved in the production of chlorine dioxide from chlorate is given by Gall (1978) and Gordon et al. (1972).

(6)

(7)

(8)

Chlorine dioxide can be prepared from chlorine and sodium chlorite through the following reactions:

(9)

(10)

The theoretical weight ratio of sodium chlorite to chlorine is 1.00:0.39 (Dowling, 1974). With available sodium chlorite (80%), the weight ratio is 1:0.30. In practice, Gall (1978) recommended a chlorite to chlorine ratio of 1:1. The excess chlorine lowers the pH, thereby increasing the reaction rate and optimizing the yield of chlorine dioxide. Dowling (1974) reported that the maximum theoretical yield of chlorine dioxide was produced when the ratio was normally maintained at a minimum of 1.0:0.5.

Alternatively, chlorine dioxide may be prepared from sodium hypochlorite (NaOCl) and sodium chlorite. The sodium hypochlorite is acidified to yield hypochlorous acid (HOCl), and the chlorine dioxide is generated according to Reaction 10. Each of the methods produces a solution containing both chlorine and chlorine dioxide.

Chlorine dioxide may also be prepared by the addition of a strong acid, such as sulfuric acid (H₂SO₄) or hydrochloric acid, to sodium chlorite as shown in the following reactions:

(11)

(12)

Although some investigators have claimed that this method produces chlorine-free chlorine dioxide, Feuss (1964) and Schilling (1956) reported that chlorine is also formed. Dowling (1974) indicated that chlorine was formed even when sulfuric acid was used.

Analytical Methods

Chlorine dioxide is one of the few stable nonmetallic inorganic free radicals (Rosenblatt, 1978). It does not contain available chlorine in the form of hypochlorous acid or hypochlorite ion (OCl^-). However, concentrations of chlorine dioxide are often reported in terms of available chlorine. The chlorine atom in chlorine dioxide has a valance of + 4. A reduction to chloride results in a gain of five electrons. In terms of available chlorine, chlorine dioxide has 263% or more than 2.5 times the oxidizing capacity of chlorine.

(13)

The weight ratio of chlorine dioxide to available chlorine is 67.45 to 35.45 or 1.9. However, in water treatment practices this increased oxidizing capacity is rarely realized. The reduction of chlorine dioxide depends heavily on pH and the nature of the reducing agent. At neutral or alkaline pH, chlorine dioxide is reduced to chlorite, a net gain of one electron. Thus, only one-fifth or 20% of its oxidizing capacity is utilized. At low pH, the chlorite (ClO₂^-) is reduced to chloride (Cl^-) releasing the remaining four available electrons.

Chlorine dioxide may be determined iodometrically (Standard Methods, 1976), amperometrically (Haller and Listek, 1948; Standard Methods, 1976), spectrophotometrically (Gordon et al., 1972), and colorimetrically (Aston, 1950; Hodgen and Ingols, 1954; Masschelein, 1966; Palin, 1957, 1960, 1967, 1970, 1974, 1975; Post and Moore, 1959; Standard Methods, 1976).

Several studies contain comparisons of various analytical methods and procedures for the measurement of chlorine dioxide. Adams et al. (1966) compared the H-acid, tyrosine, amperometric, and diethyl-p-phenylenediamine (DPD) procedures for free chlorine, chlorine dioxide, and chlorite. They reported DPD to be the most reliable, as did Dowling (1974), Miltner (1976), and the U.S. Public Health Service (Miltner, 1976). Myhrstad and Samdal (1969) noted that the DPD method (Palin, 1957, 1960) yielded consistently higher residual measurements for chlorine dioxide than those that are produced with other analytical methods. After analysis with acid chrome violet K (Masschelein, 1966), chlorine dioxide was not observed in the water of the distribution system; however, chlorite was found. The residuals that were previously interpreted as chlorine dioxide were apparently due to chlorite.

Recently, more sophisticated procedures were suggested for the determination of chlorine dioxide. Moffa et al. (1975) reported the use of electron spin resonance, and Issacsson and Wettermark (1976) described a chemiluminescent method for active chlorine compounds. Stevenson et al. (1978) reviewed electrochemical methods and presented preliminary results for a membrane amperometric probe. Under development is a sensor that shows a linear response region from about 0.5 to 10 mg/liter. The response to hypochlorous acid and chloramines was low, and the sensor does not measure chlorite or other ionized species.

No one procedure appears to possess the necessary sensitivity, selectivity, and simplicity to permit reliable determinations in the treatment of water. Each of the titration methods are prone to error because of volatilization. They are time-consuming and particularly complex when differentiation of chlorine and oxychloro species are necessary. The colorimetric procedures require strict control of pH, temperature, and reaction times and will be affected by turbidity. In addition, the selectivity of the indicators for chlorine dioxide is questionable. The direct spectrophotometric determination of chlorine dioxide at 360 nm is selective and rapid but is not sufficiently sensitive for use in water. Limited experience with the more recent procedures (chemiluminescence and membrane amperometric probe) does not permit an evaluation.

In practice, the principal distinction that must be made is that between the active biocidal species (hypochlorous acid, the hypochloride ion, and chlorine dioxide), the moderately biocidal species (monochloramine [NH₂Cl], dichloramine [NHCl₂], and nitrogen trichloride [NCl₃]), and the relatively nonbiocidal species (chlorite and chlorate [ClO₃^-] ions). This is imperative when the primary purpose for the addition of chlorine dioxide is the inactivation of microorganisms. Furthermore, when the formation of THM's is to be considered, the distinction between free chlorine and chlorine dioxide becomes important.

Biocidal Activity

Efficacy against Bacteria

Experimental data on the efficacy of chlorine dioxide as a disinfectant became available in the early 1940's. McCarthy (1944, 1945) reported that chlorine dioxide was an effective bactericide in water with a low organic content. When the levels of organic material in the water were high, chlorine dioxide was less effective.

Ridenour and Ingols (1947) reported that chlorine dioxide was at least as effective as chlorine against Escherichia coli after 30 min at similar residual concentrations. Both chlorine and chlorine dioxide residues were determined by the orthotolidine-arsenite (OTA) method. The bactericidal activity of chlorine dioxide was not affected by pH values from 6.0 to 10.0. Ridenour and Armbruster (1949) extended their observation to other enteric bacteria. The common waterborne pathogens were similarly inactivated with chlorine dioxide. They also reported that the efficiency of chlorine dioxide decreased as the temperature decreased from 20°C to 5°C.

Ridenour et al. (1949) found chlorine dioxide to be more effective than chlorine (based on OTA residuals) against bacterial endospores. They indicated that less weight of chlorine dioxide than chlorine is required to inactivate the spores of Bacillus mesentericus in either demand-free water or in waters containing ammonia. In the waters containing ammonia, chlorine had to be applied beyond breakpoint before efficient sporicidal activity was observed.

The work of Ridenour and colleagues is not discussed in depth because the small amounts of free chlorine that are produced during the generation of chlorine dioxide are not distinguished from the chlorine dioxide by the OTA method that they used for both stock solutions and residual determination. In their 1947 paper (Ridenour and Ingols, 1947), the survival measurements were ± and not quantitative. In their 1949 paper (Ridenour and Armbruster, 1949), the survival measurement was quantitative, but only one contact time was used, 5.0 min. Since chlorine dioxide is a rapidly acting disinfectant, c · t products may be misleading using this contact time.

Russian investigators (Bedulevich et al., 1953; Trakhtman, 1946) found chlorine dioxide to be a more effective or at least as effective a bactericide as chlorine. Additional data were reported for Bacillus anthracis. They observed that the efficiency of chlorine dioxide decreased as the pH of the system containing the B. anthracis increased.

Early studies on disinfectant activity are difficult to interpret because the methods of preparing chlorine dioxide invariably included the addition or production of chlorine. Analytical procedures were not sufficiently advanced to differentiate between chlorine dioxide and other oxychloro species. Thus, the quantitative analyses of stock solutions and reports of dose and residual chlorine dioxide may be in error. This suggests that the initial and residual concentrations of chlorine dioxide were probably lower than reported values and that the comparative bactericidal efficiency would suffer accordingly. In addition, the older investigations did not take into account the volatility of chlorine dioxide. Depending upon concentration and length of exposure, losses of 7% to 30% can occur within an hour.

Many of the difficulties that were encountered during the early studies were overcome in a series of studies reported by Benarde and co-workers during the mid-1960's. Their work on disinfection was based heavily on the improved methods of preparing and analyzing chlorine dioxide, which were reported by Granstrom and Lee (1958). They prepared chlorine dioxide by oxidizing sodium chlorite with persulfate (S₂O₈^2-) under acid conditions. The resulting chlorine dioxide was swept to a collection vessel by high purity nitrogen gas. Chlorine dioxide was measured spectrophotometrically at 357 nm.

Bernarde et al. (1965) compared the bactericidal effectiveness of chlorine with that of chlorine dioxide at pH 6.5 and 8.5 in a demand-free buffered system. At pH 6.5, both chlorine and chlorine dioxide inactivated a freshly isolated strain of E. coli in less than 60 s. Chlorine was slightly more effective at the lower dosages at the lower pH. At pH 8.5, chlorine dioxide was dramatically more effective than chlorine. Greater than 99% inactivation of E. coli was observed in 15 s with 0.25 mg/liter chlorine dioxide, while, under the same conditions, chlorine required almost 5 min for similar inactivation. Chlorine dioxide was significantly more efficient than chlorine in the presence of high levels of organic and nitrogenous material. Bernarde et al. (1967a) also reported that temperature affects the rate of inactivation of bacteria with chlorine dioxide. A decrease in disinfectant activity was observed as temperature decreased from 30°C to 5°C. For 99% inactivation of E. coli with 0.25 mg/liter of chlorine dioxide, 190, 74, 41, and 16 s were required at 5°C, 10°C, 20°C, 30°C, respectively. More recent work by Cronier (1977) in a clean system also demonstrated the excellent bactericidal activity of chlorine dioxide. Results of both studies are shown in Table II-6.

TABLE II-6

Concentrations of Chlorine Dioxide and Contact Times Necessary for 99% Inactivation of Escherichia coli .

Efficacy against Viruses

In the mid-1940's, there were also investigations on the virucidal activity of chlorine dioxide. Ridenour and Ingols (1946) reported that chlorine dioxide was as effective as chlorine against a mouse-adapted strain of poliovirus (Lansing). Again, their comparison was based upon OTA-determined residual levels of each disinfectant. Hettche and Ehlbeck (1953) found chlorine dioxide to be more effective against poliovirus than either chlorine or ozone. In addition to the difficulties that are associated with the preparation and measurement of chlorine dioxide and chlorine, the early virus studies were also saddled with difficult and time-consuming virus test systems.

The first definitive work on the virucidal activity of chlorine dioxide was done in Göteborg, Sweden. Warriner (1967) showed that the rate of inactivation of poliovirus 3 increased with increasing pH at pH values of 5.6 to 8.5 (see Table II-7). Similar to the action on bacteria, the viral inactivation occurred rapidly. When chlorine and chlorine dioxide were combined, the inactivation was synergistic. The inactivation with the two chlorine species together was more efficient at lower pH, but the presence of chlorine dioxide enhanced inactivation by chlorine at the higher pH.

TABLE II-7

Concentrations of Chlorine Dioxide and Contact Times Necessary for 99% Inactivation of Viruses.

Cronier (1977) compared the inactivation of E. coli (ATCC 11229), poliovirus 1, and coxsackievirus A9. All of the microorganisms that were tested were sensitive to low concentrations (< 1 mg/liter) of chlorine dioxide. Poliovirus 1 and coxsackievirus A9 were more resistant than E. coli. Similar to its bactericidal activity, chlorine dioxide was more effective as a virucide at higher pH. Cronier also reported that on a weight basis, it was similar to hypochlorous acid and better than hypochlorite ion, monochloramine, and dichloramine.

The effect of bentonite (AlO₃ · 4SiO₂ · H₂O) (added to the test as a model for turbidity) on disinfection with chlorine dioxide was studied by Scarpino et al. (1977). At turbidity levels below 2.29 nephelometric turbidity units (NTU) at 25°C, no protection was afforded to poliovirus 1 by the bentonite. However, at 3.22 and 14.1 NTU's, poliovirus inactivation was noticeably decreased. The bentonite at these levels appeared to offer protection to the virus.

Mechanism of Action

There is little information concerning the mode of action by which chlorine dioxide inactivates bacteria and viruses. Ingols and Ridenour (1948) suggested that the bactericidal effectiveness of chlorine dioxide is due to its adsorption on the cell wall with subsequent penetration into the cell where it reacts with enzymes containing sulfhydryl groups.

Benarde et al. (1967b) demonstrated that chlorine dioxide abruptly inhibited protein synthesis. The incorporation of ¹⁴C-labeled amino acids into protein by whole cells stopped within a few seconds after the addition of chlorine dioxide. Subsequently, Olivieri (1968) reported a dose-response in the inhibition of protein synthesis in bacteria that had been treated with chlorine dioxide. The site of action was localized in the soluble portion (enzymes) of the cell extracts of treated cells without affecting the integrity of the ribosomes' function in protein synthesis.

Conclusion

Chlorine dioxide is an effective bactericide and virucide under the pH, temperature, and turbidity that are expected in the treatment of potable water. It should be noted that the U.S. Environmental Protection Agency (1978) has set an interim Maximum Contaminant Level of 1 mg/liter on chlorine dioxide (U.S. Environmental Protection Agency, 1978) because of the unresolved questions on its health effects (Symons et al., 1977).

Research Recommendations

A simple, selective, and sensitive test for chlorine dioxide should be developed to monitor residual concentration.

Chlorine dioxide is currently used at several plants. A review of plant records and field studies on the stability and effectiveness of chlorine dioxide in the distribution system should be undertaken.

Studies should be directed toward evaluating the inactivation of protozoan cysts.

More information should be obtained on the mode of action by which chlorine dioxide inactivates bacteria, viruses, and cysts.

Chemistry of Iodine in Water

Iodine is the only common halogen that is a solid at room temperature, and it possesses the highest atomic weight (126.91). Of the four common halogens, it is the least soluble in water, has the lowest standard oxidation potential for reduction to halide, and reacts least readily with organic compounds.

(14)

Diatomic iodine (I₂) reacts with water to form hypoiodous acid (HOI) and iodide ion (I^-). The effect of pH on this reaction is shown in Table II-8.

TABLE II-8

Hydrolysis of Iodine at 25°C Showing the Percentage of Iodine Species at Different pH's.

The distribution of chemical species of iodine given in Table II-8 was taken from the calculations that were made by Chang (1958) from the equilibrium expression:

(15)

The value of K_h, the hydrolysis constant, is given by Wyss and Strandskov (1945) as 3 × 10^-12 at 25°C. With iodine residuals at 0.5 mg/liter, which are expected in water systems, and a pH of 5, approximately 99% of the total iodine residual is present as iodine and only 1% as hypoiodous acid. At pH 7, the two forms are present in almost equal concentrations. At pH 8, only 12% is present as elemental iodine and 88% as hypoiodous acid, which can be converted to hypoiodite ion (OI^-).

(16)

(17)

(18)

(19)

The dissociation constant, K_a, of hypoiodous acid (at 20°C) is 4.5 × 10^-13. Consequently, the dissociation of hypoiodous acid, which occurs at high pH's, is not important for practical purposes. However, as confirmed by the field studies in Florida, hypoiodous acid can form iodate ion by autooxidation at pH values above 9.

(20)

The iodate ion possesses no disinfecting ability (Marks and Strandskov, 1950).

Production and Application

Iodine may be added to a municipal water supply by several procedures. One method is to employ nonhazardous solvents and solubilizing agents such as ethyl alcohol (C₂H₅OH) and potassium iodide (KI) to overcome the low concentration of aqueous iodine stock for solution feeders. Another method produces the required concentration of iodine by passing water through a bed of crystalline iodine (saturator). This has passing water through a bed of crystalline iodine (saturator). This has been used in many small, semipublic and private home water systems. Since the maximum concentration is limited by solubility to 200-300 mg/liter at the ambient temperatures that are expected for drinking water, some physical complications would accompany the introduction of this method into the large waterworks system in view of the large saturation beds required. The iodinated anion exchange resin bed and the vaporization technique are not sufficiently developed to be considered for use in public water supplies.

In certain circumstances, potassium iodide might be combined with an oxidation reaction to release iodine. The chemistry is simple, and the persistence of the iodine that is generated may be much better than chlorine. This method has been used in swimming pools and for dechlorination purposes where a chlorine residual may be exchanged for an iodine residual, or an iodine residual may be provided where ozone is the primary disinfectant.

Analytical Methods

Both amperometric titration and leuco crystal violet (LCV) colorimetric methods give acceptable results when used to measure free iodine in drinking water. Waters containing oxidized forms of manganese interfere with the LCV method. Also, when iodide ion exceeds 50 mg/liter and chloride ion exceeds 200 mg/liter, the amperometric method is preferred. Under unusual situations, where mixtures of chlorine, bromine, and ozone occur along with iodine, the problem of separation is difficult (Standard Methods, 1976).

Biocidal Activity

Table II-9 shows the relative resistance of bacteria, viruses, and cysts to inactivation by iodine.

TABLE II-9

Comparative Values from Confirmed Experiments on Disinfecting Water with Iodine at 23°C to 30°C, at pH 7.0.

Efficacy against Bacteria

Chang and Morris (1953), summarizing the development of a universal water disinfectant tablet for the military, concluded that iodine concentrations of 5 to 10 mg/liter were effective against bacterial pathogens. Iodine was less dependent than chlorine upon the pH, temperature, contact time, and secondary reactions with nitrogenous impurities in the water. As a cysticide, iodine was poor in water with a high pH. Consequently, the tablet that was formulated contained an acid buffer to lower the pH of the water. The tablet, called ''Globaline," released 8 mg/liter of iodine, which is extremely high compared to the amount that is possibly needed for public water supplies. The tablet is not widely accepted, since color, taste, and odor problems are fairly common (O'Connor and Kapoor, 1970).

Chambers et al. (1952) investigated the effect of iodine concentration, pH, exposure time, and temperature on 13 enteric bacteria in a clean system. Their results are not completely quantitative in that the reported iodine concentration is that required to inactivate all test organisms plated out after 1, 2, and 5 min of contact. This was equivalent to approximately 99.9% activation with the procedure that they followed. There was definitely an observed pH effect. At 2°C to 5°C, some bacterial species required 3 to 4 times as much iodine for similar inactivation at pH 9.0 as was required at pH 7.5. For 5 min of exposure, the required dosage for ≥99.9% bacterial inactivation in high pH water at 20°C-26°C was always less than 1 mg/liter. A summary of their work with E. coli is given in Table II-10. This is the best available information on iodine as a bactericide.

TABLE II-10

Concentrations of Iodine and Contact Times Necessary for 99% Inactivation of Escherichia coli.

Efficacy against Viruses

Studies on the efficacy of iodine on viruses have shown that viruses are more resistant to disinfection than are vegetative cells of bacteria. Krusé (1969) compared virus and bacterial inactivation. At an iodine dose of 10 mg/liter in 0.048 mM potassium iodide at room temperature and pH 7.0, E. coli was inactivated more rapidly than f2 virus but both were reduced by 4 logs in less than 1 min. However, when the potassium iodide concentration was raised to 0.5 M, bacterial inactivation was 4 logs in 1 min while virus inactivation was only 0.5 log in 1 hr (Figure II-6).

Figure II-6

Survival of E. coli and f₂ bacteriophage reacted with 0.04 mM iodine containing (a) 0.048 mM potassium iodide (KI) at 37°C and (b) 0.5 mM potassium iodide at 0°C, and survival obtained with (c) 1 mM of a sulfhydryl reacting agent, p- (more...)

Berg et al. (1964) measured the dynamics of survival for poliovirus, coxsackievirus, and echovirus that were iodinated at pH 6.0 and at temperatures of 5°C, 15°C, and 25°C. The work of the Johns Hopkins group from 1962 to 1969 (Krusé, 1969) was primarily with the virus model f2, although some comparative poliovirus 1 work was done. Cramer et al. (1976) have shown that the mode of inactivation of these two viruses by iodine is similar. Data on the inactivation of virus by iodine are summarized in Table II-11 for both polio and f2. The results of inactivation studies by the various groups compare very favorably. At pH 6.0, iodine approaches the order of magnitude of virucidal activity of hypochlorous acid (HOCl). At the pH's likely to be maintained in the distribution system (pH 8.0), iodine is a vastly more effective virucide than combined chlorine and is not far removed from the activity range of free chlorine.

TABLE II-11

Concentrations of Iodine and Contact Times Necessary for 99% Inactivation of Polio and f₂ Viruses with Flash Mixing.

Efficacy against Parasites

Many studies on cyst inactivation have been reported, but there are discrepancies in their results due mainly to differences in the test systems that were used. The dose, pH, and temperature that were used in many of the studies are in doubt as are the different sources, cleaning methods for cysts, number of cysts used per test, and determination of viability. Stringer (1970) compared the resistance to iodine disinfection of Entamoeba histolytica cysts obtained from in-vitro culturing with those harvested from a human carrier and mixed amoebic cysts from monkeys. In water at room temperature at pH 6.5, 3 times the iodine dose was required in 10 min of exposure for 99% inactivation with wild (naturally formed) cysts compared to the cultured cysts. Large numbers of amoebic cysts from simian hosts were more readily available than cysts from human stools. They served as a reliable model for human E. histolytica.

The most extensive and reliable data on the cysticidal properties of iodine are found in the work of Chang and Stringer and their co-workers (Change and Baxter, 1955; Chang et al., 1955; Stringer, 1970; Stringer and Krusé, 1971; Stringer et al., 1975). The only quantitative comparative study of the cysticidal properties of chlorine, bromine, and iodine believed to be published is that of Stringer, who reported a 99.9% inactivation. The earlier studies of others involved "total kills" and are quite dependent on the cyst density that was used.

All investigators more or less agree that iodine is an excellent cysticide in low pH waters (< pH 4.0), suggesting that molecular iodine (I₂), rather than hypoiodous acid, is the active agent. Chang (1958) observed that when the titratable iodine dosages exceed 20 mg/liter in the presence of iodide ion, the triiodide ion (I₃^-) has a cysticidal efficiency that is equal to 1/11, 1/8, and 1/7 that of I₂ at 6°C, 25°C, and 35°C. He further calculated relative cysticidal efficiency of hypoiodous acid to be one-third that of 1₂ at 6°C and one-half that of I₂ at 20°C.

There may be the problem of ineffective iodine residual in the distribution system where cross-connection introduction of cyst contamination is a possibility. This is due to the practice (for corrosion control) of maintaining water in the distribution system at approximately pH 8.0. At the low halogen residual levels that are usually maintained, the most cysticidal form, I₂, would be less prevalent than hypoiodous acid. Water at 25°C and pH 8.0 with a total iodine residual of 0.5, 1.0, and 2.0 mg/liter would contain only 0.06, 0.2, and 0.62 mg/liter of molecular iodine. Table II-12 shows the time in minutes required for 90% inactivation of simian cysts with residuals of bromine, chlorine, and iodine in water at pH 6.0, 7.0, and 8.0 and a temperature of 30°C. At pH 8.0, iodine is inferior to free bromine and chlorine. However, in the presence of excess ammonia, with which the halogens could react, bromine appears to be more effective at pH 8.0 than do similar dosages of iodine or chlorine. Residual iodine in water at pH 8.0 has the ability to persist.

TABLE II-12

Contact Times Necessary for Low Residual Bromine, Chlorine, and Iodine in Water at 30°C to Effect 99% Inactivation of Cysts from Simian Stools.

The kinetics of iodine as a cysticide in water at 20°C to 30°C have been assembled and compared. The cultured E. histolytica cyst data of Chang and co-workers (Chang and Baxter, 1955; Chang and Morris, 1953; Chang et al., 1955) and the results of Stringer et al. (1975) present a consistent picture. For cultured cysts in water at pH 5 the c · t values for 99% inactivation calculated by Chang were 200 at 3°C, 130 at 10°C, and 65 at 23°C (Chang and Morris, 1953). Stringer et al. (1975) extended the cultured cyst data at 30°C for a range of pH given below:

These c · t values from Stringer et al. (1975) were approximately one-half the values they obtained with E. histolytica cysts from human stools and for mixed cysts from simian stools.

Mechanism of Action

The failure of a strong commercial iodine disinfectant to inactivate poliovirus (Wallis et al., 1963) led to interest in the responsible mechanism, especially the role of pH and iodide ion. Berg et al. (1964) claimed that iodine inactivation of coxsackievirus resulted from bio-molecular reaction with a single iodine molecule and that clumping of virions played a role in resistance to disinfection. Fraenkel-Conrat (1955) pointed out inconsistencies in the literature regarding the virucidal properties of iodine. Hsu (1964) and Hsu et al. (1965) clarified the mechanism of iodine on cells and virus. Hsu extracted fully active transforming DNA from iodine-inactivated Haemophilus parainfluenzae cells and just as much infectious RNA from f2 bacterial virus that had been inactivated (5 logs) by iodine as from noniodinated controls. Iodine inactivation, unlike the action of chlorine, appears to be attributable to a reaction with vital amino acids in proteins. Further experiments were conducted to determine whether sulfhydryl, tryptophanyl, histidyl, or tyrosyl groups were involved. With bacterial cells, there was a striking similarity between the kinetics of inactivation with iodine and the application of p-chloromercuribenzoic acid, a sulfhydryl reacting agent. While the active chemical species of iodine is not known, Hughes (1957), Allen and Keefer (1955), Bell and Gelles (1951), and Hsu (1964) hypothesized that the hydrated cationic iodine species (H₂OI⁺) attacks the base, first against the sulfhydroxyl groups, and is not materially affected by the presence of iodide ion as was tyrosine. When sulfhydroxyl groups are the site of inactivation, a low pH should favor the reaction. With the viruses, the sulfhydroxyl group is not involved. Evidence of tyrosine's involvement came from parallel experiments showing similar patterns of inactivation curves between iodination of f2 virus and _L tyrosine. Li (1942, 1944) had shown that the presence of iodide ion and low pH iodine (I₂) inhibited the iodination of tyrosine. Both poliovirus and f2 virus inactivation with iodine was inhibited by the iodide ion (Cramer et al., 1976). At pH 4.0, both polio and f2 viruses survived iodination whereas at pH 10 inactivation was complete. Although iodine decomposes rapidly to iodate (IO₃^-) and iodide at pH 10, flash mixing of iodine (yielding hypoiodous acid) overcomes this difficulty effectively, and the virus inactivation is complete in less than 1 min (Longley, 1964). Therefore, there is evidence that iodine action, with little or no iodide present, is effected by the modification of protein without destroying DNA or RNA (Brammer, 1963; Hsu, 1964). The mode of action of iodine in cyst penetration and inactivation has not been studied.

Conclusions

Iodine has many features that are comparable to free chlorine and bromine as a water disinfectant, but iodamines are not formed. Free iodine is an effective bactericide over a relatively wide range of pH. Field studies on small public water systems have shown that low levels of 0.5 to 1 mg/liter of free iodine can be maintained in distribution systems and that the magnitude of residual is sufficient to produce safe drinking water with no adverse effects on human health. Like other halogens, the effectiveness of iodine against bacteria and cysts is significantly reduced by high pH, but unlike bromine and chlorine it is much more effective against viruses because of the enhanced iodination of tyrosine. Currently, its use is restricted primarily to emergency disinfection of field water supplies because of its high cost and because it is difficult to apply to large systems. The possible adverse health effects of increased iodide intake for susceptible individuals in the population must also be considered.

Chemistry of Bromine in Water

Bromine was first applied to water as a disinfectant in the form of liquid bromine (Br₂) (Henderson, 1935), but it can also be added as bromine chloride gas (BrCl) (Mills, 1975) or from a solid brominated ion exchange resin (Mills, 1969). Oxidation of bromide (Br^-) to bromine can also be accomplished either chemically or electrochemically. Oxidation with aqueous chlorine gives either bromine or hypobromous acid (HOBr), depending on the ratio of chlorine to bromide. Both bromine (Liebhafsky, 1934) and bromine chloride (Kanyaev and Shilov, 1940) hydrolyze to hypobromous acid:

(21)

(22)

Molecular bromine exists in water at moderately acid pH and high bromide concentrations since the equilibrium constant of Reaction 21 is 5.8 × 10^-9 at 25°C. Like chlorine, bromine chloride has a much higher hydrolysis constant than this, so it does not exist as the molecular form in appreciable concentrations under conditions of water treatment. The ratio of molecular bromine to hypobromous acid depends on both pH and bromide concentration. From the equilibrium expression for Reaction 21:

(23)

Thus, for a solution containing 10 mg/liter bromide and a pH of 6.3, 1% of the bromine is Br₂, while at lower bromide concentrations or higher pH aqueous bromine occurs almost entirely as hypobromous acid, which is a very weak acid with a dissociation constant (for Reaction 24) of 2 × 10^-9 at 25°C (Farkas and Lewin, 1950).

(24)

The hypobromite ion (OBr^-) becomes the major form of bromine above pH 8.7 at 25°C. Lower temperature decreases both of the above equilibrium values, thereby increasing the pH range where hypobromous acid is the major chemical form of bromine in water.

Bromine and bromine chloride also react with basic nitrogen compounds to form combined bromine or bromamines (Galal-Gorchev and Morris, 1965; Johannesson, 1958; Johnson and Overby, 1971):

(25)

(26)

(27)

The observed breakpoint for ammonia (NH₃) solutions that have been treated with bromine is similar to that seen with chlorine. For bromine, this point corresponds to 17 mg/liter bromine for 1 mg/liter of ammonia nitrogen (NH₃-N). At this point, a minimum of bromamine stability occurs (Inman et al., 1976) as ammonia nitrogen is oxidized to nitrogen gas:

(28)

At bromine-to-ammonia ratios higher than this and in the acid pH range, nitrogen tribromide (NBr₃) is stable. It is the most abundant bromamine in such aqueous solutions. At lower bromine concentrations, dibromamine (NHBr₂) can be present, but it is quite unstable. Only at alkaline pH values and very high ammonia concentrations such as those found in wastewater are significant quantities of monobromamine (NH₂Br) formed. Organic bromamines are also formed, but there is little information on their forms or stability.

Production and Application

Bromine is produced by oxidation of bromide-rich brines (that contain between 0.05% and 0.6% bromide) with chlorine. Bromine is then stripped with steam or air and is collected as liquid Br₂.

Bromine chloride is produced by mixing equal molar quantities of pure bromine and chlorine (Mills, 1975). It condenses to liquid bromine chloride below 5°C at I atm pressure or above 30 psig at 25°C. In the liquid phase, more than 80% of the liquid is bromine chloride, and the remainder is Br₂ and Cl₂. In the gas phase, 40% of the bromine chloride dissociates to Br₂ and Cl₂ at 25°C.

Bromine has been applied to water as liquid Br₂. The difficulties that are encountered when handling the liquid and its corrosive nature, especially when wet, have encouraged the use of bromine chloride. Liquid bromine chloride is removed from cylinders under moderate pressure. It is then vaporized, and the gas is metered in equipment that is similar to that used for chlorine. Like chlorine, bromine chloride is shipped as the dry liquid in steel containers. Gas feeders must be made of Teflon, Kynar, or Viton plastics because bromine chloride is more reactive than chlorine with polyvinylchloride plastics.

Analytical Methods

Bromine concentrations can be measured iodometrically by procedures that are identical to those used to measure total chlorine residuals (Standard Methods, 1976). None of these procedures is capable of distinguishing free bromine (Br₂, HOBr, OBr^-) from combined bromine (bromamines) or other oxidants that are capable of reacting with iodide under the slightly acid conditions used in these procedures. However, bromate does not interfere except at low pH (Kolthoff and Belcher, 1957). UV spectroscopy can be used to measure the bromamines selectively in the presence of one another and without interference from free bromine because none of the free bromine forms except Br₃^- has strong UV absorptivities (Johnson and Overby, 1971).

Biocidal Activity

Since residual bromine was rarely measured in the studies cited here, c · t (concentration, mg/liter, times contact time, min) products have generally been calculated using the dosage of bromine added to the system.

Efficacy against Bacteria

The efficacy of bromine inactivation of bacteria has been summarized by Farkas-Hinsley (1966). Vegetative cells are readily inactivated by bromine, but reports often leave the type of bromine compound and its residual concentration as uncertain. Consequently, it is difficult to make quantitative comparisons with other disinfectants. Tanner and Pitner (1939) reported that the concentrations of hypobromous acid that are required to give ''complete kill" in 30 min are 0.15 mg/liter for Escherichia coli and 0.6 mg/liter for Salmonella typhi. Spores of Bacillus subtilis required more than 150 mg/liter of bromine. Krusé et al. (1970) found that 4 mg of bromine per liter as hypobromous acid gave 5 logs of disinfection of E. coli in 10 min at 0°C. At pH 4.5, where significant Br₂ was present, 2 mg/liter bromine at 0°C gave 4 logs of E. coli disinfection in 3 min. However, Krusé also reported that 0.1 M bromide decreased E. coli disinfection to 2 logs of inactivation from 4.5 logs at 0.001 M bromide, using 4 mg/liter bromine at pH 7.5 and 0°C in each case. This is in conflict with observations made at low pH, since high bromide should also produce bromine. The data at pH 4.5 showed 4 logs inactivation with a c · t of 6 compared to 5 logs inactivation from a c · t of 40 for hypobromous acid. Even these data, the best available, are based on the concentrations of bromine added to the solution. It is difficult to compare the various studies on disinfection by bromine, since the chemical species of bromine present in most instances have not been measured or reported.

The effect of the formation and decomposition of the bromamines on the efficacy of bacterial disinfection was first discussed by Johannesson (1958). He reported that 0.28 mg/liter of monobromamine expressed as bromine resulted in 99% inactivation in less than 1 min, while the same concentration of N-bromodimethylamine [N(CH₃)₂Br] required 12 min for 99% inactivation of E. coli. The measurements of the halogen remaining after the experiments showed very little loss.

The efficacy of bromine inactivation of spores of Bacillus metiens and B. subtilis has been studied by Marks and Strandskov (1950) and Wyss and Stockton (1947). Both pairs of investigators found that the activity was markedly pH-dependent at low pH values where molecular bromine predominates. The activity increased rapidly from pH 4 to pH 3, the lowest that was tested. The c · t required for 99% inactivation was 20-25 at 25°C and pH 3.0. At pH 6 to 8, the c · t was 225 to 375 (Wyss and Stockton, 1947). At pH 9, where the hypobromite ion starts to predominate, the c · t increased to more than 500.

Marks and Strandskov (1950) also determined that at pH 7 and 25°C monobromamine was one-half as effective as hypobromous acid against B. metiens spores. N-bromosuccinimide was 1/17, and N-bromopiperidine was 1/800 as effective. Wyss and Stockton (1947) showed the effect of changing chemical form and concentration by increasing the concentration of ammonia that was added to a 20 mg/liter bromine solution at pH 7 containing B. subtilis spores. Table II-13 shows the measured residual and the time required for 99% inactivation. Before the ammonia is added to the solution, hypobromous acid is the major chemical form. When the ammonia nitrogen reaches 1 mg/liter, nitrogen tribromide predominates. At 2 mg/liter ammonia nitrogen, the breakpoint occurs with rapid loss of combined bromine residual. At 10 and 30 mg/liter ammonia nitrogen, dibromamine is present and is less stable and effective than hypobromous acid. At 1,000 mg/liter ammonia nitrogen, monobromamine has a c · t value for 99% inactivation that is as effective as nitrogen tribromide and hypobromous acid.

TABLE II-13

The Effect of Ammonia on Bromine Concentrations Required for 99% Inactivation of Bacillus subtilis Spores at 20 mg/liter, pH 7, 25°C.

The effect of temperature on the efficacy of hypobromous acid inactivation of B. subtilis spores is shown in Table II-14 (Wyss and Stockton, 1947). The activity increases approximately 2.3 times for each 10°C rise in temperature. After measuring washed vegetative cells of B. subtilis, Wyss and Stockton also found them to be 500 times less resistant. They reported no temperature data for their vegetative cell studies.

TABLE II-14

The Effect of Temperature on Bromine Concentrations Required for 99% Inactivation of Bacillus subtilis Spores with 25 mg/liter Bromine at pH 7.0.

Efficacy against Virus

Krusé et al. (1970) reported that 4 mg/liter of bromine as hypobromous acid at pH 7 and 0°C gave 3.7 logs of inactivation of f2 E. coli phage virus in 10 min. At higher bromide concentrations and lower pH, where Br₂ becomes the principal form of bromine, the rates of inactivation increased. A c · t of 40 was required for 3.7 logs inactivation at pH 7.0 and 0°C compared to a c · t of 1 for 4.5 logs inactivation at 0°C and pH 4.8. At pH 7.5 and with a concentration of 0.1 M bromide at 0°C, a c · t of 12 yielded 5.5 logs inactivation. In these studies the residual concentrations of bromine were probably not significantly different from those in solutions that did not contain added salts. In the presence of added bromide, amines were difficult to quantify. After 2 hr, Krusé and his colleagues found no viable virus in a solution of 4 mg/liter bromine, starting from an initial titer of 1.7 × 109 plaque-forming units (PFU)/ml at 0°C and pH 7.5, when excess ammonia was present. However, the addition of bromine to phage solutions that contained excess methylamine (CH₃NH₂) and glycine (H₂NCH₂COOH) essentially stopped disinfection under these conditions. This may be due either to the formation of the N-bromo compounds or to the reduction of bromine.

Taylor and Johnson (1974) used Φx174 E. coli phage and measured concentrations of the disinfecting species in constant residual solutions. They reported that 0.32 mg/liter hypobromous acid required only 1.1 min or a c · t of 0.35 to inactivate 99% of this phage at 0°C. Compared to hypobromous acid, molecular bromine was approximately 3 times as fast, requiring Br₂ at c · t of 0.1. Nitrogen tribromide was less potent, requiring a c · t of 1.0 for 99% inactivation of this phage at 0°C.

The effect of temperature on Φx174 inactivation was measured for 0.16 mg/liter of hypobromous acid expressed as bromine at pH 7. The Arrhenius plot of In k against 1/T (where T is temperature) from 273°K to 303°K gave a slope equivalent to 37 kcal/mol or an increase in rate of inactivation of 1.9 times for a rise in temperature from 15°C to 25°C.

Sharp's group (Floyd et al., 1976, 1978; Sharp et al., 1975, 1976) has made careful studies with measured, controlled residual concentrations. They demonstrated that even under these controlled chemical conditions the degree of aggregation or clumping had a marked effect on the apparent, observed inactivation rates with reovirus and type 1 poliovirus (Sharp et al., 1975). Reoviruses, as single particles, required only 1 s for 3 logs of inactivation at pH 7 and 2°C with hypobromous acid at 0.46 mg/liter bromine. Aggregated samples gave much slower rates, especially at high levels of inactivation. At 3 logs of inactivation, the time required doubled for the same concentration conditions (Sharp et al., 1976). Poliovirus 1 was also rapidly inactivated by hypobromous acid when single virus particles were studied (Floyd et al., 1976). At 2°C and pH 7, only 3.5 mg/liter bromine or a c · t of 0.2 was required for 2 logs of inactivation. However, the rate of inactivation did not increase linearly with concentration at this temperature. Higher concentrations were much less efficient than longer exposures. At 1 mg/liter hypobromous acid, 7 s were required for 1 log of inactivation or a c · t of 0.23 for 2 logs of inactivation at 2°C and pH 7 with hypobromous acid. The inactivation rate increased only slightly with temperature at this concentration. At 10°C a c · t of 0.21 for 2 logs of inactivation and at 20°C a c · t of 0.06 was required for 99% inactivation with hypobromous acid.

The inactivation of single poliovirus particles in buffered, distilled water and constant residual concentrations for the other major bromine chemical forms have also been studied by Floyd et al. (1978). Table II-15 gives the calculated c · t required to yield 99% inactivation for these different forms near 1 mg/liter expressed as bromine. The values in Table II-15 depend on concentration except for Br₂, for which time and concentration were inversely related as normally assumed from the Watson-Chick relationship (Morris, 1975). Floyd et al. (1978) also demonstrated that dibromamine, nitrogen tribromide, and hypobromous acid were less efficient at higher concentrations while the hypobromite ion became more effective as the concentration increased. The rate for the hypobromite ion is rapid compared to the other bromine compounds. This is interesting in light of the fact the hypochlorite ion (OCl^-) is generally considered to be a much poorer disinfectant than hypochlorous acid (HOCl).

TABLE II-15

Exposure (c·t) to Various Bromine Compounds Required for 99% Inactivation of Poliovirus 1, Mahoney.

Mechanisms of Action

The pattern of bromine disinfection appears to be similar to that of chlorine. After comparing the activity of chlorine, bromine, and iodine against spores, Marks and Strandskov (1950) noted that Br₂ was 9 times more effective than hypobromous acid and that the hypobromite ion and tribromide ion (Br3^-) had very low activity. They noted that a high degree of polarity contributed to the inactivity of the ionic forms and the reduced activity of the hypohalous acid compared to the free molecular halogen. They also found that "the killing rates of bacterial spores for the hypohalous acids and probably for the molecular halogens decrease in the [following] order: chlorine, bromine, iodine."

Many workers attribute this decrease to the greater oxidation potential of the halogens with lower molecular weight. However, the polarity and perhaps the sizes of the halogens may be important in getting the disinfectants to the vital site.

Olivieri et al. (1975) demonstrated that bromine was effective in inactivating both naked viral RNA and intact virus. The primary site of inactivation of f2 phage more likely involves the reaction of bromine with the protein coat of the virus, because inactivation of RNA that was prepared from bromine-treated virus lagged significantly behind the inactivation of intact virus. The mechanism of inactivation with bromine, as with the other halogens, involves moving the disinfectant to the vital site, mass transport, and the reactivity of the bromine with that site, oxidation. The effectiveness of bromamines as disinfectants may be explained by their relatively low polarity and high reactivity. The effectiveness of the hypobromite ion as a virucide may be due to the fact the high pH's of the solutions make the virus more sensitive to the ion. Or, it could be attributed to Olivieri's observation that bromine acts primarily on the protein coat of viruses.

Summary and Conclusions

Laboratory tests show that bromine is an effective bactericide and virucide. It is more effective than chlorine in the presence of ammonia.

As a cysticide, it is highly active. Bromine is active over a relatively wide range of pH, and it retains some of its effectiveness as hypobromous acid to above pH 9.

The major disadvantage of bromine as a disinfectant is its reactivity with ammonia or other amines that may seriously limit its effectiveness under conditions that are encountered in the treatment of drinking water. Data on the effectiveness of bromine against bacteria are complicated by this reactivity and the lack of characterization of the residual species in disinfection studies.

Research Recommendations

Further research is required to quantitate the effectiveness of the various species of bromine and the bromamines, both organic and inorganic, against bacteria, viruses, and cysts, particularly with bacteria.

The reactivity of bromine and the dosages that are required to maintain effective residual in natural water systems should be investigated.

The effectiveness of various technologies for application of bromine or bromine chloride to intended drinking water should be evaluated.

The Chemistry of Ferrate in Water

Aqueous solutions of potassium ferrate are unstable and decompose to yield oxygen (O2), hydroxide (OH-), and insoluble hydrous iron oxide [FeO(OH)]:

(29)

The hydrous iron oxide is a coagulant that is commonly used in water treatment. Initial ferrate concentration, pH, temperature, and the surface character of the resulting hydrous iron oxide affect the rate of decomposition (Schreyer and Ockerman, 1951; Wagner et al., 1952; Wood, 1958). Ferrate solutions are most stable in strong base (> 3 M or at pH 10 to 11). Schreyer and Ockerman (1951) reported that dilute aqueous solutions of ferrate are more stable than concentrated solutions. The presence of other inorganic ions also affects ferrate stability in aqueous solutions.

Ferrate reacts rapidly with reducing agents in solutions (Murmann, 1974). It will also oxidize ammonia (NH₃). The rates for this reaction increase with pH (optimum pH range 9.5-11.2), molar ratio of ferrate to ammonia, and temperature (Strong, 1973). The oxidation of ammonia by ferrate below pH 9 is markedly slower than that observed for chlorine. Some information is available on the reaction of ferrates with various organic compounds (Audette et al., 1971; Becarud, 1966; Zhdanov and Pustovarova, 1967).

Preparation

Concentrated solutions of sodium ferrate (Na₂FeO₄) may be prepared electrochemically from the more common iron forms. Scrap iron can be converted to ferrate iron with a 40% efficiency (Murmann, 1974). Potassium ferrate may be prepared by wet oxidation of Fe(III) with potassium permanganate (KMnO₄). Subsequent recrystallization yields a crystalline solid with greater than 90% purity (Schreyer et al., 1950). Ferrate salts are not commercially available. Consequently, sufficient quantities for pilot or full-scale testing would have to be especially prepared.

Analytical Methods and Residuals

Aqueous ferrate solutions have a characteristic violet color that is similar to permanganate and a wavelength maximum at 505 nm in the visible portion of the electromagnetic spectrum. The molar extinction coefficient at 505 nm in 10^-4 M sodium hydroxide (NaOH) was 1,070 ± 30 1 M^-7cm^-1 (Wood, 1958). Since ferrates are unstable, residual cannot be maintained.

Biocidal Activity

Efficacy against Viruses

Waite (1978a,b) reported that the inactivation of the RNA bacterial f2 virus with ferrate appeared to be more dependent on pH than was bacterial inactivation. Considerably more viral inactivation was observed at pH 6.0 than at pH 8.0. At 1.2 mg/liter ferrate, just under 4 min were required for 99% inactivation, while for similar conditions 13 min were necessary at pH 8.0.

The disinfection efficacy of ferrate is summarized in Table II-16. The ferrate dose is given in the table, since ferrate residuals were not reported. Ferrate decomposes rapidly in aqueous solution, and the dose represents an overestimate of ferrate concentration. Nevertheless, the c · t products (concentration, mg/liter, times contact time, min) indicate the difference in efficacy relative to the test microorganism and provide a crude idea of the c · t product for ferrate.

TABLE II-16

Concentrations of Ferrate (FeO₄2^-) and Contact Times Necessary for 99% Inactivation of Escherichia coli, Streptococcus faecalis, and f₂ Virus.

Mechanism of Action

No studies have elucidated the mechanism of inactivation caused by ferrate.

Conclusions

While not as biocidal as the free halogens, chlorine dioxide, or ozone, ferrate appears to be similar to or slightly better than the chloramines as a bactericide and more active as a virucide than the chloramines. The combined application of ferrate as a disinfectant and a coagulant makes it an attractive alternate biocide. Far more information on the feasibility of large-scale preparation of ferrate and its biocidal activity (particularly under water treatment conditions) are necessary before ferrate can be given consideration as a water disinfectant for public systems.

Attainment of Elevated pH

High pH values in water are obtained by using either calcium hydroxide or sodium hydroxide (NaOH). Sodium hydroxide is a by-product of the production of chlorine by the electrolysis method. Calcium hydroxide is prepared by reacting calcium oxide (CaO) with water. The calcium oxide is produced by heating calcium carbonate (CaCO₃) to drive off the carbon dioxide (CO₂). The calcium oxide is normally prepared off the site of the water treatment plant. However, it is often slaked on site at all but the smallest plants, which purchase calcium hydroxide and use it directly. A few, very large water treatment plants produce their own calcium hydroxide by calcining their water-softening sludge.

Hydroxide (OH^-) is added to water as a water slurry containing calcium hydroxide or as a water solution of sodium hydroxide. In the usual water-softening technique, the water is flocculated for 10-30 min to promote the precipitation reactions. The precipitates are then removed by sedimentation with retention times from 1 to 6 hr.

Disinfection with high pH is readily accomplished, but the required pH values are so high that the pH must be reduced before the water is consumed. In a full-scale water treatment plant, this is accomplished in the recarbonation stage, which is a normal component of the water-softening process. However, for small institutions or private residences the pH reduction requirement would mitigate against the use of this disinfection technique.

Analytical Methods and Process Control

The pH can be reliably measured with the glass electrode at pH values of approximately 7 and at room temperatures. Special care must be taken to obtain accurate values at other temperatures and at the elevated pH values that are necessary for disinfection with this process.

Residual

The hydroxide that is used to obtain the necessary pH values enters into few side reactions in natural water after the initial formation of metallic hydroxides and the reactions with carbon dioxide. The amounts of hydroxide for these reactions can be readily computed after the water has been chemically analyzed. There is a slow additional loss of hydroxide from the water by reaction with carbon dioxide entering from the atmosphere. However, this is not a serious problem.

Biocidal Activity

The microbial inactivation under consideration in this chapter is that caused solely by high pH. When the pH of water is raised, the opportunity exists for precipitation of many compounds associated with carbonates, oxides, and hydroxides, among many others. Even in distilled water, calcium carbonate may precipitate if calcium hydroxide is used to raise the pH and if sufficient carbon dioxide enters the water during the experiment. These precipitates provide opportunities for adsorption and coagulation of microorganisms and will cause an increased removal over that obtained from the pH effect alone without the precipitate. The research cited below, unless otherwise noted, was conducted with procedures that obviated this problem or that made its effect insignificant.

Efficacy against Bacteria

Wattie and Chambers (1943) determined the times required to obtain 100% inactivation of several bacterial species at initial concentrations of approximately 1,500 organisms/ml in 20°C to 25°C dechlorinated tapwater, using calcium hydroxide for pH adjustment (see Table II-17). The inactivation was significantly less at each pH for each organism between 0°C to 1 C. At a pH of 11.5, the time for 100% inactivation of Escherichia coli was increased from 210 min between 20°C to 25°C to 355 min between 0°C to 1°C. The time for 100% inactivation of Salmonella typhi was increased from 75 min to 270 min for the same temperature change. No mention was made of difficulties from the formation of precipitates.

TABLE II-17

Contact Times Necessary for High pH Conditions to Effect 100% Inactivation of Initial Concentrations of 1,500 Organisms/ml.

Riehl et al. (1952) reported E. coli inactivations of 95% in 8 hr at pH 10.5 and 5°C, 100% in 2 hr at pH 10.5 and 15°C, and 100% in approximately 30 min at pH 10.5 and 25°C in distilled water. Calcium hydroxide was used for pH adjustment. Inactivations of 50% and 55% in 10 hr were noted for Salmonella montevideo and S. typhi, respectively, at 2°C and pH 10.6 in distilled water. Initial concentrations of organisms were approximately 1,000/ml. They also observed that higher temperatures gave more inactivation at the same pH and contact time and that the composition of the water did not markedly influence the bacterial survival.

Berg and Berman (1967) determined the inactivation of E. coli at 25°C in the dilution water medium that is described in the 11th edition of Standard Methods (1960) (see Table II-18). Sodium hydroxide was used to adjust the pH.

TABLE II-18

Inactivation of Escherichia coli by High pH Conditions at 25°C in a Dilution Medium.

Efficacy against Viruses

Wentworth et al. (1968) reported that little or no poliovius 1 was inactivated in distilled water at pH 11.2 at room temperature in 90 min when calcium hydroxide or sodium hydroxide was used to increase the pH. For poliovirus 1 in distilled water without added salts, Sproul et al. (1970) showed that no inactivation occurred in 90 min at pH values of 10.5 or less when calcium hydroxide and sodium hydroxide furnished the hydroxide. Their work was done at 21°C to 22°C. Sproul (1975) obtained poliovirus 1 inactivations in 30 min of 7% at pH 11.5, 59% at pH 11.9 in 30 min, 94% at pH 12.1 in 20 min, and 99.83% at pH 12.5 in 5 min. He used distilled water with 100 mg/liter of sodium chloride (NaCl) at 22°C to 23°C.

Berg and Berman (1967) showed the inactivation of echovirus 7 AGKP8A1 at 25°C in a dilution water medium (Standard Methods, 1960). They used sodium hydroxide to adjust pH (see Table II-19). From his work with the f2 bacteriophage, Donovan (1972) suggested that a calcium-virus complex was formed and that a large part of the observed decrease in titer at pH 11.5 was caused by an aggregation of the virus. Electron micrographs and chemical evidence supported the observation.

TABLE II-19

Inactivation of Echovirus 7 by High pH Conditions at 25°C in a Dilution Medium.

Mechanisms of Action

The inactivation mechanism of high pH on bacteria has not been examined. There is ample evidence to show that the poliovirus is inactivated by the disruption of the capsid and a loss of the RNA to the water (Boeye and Van Elsen, 1967; Maizel et al., 1967; Van Elsen and Boeye, 1966). Boeye and Van Elsen (1967) suggested that, at pH 10 and at elevated temperatures (≥30°C), the RNA was released from the capsid in a degraded form or that it was quickly and extensively broken down after release. The applicability of this mechanism to enteroviruses other than poliovirus is unknown.

Conclusions

The utilization of high pH conditions for disinfection of water is feasible, but higher pH values than are normally used in water treatment and long contact times are required. At pH values of approximately 10.5, bacteria are inactivated in up to 600 min, but viruses are probably unaffected. A pH value of 12.0 to 12.5 and a contact time of approximately 30 min can probably yield 99.0% to 99.9% inactivation of most bacteria and of certain viruses. This process has a drawback: the pH that is necessary for effective disinfection must be reduced before the water can be consumed. Use of the high pH as the sole means of disinfection is not recommended. In many situations where water is softened by the precipitation process and where the water is of poor biological quality, the disinfection potential of this process could be used by increasing the pH above its present values.

Research Recommendations

Although this method has not been used deliberately for disinfection in the past, additional work on this method may be warranted since high pH conditions are attained in many treatment plants. Studies are needed to determine the pH values required for inactivation of protozoans, a broader and more representative group of viruses, and a larger group of enteric bacteria.

Analytical Methods

Analytical methods for hydrogen peroxide are based on oxidation or reduction with potassium permanganate (KMnO₄), potassium iodide (KI), or ceric sulfate [Ce(SO₄)₂] (Chadwick and Hoh, 1966). A colorimetric procedure based on the oxidation of titanium sulfate (TiOSO₄) (Snell and Snell, 1949) is sensitive to about 1 mg/liter in the absence of other oxidizing agents.

Production and Application

Commercially produced hydrogen peroxide is available in aqueous solution, usually ranging from 30% to 90%. It is stabilized during manufacture by addition of such compounds as sodium pyrophosphate (Na₄P₂O₇), acetophenetidin (CH₃CONHC₆H₄OC₂H₅), or acetanilide (CH₃CONHC₆H₅). As the concentration of hydrogen peroxide is decreased, the concentration of stabilizers is typically increased. Experience in the waterworks industry using hydrogen peroxide is nonexistent. However, hydrogen peroxide is widely used as a bleaching agent in making cotton textiles or in wood pulping (Chadwick and Hoh, 1966). Presumably, a concentrated solution would be diluted and applied with a chemical metering pump. Careful attention to safety in handling would be required because of the possibility of fire or explosion.

Biocidal Activity

The few pre-1965 references on disinfection by hydrogen peroxide have been summarized by Yoshpe-Purer and Eylan (1968). Although its bactericidal activity was indicated, a need for catalytic Fe²⁺ or Cu²⁺ was reported. Yoshpe-Purer and Eylan (1968) worked with Escherichia coli, Salmonella typhi, and Staphylococcus aureus in pure culture and as a mixture of bacteria. They also studied the effect of hydrogen peroxide concentrations (from 30 to 60 mg/liter), contact time (10 to 420 min), and initial concentration of organisms. Without ever measuring residuals, they concluded that bacterial inactivations were relatively slow, that E. coli was more resistant to hydrogen peroxide than S. typhi or S. aureus, and that the required inactivation time was increased as the initial concentration of organisms was increased. Regardless of the hydrogen peroxide concentration or the type of organism used , all tests were claimed to be ''sterile" in 24 hr. Table II-20 summarizes some of these data.

TABLE II-20

Contact Times Necessary for Hydrogen Peroxide (H₂O₂) to Effect 99% Inactivation of Various Concentrations of Escherichia coli, Salmonella typhi, and Staphylococcus aureus.

Toledo et al. (1973) studied inactivation of S. aureus and spores of several species of Bacillus and Clostridium with 25.8% (258,000 mg/liter) hydrogen peroxide. The time to reduce S. aureus by 6 logs was 1 min, whereas reduction of spores by 99% required from less than 1 min to approximately 17 min, depending on the species. Increasing the hydrogen peroxide concentration up to 41% (or 410,000 mg/liter) or increasing temperature from 24°C to 76°C significantly reduced the required inactivation time.

Several studies on virus inactivation by hydrogen peroxide have been reported. Lund (1963) obtained a 99% inactivation of poliovirus (Saukett strain) in about 6 hr with 0.3% (3,000 mg/liter) hydrogen peroxide. Mentel and Schmidt (1973) worked with rhinovirus (types 1A, 1B, and 7). They found that while a 1.5% (15,000 mg/liter) concentration required approximately 24 min for 99% inactivation, equivalent inactivation occurred in 4 min with a concentration of 3% (30,000 mg/liter).

Information is lacking on the effect of hydrogen peroxide on protozoa and helminths in water.

In none of the studies cited above is there any indication that the dosage of hydrogen peroxide was measured other than by the volumetric addition of hydrogen peroxide. No measurements of residual were reported, although Yoshpe-Purer and Eylan (1968) claimed that a residual was present for up to 13 days as indicated by inactivation of added doses of bacteria.

Mechanism of Action

No studies have specifically identified the mechanism of action of hydrogen peroxide. Spaulding et al. (1977) believed that the hydrogen peroxide molecule itself was not responsible for the action but, rather, that the free hydroxyl radical (HO · ) that it produced was the specific inactivating agent. They claimed that the catalytic effect of iron or copper ions supported this theory. Yoshpe-Purer and Eylan (1968) also reported that free radicals were important but that sufficient catalyzing metal ions were available either from the tap water that they used or from the bacterial cells themselves.

Conclusions

Because of its relatively high cost and the high concentrations that are required to achieve disinfection in reasonable time, hydrogen peroxide is not a generally satisfactory disinfectant for drinking water.

Research Recommendations

If further research is to be conducted, the stability of hydrogen peroxide and its residual effect in the presence of organic material and other substances in water or distribution systems should be investigated. Parallel studies using chlorine and hydrogen peroxide, separately and in conjunction, should be conducted.

Application

Laboratory studies of destruction of microorganisms are generally conducted by exposing small containers holding the test suspensions to the shielded source of ionizing radiation. However, in a practical application, the shielded source must be designed to permit relatively thin sheets of flowing liquid to be exposed. This is particularly important for high-energy electrons that have a less penetrating power than gamma rays. The MIT study (Massachusetts Institute of Technology, 1977) included a field demonstration in which a plant treating approximately 380,000 liters/day (minimum dosage of 400,000 rads) was designed and operated. The energy source was a 50-kW electron accelerator operated at 850,000 V. Sludge was pumped under the electron beam through a drum system that produced a layer that was 1.2-m wide and 2-mm thick. The entire operation was conducted in a concrete vault to shield the operators against X rays, which are produced incidentally. Shielding of workers is a major requirement both for gamma ray and high-energy electron sources.

Analytical Methods and Residual

No residual is produced by ionizing radiation. Consequently, dosage has been measured exclusively. The MIT (1977) report and Silverman and Sinskey (1977) summarized analytical methods including calorimetry, photoluminescent dosimetry, colorimetry, and the use of ionization chamber instruments.

Biocidal Activity

Efficacy against Bacteria

The first study of ionizing radiation in the treatment of water was reported by Dunn (1953). In addition to reviewing the literature and providing some discussion on ionizing radiation, he studied the use of a 1,000-Ci source of cobalt-60 (providing gamma rays) and a Van de Graaff generator (providing high-energy electrons). Working with natural waters and wastewaters that were not chemically or bacteriologically characterized or controlled, he exposed samples to a constant radiation flux (2,000 R/min from the cobalt-60 or electrons from 3-MV operation of the Van de Graaff generator). Varying only the time of exposure, Dunn found that it took 0.125 min (dosage 250,000 R) with the cobalt-60 source to reach 95.4% to 99.999% inactivation of total initial numbers of bacteria (as measured by a plate count at 37°C).

Ridenour and Armbruster (1956) studied the effect of a 10-kCi source of cobalt-60 (3,000 R/min) on a variety of natural waters, wastewaters, and pure cultures (approximately 2 × 107 organisms per milliliter) of 10 different organisms. They found that a dosage of 100,000 roentgen equivalent physical (rep) reduced the count of all species tested by > 99%. (In order of increasing resistance, the species were: Enterobacter aerogenes, Escherichia coli, Shigella flexneri, Salmonella typhi, S. sonnei, Salmonella sp., Staphylococcus aureus, S. paratyphi B, Streptococcus faecalis, and Bacillus subtilis.) With river water, a dosage of from 50,000 to 100,000 rep reduced the total bacterial count and the coliform index by at least 99%, but 150,000 rep were required to reduce the streptococcus index by 99%. Varying the pH (5.0, 7.0, and 8.5) had no effect on the rate of inactivation.

Also using cobalt-60 (1,100 Ci), Lowe et al. (1956) exposed pure cultures of bacteria that were suspended in double distilled water and sterile settled sewage. The concentration of microorganisms was approximately 1 × 10⁶/ml. Although they did not report exposure time, they summarized their exposure data in rads (radiation absorbed dose). (See Table II-21).

TABLE II-21

Dosages of Cobalt-60 Irradiation Necessary for 99% Inactivation of Microorganisms in Distilled Water.

There have been more studies on the disinfection of wastewater than on disinfection of drinking water. Ballantine et al. (1969) and Compton et al. (1970) considered ionizing radiation not as good as other available methods but, with increased emphasis on water reclamation (Eliassen and Trump, 1973) and the use of high-energy electrons (Massachusetts Institute of Technology, 1977; Wright and Trump, 1956), the prospects for practical application seem improved. In the MIT (1977) study, which is the most systematic one conducted to date, washed cells or spores were suspended in 0.067-M phosphate buffer (1 × 10⁸ cells/ml) and exposed to high-energy electrons from a Van de Graaff generator. The approximate dosages, in rads, required for a 99% inactivation of the organisms tested were: Escherichia coli (K12), 38,000; Salmonella typhimurium (LT2), 24,000; Micrococcus sp., 35,000; Aspergillus niger (spores), 78,000; S. typhimurium (24), 100,000; Streptococcus fecalis, 300,000; and Clostridium perfringens (spores), 400,000. The investigators found that the solids in the wastewater had no effect on disinfection.

Of the factors affecting disinfection, oxygen was most important. For example, for E. coli K12, the dosage required for 99% inactivation was more than doubled when an atmosphere of air was replaced by one of nitrogen, and almost halved when the air was replaced by an oxygen-rich atmosphere.

Mechanism of Action

In a review of the literature, Silverman and Sinskey (1977) summarized the mode of action of disinfection by ionizing radiation. Two effects are recognized: the direct effect, in which the primary cellular target is DNA that is damaged by the energy released from the ionizing radiation, and the indirect effect, which is associated with the production of such substances in the cell menstruum as hydrogen peroxide (H₂O₂), organic peroxides, and free radicals. Oxygen is important because it reacts with electrons and radicals, which in turn react to form hydrogen peroxide or ozone (O₃).

Conclusions

Ionizing radiation can disinfect water effectively; however, the large source, shielding, and relatively thin exposure layers that are required create difficult engineering and safety problems. The complex technology may limit application to large facilities that can provide for adequate safeguards. The absence of residual disinfection is also restrictive.

Research Recommendations

Any further research should address practical methods of delivering the ionizing radiation to the water mass. Basic engineering design, construction, and operation data should be developed.

Analytical Methods and Residual

Concentrations of potassium permanganate can be determined readily by iodometric titration using sodium thiosulfate (Na₂S₂O₃) as a titrant or by direct titration with ferrous sulfate. Manganese can be determined by atomic absorption spectrophotometry, or the permanganate ion (MnO₄^-) can be measured colorimetrically (Standard Methods, 1976). The minimum detectable concentration of manganese by the colorimetric procedures, using a 100-ml sample, is 50 µg/liter. A concentration of 0.05 mg/liter or greater can be detected visually by the pink color that is imparted to the water (Welch, 1963).

Production and Application

Crystalline potassium permanganate is highly soluble in water (2.83 g/100 g at 0°C). In waterworks, it is prepared usually as a dilute solution (1% to 4%) and applied with a chemical metering pump (American Water Works Association, 1971). It also may be added as a solid using conventional dry-feed equipment.

Reduction of the permanganate ion produces insoluble manganese oxide (Mn₃O₄) hydrates. To prevent distribution of turbid water or of water that will cause unsightly staining of plumbing fixtures, potassium permanganate most often is applied as a pretreatment that is followed by filtration. For example, addition of potassium permanganate to a finished water to maintain a residual in a distribution system is unacceptable because of the pink color of the compound itself or the brown color of the oxides.

Biocidal Activity

Cleasby et al. (1964) and Kemp et al. (1966) summarized the few scientific references to disinfection by potassium permanganate that were published earlier than 1960. Although some bactericidal activity was indicated, no quantitative data were presented, and the early reports disagreed as to its value.

The most systematic study of disinfection by potassium permanganate has been made by Cleasby et al. (1964). They worked exclusively with Escherichia coli (prepared as a lactose broth culture) and studied the effect of potassium permanganate at doses of 1 to 16 mg/liter at pH values of 5.9, 7.4, and 9.2, temperatures of 0°C and 20°C, and contact times of 4 to 120 min. They concluded that bacterial inactivation was relatively ineffective, but slightly better at the higher temperature, and that increasing pH decreased disinfection rates. Table II-22 summarizes in tabular form their data, which were presented originally in graphic fashion.

TABLE II-22

Concentrations and Contact Times Necessary for Potassium Permanganate to Effect 99% Inactivation of a 48-Hr Escherichia coli Lactose Broth Culture.

At least two patent applications involve the use of potassium permanganate as a disinfectant for water in swimming pools (Heuston, 1972; Seidel, 1973). Seidel's patent covered use of a concentration of 0.1 to 0.2 mg/liter in a recirculating system including filtration through quartz gravel. She claimed removal of bacteria and algae, but the specific merit of potassium permanganate in her system cannot be determined from the patent application. Heuston (1972) proposed a tablet containing 0.001 g of potassium permanganate per 0.2 g tablet to give a dose of 1 mg/liter. Because the tablet includes potassium iodide (KI) and other oxidizing agents, it is impossible to assess the specific killing action of potassium permanganate.

A number of studies on inactivation by potassium permanganate have been reported. However, these dealt largely with the mode of action of potassium permanganate in viral inactivation (Lund, 1963, 1966) or at efforts to control human diseases (Peretts et al., 1960; Schultz and Robinson, 1942; Wagner, 1951), animal diseases (Derbyshire and Arkell, 1971), or plant diseases (Eskarous and Habib, 1972; Hughes and Steindl, 1955). They provide few clear-cut quantitative data on biocidal activity of the compound.

Information is lacking on the effect of potassium permanganate on protozoans and helminths in water.

Mechanism of Action

The mechanism of action of potassium permanganate has not been definitively identified. From the studies that have been conducted (Lund, 1963, 1966) it may be presumed that it exerts its disinfection activities by oxidizing compounds that are involved in essential cellular functions. However, Lund (1963) questioned the suggestion that the oxidative process was the mechanism for poliovirus inactivation, but was unable to exclude it as a possibility.

Conclusions

The relatively high cost, ineffective bactericidal action, and aesthetic unsuitability of maintaining a residual in the distribution system make potassium permanganate a generally unsatisfactory disinfectant for drinking water.

Research Recommendations

Additional studies on potassium permanganate are hardly warranted, but in the unlikely event that further research on its biocidal activities is conducted, it would be desirable to study the effect of organic material on its disinfection and the effect of the compound on organisms other than E. coli. To relate disinfection efficiency to more conventional practices, parallel studies using chlorine should be conducted.

Production and Application

In water treatment, silver has been applied principally by dissolving the metal or by incorporating a silver compound in a filter medium, often an activated carbon filter. Romans (1954) has described older processes and patents. Davies (1976) reviewed processes for treating swimming pool water in which silver was added as a soluble salt and the silver ions were kept oxidized by the addition of persulfate; metallic silver was deposited on an activated carbon filter; silver was released from solid silver electrodes, which were alternately made anodic and cathodic; silver was released from an activated carbon filter containing silver and supplemented by passing the water through anodic silver screens; and silver was released from a pair of silver-copper alloy electrodes by applying a low reversing voltage.

During the past 6 yr, dozens of patents have been issued in Germany, Spain, South Africa, India, and the United States for devices or systems for adding silver to home drinking water systems or swimming pools. The home systems are most frequently combinations of activated carbon to remove tastes and odors and silver to prevent bacterial growth on the filter.

Because of the low solubility of silver, the dose is usually less than 50 µg/liter. Presumably this is reduced rapidly because of the adsorption of silver to surfaces, but there is no information on the silver residual in treated water. However, the maximum contaminant level (MCL) is 0.05 mg/liter (U.S. Environmental Protection Agency, 1975).

Analytical Methods

Until recently there have been no satisfactory techniques for measuring silver at the µg/liter level. Using a dithizone colorimetric method, the minimum detectable quantity of silver is 200 µg/liter (Standard Methods, 1976). Using an atomic absorption spectrophotometric method, the detection limit is 10 µg/liter (U.S. Environmental Protection Agency, 1974), and, if a heated graphite furnace is used, the detection limit is reduced to 0.005 µg/liter (Rattonetti, 1974).

Biocidal Activity

One of the earliest studies was conducted by Just and Szniolis (1936), who found that 100 µg of silver per liter disinfected water within 3 to 4 hr, that silver nitrate and metallic silver were equally effective, and that histopathological changes occurred in rats that were given water containing 400-1,000 µg silver/liter for 100 days. Other systematic studies of silver disinfection have been conducted by Renn and Chesney (1953-1956), Wuhrmann and Zobrist (1958), and Chambers and Proctor (1960). These studies have been summarized by Woodward (1963). Table II-23, adapted from Wuhrmann and Zobrist (1958), shows that in the concentrations used, silver acted slowly, but that there were increases in the rate of bacterial inactivation with increasing temperature and pH. Chambers and Proctor (1960) obtained similar results, but the inactivation rates of Renn and Chesney (1953-1956), who used comparable concentrations of silver, were faster.

TABLE II-23

Concentrations and Contact Times Necessary for Silver to Effect 99.9% Inactivation of Escherichia coli in 0.005 M Phosphate Buffer.

Both Wuhrmann and Zobrist (1958) and Chambers and Proctor (1960) found that the presence of phosphate (at 60 mg/liter) slowed bacterial inactivation. Because phosphate is typically found at much lower concentrations in drinking water, this observation has little practical significance, but it may account for disparate results in some laboratory studies.

Increasing water hardness slowed bacterial inactivation. According to Wuhrmann and Zobrist, an increase of 3 min was required to achieve 99.9% bacterial inactivation (at 20°C and pH 7.0) for each 10 mg/liter increase in hardness. Likewise, chlorides interfered with the action of silver. At 10 mg/liter, they increased the inactivation time by 25% and at 100 mg/liter, by 70%.

The source of silver seemed to be irrelevant to the inactivation rate because silver either added as silver nitrate or dissolved from metallic silver gave approximately equal results.

In most studies of the disinfection of water with silver, Escherichia coli has been the test organism. Wuhrmann and Zobrist also tested a Salmonella species and found it to be at least as sensitive as E. coli. Yasinskii and Kuznetsova (1973) observed that 90 µg/liter inactivated Vibrio comma (1 × 10⁶ cells/ml) in 30 min or 22.5 µg/liter in 60 min.

Fair (1948) and Harrison (1947) suggested that silver salts were ineffective against cysts of Entamoeba histolytica, but the data provided in these reports were limited. By contrast, Newton and Jones (1949) observed that electrolytically produced silver in tap water gave greater than 99% cyst inactivation in 1 hr and highly variable residual concentrations that ranged between 17 and 33 mg/liter at pH's of 9.0 to 9.8. When silver nitrate in distilled water at 0.5 mg/liter was used, 4 to 6 hr were required for a 90% to 99% inactivation. At 5 mg/liter, a > 99% inactivation occurred in 3 hr, and at 30 mg/liter, a > 99% inactivation occurred in I to 2 hr.

Chang and Baxter (1955) used 150 mg/liter of silver nitrate (95 mg/liter Ag⁺) with contact lines of 1 to 6 min. They concluded that cysticidal activity was only moderate compared to that of iodine. Apparently, there are no data on silver as a virucidal agent in water except for one study by Lund (1963), who decreased poliovirus activity by 2.5 logs in 4 hr with a concentration of approximately 68 mg/liter.

A good general review of silver and its compounds and a more thorough medical examination of their applications was published by Grier (1977), who concluded that there would be a ''significant place for silver compounds in the prevention and treatment of at least some bacterial diseases."

Mechanism of Action

Romans (1954) summarized the mechanism of so-called oligodynamic activity: "However, there is much difference of opinion as to the form of the active principle, as to the mechanism of its action, and the value of the results obtained. Some authorities believe that the active form is a positively charged ion, some think it is a complex ion, others a salt and still others think it acts by formation of proteinates or merely as a catalyst. These are only a few of the ideas that have been expressed with formidable experimental support."

Chang (1970) attributed a direct action to silver in the nonreversible formation of silver-sulfhydryl complexes that could not function as hydrogen carriers:

(30)

(31)

He considered silver to be bacteriostatic as well as bactericidal. This would explain the relatively long contact times that are required for antibacterial activity at the concentrations that are normally used.

Zimmerman (1952) demonstrated that low concentrations of silver do not enter the cell but are adsorbed onto the bacterial surface just as silver tends to be adsorbed on other surfaces (Chambers and Proctor, 1960). According to Chang (1970), the adsorbed silver ions must immobilize the dehydrogenation process because bacterial respiration takes place at the cell surface membrane.

Conclusions

As a disinfectant, silver may be applicable to home treatment systems and swimming pools. These applications may be more effective in keeping filters free from bacteria than in actually inactivating organisms that are suspended in the water. This is because of the relatively low rate of solution of silver in water and its low biocidal activities.

Woodward (1963) ably summarized the situation: "The high cost of silver will limit it to specialty uses. The fact that silver does not impart taste, odor, or color to water makes it attractive for use. Its slow bactericidal action, although a disadvantage in some situations, may be an advantage in others, particularly where water is stored for long times before use, as on shipboard. Until some of the uncertainties regarding silver are resolved, it would be prudent to use it as a drinking water disinfectant only in situations where substantial factors of safety can be provided and where the bactericidal effectiveness of the procedure can be monitored."

Silver and its compounds are weak, costly disinfectants that are unsuitable for use in municipal drinking water supplies. To achieve acceptable disinfection in a reasonable time would require concentrations exceeding the MCL of 0.05 mg/liter.

Research Recommendations

Now that adequate chemical analytical techniques are available for measuring low concentrations of silver, it would be desirable to conduct disinfection studies in which both dose and residual are measured accurately. To relate disinfection efficiency to more conventional practices, parallel studies using chlorine should be conducted.

Dose

The dose, D, of electromagnetic radiation that is applied to a solution is commonly measured as the intensity of radiant energy input, I₀, at the lamp surface or at some given distance from the lamp (I₀ being expressed as µW/cm²), multiplied by the time of exposure, t, in seconds (Luckiesh and Holladay, 1944):

(32)

Another measurement of dose is chemical actinometry (Calvert and Pitts, 1966). In this method, a photochemical reaction with a known quantum yield is used to measure the intensity (quanta per second) of light that is absorbed by the actinometry solution. Knowing the time of exposure, the volume of solution, and the volume of sample analyzed, the moles of photochemical reaction per unit of time can be related through the quantum yield to the Einsteins or moles of photons per second that are absorbed by the fluid. From the wavelength of radiation, l, and the area of lamp surface, the average intensity at the lamp surface can be determined.

(33)

Neither of these measurements of energy input, I₀, per unit area of lamp exposed considers the change in intensity through the depth of exposed fluid and the solid angle over which this energy interacts. Recently, attempts have been made to correct the decrease in intensity with depth in studies of wastewater disinfection where this problem is acute (Roeber and Hoot, 1975; Severin, 1978; Venosa et al., 1978). These studies have shown that the usual water quality indices, such as chemical oxygen demand (COD), color, and turbidity, do not adequately predict the loss of intensity through the solution. The direct measurement of UV transmittance has been successful. In spite of these difficulties, Huff et al. (1965) suggested that-color at a maximum level of 5 units, iron at 3.7 mg/liter, and turbidity up to 5 units did not decrease treatment efficiency below acceptable limits in a unit with a maximum water depth of approximately 7.5 cm.

Biocidal Dose

The biocidal dose of UV energy consists of the intensity of UV energy that is absorbed at the reactive site within the microbe over the time of interaction. The biocidal dose is then a function of the energy input from the UV source into the solution, dispersion of the energy as a function of distance from the source, the depth of the fluid between the organisms and the source as well as its absorptivity, and, finally, the losses and reflection of UV light within the contactor. All of these factors determine the actual intensity of radiation that is available to the microorganism at any one point within the contactor (Childs, 1962; Luckiesh and Holladay, 1944; Luckiesh et al., 1944).

Quite early in the study of the batericidal action of UV light, Gates (1929) found that the relation between the intensity of incident energy and time required for bacterial destruction were not of equal importance in determining experimental disinfection. However, Hoather (1955) found that when one corrects for the transmitted intensity of radiation, taking into consideration the absorptivity of the water and dispersion as a function of distance from the source, the required time of exposure is inversely proportional to the calculated intensity of radiation penetrating the water for a given degree of inactivation. Finally, Oliver and Cosgrove (1975) used chemical actinometry and a pulsed laser to show that their coliform and streptococcal inactivation in wastewater depended only on the total dose that is delivered and not on the dose rate or intensity of light that impinged on the sample. Thus, the same number of µW s/cm² provided the same disinfection regardless of the time and intensity that were used to produce that dosage.

Bacterial spores have also been studied with chemical actinometry. For certain stocks, the rate at which the energy was deposited affected the degree of response (Powers et al., 1974). Powers and his colleagues observed that the effect of the dose rate was not due to geometric factors or intensity variations with depth and solution absorptivity. This explains many earlier results. They did find that dependence on dose rate was seen only for some of their stocks of spores, it was not as dramatic with monochromatic-filtered light of a wavelength of 253.7 nm, and it was observed only at very low doses. The explanations usually given for these low-dose effects are photochemical back reactions, enzymatic repair, and photoreactivation mechanisms (Deering and Setlow, 1963; Harm, 1968).

Residual

UV radiation produces no residual. Therefore, monitoring and control of disinfection efficiency are more difficult for UV than for chemical disinfectants.

However, this is not a major problem, because disinfection can be controlled by adjusting the contact time and the UV energy that is transmitted through the solution. The major disadvantage is the lack of a tracer for ensuring the integrity of the distribution system.

Biocidal Activity

UV disinfection follows Chick's Law of kinetics (Luckiesh and Holladay, 1944):

(34)

where N is the number surviving at time t of an original population N₀ that has been exposed to an intensity I. Q is the dose for 1 log survival, one lethal unit exposure, or a lethe.

The variables discussed above interfere with the accurate determination of the average intensity and time of exposure in a practical UV contactor. Consequently, there is considerable disagreement in the literature concerning the absolute magnitude of Q for the microorganisms that have been studied.

Of the Gram-negative bacteria, Escherichia coli is consistently more resistant to disinfection by UV than are the Salmonella and Shigella species that have been studied (Cortelyou et al., 1954; Kawabata and Harada, 1959). The vegetative forms of the Gram-positive bacteria were more resistant than E. coli. Kawabata and Harada found Streptococcus faecalis nearly 3 times more resistant than E. coli, while Bacillus subtilis was 4 times more resistant. The spores of B. subtilis were 6 times more resistant than E. coli.

Luckiesh and Holladay (1944) reported that a dosage of 2,400 µW · s/cm² at a wavelength of 254 nm produced one lethe of disinfection of E. coli. Huff et al. (1965) studied a two-lamp shipboard disinfection unit with an approximately 75-cm long, 20-cm diam. unbaffled stainless steel cylinder. Their dosages were generally quite high, ranging from 3,000 to 11,000 µW ··s/cm² at a maximum water depth of approximately 19 cm. At the lowest dosage, there was a 0.02% to 0.04% survival of E. coli, which, assuming Chick's Law, gives a dosage of less than 400 µW · s/cm²/lethe. They also found approximately 2 logs less spore inactivation with the same dose as required for vegetative cells. They reported more than 4 logs of inactivation of polio-, echo-, and coxsackievirus with 4,000 µW · s/cm². Based on these results, the U.S. Department of Health, Education, and Welfare issued a policy statement on April 1, 1966, stating criteria for the acceptability of UV disinfecting units as a minimum dosage of 16,000 µW · s/cm² with a maximum water depth of approximately 7.5 cm.

Morris (1972) has compared the dose of 254-nm UV energy required to inactivate a range of microorganisms suspended in droplets of buffer solution on an aluminum surface. Their results are shown in Table II-24 expressed as µW · s/cm²/lethe.

TABLE II-24

Ultraviolet Energy Necessary to Inactivate Various Organisms.

Required dosages reported by Morris (1972) are similar in magnitude to those found by Huff et al. (1965). Both Huffs values and those of Morris (1972) are lower than those of Luckiesh and Holladay (1944), partly because of the additional dose of UV that was provided by reflections from the contactor and aluminum surface, which produce an estimated dosage approximately twice the values given. No studies of the action of UV light on parasitic helminths and protozoa in the treatment of drinking water appear to have been reported.

Mechanism of Action

Inactivation by UV light is believed to act through the direct absorption of UV energy by the microorganism, causing a molecular rearrangement of one or more of the biochemical components that are essential to the organism's functioning. The major site of UV absorption in microorganisms is the purine and pyrimidine components of the nucleoproteins. Since the relative efficiency of disinfection by UV energy as a function of wavelength follows the absorption spectrum of these chromophore groups and because the first law of photochemistry states that the light that is absorbed by a molecule can produce photochemical change, the mechanism of action of UV probably occurs through the progressive biochemical change that is produced primarily in the nucleoproteins. Witkin (1976) has reviewed both UV mutagenesis and inducible DNA repair. The major specific mechanisms that have been suggested for UV damage involve the reversible formation of pyrimidine hydrates and pyrimidine dimers. Breaks in the bonding structure, known as nicking, also occur. The older literature has been reviewed by Reddish (1957).

The repair of damaged nucleoprotein with light of a wavelength longer than that of the damaging radiation is commonly referred to as photoreactivation (Witkin, 1976). This process, originally identified by Kelner (1949), has been studied extensively. It occurs in the visible wavelength range from 300 to 550 nm (Jagger, 1960; Kelner, 1949). In addition, repair of nucleoproteins that have been damaged by UV radiation can also occur in the dark. The mechanisms of repair of damage to DNA caused by UV radiation are generally thought to involve induced enzymes (Witkin, 1976), but repair has also been observed in inactive systems for which photochemical back reactions have been suggested (Powers et al., 1974).

Conclusions

UV light produces no residual. Therefore, if a residual is desired, another disinfectant must be used. Current technology is limited to the use of UV light in small systems. Requirements for equipment maintenance limits widespread adoption of UV light for drinking water disinfection.

Conclusions

Table II-28 provides the current status of theoretically possible methods for disinfecting drinking water. To derive the conclusions in the table, the specific biocidal activity data summarized in Table II-27 were considered as well as information (or lack of it) on the practical application and reliability of the methods.

TABLE II-28

Status of Possible Methods for Drinking Water Disinfection.

Chlorination, ozonization, and the use of chlorine dioxide come closest to meeting the criteria desired. At this time none of the other possibilities considered can substitute for techniques presently used to disinfect drinking water.

The ultimate choice between methods will require weighing the characteristics detailed in this evaluation against the nature of byproducts to be expected from the use of the particular method and the potential toxicities of the by-products. Evaluations of these aspects of drinking water disinfection, also carried out by the Safe Drinking Water Committee, are reported in parallel to this study and should be consulted before final conclusions are drawn.