Environ. Sci. Technol. 2008, 42, 2582–2589
Stoichiometry of Coagulation Revisited J. Y. SHIN,† R. F. SPINETTE,‡ AND C . R . O ’ M E L I A * ,‡ Geography and Environmental Engineering, The Johns Hopkins University, Baltimore, Maryland 21218
Received June 22, 2007. Revised manuscript received November 8, 2007. Accepted December 14, 2007.
The roles of particles and natural organic matter (NOM) in determining coagulant (alum) doses in potable water treatment were investigated at two pH conditions (6 and 7). The concentrations of NOM and colloidal silica particles in raw water were systematically varied separately and in combination, and the impacts of these two classes of contaminants on the minimum effective alum doses were investigated using observations of turbidity and dissolved organic carbon (DOC) in laboratory jar tests. At both pHs, coagulant requirements for the removal of these contaminants by sedimentation and filtration were dominated by the DOC concentration in the raw water. The presence of low NOM concentrations (0.75–1.5 mgofC/L)decreasedtheminimumeffectivealumdosedramatically for waters low in silica particles, possibly by promoting the precipitation of aluminum hydroxide and/or Al-NOM solids, whose removal would otherwise be limited by low collision opportunities. Strong stoichiometric relationships were observed between DOC and coagulant demand at both pHs regardless of silica particle concentration. Silica contributed to coagulant demand only at very high particle concentrations.
Introduction Coagulation by hydrolyzing metal salts has been widely and extensively investigated for the removal of turbidity and natural organic matter (NOM) in potable water treatment. In practice, the concentration of particles in water has long been considered to be the dominating, if not sole, factor that determines the appropriate chemical conditions for coagulation and filtration. Other studies, however, have also recognized the importance of natural organic matter in water treatment and its role in determining the conditions for effective design and operation of coagulation, sedimentation, and filtration systems. The research progress in the area of coagulation of particles and NOM benefited from two studies conducted independently in the 1960s. Both of these studies addressed the experimental and theoretical aspects of the stoichiometry of coagulation by metal salt coagulants. The first study by Black et al. (1) investigated the coagulation of color causing materials, i.e., NOM, by ferric sulfate and established the existence of a stoichiometric relationship between the concentration of color and the corresponding coagulant dose required for its removal. The second study by Stumm and O’Melia (2) focused on the coagulation of colloidal silica by ferric chloride in the absence of NOM and * Corresponding author fax: 410-516-8996; e-mail: omelia@ jhu.edu. † Washington Suburban Sanitary Commission, Laurel, MD. ‡ The Johns Hopkins University. 2582
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 7, 2008
suggested a conceptual model relating the effects of particle concentration on the mechanisms of coagulation and on the coagulant demand required for the removal of turbidity. While research in the areas of coagulation of particles and NOM has led to significant advances over the past 4 decades, the majority of these studies, including the two stated above, has focused separately on the reactions between the coagulant and only one of these contaminants. The relative importance of colloidal particles and NOM in controlling coagulant requirements when they are both present in water has usually been overlooked. For a system consisting of multiple contaminants such as inorganic colloidal particles and NOM to be treated by iron(III) or aluminum salts, the overall reaction among these constituents can be considered to involve competing reactions of hydroxide and other organic/inorganic ligands for complexation with free metal ions and their hydrolysis products. The pathways by which these contaminants are removed therefore depend on numerous factors such as the speciation of hydrolysis products, the presence and the relative reactivity of inorganic/organic constituents in complexation with aluminum or Fe(III) species, the kinetics of hydrolysis and of the reactions between the coagulant and other ligands, and the rates of mass transport among these constituents. In this context, it was hypothesized that the relative abundance of colloidal particles and NOM will have significant influence on (a) the mechanisms for the removal of these contaminants by hydrolyzing metal salts, (b) the control of coagulant demand, and (c) the design, operation, and performance of solid–liquid separation processes following coagulation. In this study, the relationships between the coagulant and the concentrations of colloidal particles and NOM were investigated by determining the minimum effective coagulant doses when they are present in mixtures. Using these observations, this study aimed at assessing the mechanisms responsible for the removal of these constituents by aluminum salts and determining the controlling factors for the coagulant demand. Implications were made for the performance and operation of subsequent solid–liquid separation processes. Coagulation of Colloidal Particles. Removal of negatively charged colloids can be accomplished by coagulation using ferric or aluminum salts which, upon addition into water, undergo a series of hydrolysis reactions. Positively charged Al or Fe(III) species produced from hydrolysis reactions act as effective coagulants for the removal of negatively charged particulate matter. Two removal mechanisms have been identified: (a) charge neutralization of negatively charged particles by positively charged metal hydrolysis species followed by aggregation of the destabilized particles and (b) formation of flocs composed of metal hydroxide precipitates accompanied or followed by sweep flocculation of colloidal particles. The concentration of colloids is a critical factor that determines the predominant mechanism for removal. Two different coagulation regimes were recognized depending on the concentration of colloidal particles (2). High colloid concentrations provide sufficient contact opportunities for aggregation of destabilized particles and hence require coagulant doses minimally sufficient to induce charge neutralization. In this case, the interaction between the cationic metal hydrolysis species and negatively charged colloids at the effective coagulant dose is stoichiometric. Once 10.1021/es071536o CCC: $40.75
2008 American Chemical Society
Published on Web 02/20/2008
destabilized, these colloids can be aggregated if adequate interparticle contacts are provided, usually by means of mixing. In contrast, at low particle concentrations, the contact opportunities between particles are limited by low solid concentrations. Under these conditions, removal must be achieved by sweep flocculation, which typically requires coagulant doses greater than those that would be required for charge neutralization. In this regime, the coagulant dose required for sweep flocculation of particles is not a “stoichiometric” function of particle concentration but is mostly determined by chemical conditions that influence metal hydroxide precipitation including the coagulant dose, pH, alkalinity, and temperature. Coagulation of Natural Organic Matter. Edzwald (3) suggested that the following four reactions are involved in the removal of NOM by aluminum coagulants: (a) complexation of soluble aluminum species with NOM, particularly reactions with humic and fulvic acids that act as strong organic ligands for aluminum; (b) aluminum neutralization reactions by inorganic ligands, such as OH-, in competition with organic ligands; (c) direct precipitation of Al-NOM particles following complexation of Al species with NOM molecules; and (d) adsorption of NOM or Al-NOM complexes on amorphous Al(OH)3(s) which precipitated after coagulant addition. Regardless of the mechanisms controlling NOM removal, the amount of aluminum salt required to bring about removal of NOM is stoichiometrically related to NOM concentration (1, 4). Complexation-precipitation occurs in the presence of soluble cationic aluminum hydrolysis species. These soluble monomeric or polymeric species react with anionic functional groups on NOM to precipitate as an aluminum-NOM complex (5). Adsorption onto aluminum hydroxide precipitates (Al(OH)3(s)) is an important pathway for the removal of NOM under conditions favoring Al(OH)3(s) precipitation (6–8). Adsorption occurs by surface complexation or ligand exchange (9) of NOM molecules with surface groups on Al(OH)3(s). The extent of adsorption depends on the type of functional groups on the NOM and the chemical conditions under which Al(OH)3(s) precipitation occurs, such as pH. A third situation is possible when the solution chemistry, primarily the pH and temperature of coagulation, allows the presence of both soluble polymeric aluminum species and solid metal hydroxide precipitates simultaneously. Polynuclear aluminum species react with NOM and neutralize or substantially reduce the negative charge on the NOM, promoting its adsorption onto Al(OH)3(s) (10). The removal of NOM by this combined mechanism is considered more efficient than by adsorption or complexation-precipitation alone.
Materials and Methods Model Waters. The concentrations of colloidal silica particles and dissolved organic carbon (DOC) were varied to investigate the effects of their concentrations on coagulant demand separately and in combination. Silica concentration was varied from 0 to 200 mg/L in the presence of added DOC and from 0 to 600 mg/L without DOC. The concentration of DOC was varied from 0 to 6 mg of C/L. In all jar test studies, 10-3 M NaCl was added as background electrolyte and NaHCO3 was added as buffer (10-4 M at pH 7 and 2 × 10-5 M at pH 6). After the addition of colloidal particles and NOM, the water was equilibrated for 18 ( 2 h to allow NOM molecules to partition onto silica particle surfaces. Water collected from the Great Dismal Swamp National Wildlife Refuge in southeastern Virginia (referred to herein as Dismal Swamp water) was used as a source of natural organic matter. The DOC and specific ultraviolet light absorbance (SUVA) of this water were 110.9 mg of C/L and
5.36 L/(mg · m), respectively. The charge of the NOM as a function of pH was determined using titrations with 0.1 N HCl and 0.1 N NaOH and the measurement of major metals and anions in the NOM solution. Titrations were made using two dosimetry instruments and a pH meter interfaced to a computer. Detailed characteristics of the Dismal Swamp water are listed in Supporting Information Table S1. Colloidal silica particles (Snowtex-ZL, Nissan Chemical Industries, Tokyo, Japan) were used as a source of turbidity in the model waters. Dynamic light scattering using photon correlation spectroscopy indicated that the particles have a hydrodynamic diameter of 129 nm (Brookhaven Instruments Co., BI-200SM and BI-9000AT). Jar Test Procedure. Jar tests were conducted using model raw waters prepared using colloidal silica particles and NOM. All jar tests were conducted at pHs 6 and 7. Aluminum sulfate (alum, Al2(SO4)3 · 18H2O) was used as the sole coagulant. Nine to fifteen jars containing 150 mL aliquots of model raw water were used for each jar test. Alum was added at increments of 0.2-2 mg/L to obtain accurate estimates of the minimum effective alum dose (MEAD). pH was monitored during alum addition and immediately adjusted to the desired pH using small amounts of dilute (0.1-0.001 N) NaOH and HCl. Samples were then mixed at 100 rpm for 2 min to provide additional rapid mixing, after which the speed was reduced to 25 rpm for 30 min to provide flocculation. The paddles were then removed, and the contents of the jars were allowed to settle quiescently for 60 min. After sedimentation, a 50 mL sample of the settled water from each jar was gently withdrawn from just below the water surface using a wide mouth pipet and analyzed for pH and settled turbidity. The samples were then filtered through a 1.2 µm glass fiber filter (Whatman GF/C, prewashed with distilled/deionized water) and analyzed for filtered turbidity and DOC. MEADs were determined by plotting settled turbidity, filtered turbidity, and DOC as functions of alum dose. Examples of these plots at both pHs are provided in Figure 1. The alum dose where the most dramatic decrease in settled turbidity occurs was defined as MEAD for settled turbidity. MEADs for filtered turbidity and DOC were defined in a similar manner. In most cases, as shown in Figure 1a, further increasing the alum dose beyond the MEAD neither improved nor impaired the removal of turbidity and DOC. In some cases, especially without the presence of NOM at pH 7 and at low DOC at pH 6 (Figure 1b), overdosing of alum resulted in restabilization of the suspension and deterioration in settled water quality, followed by a gradual decrease in settled turbidity at high alum doses. In such cases, the doses where the initial decreases in turbidity and DOC occurred were defined as the MEADs. The results shown in Figure 1b are consistent with the classical view of particle coagulation: (a) low doses of aluminum are not effective, (b) adding more alum to a minimum effective dose provides charge neutralization and flocculation is rapid enough to provide settleable flocs, (c) increasing the dose further reverses the charge on the silica particles and produces a stable suspension, and (d) further increase in dose produces a substantial quantity of Al(OH)3(s) that enmeshes positively charged particles in a settleable aggregate/sweep floc.
Experimental Results Coagulation of Colloidal Particles without NOM. The changes in the MEAD for the reduction of settled turbidity and filtered turbidity were observed as a function of silica particle concentration in the absence of NOM. The results are shown in Figure 2. At both pHs, the correlation between silica concentration and the MEADs for settled turbidity showed two distinctively different patterns depending on the range of silica particle concentrations. First, at low silica particle concentration, the MEAD for settled turbidity was VOL. 42, NO. 7, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2583
FIGURE 1. Definitions of minimum effective alum doses (MEADs) for the removal of settled turbidity, filtered turbidity, and DOC: (a) pH 7 (DOC ) 0.75 mg of C/L; silica ) 30 mg/L); (b) pH 6 (DOC ) 1.35 mg of C/L; silica ) 15 mg/L).
FIGURE 2. Minimum effective alum doses for settled and filtered turbidity as a function of silica particle concentration (DOC ) 0 mg of C/L). high and was not affected by silica concentration. Second, at high silica particle concentration, the MEAD increased in direct proportion with increasing silica particle concentration. At both pHs, the stoichiometric relationship between silica and the MEAD based on settled turbidity started to occur at a silica concentration of 15 mg/L, which indicates that interparticle contact opportunities were satisfied at this concentration. The general profile of MEAD for settled turbidity as a function of silica particle concentration is consistent with the mechanisms proposed by Stumm and O’Melia (2). In the low silica regime (silica < 15 mg/L) the removal of silica particles is primarily achieved by sweep flocculation. For these waters, the contact opportunities for effective flocculation and settling are limited by low particle concentrations, resulting in alum doses far greater than those which 2584
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 7, 2008
FIGURE 3. Minimum effective alum doses for the removal of settled turbidity at varying NOM: (a) pH 7; (b) pH 6. would be required for charge neutralization. At higher silica particle concentrations where interparticle collisions are not limited, the high dose required for sweep flocculation of low silica waters is replaced by the lower alum dose required to bring about charge neutralization of the negatively charged surface of the silica particles. Role of NOM in Collision-Limited Systems. The presence of NOM in model waters with low particle concentrations affected the MEADs for settled and filtered turbidity appreciably. Addition of very small amounts of NOM to these waters significantly lowered the MEADs for settled turbidity, in contrast to waters without NOM that required high coagulant doses for removal by sweep flocculation. The MEADs for the removal of settled turbidity are plotted in Figure 3 as a function of initial DOC concentration. At pH 7, the MEADs based on settled turbidity for waters containing low silica particle concentrations (
