Anal. Chem. 2001, 73, 4815-4820
Characterization of Submicrometer Aqueous Iron(III) Colloids Formed in the Presence of Phosphate by Sedimentation Field Flow Fractionation with Multiangle Laser Light Scattering Detection Matthew L. Magnuson,* Darren A. Lytle, Christina M. Frietch, and Catherine A. Kelty
Office of Research and Development, National Risk Management Research Laboratory, Water Supply and Water Resources Division, Treatment Technology Evaluation Branch, United States Environmental Protection Agency, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268
Iron colloids play a major role in the water chemistry of natural watersheds and of engineered drinking water distribution systems. Phosphate is frequently added to distribution systems to control corrosion problems, so iron-phosphate colloids may form through reaction of iron in water pipes. In this study, sedimentation field flow fractionation (SdFFF) is coupled on-line with multiangle laser light scattering (MALLS) detection to characterize these iron colloids formed following the oxygenation of iron(II) in the presence of phosphate. The SdFFF-MALLS data were used to calculate the hydrodynamic diameter, density, and particle size distribution of these submicrometer colloids. The system was first verified with standard polystyrene beads, and the results compared well with certified values. Iron(III) colloids were formed in the presence of phosphate at a variety of pH conditions. The colloids’ hydrodynamic diameters, which ranged from 218 ( 3 (pH 7) to 208 ( 4 nm (pH 10), did not change significantly within the 95% confidence limit. Colloid density did increase significantly from 1.12 ( 0.01 (pH 7) to 1.36 ( 0.02 g/mL (pH 10). Iron(III) colloids formed at pH 10 in the presence of phosphate were compared to iron(III) colloids formed without phosphate and also to iron(III) colloids formed with silicate. The iron(III) colloids formed without phosphate or silicate were 0.46 g/mL more dense than any other colloids and were >6 times more narrowly distributed than the other colloids. The data suggest competitive incorporation of respective anions into the colloid during formation. The presence of iron-containing colloids and particles in water is widely understood to play a role in many natural processes. In drinking water distribution systems, they are present in the source water and may be formed from corrosion of iron pipes, resulting in consumer complaints. Phosphate is frequently added to control lead and copper corrosion and may provide protection against iron * Corresponding author. Phone: 513-569-7321. Fax: 513-569-7658. E-mail:
[email protected]. 10.1021/ac010702m Not subject to U.S. Copyright. Publ. 2001 Am. Chem. Soc.
Published on Web 09/18/2001
release. The nature of iron-phosphate colloids has been the subject of research at both the fundamental and engineering level.1-6 Understanding the nature and properties of iron(III) colloids formed following the oxidation of iron(II) in the presence of phosphate may lead to better corrosion control and a reduction of consumer complaints. Such work may also have implications for flushing distribution systems to remove colloidal corrosion products. Flushing needs to be carefully controlled to balance efficiency versus cost, so models have been developed to optimize flushing, particularly in distribution system dead ends.7 Accurate modeling of flushing regimes involves knowledge of certain physical properties, such as hydrodynamic diameter, solution conformation of the colloid, density, and particle size distribution. This basic information is lacking in the literature, most likely due to the technical challenges in the measurement of these parameters for submicrometer colloids. Accordingly, current models rely on empirical estimates7 for the calculations, introducing uncertainty into the modeling results. The formation of iron colloids is a complex issue, involving oxidation and hydrolysis of the elemental iron through a series of steps to form primary nucleation particles (PNPs). The PNPs may dissolve, grow into stable colloids, or, in the case of unstable colloids, flocculate. PNPs and stable colloids are typically in the submicrometer size range and require sophisticated analytical techniques for characterization, while flocculate is much larger, visible to the unaided eye. The PNPs of stable iron(III) colloids has been studied by preparing iron(III) hydroxide colloids in the presence of tris(hydroxymethyl)aminomethane base with characterization using photon correlation light scattering (PCS)3 and (1) Gshwend, P. M.; Reynolds, M. D. J. Contam. Hydrol. 1987, 1, 309-327. (2) Buffle, J.; De Vitre, R. R.; Perret, D.; Leppard, G. G. Geochim. Cosmochim. Acta 1989, 53, 399-408. (3) Von Gunten, U.; Schneider, W. J. Colloid Interface Sci. 1991, 145, 127139. (4) Leppard, G. G. Analyst 1992, 117, 595-603. (5) He, Q. H.; Leppard, G. G.; Paige, C. R.; Snodgrass, W. J. Water Res. 1996, 30, 1345-1352. (6) Taillefert, M.; Bono, A. B.; Luther, G. W., III. Environ. Sci. Technol. 2000, 34, 2169-2177. (7) Friedman, M.; Sherwin, C.; Hiltebrand, D. Proceedings on CD-ROM, Water Quality Technology Conference; Salt Lake City, UT, November 5-9, 2000.
Analytical Chemistry, Vol. 73, No. 20, October 15, 2001 4815
electrovoltammetric methods.6 The PNPs, which have also been observed with electron microscopy,5 are in the 1-10-nm size range. Stable iron-containing colloids, with sizes intermediate between the PNPs and flocculate, have also been characterized by transmission electron microscopy (TEM).1,2,4,5 These reports have been with iron colloids prepared in a variety of methods in the laboratory3-6 or sampled from natural systems,1,2,4 which reduces systematic comparison of the results and elucidation of important physicochemical parameters. None of the conditions is similar to those expected in a drinking water distribution system. Accordingly, the purpose of this study is to characterize iron(III)-containing colloids in the laboratory under conditions representative of common practice in a drinking water distribution system. While both TEM1,2,4-6 and PCS3 have been used to characterize these colloids, fundamental aspects of their operational principles limit the data produced. For example, in TEM, the colloids must be fixed in some manner prior to analysis, complicating the analysis and interpretation since the fixing media may not actually be representative of aqueous solution. PCS has the advantage that the measurement can be performed in solution. Complications arise8 since PCS utilizes “dynamic light scattering”, which is subject to a variety of instrumental artifacts and experimental difficulties. (Even minute amounts of common laboratory dust can cause large error, since PCS cannot distinguish dust from the analyte.) As a consequence, PCS experiments do not reliably resolve particle sizes that differ by less than 50%, biasing estimates of particle size distribution. Accurate particle size distribution may be desirable because it may be significant for modeling purposes as well as for inferring the formation mechanism(s) of the colloids. Better characterization results when the colloids are first separated by size. Separation can also reduce the problem with dust and other contamination when PCS is used. High-resolution separation in the colloid size range is possible through the use of field flow fractionation (FFF), in which particles of different sizes elute from the instrument at different times in response to two opposed fields applied to the sample.9 There are several variations of field flow fractionators, depending on the nature of the applied fields, and they have found applications in environmental, polymer, colloidal, and macromolecular analysis.9 A variation known as sedimentation field flow fraction (SdFFF) has been combined with PCS to determine physical characteristics of polymer beads10 and also diesel soot particles.11 In these applications, fractions are collected at different elution times and analyzed off-line by PCS.10,11 However, since a sufficient volume needs to be collected for the off-line PCS analysis, a distribution of particle sizes exists within the sample, so the results contain an inherent bias. To provide on-line detection, and circumvent the difficulties of off-line PCS analysis, another light scattering technique known as multiangle laser light scattering (MALLS) can be used with separation techniques. An additional advantage of MALLS,12 “static light scattering” is employed, which, based on the MALLS calculations, can provide more realistic particle size distributions (8) Ostrowsky, N. Chem. Phys. Lipids 1993, 64, 45-56. (9) Giddings, J. C. Science 1993, 260, 1456-1465. (10) Caldwell, K. D.; Jones, H. K.; Giddings, J. C. Colloids Surf. 1986, 18, 123131. (11) Kim, W.-S.; Park, Y. H.; Shin, J. Y.; Lee, D. W. Anal. Chem. 1999, 71, 32653272. (12) Wyatt, P. J. Anal. Chim. Acta 1993, 272, 1-40.
4816
Analytical Chemistry, Vol. 73, No. 20, October 15, 2001
than PCS. Recently, MALLS was been combined with a technique called flow field flow fractionation (FlFFF) to characterize several standard polymer samples.13,14 Polymers of all sorts have been the subject of MALLS investigation, and MALLS experiments have been worked out in terms of polymer and polymer-related issues. Iron colloids and substances like themsreaction products of unknown compositionshistorically are not characterized by MALLS. It seemed reasonable, however, that these colloids should be amenable to MALLS, so this approach was investigated. In this study, SdFFF, which provides better resolution that FlFFF and also allows determination of density,9 is combined with MALLS to determine key parameters for hydrodynamic models of flushing of the iron-phosphate colloids formed at pH values and phosphate concentrations used in some drinking water distribution systems for corrosion control. The results from iron-phosphate colloids are compared with iron-hydroxide colloids and also iron-silicahydroxide colloids. THEORY Sedimentation Field Flow Fractionation. The theory of SdFFF has been extensively discussed elsewhere.9,15,16 It is useful to review a few salient points for the purpose of this investigation. Separation in SdFFF is achieved by imposing a gravitational field perpendicular onto particles entrained in a laminar flow. Precise mathematical relationships describe the interaction between the flow field and the gravitational field and the physical characteristics of the particles. The result is that particles elute from the SdFFF unit at different retention times, tr. In particular:9,15,16
tr ) π∆FwGd3Vo/36kTVf
(1)
where k is the Boltzmann constant, G is the gravitational constant, T is the temperature, w is the channel width, ∆F is the difference in density between the carrier liquid and the particle, and d is the diameter of the spherical particle. Vo is the void volume of the SdFFF channel, and Vf is the volumetric flow rate through the channel. If the particle diameter is known, ∆F can be calculated and vice versa. The intensity versus elution time plot is referred to as a fractogram. Multiangle Laser Light Scattering. The theory of operation of MALLS and the derivation of the mathematical relationships has been extensively discussed elsewhere.12 Several mathematical models relate the intensity of scattered light to the particle size, based on the geometry of the particle in solution. The geometry of the particle may be inferred by fitting the data to several models. The model with the smallest error is used for data analysis. The hydrodynamic diameter of a sphere determined by multiangle laser light scattering is given by
d)
x203〈R
〉
2 G
(2)
where 〈RG2〉 is the mean square “radius of gyration” calculated (13) Thielking, H.; Roessner, D.; Kulicke, W.-M. Anal. Chem. 1995, 67, 32293233. (14) Thielking, H.; Kulicke, W.-M. Anal. Chem. 1996, 68, 1169-1173. (15) Giddings, J. C.; Karaiskakis, G.; Caldwell, K. D.; Myers, M. N. J. Colloid Interface Sci. 1983, 92, 66-80. (16) Giddings, J. C. J. Chromatogr. 1976, 123, 3-16.
Figure 1. Diagram of the experimental setup for sedimentation field flow fractionation (SdFFF) with on-line multiangle laser light scattering (MALLS) detection.
from the light scattering results. In this manner, the hydrodynamic diameter is independent of the molecular weight. In principle, the molecular weight may be calculated from light scattering results,12 and the calculation of molecular weights by MALLS has proven appropriate for polymers.13,14 However, it is not clear that the wave front of the light is not altered by a potential lattice structure of the type of samples under study here, thereby prohibiting calculation of molecular weight. By measuring the diameter by MALLS (eq 2), it is possible to calculate the particle density via the SdFFF retention time (eq 1). In practice, these calculations are performed within the data processing software for the instrument by using the point of maximum intensity within the fractogram. Particle size distributions are also be directly calculated from the light scattering data based on fundamental relationships.17 EXPERIMENTAL SECTION Apparatus. Figure 1 diagrams the combined SdFFF-MALLS system used in these experiments. The SdFFF instrument was an S-101 particle/colloid fractionator, manufactured by FFFractionation LLC (Salt Lake City, UT). This unit has a 89.5-cm-long channel made out of Hastalloy C alloy. The channel thickness was 0.0254 cm, and the channel width was 2.0 cm. The pump used for the SdFFF system was a Dionex GMP2 dual-piston gradient pump (Sunnyvale, CA). The exact flow rate was determined gravimetrically with a stopwatch. An additional pulse dampener was used (Lo Pulse, Alltech Associates, Deerfield, IL) followed by an additional bubble trap (FFFractionation LLC). Pressure pulses and bubbles degrade system performance. An in-line solvent filter (Millipore, Bedford, MA) with a 100-µm membrane filter (Millipore VVLP02500) was place after the pump to filter the carrier, which reduces detector background. The tubing carrying the effluent from the SdFFF unit was directed into the MALLS detector, a Dawn Eos (Wyatt Technology, Santa Barbara, CA.) The Dawn Eos has a 685-nm, 30-mW laser and was equipped with a K5 flow-through cell. All instruments were calibrated according to the respective instruction manuals. Software supplied by the respective manufacturers was used for instrument operation and data analysis. Calculations within the Astra software for the Dawn Eos was supplemented with a “particles module”, supplied by the manufacturer. Preparation of Iron(III) Colloids. In the following discussion, iron(III) colloids formed in the presence of hydroxide and phosphate will be referred to as iron-phosphate colloids. Likewise, iron colloids formed in the presence of only hydroxide will be referred to as iron-hydroxide colloids, and colloids formed in the presence of hydroxide and silicate will be referred to as iron(17) Kerker, M.; Farone, W. A.; Matijevic, E. J. Opt. Soc. Am. 1963, 53, 758764.
silicate colloids. It should be noted, in this paper, some descriptions of these colloids, as is the convention, will refer to them as particles, e.g., in the term “particle size distribution”. Particle/colloid generation studies were conducted in a 1.2-L glass reaction vessel. The top of the cell contained several ports which supported a pH meter, mechanical stirrer, and a dissolved oxygen/temperature electrode. Other ports allowed gas feed, sampling, and acid or base injection. A computer softwarecontrolled dual-titration system (Schott Gera¨te) was used to adjust the initial pH and rapidly compensate for pH changes brought about by the addition of iron(II) and chemical reactions. The computer software recorded pH and titrant volumes. Experiments were initiated by adding 1 L of MilliQ water (building deionized water passed through a Milli-Plus cartridge deionized water system; Millipore Corp., Bedford, MA) to the reaction cell. A gas mixture of 12.2%/87.8% O2/N2 (v/v) was bubbled through the water until the dissolved oxygen concentration stabilized. The gas tube was then positioned just above the water surface. An appropriate amount of sodium bicarbonate was then added to the water as well as sodium phosphate (Na3PO4‚H2O) (Mallinckrodt, Phillipsburg, NJ) or sodium silicate (PQ Corp., Valley Forge, PA), depending on the experiment. The titration system was programmed to the desired pH and started. Dilute 0.6 N HCl (Mallinckrodt) and 0.5 N NaOH (Fisher, Fairlawn, NJ) were used to chemically adjust feedwater pH. After the pH stabilized, ferrous sulfate (FeSO4‚7H2O) (Fisher) was added to give an initial iron(II) concentration of ∼5 mg/L. Samples were drawn out of the cell 20 min after complete oxidation of iron(II) had taken place. It was preliminarily investigated whether the completeness of the oxidation corresponded to complete growth of the colloids in this reactor. Sizing experiments indicated that particle size did not change over the course of several days following reaction, indicating complete colloid growth. As a matter of practice and convenience, solutions used in this study aged at least a day before analysis. Materials. Standard polystyrene particles were obtained from Duke Scientific (Palo Alto, CA) with nominal diameter of 343 ( 9 nm, certified traceable to National Institute of Standards and Technology. The SdFFF carrier solution, made fresh daily, was deionized water containing 500 mg/L sodium dodecyl sulfate (SDS, Sigma, St. Louis, MO) and 100 mg/L sodium azide (Sigma) as a bactericide. The pH of the carrier solution was adjusted with potassium hydroxide (Fisher Scientific, Fairlawn, NJ) to the desired value. Sodium sulfate (GFS, Columbus, OH) was added to the desired concentration to investigate the effect of ionic strength on retention behavior. The reservoir containing the carrier solution was degassed with helium. Procedures. The polystyrene particle standard was dispersed to 0.001% (w/w) in deionized water. To preconcentrate the ironcontaining colloids for analysis, a 15-mL polypropylene centrifuge tube was filled with the sample and centrifuged at 4500g for at least 1 h. Most of the solution was drawn off with a pipet following centrifugation, leaving
