Environ. Sci. Technol. 2002, 36, 4288-4294
Trace Metal Leaching Behavior Studied through the Use of Parametric Modeling of Water Borne Soil Particles Fractionated with a Split-Flow Thin Cell MATTHEW L. MAGNUSON,* KEITH C. KELTY, AND CATHERINE A. KELTY United States Environmental Protection Agency, Office of Research and Development, National Risk Management Research Laboratory, Water Supply and Water Resources Division, Treatment Technology Evaluation Branch, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268
Leaching of particle-bound metals affects the ability of settling ponds and other engineered structures to remove metallic pollutants, and leaching behavior is related to particle size. In this investigation, water borne soil particles were leached and fractionated with a split-flow thin cell, and the metal loadings were quantified as a function of particle size. For comparison of the metal-loading curves, different empirical modeling procedures were investigated to convert the data to a precise functional form suitable for quantitative comparison of changes in differential loading as a function of particle size. Results of this investigation are presented for a soil sample before and after leaching caused by simulated acid rain conditions. Following simulated acid rain leaching, the shape of the differential distribution curves change, and these changes reflect the particle size mediated leaching behavior. For the soil used in this demonstration, simulated acid rain leaching shifted the differential loading toward larger particle sizes, and the magnitude of the shift varied significantly among the metals. Because settling rate decreases as the square of particle size, this could potentially affect management decisions for settling ponds receiving these particles. The high precision afforded by the analysis allows the development of insight into the leaching mechanisms through comparing “partial” acid rain leaching with “total recoverable” leaching by EPA Method 3050.
Introduction The transport of metallic pollutants through water systems is often attributed to adsorption of the pollutant onto suspended particles in the water (1, 2). These polluted particles are important in urban drainage areas such as roadways (3-5) because storm events can flush the particles from their site of generation and disperse them through the watershed, potentially contaminating vast areas. The pollutant-laden particles may be removed by settling ponds and other engineered structures based on the settleability of the particles. From a pollutant transport and stormwater man* Corresponding author phone: (513)569-7321; fax: (513)569-7658; e-mail:
[email protected]. 4288 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 20, 2002
agement standpoint, it is important to know how well the settling pond is expected to remove a particle bound pollutant. Metallic pollutant transport may be affected by leaching, i.e., by acidic rainfall. As metal leaches from the particle, the efficiency of a settling pond to remove the pollutant may change. Therefore, it is useful to study how leaching affects particles with differing settleabilities. While bulk leaching experiments with the entire sample may provide some data about overall leaching, they are not necessarily helpful to discern how metals are leached from particles of differing settleabilities. Accordingly, it is useful to fractionate the sample by settleability and study the leaching of the fractionated material, to correlate particle settleability with pollutant loading. Experimentally, the investigation of the correlation between particle settleability and pollutant loading is complicated by the size of the particles that are associated with the greatest metallic contamination. These particles typically have nominal diameters < 50 µm (5). Several experimental approaches to this problem have been discussed recently (6). One promising technique was splitflow thin-cell (SPLITT) fractionation, which provides high resolution correlation between metal loadings and particle settleability. This type of high resolution analysis is technically difficult to perform using other fractionation techniques. Several advantages of SPLITT fractionation were demonstrated (6). Namely, the technique has advantages in terms of rapid analysis, low sample consumption, and providing high resolution of particles of different sizes. SPLITT fractionation was shown to characterize the type of metalpolluted particles that are expected to enter a settling pond, potentially affecting settling pond management decisions. As far as settling pond management is concerned, even with high quality data generated by appropriate modern analytical techniques, difficulty in data analysis may arise, chiefly because it is experimentally impossible to obtain pollutant loadings at each particle settleability. To simplify and improve data analysis, it is scientifically desirable to generate a distribution of the metal loading as a function of particle size. Parametric modeling has successfully been used to describe the particle size distribution of soils (7-11), suggesting that parametric modeling might be applied to help investigate the loading of metal bound to these particles. The use of models may help to overcome the difficulties described above by precisely smoothing the experimental data and interpolating between data points. The present study initially investigates procedures for using parametric models to obtain a precise mathematical description of metallic distribution on fractionated soil particles. These models were then applied to the ultimate goal of this work, namely an improved capability to correlate leaching of metals, such as that caused by simulated acid raid, with changes in particle settleability. To our knowledge, this paper represents the first investigation to apply parametric models to further the scientific understanding of metal loading of fractionated soil particles.
Theory Based on the type of information desired, it is useful to express the metal loading of settleable particles as a cumulative distribution plot and/or a differential distribution plot. Many particle fractionation techniques, including SPLITT fractionation, produce cumulative data. While conversion to the differential form may be accomplished numerically, accurate numerical differentiation requires a large number of experimental points, often more than may be conveniently 10.1021/es0114815 Not subject to U.S. copyright. Publ. 2002 Am. Chem.Soc. Published on Web 09/07/2002
obtained. This limits the quality of the results. Accurate mathematical description of the experimental, cumulative curve may be used, therefore, to obtain a better differential distribution. In modeling cumulative plots, it is useful to borrow from experience with modeling cumulative plots of particle size distribution in soils, which is itself a relatively unexplored field (7), probably due to difficulties in curve fitting data prior to the relatively recent advent of computerized data analysis. Models have been developed to treat particle size distribution (8-11), along with other specialized applications such soil texture schemes (7, 12, 13), moisture retention (14, 15), and other interesting properties of soils. No one approach is appropriate for every application, but some of the schemes for other areas of soil physics seemed like a good starting point for the parametric modeling/ interpolation of pollutant loading as a function of particle settleability. It is important to point out that the modeling schemes are largely empirical because the physical origin of the particles is most likely the result of an intractable combination of complex physical processes. Probably the most popular expression for particle size distribution is the log-normal distribution and its variant, the cumulative log-normal distribution (8, 9, 13). Other mathematical expressions, such as polynomials (7), growth functions (10), spline fits (14), fractal analysis (11, 12), and others have been used. Specialized modeling involving cubic splines, which are essentially piecewise cubic functions, work very well in very general cases but add computational complexity when applied to the soils of interest here, in which cumulative metal loading tends to increase monotonically with particle settleability in the size range of interest. Therefore, one goal in choosing mathematical models is to be able to apply the model using readily available curve fitting routines, available with most commercial computer spreadsheet applications. The functional forms investigated in this study are the cumulative log-normal, the Gompertz function, polynomial, and fractal power law. The cumulative lognormal distribution is given by
where a1, ..., an are the fitted shape parameters of the curve. Several suggestions for fractal approaches appear in the literature for particle size distribution in soil (11, 12). The fractal form investigated here is power scaling of the type
collected from pastureland that was frequently flooded from a nearby creek (16). These types of particles may also represent the types of particles that may enter a settling pond because the soil was created under conditions similar to those expected from a stormwater event. The samples were added at a concentration of 0.1% (w/w) to deionized water. Samples were then passed through a 40 µm stainless steel screen to remove oversized particles and any particles less dense than water, which cannot be fractionated on the basis of their settling rates. This screening was carefully performed to avoid depositing sample on the screen, and all samples are expected to be equally affected by this procedure. NIST 2710 was chosen to investigate leaching by simulated acid rain because the certificate of the material (16) indicated that it was amenable to leaching experiments. Therefore, the sample was acidified beforehand and leached at room temperature for at least 24 h at pH 4.3 by adding nitric acid. Twenty-five microliters of chloroform was added to the solution to prevent microbial growth. This procedure was designed to leach metals already present in the soil, i.e., no additional metals were added to the native sample. This sample will be referred to here as the “simulated acid rain” leached soil. This is similar to the pH of the most acidic rainwater expected in the continental United States (17) and slightly above that expected in Europe (18). Fractionation of Soil Particles by Relative Settling Diameter. Fractionation of particles by their settling ability was accomplished by the use of split-flow thin-cell (SPLITT) fractionation (19-21), which has been primarily used for environmental particles as a preparative separation technique (22, 23). For the fractionation of several diverse particle types, including soil, that would be expected to enter settling ponds, SPLITT fractionation provided higher resolution in a shorter experiment time than other fractionation techniques and required small sample sizes (6). In the SPLITT technique, small particles (
