Experimental Determination of Effective Diffusion Parameters in the Matrix of Fractured Till

a solute diffuses may cause the observed, or effective diffusion coefficient (De) to be lower than the diffusion Diffusion is often the dominant mode of solute transport in soils coefficient of a compound in water (D0) (Rao et al., when advection is minimal. This paper describes the application of a radial diffusion cell method to estimate the effective diffusion coeffi1980). Moreover, exclusion of a solute from some pore cient (De) and effective diffusive porosity ( De) for use in solute transclasses may reduce the effective diffusive porosity ( De), port models for fractured-porous media. Twenty-four experiments or that portion of a porous medium that is accessible were conducted for 28 d using three conservative solutes (Br, PFBA, to diffusing molecules (van der Kamp et al., 1996). These and PIPES) on eight late Wisconsinan and Pre-Illinoian till samples effective parameters are likely to be specific to each from Iowa. The mean value of the total porosity ( T) of the till samples soil–solute combination, and therefore require direct was 30.0%. Concentrations of the three tracers in the reservoir demeasurement for the solute and material of interest. creased with time and eventually approached equilibrium concentraMathematical solutions to the diffusion problem are tions. A model simulated the observed concentration data and the well established (Crank, 1975). In addition, models are modified goodness-of-fit (d1) values ranged from 0.878 to 0.950. Mean now capable of simulating solute diffusion coupled with values of De from the model were 28.3 (Br ), 26.5 (PFBA), and 21.6% (PIPES) and there were significant differences in De among advection through macroporous soil and fractured till the three tracers (p 0.05). Mean values of De were 5.6 10 10 m2 (Therrien and Sudicky, 1996; Toride et al., 1999). These s 1 (Br ), 2.9 10 10 m2 s 1 (PFBA), and 1.3 10 10 m2 s 1 (PIPES). models require that effective diffusion parameters be Values of De differed significantly by compound and were significantly provided as input. Laboratory methods, including the different (p 0.05) from the aqueous diffusion coefficient (D0). Calcuhalf-cell, the reservoir-cell, and radial diffusion cell, lated mean values of the first-order mass exchange coefficient ( ) have been proposed for determination of De. The most were 8.4 10 7 (Br ), 4.1 10 7 (PFBA), and 1.6 10 7 s 1 (PIPES); widely used is the half-cell method, where two cells, they differed by compound (p 0.05) and generally decreased with one spiked with a solute, are pressed together, allowing increasing molecular weight of the tracer. This study confirmed that diffusion to take place (Li and Gregory, 1974; Robin et the radial diffusion cell method is an efficient method to estimate al., 1987). Work by van Rees et al. (1991), however, effective diffusion parameters necessary to accurately model solute transport in fractured till and soil. demonstrated that the process of sectioning a soil column may cause errors in De estimates. As an alternative, they proposed a reservoir-cell method, where a reservoir of water spiked with a solute is allowed to diffuse into D is the predominant mode of solute transan adjacent soil column. This method was shown to be port in soils where advection is negligible (Sawatsky more accurate and less labor intensive than the half-cell et al., 1997). Diffusion may also have a strong influence method. A review of these and other similar methods is on solute concentration within mobile pores of a dual presented by Shackelford (1991) and Flury and Gimmi porosity medium or in soil of low permeability containing (2002). A more recent modification to these standard fractures or macropores. A small mass exchange by diffumethods is the use of a radial diffusion cell (Novakowski sion from the mobile to the immobile region (or vice versa) and van der Kamp, 1996; van der Kamp et al., 1996), is likely to cause a large change in solute concentration which consists of a small, cylindrical reservoir drilled within the mobile region (Coats and Smith, 1964). Alinto a saturated soil or till core along its axis. The sample though diffusion may take place at the pore scale, when reservoir is filled with a solution of known tracer concendiffusion controls the concentration of solutes in the tration, which is then monitored at discrete times. Effecmobile region (e.g., fractures), diffusion may influence tive diffusion parameters (De and De) are estimated by concentration tens or hundreds of meters downgradient fitting a radial diffusion model to the time-concentration (McKay et al., 1993). Quantification of solute diffusion curve generated from the experiment (Novakowski and is therefore critical to our overall understanding of solvan der Kamp, 1996). Among the benefits of the method ute transport, most notably in fractured soils. are (i) ease of sample collection and preparation, (ii) Diffusion is a function not only of the solute but also minimal sample disturbance, (iii) no sectioning of the of characteristics of the fractured-porous medium. For soil sample, and (iv) that an estimate of De may be obexample, the tortuosity of pore throats through which tained. We describe a series of experiments on eight samples M.F. Helmke, Versar, Inc., 6850 Versar Center, Springfield VA 22151; of unfractured late Wisconsinan and Pre-Illinoian till from W.W. Simpkins, Dep. of Geological and Atmospheric Sciences, Iowa Iowa where the radial diffusion cell method was applied State Univ., Ames IA 50011; R. Horton, Agronomy Dep., Iowa State to estimate diffusion properties. Diffusion parameters Univ., Ames, IA 50011. Received 13 June 2003. Original Research Paper. *Corresponding author ([email protected]). Abbreviations: BTC, breakthrough curve; DML, Des Moines Lobe; IES, Iowa Erosion Surface; MCL, maximum contaminant level; MSEA, Published in Vadose Zone Journal 3:1050–1056 (2004).  Soil Science Society of America Management Systems Evaluation Area; SIDP, Southern Iowa Drift Plain. 677 S. Segoe Rd., Madison, WI 53711 USA