SSSAJ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rennert, T.
Right arrow Articles by Greef, K.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Rennert, T.
Right arrow Articles by Greef, K.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Rennert, T.
Right arrow Articles by Greef, K.
Related Collections
Right arrow Soil Chemistry
Soil Science Society of America Journal 67:756-764 (2003)
© 2003 Soil Science Society of America

DIVISION S-2—SOIL CHEMISTRY

Sorption and Transport of Iron-Cyanide Complexes in Goethite-coated Sand

Thilo Rennerta, Tim Mansfeldt*,a, Kai U. Totscheb and Karin Greefa

a Arbeitsgruppe Bodenkunde und Bodenökologie, Fakultät für Geowissenschaften, Ruhr-Universität Bochum, D-44780 Bochum, Germany
b Lehrstuhl für Bodenkunde, TU München, D-85350 Freising-Weihenstephan, Germany

* Corresponding author (Tim.Mansfeldt{at}ruhr-uni-bochum.de)


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Iron-cyanide complexes are present in soil and ground water because of anthropogenic inputs. We studied the sorption and the transport of the complexes ferrocyanide, [Fe(CN)6]4-, and ferricyanide, [Fe(CN)6]3-, in goethite-coated sand in column experiments under saturated conditions as influenced by flow velocity and flow interruption. Isotherm parameters obtained from batch experiments of Fe-cyanide complex sorption on goethite were used to simulate breakthrough curves in goethite-coated sand. The breakthrough curves of ferrocyanide were inversely modeled. The transport of both complexes was retarded and rate-limited. Only ferricyanide breakthrough curves revealed concentration drops after flow interruption. Simulations with batch parameters roughly reflected breakthrough curves of ferricyanide, but not of ferrocyanide. Ferricyanide sorption and desorption could not be described with the same isotherm indicating hysteresis. Since the sorption fronts of ferrocyanide breakthrough curves revealed the formation of a shoulder, it was concluded that ferrocyanide sorption in column experiments could not be described by a single isotherm, which is based on a singular sorption process. Therefore, sorption of ferrocyanide on goethite was assumed to be influenced by more than one sorption mechanism. Inverse modeling of ferrocyanide breakthrough data using the Langmuir isotherm resulted in erroneous sorption maxima.

Abbreviations: ADE, advection-dispersion equation • pvs, pore volumes


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THE FE-CYANIDE COMPLEXES ferricyanide, [Fe(CN)6]3-, and ferrocyanide, [Fe(CN)6]4-, are from anthropogenic sources in soils (Fuller, 1985). Their presence in soils and aquifers is caused by the deposition of industrial refuse such as purifier wastes originating from coal gasification (Shifrin et al., 1996), blast furnace sludge originating from pig iron production (Mansfeldt and Dohrmann, 2001), or paper de-inking sludge originating from paper recycling (Mansfeldt, 2001). An additional input to the soil environment is given by the use of road salt which contains Berlin Blue, Fe4[Fe(CN)6]3, or Na4[Fe(CN)6] as anticaking agents (Paschka et al., 1999).

The mobility of Fe-cyanide complexes in soils on sites of former manufactured gas plants, on which purifier wastes have been deposited, is governed by pH and redox potential, both influencing the dissolution and precipitation of Berlin Blue (Meeussen, 1992). It is the most important CN-phase in these soils (Meeussen et al., 1994; Mansfeldt et al., 1998). If, however, Fe-cyanide complex concentrations are too small for precipitation, sorption of these anions on the solid soil matrix may be the main immobilizing process (Meeussen, 1992).

The sorption of both Fe-cyanide complexes on goethite, {alpha}-FeOOH, as a model sorbent in soil was investigated by Rennert and Mansfeldt (2001)( 2002b) in batch experiments. They proposed various sorption mechanisms for the Fe-cyanide complexes, for ferrocyanide a combination of inner-sphere surface complexation and surface precipitation of a Berlin-Blue-like phase and for ferricyanide outer-sphere and weak inner-sphere surface complexation. The sorption of ferricyanide on goethite was also investigated by Theis et al. (1988) using batch experiments as well as experiments with small columns. They deduced outer-sphere complexation of ferricyanide from their results. Fuller (1985) used soil material in column experiments to examine the sorption of ferricyanide. The most important property governing sorption was soil pH followed by the contents of clay minerals and Fe oxides. Because of early breakthrough in column experiments, Ghosh et al. (1999) suggested nonreactive transport of ferrocyanide in neutral sandy aquifer material. In batch experiments with uncontaminated soils, the sorption of both Fe-cyanide complexes was promoted in acid soils containing large amounts of Al and Fe oxides, clay minerals, and, in some cases, soil organic matter (Rennert and Mansfeldt, 2002a). The release of Fe-cyanide complexes from a contaminated former manufactured gas plant site soil in column experiments under dispersive flow conditions was strongly rate-limited (Weigand et al., 2001). Their results indicated that both the dissolution of Berlin Blue and the desorption of Fe-cyanide complexes was rate-limited. In none of these studies was the sorption and transport of Fe-cyanide complexes modeled numerically.

The main aim of this study was to investigate the sorption and the transport of ferricyanide and ferrocyanide in goethite-coated sand in columns under saturated conditions as influenced by flow velocity and flow interruption. Goethite-coated sand was used, because it may be regarded as a model of a subsurface soil horizon lacking soil organic matter. Furthermore, sorption parameters obtained from batch experiments with goethite were used in numerical simulations of the breakthrough of the Fe-cyanide complexes in goethite-coated sand.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Preparation and Properties of Goethite and Goethite-Coated Sand
Goethite-coated sand was used as model porous medium. Goethite was prepared according to Schwertmann and Cornell (1991) by precipitating ferrihydrite from alkaline solution. Ferrihydrite was converted to goethite by aging for 60 h at 343 K. The product was cleaned by pressure filtration. The point of zero charge of goethite was 8.3, and its specific surface area measured by N2 adsorption was 30 m2 g-1 (Rennert and Mansfeldt, 2001). We used acid-washed sea sand (Riedel-de Haën, Seelze, Germany), which contained 70% medium sand (0.2–0.63 mm) and 30% fine sand (0.063–0.2 mm). The sand was heated to 1100 K to destroy organic matter. After cooling, the sand was suspended (1.8 g mL-1) and a goethite suspension (0.3 g mL-1; pH 2.5) was added to the sand suspension. Highly coated sand was obtained by this procedure (Scheidegger et al., 1993). In contrast to Scheidegger et al. (1993), the goethite-sand suspension was left standing for 60 h at room temperature and occasionally stirred. Then excess goethite was removed by wet sieving (0.63 mm sieve). Subsequently, the goethite-coated sand was freeze-dried after washing with deionized water. It contained 369 µg goethite g-1 sand. Similarly, Filius et al. (1999) obtained 440 µg goethite g-1 sand using the Scheidegger et al. (1993) method. As shown by x-ray photoelectron spectroscopy (XPS) and specific surface area analysis, pure goethite and goethite coating the sand do not differ (Scheidegger et al., 1993). Total Fe concentrations were determined by atomic absorption spectroscopy (PE 3100, Perkin Elmer, Überlingen, Germany) after microwave digestion (MWS-1, Berghof, Eningen, Germany) of the goethite-coated sand with 65% (v/v) HNO3. The specific surface area (Brunauer-Emmett-Teller [BET]) of the goethite-coated sand was <1 m2 g-1. The pH of the goethite-coated sand measured in 0.01 M CaCl2 with a substrate to solution ratio of 1:2.5 was 4.9.

Column Experiments
We used columns made of polyvinyl chloride (PVC) (emc, Erfurt, Germany) with a length L of 10 cm, an i.d. of 4 cm, and a cross-sectional area, A, of 12.57 cm2. The columns were packed to uniform bulk density (approximately 1.6 g cm-3 in all experiments) with goethite-coated sand. At the bottom and at the top, the columns were capped with ceramic porous plates. The background electrolyte 0.01 M NaNO3 was fed to the column using a peristaltic pump (Ismatec MCP with a CA 4 pump head, Ismatec Laboratoriumstechnik, Wertheim, Germany) to obtain saturated conditions. Before the experiments, the background solution and all other feeding solutions were degassed. The column was saturated at a flow rate Q = 0.3 cm3 min-1 from bottom to top to prevent air entrapment.

The flow regime was characterized by applying the tracer Cl- at three flow rates (Q = 0.5; 1.0; or 1.5 cm3 min-1) as pulse inputs of 0.01 M NaCl added to the background electrolyte. In one tracer experiment, Cl- was applied by continuous feed. Aliquots of 2 mL of the effluent were collected in a fraction collector (Linear II, Reichelt Chemietechnik, Heidelberg, Germany). Chloride concentrations were determined by ion chromatography with electrical conductivity detection (DX 500, Dionex, Idstein, Germany) after appropriate dilution. A tracer breakthrough was performed before all experiments with Fe-cyanide complexes. After the tracer experiments, 12 to 25 pore volumes of the background electrolyte were used to leach Cl- from the column.

For the sorption and transport experiments of Fe-cyanide complexes, solutions of K3[Fe(CN)6] or K4[Fe(CN)6] (reagent grade, Riedel-de Haën, Seelze, Germany) containing 0.05 mM [Fe(CN)6] in the background solution were used. Pulse inputs of a Fe-cyanide complex solution were fed to the column at three flow rates (Q = 0.5; 1.0; or 1.5 cm3 min-1). The number of pore volumes, pvs, of a Fe-cyanide complex solution added varied between the experiments: ferricyanide, 2.5 pvs (Q = 0.5 cm3 min-1); 3.0 pvs (Q = 1.0 cm3 min-1); 6.9 pvs (Q = 1.5 cm3 min-1); ferrocyanide, 2.7 pvs (Q = 0.5 cm3 min-1); 3.1 pvs (Q = 1.0 cm3 min-1); 5.5 pvs (Q = 1.5 cm3 min-1). Then the Fe-cyanide complex solution was replaced with the CN-free background electrolyte solution of which 1.3 pvs were applied. Subsequently, the flow was interrupted for 16 to 20 h. After the interruption, flow was continued with the background electrolyte. When investigating the sorption of ferricyanide at Q = 0.5 cm3 min-1, the flow was interrupted for four times ranging from 1 to 96 h after 1.6, 1.8, 2.0, and 3.3 pvs. Again, fractions of the effluent solutions were collected. The pH was measured potentiometrically with a WTW pH 90 (Wissenschaftlich-Technische Werkstätten, Weilheim, Germany) and an INLAB 406 electrode (Ingold, Steinbach, Germany). Cyanide was determined according to Mansfeldt and Biernath (2000) using a micro-distillation technique (MicroDistiller, Eppendorf-Netheler-Hinz, Hamburg, Germany). All experiments with ferrocyanide were performed in darkness.

Data Analysis
Bulk density, {rho}b, and volumetric water content, {theta}, were measured directly. The latter was determined gravimetrically after each experiment. Mean pore-water velocity, {nu}, was calculated from the flow rate Q [{nu} = Q/(A x {theta})]. The breakthrough data are given as reduced variables. Reduced concentrations (c/c0) were given by the ratio of effluent and influent concentrations. The number of pore volumes eluted (pv), that is, reduced time, was calculated by

[1]
where q = Q/A, Darcian flow, and tbt = breakthrough time of the respective concentration. The hydraulic regime was characterized by inverse modeling of the tracer breakthrough data.

In the following, [L], [M], and [T] indicate units of length, mass, and time, respectively. The computer code CXTFIT 2.1 (Toride et al., 1995) was used to determine the dispersion coefficient D [L2 T-1] and the retardation factor R by fitting the breakthrough data of Cl- to the advection-dispersion equation, ADE,

[2]
where c is the flux averaged concentration of the liquid phase [M L-3], s is the concentration of the sorbed phase [M M-1], D is the dispersion coefficient [L2 T-1], {theta} is the volumetric water content [L3 L-3], Jw is the volumetric water flux density [L T-1], and {rho}b is the bulk density [M L-3].

Chloride adsorption by the solid phase is described by a linear isotherm as

[3]
where Kd [M-1 L3] is a distribution constant. Assuming steady state flow, Eq. [2] is rearranged to Eq. [5] using Eq. [3] and [4],

[4]

[5]
where {nu} = Jw/{theta} is the mean pore-water velocity.

Tracer breakthrough derived D values of a given column were used for the simulation of Fe-cyanide complex breakthrough curves in this column.

Simulations of the sorption and the transport of Fe-cyanide complexes were performed with the numerical computer code CARRY 6.0 (Totsche et al., 1996). The water movement in CARRY is described by the ADE, Eq. [2]. CARRY operates with a bulk solid phase composed of different sorption sites, which can be discriminated in terms of bulk density fractions. In all simulations, goethite was regarded as the sorbent. Therefore, the bulk density fractions of goethite (given by {rho}Gt = massGt/(L x A), where Gt indicates goethite) were used in the simulations as the reactive bulk density fractions. CARRY input files require values of dispersivity {lambda} [L] ({lambda} = D/{nu}) and tortuosity {tau} ({tau} {approx} {theta}/{theta}sat2/3, where {theta}sat is the saturated water content [Millington and Quirk, 1961]). To describe the retention of Fe-cyanide complexes during transport, data of the sorption of Fe-cyanide complexes on goethite adapted from Rennert and Mansfeldt (2001) were used.

Rennert and Mansfeldt (2001) performed batch sorption experiments as influenced by pH (ranging from 3.5 to 9) in the presence of the background electrolyte 0.01 M NaNO3. Sorption isotherms were calculated according to a model by Barrow (1999). The isotherms showed that the sorption of both complexes decreased with increasing pH. Unlike ferricyanide, ferrocyanide was sorbed above the point of zero charge of the goethite surface indicating stronger sorption. Sorption kinetics were investigated in the time range from 1 to 120 h. Ferrocyanide sorption was complete after 1 h, whereas that of ferricyanide continued and was finished after 24 h. Ferricyanide sorption decreased with increasing ionic strength in contrast to ferrocyanide sorption, which was not affected by varying concentrations of the background electrolyte NaNO3. Desorption of ferricyanide induced by raising the pH to 9.3 was quick and complete. In contrast, desorption of ferrocyanide was rate-limited and incomplete after 240 min.

For the present study, their isotherm data were interpreted with the Langmuir isotherm,

[6]
where smax is the sorption maximum [M M-1], and KL is the Langmuir constant [M-1 L3]. As both isotherm parameters smax and KL for both Fe-cyanide complexes decreased exponentially with increasing pH, values of the parameters at each equilibrium pH in column experiments were calculated. These values were smax = 1.67 x 10-3 g g-1 and KL = 176862 g-1 cm3 for ferricyanide (pH 6.3) and smax = 2.79 x 10-3 g g-1 and KL = 384326 g-1 cm3 for ferrocyanide (pH 6.2) in the column experiments. These parameters were used to simulate Fe-cyanide complex breakthrough curves with CARRY. Sorption, s, in Eq. [2] of the complexes was described by Eq. [6] instead of Eq. [3] to combine sorption and transport. Sorption on all surfaces sites was assumed to be rate-limited (first-order kinetics), and the forward (sorption) rate parameter kf (given by kf = KL x kb, where kb is the backward [desorption] rate parameter) varied between 10-4 and 10-2 cm3 g-1 min-1 in the simulations. Consequently, the backward (desorption) rate parameter kb varied between 5.7 x 10-10 and 5.7 x 10-8 min-1 for ferricyanide and between 2.6 x 10-10 and 2.6 x 10-8 min-1 for ferrocyanide.

Inverse modeling of the ferrocyanide breakthrough data was performed using the computer code MCMFIT (Bajracharya and Barry, 1995). It estimates parameter values by nonlinear least-squares fitting for nonlinear adsorption of a single solute species with one-dimensional transport. In MCMFIT, a general adsorption isotherm by Barry (1992) is used (Eq. [7])

[7]
which represents the Langmuir isotherm (Eq. [6]) with {alpha}1 = smax and {alpha}2 = KL, when {alpha}4 = -1 and {alpha}3 = 1. In the modeling procedures, transport was described by the ADE, Eq. [2], and sorption was described by Eq. [7] with {alpha}1 = smax and {alpha}2 = KL, when {alpha}4 = -1 and {alpha}3 = 1. The parameters KL, smax, and kf were fitted, whereas physical parameters such as {nu} and D were kept constant, because they were calculated from direct measurements ({nu}) or taken from tracer breakthroughs (D).


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Breakthrough of Chloride
Measured data and modeled breakthrough curves of the tracer Cl- in goethite-coated sand are presented in Fig. 1 . Tracer breakthrough curves were symmetrical, indicating linear equilibrium sorption and physical equilibrium, that is, the absence of preferential flow. The position of the desorption fronts differed, because the number of pore volumes added ranged from 0.49 to 2.44. Parameters resulting from fitting Cl- breakthrough data to the ADE (Eq. [2]) are shown in Table 1. All curve fittings resulted in high correlation coefficients r (given as their squares in Table 1). As can be seen from the retardation factors R ranging from 0.93 to 1.11, Cl- was nearly an inert tracer. Column Peclet numbers (Pe = {nu}L/D) ranged from 82 to 178. In spite of these variations, a strongly advective flow regime was indicated in all experiments. The D values were used in simulations of Fe-cyanide complex breakthrough presented below.



View larger version (24K):
[in this window]
[in a new window]
  		Fig. 1. Breakthrough curves of Cl- in goethite-coated sand at three pore-water velocities. Breakthrough curves are numbered according to Table 1. Symbols indicate measured data, the lines result from fitting the data to Eq. [2] using CXTFIT.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Experimental parameters of Cl- breakthrough curves in goethite-coated sand and results of fitting the breakthrough data to the advection-dispersion equation (Eq. [2]) using CXTFIT 2.1. The parameters R and D were fitted.

 
Breakthrough of Ferricyanide
Measured data and simulated ferricyanide breakthrough curves in goethite-coated sand at three pore-water velocities are shown in Fig. 2 . The experimental parameters of all Fe-cyanide complex breakthrough curves are summarized in Table 2.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 2. Breakthrough curves of ferricyanide in goethite-coated sand at three pore-water velocities (a, {nu} = 0.104 cm min-1; b, {nu} = 0.215 cm min-1; c, {nu} = 0.305 cm min-1). Lines result from simulations using batch-derived Langmuir isotherm parameters (Eq. [6], smax = 1.67 x 10-3 g g-1, KL = 176862 g-1 cm3) with four different forward rate parameters kf using CARRY. Arrows indicate the beginning of a flow interruption.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Experimental parameters of Fe-cyanide complex breakthrough curves in goethite-coated sand.

 
At all pore-water velocities, ferricyanide revealed early breakthrough (at pv < 1). The sorption fronts were steep, and the maximum concentration c/c0 = 1 was reached quickly (after about 1.5 pv). In the absence of preferential flow, early breakthrough can be attributed to nonlinear sorption in addition to dispersion. The steep rise was enhanced by the strongly advective flow regime as well. The positions of the sorption fronts did not depend on the pore-water velocity. At the desorption fronts, the reduced ferricyanide concentrations increased after flow interruption. This indicates rate-limited desorption of sorbed ferricyanide. Then the concentrations quickly dropped. Ferricyanide was not completely leached from goethite at the smallest pore-water velocity (Table 2).

Rate-limited sorption should induce a dependence of the positions of the breakthrough curves on flow velocities. However, the flow velocities were large, which might explain the lack of dependence observed. The large flow velocities also caused the quick initial breakthrough, because there was not enough time to attain an equilibrium, hence illustrating kinetic nonequilibrium of ferricyanide sorption.

Figure 2a shows the effect of flow interruption on ferricyanide concentrations when ferricyanide was continuously fed to the column (the first three flow interruptions). The reduced concentration decreased after the flow interruption, but returned quickly to the initial values before the flow interruption began. The decrease was caused by rate-limited sorption (Brusseau et al., 1989b). Therefore, sorption as well as desorption of ferricyanide on goethite-coated sand were rate-limited.

In batch experiments, the sorption of ferricyanide on goethite reached apparent equilibrium after 24 h (Rennert and Mansfeldt, 2001). In contrast, the kinetics of ferricyanide sorption in the column experiments presented here differed from those in batch experiments, as the drop in the reduced concentrations was larger after the flow interruption which lasted 96 h than that after the flow interruption which lasted 20 h.

For all parts of Fig. 2, measured data were reflected by the simulated breakthrough curves using Langmuir isotherm parameters except for the data after the flow interruptions. Simulations of the breakthrough curves were performed at constant pH (6.3). This was an adequate simplification, because the effluent pHs were in the narrow range of 6.0 to 6.4. The speciation of ferricyanide (all dissociation constants of ferricyanic acid are <1 (Jordan and Ewing, 1962)), and the surface properties of goethite are unlikely to vary dramatically within this narrow pH range.

Especially the quick rise of concentrations after the flow interruptions was not reflected by the simulations. However, the simulation with the forward rate parameter kf set to 10-3 cm3 g-1 min-1 (Fig. 2a) described the measured sorption front sufficiently and was in good agreement with the measured smallest concentrations after the second and third flow interruption. The effects of flow interruption on the concentrations were poorly simulated at higher flow velocities as shown in Fig. 2b and 2c again indicating that the flow velocities were large compared with the rate of sorption.

When flow was not interrupted, the simulations with Langmuir parameters derived from batch experiments with kf = 10-3 cm3 g-1 min-1 were approximate descriptions of the actual breakthrough curves for all pore-water velocities. All sorption fronts were reflected by simulations with the parameters mentioned above. The effects of flow interruption at the desorption fronts were not apparent in the simulations. The larger was the forward rate parameter kf used in the simulations, the larger was the increase of the concentrations after the flow interruption at the desorption fronts. However, the larger was kf, the poorer were the simulations of the sorption fronts. This in turn indicates that the backward rate parameters kb to simulate the desorption front must be larger than those calculated from the forward rate parameter kf and the Langmuir constant KL. In this case, KL is not constant because of differences in sorption and desorption indicating kinetic nonequilibrium of ferricyanide sorption.

Because of the effect of flow interruption, early breakthrough, and different sorption and desorption kinetics, the transport of ferricyanide in goethite-coated sand was affected by rate-limited and nonlinear sorption.

Breakthrough of Ferrocyanide
Measured data and simulated ferrocyanide breakthrough curves in goethite-coated sand at three pore-water velocities are shown in Fig. 3 . Similar to ferricyanide, ferrocyanide revealed early breakthrough. For all pore-water velocities, the measured breakthrough curves were not symmetrical. Except for the BTC shown in Fig. 3c, the breakthrough of ferrocyanide was not complete, as the maximum concentration c/c0 = 1 was not reached. This indicates rate-limited sorption. Again, different pore-water velocities had no effect on the breakthrough curves, as the positions of the steep rise of the sorption fronts were identical. The sorption fronts could be separated into three parts: a steep rise (0 to {approx} 1.5 pv); a plateau (1.5 to {approx} 2 pv); and subsequent converging on the maximum concentration (>2 pv). The combination of these three parts may be designated as a shoulder. Shoulder formation suggests complex and nonsingular sorption phenomena. The desorption fronts dropped quickly to low concentrations and revealed slight tailing. The flow interruptions had no effect on the shape of all breakthrough curves. The sorption as well as the desorption fronts revealed slight tailing. Tailing of a sorption front can be attributed to physical and/or chemical nonequilibrium (Brusseau et al., 1989a). Physical nonequilibrium affects the transport of both reactive and nonreactive solutes because of diffusive mass transfer between nonadvective and advective flow regions. Peclet numbers of Cl- breakthrough curves always indicated advective flow (Table 1). Furthermore, breakthrough of only slightly retarded Cl- did not show any tailing (Fig. 1). Therefore, physical nonequilibrium can be excluded and the tailing of the ferrocyanide breakthrough curves is caused by rate-limited and/or nonlinear sorption.



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 3. Breakthrough curves of ferrocyanide in goethite-coated sand at three pore-water velocities (a, {nu} = 0.104 cm min-1; b, {nu} = 0.209 cm min-1; c, {nu} = 0.305 cm min-1). Lines result from simulations using batch-derived Langmuir isotherm parameters (Eq. [6], smax = 2.79 x 10-3 g g-1, KL = 384326 g-1 cm3) with four different forward rate parameters kf using CARRY. Arrows indicate the beginning of a flow interruption.

 
The measured concentrations after flow interruption did not indicate rate-limited desorption of sorbed ferrocyanide. However, this does not prove that rate-limited desorption is absent, because the duration of the flow interruptions might not have been adequate to observe an effect on measured concentrations.

All breakthrough curves simulated with parameters derived from batch experiments failed to reflect the measured data properly. Because of the same reasons as pointed out for the simulations of ferricyanide breakthrough curves, the simulations of ferrocyanide breakthrough curves were performed at constant pH. In all simulations, it was assumed that sorption of ferrocyanide occurs on homogeneous sorption sites, and that sorption is rate-limited. Measured ferrocyanide breakthrough was later than predicted by batch-derived parameters indicating greater sorption in column experiments or that sorption was rate-limited in column experiments. The shoulder was not reproduced by any simulation. This indicates that ferrocyanide sorption in column experiments was not predicted by parameters obtained in batch experiments, although the sorbent was the same.

Apart from the occurrence of a shoulder, inverse modeling of ferrocyanide sorption demonstrated the inapplicability of the Langmuir isotherm. Fitted parameters are given in Table 3 and modeled breakthrough curves in Fig. 4 . These breakthrough curves reflected the measured data except for the shoulder, and therefore it was not possible to fit the whole breakthrough curves using Langmuir isotherms. Although the fittings resulted in large values of r2, the fitted sorption maxima smax indicated that the measured breakthrough curves were not adequately described. Considering the specific surface area of goethite, the fitted sorption maxima (Table 3) were in the range of 236 to 617 µmol m-2. These values are erroneous, as can be seen from a comparison with literature data obtained in batch experiments: the sorption maxima of ferrocyanide on goethite is 1.6 µmol m-2 (Rennert and Mansfeldt, 2001); the sorption maximum of strongly sorbed phosphate on goethite is 2.51 µmol m-2 (Schwertmann, 1988). As a consequence, the fittings presented show that significant correlation does not imply causal connection.


View this table:
[in this window]
[in a new window]
  		Table 3. Langmuir isotherm parameters resulting from fitting ferrocyanide breakthrough data to Eq. [6] using MCMFIT. Pore water velocity was not fitted.

 


View larger version (13K):
[in this window]
[in a new window]
  		Fig. 4. Breakthrough curves of ferrocyanide in goethite-coated sand at three pore-water velocities (a, {nu} = 0.104 cm min-1; b, {nu} = 0.209 cm min-1; c, {nu} = 0.305 cm min-1). Lines result from fitting the data to the Langmuir isotherm (Eq. [6]) using MCMFIT. Fitted parameters are given in Table 3.

 
In further calculations, we fitted all parameters of the general isotherm (Eq. [7]), but all these calculations (not shown here) did not fit the shoulder. This additionally indicates the failure of a single isotherm approach in describing the measured ferrocyanide breakthrough curves.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
CONCLUSIONS
 REFERENCES
 
The sorption and transport behavior of the Fe-cyanide complexes differed as summarized in Table 4.


View this table:
[in this window]
[in a new window]
  		Table 4. Comparison of ferricyanide and ferrocyanide sorption and transport behavior in goethite-coated sand.

 
The batch-derived parameters yielded approximate descriptions of the breakthrough curves except for the data after the flow interruptions. The failure of simulations with batch-derived parameters to describe the effects of flow interruption on measured breakthrough data may be caused by the necessity to use larger backward rate parameters for desorption which differ from those calculated from the forward rate parameters and the Langmuir constant. These different rate parameters result in sorption and desorption fronts, which are not congruent, possibly caused by hysteresis. Differences in the transport of ferricyanide to the goethite surface in the case of sorption and from the goethite surface in the case of desorption would result in hysteresis, thus in different sorption and desorption fronts, respectively. Another possibility of the occurrence of hysteresis is structural change of the goethite surface due to sorption.

Our results agree with the findings of Theis et al. (1988) mostly. In their study ferricyanide revealed early breakthrough in small columns filled with goethite. They also found a single reaction, which was rate-limited because of surface mass transfer. Desorption was quick and complete, and hysteresis was not observed. However, their breakthrough experiments were not complete, as the desorption fronts were not investigated. Therefore, differences between the sorption and desorption fronts, which indicate hysteresis, could not be observed in their investigations.

As for ferricyanide, the ferrocyanide sorption fronts differed from the desorption fronts. This indicates hysteresis for ferrocyanide sorption, too. As mentioned before, Langmuir isotherm parameters obtained from batch experiments and from inverse modeling of breakthrough data did not reflect measured ferrocyanide breakthrough curves in goethite-coated sand. The problem in both approaches was to describe the shoulder after 1.5 pv, which indicated chemical nonequilibrium. Possible reasons of shoulder formation are (i) sorption of more than one sorbate on the surface or (ii) sorption of one sorbate by more than one sorption mechanism. Shoulder formation because of heterogeneous sorbate solutions has been shown and discussed by Weigand and Totsche (1998) for the transport of dissolved organic matter, DOM, in goethite-coated sand. As DOM comprises a continuum of different organic substances, a fractionation of DOM during the transport into subcomponents differing in their reactivities occurs. However, ferrocyanide in the sorbate solution is not fractionated in our experiments, because (i) at experimental pH ferrocyanide was completely dissociated (the dissociation constants for ferrocyanic acid are pK4 = 4.2; pK3 = 2.2; pK2 = pK1 < 1 (Jordan and Ewing, 1962)), and (ii) ferrocyanide was not partially decomposed to CN-, because the experiments were performed in the absence of light. It has been established that decomposition of aqueous ferrocyanide occurs during irradiation by UV light or in the presence of a catalyst only (Rader et al., 1993). Therefore, fractionation of ferrocyanide cannot explain the shoulder formation. The other aspect of shoulder formation is the presence of different sorption mechanisms. Rennert and Mansfeldt (2001)(2002b) proposed a combination of inner-sphere surface complexation and precipitation of a Berlin-Blue-like phase to explain the ferrocyanide sorption mechanism on goethite in the acidic range. However, the experimental pH was > 6. At this pH, dissolution of goethite resulting in ferric ions in solution is negligible. This cation is required for the precipitation of Berlin Blue. Therefore, the formation of Berlin Blue during the experiments is excluded.

Nevertheless, the presence of different sorption mechanisms is a possible reason for the failure of batch-derived parameters in predicting breakthrough curves. At least one of these sorption mechanisms should be rate-limited. If one of these sorption mechanisms is coprecipitation or precipitation, the goethite surface is changed because of sorption, which might explain hysteresis. The main questions arising from this conclusion are: what are these different sorption mechanisms; what are the relative proportions of these mechanisms in explaining the extent of sorption; and how can they be quantified in terms of sorption parameters to simulate breakthrough curves numerically?

In many studies batch-derived parameters were used to simulate sorption and transport observed in column experiments. A comparison of batch and column derived Cd and methylene blue sorption data is given by Bürgisser et al. (1993). The sorption of neither substance was influenced by kinetics. The results of both experimental techniques were in good agreement using Freundlich isotherms. Similar naphthalene sorption coefficients measured by batch and column methods were found by MacIntyre et al. (1991). The agreement among coefficients was attributed to fast kinetics and linear sorption. Sorption nonequilibrium during transport can result in lower R values compared with those estimated from a batch isotherm (Brusseau et al., 1991). Retardation of U(VI) in subsurface media was underestimated by batch-derived adsorption isotherms (Barnett et al., 2000). Simulating the transport of U(VI) with a fractional order kinetic model yielded a good approximation of observed breakthrough data.

These examples show that the transfer of batch-derived parameters to column experiments results in good agreement if the sorption processes are identical in both approaches and if these sorption processes can be described in terms of isotherms. Conversely, if unknown sorption kinetics and hysteresis occur in column experiments, the prediction of breakthrough curves based on batch-derived data should fail.

The nonreactivity of ferrocyanide in neutral sandy aquifer material assumed by Ghosh et al. (1999) may be caused by a strongly advective flow regime. As they did not give any information on the dispersivity of the system nor on sorption parameters, nonreactive behavior of ferrocyanide could not be inferred from their data. Rate-limited ferrocyanide sorption causing early breakthrough combined with a strongly advective flow regime would yield a ferrocyanide breakthrough curve as they presented. Generally, goethite-coated sand at pH > 6 as used in this study can be regarded as a kind of neutral sandy aquifer material which retarded both Fe-cyanide complexes.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
REFERENCES
 
The transport of both Fe-cyanide complexes in goethite-coated sand is characterized by rate-limited and nonlinear sorption. In the case of ferricyanide, the sorption on goethite-coated sand in column experiments is only roughly reflected by batch-derived Langmuir isotherm parameters, because the effects of flow interruption on breakthrough curves are poorly simulated using these parameters. This might be caused by different sorption and desorption kinetics resulting in hysteresis. A model including hysteresis should be used to simulate ferricyanide breakthrough curves in goethite-coated sand rather than a single isotherm approach.

The transfer of batch-derived sorption parameters on breakthrough curves derived from column experiments fails in the case of ferrocyanide. This might be caused by the presence of different sorption mechanisms implying the necessity to use more than one approach to describe ferrocyanide sorption during transport. A single sorption isotherm as in batch experiments cannot be used to describe the sorption of ferrocyanide on goethite in column experiments. Further research is necessary to enlighten the sorption processes of ferrocyanide during transport.


   ACKNOWLEDGMENTS
 
This study was part of the Priority Program 546 "Geochemical processes with long-term effects in anthropogenically affected seepage- and ground water." Financial support was provided by Deutsche Forschungsgemeinschaft. We thank Dr. K. Bajracharya, Department of Natural Resources and Mines Queensland, Indooroopilly, Australia, who contributed the MCMFIT computer code. The goethite was prepared by H. Biernath, Ruhr-Universität Bochum.

Received for publication February 5, 2002.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 CONCLUSIONS
 REFERENCES
 




This article has been cited by other articles:


Home page
 Soil Sci.Home page
J. M. Phillippi, V. A. Loganathan, M. J. McIndoe, M. O. Barnett, T. P. Clement, and E. E. Roden
Theoretical Solid/Solution Ratio Effects on Adsorption and Transport: Uranium(VI) and Carbonate
Soil Sci. Soc. Am. J., March 12, 2007; 71(2): 329 - 335.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (3)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rennert, T.
Right arrow Articles by Greef, K.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Rennert, T.
Right arrow Articles by Greef, K.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Rennert, T.
Right arrow Articles by Greef, K.
Related Collections
Right arrow Soil Chemistry

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Journal of Plant Registrations Journal of
Environmental Quality		 The Plant Genome