| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Dep. of Soil Science, Inst. of Ecology, Berlin Univ. of Technology, Salzufer 11-12, D-10587 Berlin, Germany
b Dep. of Soil Protection, Inst. of Ecology, Berlin Univ. of Technology, Salzufer 11-12, D-10587 Berlin, Germany
c Dep. of Geochemistry, Hydrogeology and Mineralogy, Freie Universität Berlin, Malteser Straße 74-100, D-12249 Berlin, Germany
d Dep. of Soil Science, Inst. of Ecology, Berlin Univ. of Technology, Salzufer 11-12, D-10587 Berlin, Germany
* Corresponding author (katrin.ilg{at}tu-berlin.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: CDE, convection–dispersion equation.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
To date, colloids from natural soils are sampled by extracting soils in the laboratory with various solutions and separating the dissolved and colloidal phases afterward (Rhea et al., 1996; Sequaris and Lewandowski, 2003). Most of these methods, however, produce artifacts, e.g., by shaking, and therefore provide only a relative measure of colloid concentrations. They do not reflect the actual concentrations and mobility of colloids in soil solution under field conditions. Under field conditions, various kinds of lysimeters are used to collect colloids via seepage water sampling (Gasser et al., 1994; Kaplan et al., 1997; Worall et al., 1999). The most frequently used lysimeter types are wick samplers, suction plates, and zero-tension lysimeters. As Thompson and Scharf (1994) noted, however, not all types of lysimeters are equally suited for colloid sampling. Filtration and trapping of colloids in porous suction plates or cups with an average pore diameter of a few micrometers as well as in fiberglass wicks may tamper colloid sampling (Grossmann and Udluft, 1991; Czigany et al., 2005). Additionally, installation of lysimeters may disturb the surrounding soil and lysimeters, especially zero-tension lysimeters, and might influence the flow of water in the soil (e.g., Abdou and Flury, 2004), which in turn might affect colloid transport. Thompson and Scharf (1994) recommended specially constructed zero-tension lysimeters to sample and monitor colloid concentrations in the field. These lysimeters consist mainly of a sampling cup, which is connected to the soil surface by a tube and covered by a 150-
m polyester membrane, and are installed below an excavated, and afterward replaced, soil core. They did not demonstrate the superior performance of their system compared with others, however. Czigany et al. (2005), e.g., found that fiberglass wicks are also suitable for colloid sampling, but only under certain conditions, particularly at high pH.
The objective of our study was to quantify the colloid-sampling efficiency of five different lysimeter systems (1.2-µm membrane, 10-µm membrane, porous plate, wick, and zero tension) under unsaturated conditions. We used 59Fe-labeled goethite as a colloid.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
m, polyester (Pall, Portsmouth, UK); (ii) membrane, pore size 10 m, polyester (Franz Eckert GmbH, Waldkirch, Germany); (iii) membrane, pore size 200 m, polyamide (Franz Eckert GmbH, Waldkirch, Germany); (iv) porous plate, nominal maximum pore size 16 m, thickness 5 mm, sintered borosilicate (no. 50904, ROBU, Hattert, Germany), hydraulic conductivity = 17 mm h–1; and (v) wick, 1.3-cm diameter, length 50 cm, glass fiber, 8-m diameter (no. 1381, Pepperell Braiding Co., East Pepperell, MA).
The collection systems were fixed at the bottom of Plexiglas columns with a length of 30 cm and an inner diameter of 8 cm (Fig. 1
). A 30-cm hanging water column in a silicone tube (0.3-cm i.d.) generated suction for porous plates and 1.2- and 10-m membranes. The volume of water in the hanging tube was negligible compared with the volume of water in the columns. Therefore, the tubes did not influence transport parameters of the tracer or colloids. Fiberglass wicks are self-priming and act as a hanging water column according to the height difference of the wick without external application of suction (Boll et al., 1992). The 200-m membrane lysimeters drained freely into the sampler and are denoted as zero-tension lysimeters in the following. We tested the 20 columns in two runs with different colloid concentrations. One replicate of each lysimeter type appeared in both runs and the remaining three replicates were randomly distributed.
|
M NaH2PO4 solution for 24 h. A pH of 6.7 was adjusted with NaOH. On top of the columns, an irrigation system was installed, which consisted of a Plexiglas block containing 45 cannulae with an inner diameter of 0.4 mm at the bottom side (Fig. 1). Peristaltic pumps fed the columns with solution. An automatic sampler collected column outflow.
Colloid Suspension
We synthesized goethite by oxidation of Fe(II) (FeSO4•7H2O) with H2O2 at a pH of 7. Goethite was washed until the water had an electrical conductivity of 10 S cm–1 and freeze-dried, and goethite aggregates were ground. X-ray diffraction analysis (D5050, Siemens, Germany) showed a pure goethite without contamination. Electron microscopic pictures showed crystals with an average size of 80 nm (S520, Hitachi America, Brisbane, CA). The specific surface area was 144.8 m2 g–1 (Autosorb-1, Quantachrome Instruments, Odelzhausen, Germany).
Two batches of goethite were irradiated with neutron radiation for 55 h in the research reactor of the Hahn-Meitner-Institut (Berlin, Germany) and allowed to decay for 3 d. Upon irradiation, 56Fe was transformed to 59Fe to some extent. Decay of 59Fe to stable Co59 emits ß– and 
-radiation. The half-life of 59Fe is 45.1 d (Metcalfe, 1976). Resulting activities after irradiation were 1.1 · 107 (first batch) and 1.6 · 106 Bq (second run).
We suspended the two batches of goethite in 1 and 0.5 L of 80 M NaH2PO4 and kept them in an ultrasonic bath for 30 min to break up agglomerates. Following ultrasonic treatment, the first batch of goethite suspended in NaH2PO4 solution was stirred for 24 h. Because of technical reasons, the second batch of goethite suspension was shaken end-over-end for 15 h. Filtration with a 1.2-m cellulose acetate filter (Sartorius, Göttingen, Germany) removed all particles not defined as colloids (Kretzschmar et al., 1999). The resulting suspensions had an activity of 12.3 Bq mL–1 in the first batch (equivalent to 0.1 mg goethite L–1) and 338 Bq mL–1 in the second batch (equivalent to 10 mg goethite L–1). For a nonradioactive, equally treated goethite, we measured an average particle size of approximately 400 nm after filtration (High Performance Particle Sizer, Malvern Instruments, Malvern, UK).
The resulting adsorption of phosphate shifts the isoeletric point of goethite from pH 7 or 8 down to approximately pH 4 (Stumm and Sigg, 1979). Therefore phosphate-covered goethite colloids were negatively charged in our experiment, which we conducted at pH 6.5 to 7 (Puls and Powell, 1992). The zeta-potential of a nonradioactive, equally treated goethite at pH 6.5 was –29.8 mV (DTS 5200 Zetasizer 2000, Malvern Instruments, Malvern, UK). We supposed the negatively charged colloids to be repelled from the phosphate-covered sand matrix and therefore to be more mobile than goethite colloids without adsorbed phosphate.
Column Experiment
We irrigated the columns in a steady-state, unsaturated-flow modus with an average irrigation rate of 58 mm h–1 (in the pre-experiment, 2.5 mm h–1). Irrigation rates ranged from 51 to 65 mm h–1. Before application of the goethite suspension, columns were irrigated with 80 M NaH2PO4 solution until constant water contents and flow rates were established for each individual column. We applied 26 mg NO3 L–1 in the first run and 99 mg NO3 L–1 in the second run as a conservative tracer in the colloid suspension. Irrigation for application of the tracer and colloids took 5 min, after which we continued irrigation with the background electrolyte (80 M NaH2PO4). Each run lasted 1.5 h. During the first third of each run, samples were collected every minute, during the second third every 2 min, and during the last third every 6 min.
The outflow was determined gravimetrically for each sample. We detected NO3– with a photometer at a wavelength of 203 nm (Specord 200, Carl Zeiss Technology, Jena, Germany). The detection limit was 0.32 mg NO3 L–1. The radioactivity of the samples was determined with a Ge detector (Canberra Industries, Meriden, CT) and the data processed with an analog–digital converter (7070 ADC, FAST Comtec Datensysteme GmbH, Oberhachingen, Germany) and MCD/PC operating software (FAST Comtec Datensysteme). The detector recorded -radiation at 1098 and 1292 MeV. We considered only the signal at 1098 MeV, because transition probability was larger than for the peak at 1292 MeV. Integration time of the first run was 1200 s and of the second run only 500 s because of the larger activity.
After the irrigation experiment, we selected one column of each lysimeter type from the second run with large colloid concentrations to separate the sand into three layers—upper, middle and lower part—to determine the radioactivity. Similarly, one wick was cut into three sections. A defined amount of active goethite solution mixed with uncontaminated sand or wick was used for calibration. Unfortunately, it was not possible to determine the activity of porous plates and membranes because the inhomogeneous distribution of 59Fe hindered proper calibration. Sand and wick samples were measured in a Ge detector (Canberra Industries, Meriden, CT) using an integration time of 500 s.
Adsorption of Goethite to Sand
To check if goethite colloids sorb to the quartz sand, samples of 7 g of phosphate-covered sand were shaken for 15 h end-over-end with 50 mL of phosphate-covered goethite suspensions with increasing concentrations (5–400 Bq mL–1). Afterward, we decanted the solution and measured the activity in the supernatant. Adsorption of goethite colloids to sand was negligible.
Calculations and Statistical Evaluations
We conducted a stationary experiment with constant-flow conditions. Transport of solutes through a homogeneous porous medium can be described with the convection–dispersion equation (CDE) (Wierenga and van Genuchten, 1989; Vanclooster et al., 1993). Deposition of colloids can be described by an additional first-order deposition term. Colloid deposition occurs either directly at the water–solid or the air–water interface. Commonly, remobilization back to solution is small and can be neglected (Schäfer et al., 1998). The CDE for colloid transport with a first-order deposition term is:
 | [1] |
is the average transport velocity (mm h–1), and d is the coefficient of deposition (h–1). We fitted D and for NO3– and D, , and d for colloids. All parameters were fitted simultaneously using the measured breakthrough curves of the tracer and the colloids with the software CXTFIT 2.1 (Toride et al., 1998). Statistic calculations were done using STATISTICA 6.0 software (StatSoft, Tulsa, OK). All data were tested for normality with the Shapiro–Wilk W test (Shapiro et al., 1968) and for homogeneity of variances using the Levene's test (Sachs, 1982, p. 498). We checked the significance of differences between mean values with the Tukey test (Winer et al., 1991, p. 351–354). Not normally distributed data were tested with the nonparametric Kruskal–Wallis H test (Winer et al., 1991, p. 1028–1029) and the Mann–Whitney U test (Dineen and Blakesley, 1973). Parameters of NO3– and goethite transport were compared with the parametric paired t-test and the nonparametric paired Wilcoxon test (Sprent and Smeeton, 2001, p. 124–129). We used a level of significance of P < 0.05 for all tests.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
m membrane lysimeters failed, probably because of an incorrect application of tracer and colloids. No NO3– and no 59Fe could be detected in the outflow of this column, although water flowed through the column. Therefore we took only three lysimeters of this type into account. In one 10-m membrane lysimeter and one porous plate lysimeter, water did not flow through continuously, therefore only a few samples were collected.
Water Regime and Nitrate Transport
Nitrate breakthrough curves were fitted with CXTFIT with an average R2 of 0.72 ± 0.24 for 10-µm membrane, 0.98 ± 0.01 for wick, 0.86 ± 0.06 for zero-tension, 0.92 ± 0.08 for porous plate, and 0.95 ± 0.02 for 1.2-µm membrane lysimeters. Exemplary breakthrough curves and fits of every lysimeter type for both applied colloid concentrations are shown in Fig. 2
and 3
. A pronounced tailing for NO3– could be observed, but a mobile–immobile model approach (Cherrey et al., 2003) did not improve the fit. Average NO3– recovery was 102%, ranging from 90 to 108% with one exception of 145% for a 10-µm membrane column in the first run (Table 1). We ascribe recoveries above 100% to irregularities of peristaltic pumps, which probably delivered slightly more than the calculated amount of solution. The recoveries did not differ significantly between the lysimeter types. The coefficients of dispersion varied between 970 and 2910 mm2 h–1. Transport velocity ranged between 156 and 415 mm h–1 (Table 1). Transport parameters of NO3– did not differ significantly between the two runs.
|
|
|
|
|
and dispersion coefficients D were not significantly different for the smaller and larger colloid concentration; however, the recovery and hence the first-order deposition rate coefficient was (Table 2). For the small colloid concentration, was significantly smaller and therefore recovery larger than in the second run. Further, functioning of the lysimeter types was different at different colloid concentrations: while recoveries were similar for all lysimeter types for the small colloid concentration, results differed after application of the larger colloid concentration. Membranes of 10-m pore size and zero-tension lysimeters showed above-average recoveries of 59Fe in drainage, whereas recoveries of wick, 1.2-m membrane, and porous-plate lysimeters were below average (Table 2). By measuring the 59Fe activity in different segments of the columns of the second run, we established depth profiles, which illustrated the retention of colloids in the sand (Fig. 5 ). In all examined columns, the majority of colloids was retained within the upper 0 to 3 cm of the sand column. In the zero-tension lysimeter column, we detected no colloids in the lowest depth interval (Fig. 5e, 6–9 cm). The total recovery of 59Fe (sand, leachate, and wick) was 99 ± 19%.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
We assume a gradient of water content between the upper and lower parts of the columns, because of the steep water retention curve of the coarse sand. Water contents directly above the lysimeters are influenced by the lysimeters themselves, which is a natural characteristic of the collection systems and therefore also valid under field conditions. Therefore, the effects of differing hydraulic conditions in the sand columns and the effects of the lysimeters themselves on colloid transport cannot be separated from each other. In contrast to other systems, in a zero-tension lysimeter system water has to accumulate above the collection system until atmospheric pressure is reached before water can flow off (Abdou and Flury, 2004). But also the systems with membranes and porous plates influence the water content above the lysimeter, depending on the conductivity of the membranes and the plate compared with the conductivity of the sand. Due to the low unsaturated hydraulic conductivity of the coarse sand, the influence of the lysimeter on the hydraulic conditions is not as pronounced as in finer sands or silty materials.
Colloids
In the pre-experiment, all colloids were retained in the column because of (i) the length of the column (25 cm) or (ii) the small grain size of the sand. This confirms the efficiency of slowly percolating sand filters for the removal of colloids (Gimbel and Nahrstedt, 2004).
In our experiment with coarse sand and short columns, however, an accelerated transport of goethite compared with NO3– was observed, independent of the sampling system used (Fig. 4). According to Keller et al. (2004) and Sirivithayapakorn and Keller (2003), accelerated transport can be assigned to a size-exclusion effect: colloids move faster through porous media because they travel through wider pores compared with a dissolved tracer. The effects are small, however, and the statistical evaluation of transport parameters did not confirm significant differences of transport velocities and dispersion coefficients between goethite and NO3–, which we ascribe to the homogeneous pore system and the shortness of columns used in the experiment. Tailing of NO3– and colloid breakthrough curves cannot be caused by reactions with sand, because NO3– behaves like a conservative tracer and adsorption of goethite by the sand was tested and can be excluded. The exchange of NO3– and goethite between mobile and immobile regions of the pore system can be another reason for tailing (Cherrey et al., 2003); however, application of a mobile–immobile model of solute transport did not improve the goodness of fit.
We observed smaller recoveries at large colloid concentration and larger recoveries at small colloid concentration, although deposition should be independent of the applied colloid concentration and is assumed to follow a first-order kinetic rate law (Kretzschmar et al., 1997). Differing recoveries occur if hydraulic conditions, e.g., the flow rate, vary: larger flow rates cause larger water contents within the columns and therefore a smaller deposition of colloids (Cherrey et al., 2003; Mays and Hunt, 2005). But none of the hydraulic parameters differed between the two runs (Table 1). Li et al. (2004) observed a decreasing colloid deposition with increasing transport distance. This finding would correspond to our result if the effect is related to decreasing colloid concentrations along the transport path. Colloid concentrations may decrease due to straining, i.e., the deposition of colloid particles in downgradient pores, which are too small to allow particle passage (Bradford et al., 2004). The effect might be enhanced by pore clogging, because initially deposited colloids make the pores smaller and smaller. An effect of differing particle sizes on varying deposition of colloids in the two runs cannot be excluded, because we were not able to check the particle size of radioactive goethite colloids. Differing particle size distributions would strongly influence colloid migration (Bradford et al., 2004). Furthermore, it cannot be ruled out that the stability of the colloidal suspension was lowered due to the larger colloid concentration or slightly larger ionic strength in the second run because of the larger NO3– concentration (Kretzschmar and Sticher, 1997; Saiers and Lenhart, 2003).
Differences between lysimeter systems did not appear at small colloid concentrations and small deposition rate. Obviously filtration caused by different lysimeter systems did not significantly influence the outflow of colloids under these conditions. Large colloid concentrations and large deposition rates, however, might have increased filtration by membranes, porous plates, and wicks.
Compared with the porous plate, for the 1.2-m and the 10-m membrane lysimeters it is not very likely that relevant concentrations of colloids were deposited directly in the thin membranes. In the lowest depth segment, the 1.2-m membrane system showed the largest colloid concentration compared with the other systems (Fig. 5c). We assume that colloids accumulate at the surface of the membrane because of a filtration effect of the lysimeter. For the 10-m membrane, colloid concentration in the lowest depth segment was smaller and recoveries in the outflow larger because it did not filter as much colloid as the 1.2-m membrane (Fig. 5a). In the second run we had only one 10-µm membrane column. Anyway, we assume that the result is reliable because standard deviations of other column replications in this run were small compared with the measured recovery. The retention of colloids in fiberglass wicks corresponds with the findings of Czigany et al. (2005), who observed a recovery <5% for ferrihydrite transport through wicks at pH 7. They ascribe colloid retention to physicochemical deposition inside the wick, which was probably the reason for fragmentary breakthrough curves of wick samplers in our study. The zero-tension lysimeter columns retained no colloids in the lowest column segment at all. As we argued above, water saturation is larger in the lower part of this column. Under saturated conditions, colloid deposition is smaller and colloids are more mobile (Cherrey et al., 2003). Additionally no barrier as, e.g., a porous plate or a wick, retained colloids. This might explain the larger recovery of zero-tension lysimeters in the outflow compared with 1.2-m membrane, porous-plate, and wick lysimeters in the second run.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
Received for publication April 10, 2006.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. Siemens, K. Ilg, H. Pagel, and M. Kaupenjohann Is Colloid-Facilitated Phosphorus Leaching Triggered by Phosphorus Accumulation in Sandy Soils? J. Environ. Qual., October 23, 2008; 37(6): 2100 - 2107. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Weihermuller, J. Siemens, M. Deurer, S. Knoblauch, H. Rupp, A. Gottlein, and T. Putz In Situ Soil Water Extraction: A Review J. Environ. Qual., October 24, 2007; 36(6): 1735 - 1748. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
