| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Lehrstuhl Bodenkunde, Technische Univ. München, D-85350 Freising, Germany
* Corresponding author (kai.totsche{at}uni-jena.de)
Received 24 April 2005.
 |
ABSTRACT |
|---|
 TOPEXECUTIVE SUMMARY ABSTRACT INTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Abbreviations: DOC, dissolved organic carbon • DOM, dissolved organic matter • EC, electrical conductivity • NAPLs, non-aqueous phase liquids • FAU, Formazine Attenuation Units • PAH, polycyclic aromatic hydrocarbons • pv, pore volumes • TSTR, two-site/two region
|
INTRODUCTION |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Dissolution according to Raoult's law as the dominant release process might also be challenged by the presence of organic and inorganic colloids and suspended particles. These materials have been shown to substantially affect the mobility and transport of PAHs (Chiou et al., 1986; Grolimund et al., 1996; Villholth 1999; MacKay and Gschwend 2001; Kim and Corapcioglu 2002; Kögel-Knabner et al., 2000; Totsche and Kögel-Knabner 2004). Different processes result in the release, transport, and redistribution of colloids and particles, like changes of the solution's pH or the ionic strength, or the increase of hydrodynamic forces due to infiltration of rainwater (McDowell-Boyer 1992; Ryan and Gschwend 1994; Ryan and Elimelech 1996; Kretzschmar and Sticher 1997; Bunn et al., 2002). A qualitative and quantitative understanding of the controls of colloid/particle release and transport and the possible kinetic limitations is therefore an essential prerequisite for the understanding of PAH fate at contaminated sites.
We present a study on the release and transport of PAHs, DOC, and particles from NAPL-contaminated gravelly soil material. The materials originate from an abandoned industrial site, which was contaminated with aged tar oils. Expecting the mobilization and transport of even larger particles due to the macroporosity of this gravelly material, particles up to the size of 200 µm were investigated. To distinguish between PAHs associated with large colloids or suspended particles on one hand, and PAHs associated with small colloids or in dissolved form on the other hand, column effluent was filtered at 0.7 µm. As possible kinetic limitations to the release can only be detected within a small range of the ratio of the mass-transfer timescale to the transport timescale, an experimental design introduced by Wehrer and Totsche (2003, 2005) was used which employs two different flow rates and at least two flow-interrupts of different duration.
|
MATERIAL AND METHODS |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
At present, the site is prepared for building and construction. This includes excavation, filling, levelling, and compaction. Such activities will result in the disruption of the integrity of the NAPL interfaces. Concomitantly new NAPL surfaces are formed and exposed while others are coated with mineral soil material. Thus, it is expected that the construction activities will have a severe effect on the release of PAHs from disturbed residual NAPLs.
Physical and Chemical Properties of the Soil Material
Particle size analysis was done by sieving (>2 mm), wet sieving (sand fractions), and a sedimentation method with X-ray attenuation measurement for the silt and clay fractions [Sedigraph 5100, Micrometrics GmbH, Moenchengladbach, Germany]. The fraction (<2 mm) was analyzed for contents of organic carbon (C) and carbonate in duplicate by dry combustion [CN-Analyzer Vario EL, Elementar, Germany]. The contents of total and inorganic C were measured in air-dried samples and after ignition at 550°C, respectively. The organic C content was calculated from the difference between total and inorganic C content. The pH values were determined in deionized water and in 0.01 M CaCl2 solution (Avery and Bascomb, 1974). The oxalate and dithionite soluble iron (Fe) and manganese (Mn) were extracted according to Schwertmann (1964) and Mehra and Jackson (1960) and quantified by ICP–OES [Vista Pro CCD Simultaneous, Varian, Germany]. Bulk densities were determined with an excavation method as described by Blake and Hartge (2002). The physical and chemical properties of the soil material are given in Table 1.
|
Column Study
The release of PAHs, DOC, and suspended matter was studied with packed soil columns under water-saturated flow conditions. Undisturbed sampling was not possible due to the large amount of coarse gravels with particle diameters >5 cm. The soil materials were manually excavated below a contaminant source area in a depth of 0.6 to 0.8 m. Particles with diameters larger than 3 cm were removed. To consider the effect of construction activities on the integrity of the residual NAPL, the sample pretreatment comprised air-drying and homogenization. The packing procedure resulted in homogeneous soil columns with bulk densities of 2 g cm–3.
A sketch of the soil column system is given in Fig. 1 . To minimize PAHs sorption, the columns (height: 15 cm; i.d.: 9.4 cm) and the porous plates [200 µm mesh, emc GmbH, Germany], which were used as bottom and top capping, and all tubing are made of stainless steel. The solution storage bottle and the fraction collector test tubes are made of glass. A peristaltic pump [Ismatec, Gattbrugg, Switzerland], installed upflow of the columns, was used to feed the solution. Column effluent was collected with a fraction collector [Spectrum, Houston, TX]. The experiments were run at 20°C in a climatic chamber.
|
Applied Flow Scheme and Analysis of Transport Properties
Mass transfer of PAHs to the liquid phase may be kinetically limited and can be detected by flow-interrupts (Brusseau et al., 1997). In short-term column experiments, however, flow-interrupts are only effective within a small range of the ratio of the mass-transfer timescale to the transport timescale. Wehrer and Totsche (2003), (2005) showed that the detectability of rate-limited mass transfer can be improved if two columns are used which are percolated at sufficiently different mean pore water velocities and if the flow is interrupted at least twice.
We kept a constant mean volumetric flow rate of 11 mL h–1 for the slow column and 50 mL h–1 for the fast column (i.e., 0.044 and 0.20 pore volumes per hour, resulting in mean pore water velocities of 6.4 and 30 mm h–1). For both columns, six pore volumes were exchanged before the flow was interrupted for 1 d. Flow was resumed for another five to six pore volumes before it was interrupted a second time for 5 d. Finally, six pore volumes were exchanged, again.
Analysis of the transport regime was done with chloride (10–2 M NaCl) as nonreactive tracer. To check the homogeneity and reproducibility of the packing procedure, we ran a third column prepared in the same way as the two others and ran a tracer transport experiment. A flow-interrupt was also imposed on the breakthrough of the tracers to check whether physical or chemical processes are responsible for the rate limitations. Column dispersivities 
= D/v (D: coefficient of dispersion; v: mean pore water velocity) were obtained by fitting of the chloride breakthrough curves using the local equilibrium assumption and the Two-Site-Two-Region models in comparison (Parker and Van Genuchten, 1984).
Analytical Methods
The chemical analysis of the effluent comprised the determination of pH [ion-sensitive electrode, SenTix 41, WTW, Weilheim, Germany] and electrical conductivity [TetraCon 625 conductivity cell, WTW, Germany]. The breakthrough curve of chloride was measured with an ion-selective electrode [Ionplus Chloride, Thermo Electron, Waltham, MA]. Dissolved organic carbon was determined as nonpurgeable organic C using a TOC-Analyzer [5050A, Shimadzu, Japan] after filtration <0.45 µm and acidification.
Turbidity was determined by spectral absorption measurement at 860 nm [Cary 50 UV-Vis Spectrophotometer, Varian, Darmstadt, Germany] in samples shaken horizontally for 10 s after allowing 1-min settling time, and given as Formazine Attenuation Units (FAU). The adsorption at 254 nm, which is a relative measure for the aromaticity of DOC, was determined to calculate SUVA254, which is defined as the UV absorbance divided by the DOC concentration.
Before the extraction of PAHs, the effluent was filtered with fiberglass filters with a mesh size of 0.7 µm [GF 92, Schleicher & Schuell MicroScience GmbH, Germany]. This pore size represents the smallest commercially available glass-fiber fiberglass filter. The extraction of the 16 EPA-PAH priority pollutants in the filtrate was done with solid-phase (Chladek and Marano, 1984). The PAHs in the retentate (particle fraction 0.7–200 µm) were extracted according to Hartmann (1996). The concentrations in the fraction 0.7–200 µm are given in µg L–1: The detected PAH masses are based on the sample volume which passed the 0.7 µm filter.
The PAHs were analyzed using a gas chromatograph coupled to a mass selective spectrometer [GC 8000, MD 800, Fisons Instruments, Beverlly, MA] supplied with a DB 5 MS column [internal diameter 0.25 mm, film thickness 0.25 µm; J. and W. Scientific, Folsom, CA]. The oven temperature program was as follows: 1 min 85°C, 85 to 160°C (15°C min–1), 160 to 300°C (5°C min–1), 300°C (15 min); injector temperature: 280°C; splitless injection. The PAHs were quantified with a mixture of seven deuterated PAHs [PAH surrogate cocktail, Cambridge Isotope Laboratories Inc., Andover, MA]. PAHs recoveries of the internal standard were quantified by adding an external standard [Perylene D-12, Supelco, Sigma-Aldrich, Muenchen, Germany].
The analyses of the PAHs were done twice. The analytical results plotted in the graphs are given by means of two individual determinations with error bars indicating maximum and minimum of the measured concentrations of the PAHs. The difference between the two measurements was generally small.
Data Evaluation
Parameters characterizing the rate-limited release of DOC and PAH were analyzed by data obtained from the breakthrough curves. The effective mass transfer coefficient keff [T–1] was calculated according to Münch et al. (2002):
 | [1] |
[L3 L–3], Ceq [M L–3], Cact [M L–3], and Ci [M L–3] denote the duration of the flow-interrupts, the volumetric water content, the equilibrium concentration, the concentration after the flow was resumed, and the effluent concentration before the flow-interrupt, respectively. The parameters Ceq and keff were calculated using the concentration before and after the flow-interrupts by inverse simulations with Mathcad 2000 Professional [Mathsoft Inc., Muenchen, Germany].
The ratio of the reaction-time scale to transport-time scale is described by the dimensionless Damköhler number Da [–]. It is used as a measure for the degree of nonequilibrium (Bahr and Rubin, 1987):
 | [2] |
 | [3] |
|
RESULTS AND DISCUSSION |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
|
Figure 3 shows the course of pH and electrical conductivity (EC) of the slow and fast column. The pH values range from 7.1 to 7.8 for the slow column and from 6.6 to 7.4 for the fast column. Compared to the input pH of 7.0, the slow flow conditions result in an increased pH over the whole experiment. The fast column shows no difference to the input pH after 3 pv were exchanged. No response to the short flow-interrupt is observed for both columns, but the longer stop leads to increased values after flow was resumed.
|
The effluent pH should be affected mainly by the dissolution of carbonates and by the release of organic matter which results in the formation of dissolved organic matter (DOM). While the first would result in a pH increase compared to the unbuffered and neutral input solution, the release of organic matter should lower the pH due to a deprotonation of, for example, carboxyl-functional groups. For both columns, effluent pH is higher than the pH of the input solution, indicating that the effect of the dissolution of carbonates outbalances the effect of DOM formation. The higher effluent pH of the slow flow column points to the fact that the dissolution of carbonates and the development of the pH is rate limited. The longer residence time in the slow column results in a prolonged time for the dissolution process. Rate-limited dissolution is also corroborated by the marked pH raise in the fast column at the longer flow-interrupt (120 h). Under flow conditions, however, the equilibrium pH of the soil (pHH20 = 9.0, Table 1) is never reached. Compared to the flow through system, where the solid/solution ratio is smaller than one, the equilibrium pH of soils is measured in a dilute aqueous suspension with a solid/solution rate in between 1:1 and 1:2.5. Dilution is known to increase pH.
The EC is controlled by the level of the EC in the input solution and by possible interactions of the solution with the soil material. The release and dissolution of carbonates should result in an increase of the effluent EC. The formation of DOC, however, will not inevitably result in an increase of EC. It rather depends on the presence and amount of functional groups and the protonation state of DOC. For both columns, the effluent EC is up to 130 µS cm–1 higher than that of the input solution (1160 µS cm–1). Together with the concomitant pH raise, we can conclude that the dissolution of carbonates contributes most to the EC of the effluent and that dissolution of organic matter rich in ionic moieties is of minor importance. The higher EC level of the slow column is the consequence of rate limited release, which affects the carbonates, the pH, and, of course, also the EC. This, again, is supported by the marked response to the longer flow-interrupt.
The fact that the EC declines for both the fast and the slow column after 6 and 10 pv have been exchanged, indicates that the pool of readily dissolvable carbonates is exhausted. From now on, the effluent EC is mainly controlled by the level of the input solution.
Release of Dissolved Organic Carbon
Initially after the onset of the flow, maximum DOC concentrations of 20 and 16 mg L–1 were observed for the slow and the fast column, which are rapidly declining within the first 2 pv (Fig. 4
). Then, a more constant and slightly declining DOC effluent level is observed which lasts for the rest of the experiment. In general, the slow flow column strays in excess of the fast column. Again, the short flow-interrupt has no significant effect on the DOC effluent concentrations while the longer flow-interrupt does so. At the end of the experiment, effluent concentrations are reduced to a constant level of about 12 mg L–1 for the slow and 4 mg L–1 for the fast column.
|
The first flush is followed by a constant period of DOC export, which is due to a fraction of DOC released under rate-limited conditions. The fact that this fraction shows slightly declining effluent concentrations reflects the situation that a finite pool of DOC is continuously exhausted.
Weigand and Totsche (1998) explained the two-step release by the existence of at least two chemically different fractions of DOC in soils. One fraction is nonreactive with respect to the interactions with the soil matrix while the other is controlled by rate limited release characteristics. The assumption of at least two differently reactive DOC fractions is also supported by Wehrer and Totsche (2005) and our own measurements of SUVA254. The UV radiation near 254 nm is absorbed by C = C double bonds. Any changes of UV absorbance are a measure for the relative change of the content of such moieties (Chin et al., 1994). The course of SUVA254 shows initially increasing values and decreasing or constant values after the flow interrupts in each column (data not shown). The content of C = C double bond moieties is generally greater at the higher flow velocity. However, although the different moieties of DOC show different release behavior, SUVA254 does not represent either the reactive or nonreactive fraction (Weishaar et al., 2003).
Except for the first flush, the response of DOC is the same as observed for pH and EC. Again, rate-limited transport explains the observed release behavior which is supported by the higher yield at the slower flow velocity and the markedly increased effluent concentration after the longer flow-interrupt. The estimation of rate parameter keff and equilibrium concentration Ceq reveals the same keff values for the slow and the fast column. The calculated values are given in Table 3. They are in a similar range as those published by Münch et al. (2002) (keff = 1.6 x 10–3 h–1, Ceq = 94 mg L–1). However, they found a three times higher equilibrium concentration. The calculated R-dependent Damköhler numbers Da' for the slow and the fast column are 0.021 and 0.005, respectively (Table 3). These low Damköhler numbers indicate nonequilibrium conditions, that is, slow reactions.
|
|
The observation that the slow column produces higher turbidity than the fast one was not expected. This rather implies that the particles and colloids which cause turbidity are released under rate-limited conditions as reported among others by Lægdsmand et al. (1999) for undisturbed soil cores. The decrease of turbidity during the longer flow-interrupt, however, contradicts rate-limited release. Other counterproductive processes seem to outbalance the rate limited release, for example, the destabilization of the colloidal solution during the flow interrupt. This might even be enforced by the increase of the EC which would result in a more efficient coagulation followed by sedimentation during no-flow conditions.
Course of the Particles in the Retentate (Fraction 0.7–200 µm)
At the onset of the flow, the release of particles in the size fraction 0.7 to 200 µm shows initially higher values (75 mg L–1) for the fast column than for the slow column (53 mg L–1) (Fig. 5). The initially high concentrations rapidly decline during the exchange of the first two pore volumes within both columns. Subsequently, effluent concentrations increase and reach values of 32 and 38 mg L–1 for the slow and the fast column, respectively. This wavelike raise and fall of the concentrations of the particles lasts until the very end of the experiment. Both the short and the longer flow interrupt have no effect on the effluent concentration. Compared to the slow column, higher effluent concentrations are observed only for the first two pore volumes for the fast column. Thenceforward, either equal or even smaller concentrations are found for the fast column.
As already discussed for DOC, the initially high release is caused by the export of particles and larger colloids which were formed during pretreatment of the soil. Drying and homogenization are known to result in the formation of loosely attached particles. Upon onset of flow, these materials are exported to produce the first flush. This situation seems to be quite typical for soils at (abandoned) industrial and urban sites. Here, the ongoing disturbance due to construction activities including excavation, crushing, translocation, and backfilling of soil materials results in disintegration of profile build-up, soil structure, and aggregates (Totsche et al., 2003a). It is to be expected that a large amount of particulate materials are relocated in the unsaturated zone and transported into deeper layers. The fact that the first flush wears off after two pore volumes have been exchanged suggests that these materials originate from a finite pool which is exhausted within a limited space of time.
The export of particles in the larger size fraction shows no clear dependence on flow velocity and no marked response to the flow-interrupts, indicating that the release is not that much dependent on the residence time. This contrasts the findings we have for EC and DOC which were found to be sensitive to both the flow velocity and the flow-interrupts. From these observations we conclude that release and mobility of particles in the large-size fraction is predominantly independent of pH and EC. A weak correspondence of the large fraction with turbidity is found. As turbidity measurement is most sensitive for colloids/particles in the size of a few micrometers (Gippel, 1995), obviously, larger particles dominate in the fraction 0.7 to 200 µm.
Another notable finding is that in the beginning the fast column produced more particles, while in the end the slow column had higher particle concentrations in the effluent. This can neither be observed for the turbidity nor for DOC. We believe that the higher flow velocity is more effective and relevant for the larger size particles than for DOC, EC, and turbidity. This implies that for the larger size fraction we have to consider a third process which affects their release behavior, that is, particle detachment and transport due to higher shear forces. This is supported also by the observation that the mean pore velocity affects only the first flush, which is higher at the higher flow velocity. The pool of particles which can be hydraulically mobilized is limited, which can be concluded from the fact that the slow column produces higher particle concentrations than the fast one.
Release of PAH in the Filtrate <0.7 µm and Comparison to Raoult's Law
Initial export of PAH is characterized by small but increasing concentrations for both columns with generally larger concentration for the slow column (Fig. 6
). Marked response of the PAH breakthrough was observed for both flow-interrupts with a more expressed reaction for the fast column. The observed maximum concentrations after flow has been resumed, that is, 145 µg L–1 in the slow and 142 µg L–1 in the fast column, were almost independent of the duration of the flow-interrupt. At the end of the experiment PAH effluents showed constant concentrations of 80 µg L–1 (slow) and 20 µg L–1 (fast), respectively.
|
Another observation is that the maximum PAH concentrations reached after the flow-interrupt does neither depend on the duration of the flow-interrupt nor on the flow velocity. In the fast column, the longer flow-interrupt results in almost the same effluent concentration as in the slow column. Moreover, during the longer flow-interrupt the PAH concentration raise is small in the slow column and independent of the duration of the flow-interrupt. This indicates that equilibrium is almost accomplished in the slow column and that the flow-interrupts, even in the fast column, results in almost complete equilibration.
Rate-limited PAH release is also supported by the fact that effluent PAH concentrations increase until the first flow-interrupt. Possible processes are diffusion-limited desorption from the solid phase or the rate-limited dissolution from the aged NAPL phase. Rate-limited mass transfer from immobile to mobile regions could be excluded as no indications for this are observed for the conservative tracer.
A comparison of the release of DOC and EC with the PAH in the filtrate reveals a strong correspondence for the remaining part of the breakthrough after the first flow-interrupt. Although PAH in the small fraction do not show a first flush, a mobilizing effect of DOC fractions might be assumed. While the readily available and nonreactive fraction of DOC which caused the first flush does not affect the transport of PAH, the DOC fraction characterized by rate-limited release might contribute or even control the rate-limited export of PAH. A mobilization of PAH due to the presence of co-solvents or natural biosurfactants could be excluded. Measurements of the surface tension of the seepage water collected from the investigated site showed no decreased values.
The calculated rate parameter keff and the equilibrium concentration Ceq of PAH are the same for both flow velocities within the found errors (Table 3). The ratios of reaction time scale to transport time scale expressed as the R-dependent Da' are 0.022 for the slow and 0.005 for the fast column (Table 3). Again, the low Damköhler numbers indicate non-equilibrium conditions.
To test the applicability of Raoult's law, aqueous equilibrium concentrations of the single PAH are calculated and compared to the measured concentrations in the filtrate. According to Raoult's law the concentration of a solute in the aqueous phase is controlled by its mole fraction in the NAPL and its aqueous solubility. The equilibrium concentrations were calculated using the PAH concentrations in the organic mixture, the measured molecular weight of 295 g mol–1 and the subcooled liquid solubilities of the PAH. To cover a span of coal tar types, we assumed minimum and maximum coal tar molecular weights between 230 and 780 g mol–1 (Lee et al., 1992). The molar fractions of PAH were calculated based on the concentrations in the soil material which were scaled to the mean coal tar content of the soil (1.0 g kg–1). The subcooled liquid solubilities were calculated after Peters et al. (1997) using aqueous solubilities from Mackay and Shiu (1977) and Walters and Luthy (1984).
Figure 7 shows the calculated aqueous equilibrium concentrations after Raoult and the measured concentrations in the filtrate. The comparison reveals no similarity of measured and calculated concentrations. For the identified 14 PAH (Naphthalene and Phenanthrene are below their detection limit), only the concentrations of Benz[a]anthracene, Chrysene and Dibenz[a,h]anthracene are in the range of the calculated values. The uncertainty of our estimation affects only the absolute values and not the pattern of the PAH. From these results we conclude that Raoult's law does not solely explain the observed concentrations and that additional processes control the release of PAH. A possible explanation could be the rate-limited formation and mobilization of small NAPL-"droplets" in the size of small colloids, which has been discussed, for example, by Pumphrey and Chrysikopoulos (2004) or the detachment of small fragments of the NAPL source material.
|
|
With the lasting flow, the first-flush wears off and so does the excess export of the particle-associated PAH in the fast column. The later stage of the experiment, in particular after eight pv have been exchanged, is dominated by rate-limited release of particle-associated PAH.
As already discussed, the effect of the higher flow velocity is more effective and relevant for the particles and such for the particle-associated PAH. The export of the particle-bound PAH is thus affected by processes already discussed for the release of particles: First flush and particle detachment and transport due to higher shear forces and, to a minor extent, also rate-limited release of particle-associated PAH. The role of the individual process is thereby controlled by the size of the respective pools.
The comparative analysis of the pattern of the PAH in the soil, the filtrate, and the retentate should shed further light on the governing release and transport pathways. If PAH export is controlled by the release and transport of the NAPL source material in the form of droplets or fragments, we would expect the same distribution pattern of the PAH in the different fractions. We compare the PAH pattern of the soil material, the retentate (>0.7 µm), and the filtrate (<0.7 µm). The results are given in Fig. 9 . Indeed, the PAH patterns are almost the same for the retentate and filtrate of both columns and for the soil material. All distributions show dominance of the higher molecular-weight PAH (>3-rings) which account for about 90% of the total PAH. The fact that the PAH patterns are similar in all fractions, independent of flow velocity and particles size, suggests that NAPL transport mainly occurs in the form of particles of different sizes. Small NAPL droplets or fragments <0.7 µm are mobilized under the rate-limited process as discussed above. The NAPL particles in the size of 0.7 to 200 µm are released by the three processes, that is, the first flush, the hydraulic mobilization and, to a lesser extent, the rate-limited formation and release.
|
|
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
The pH, EC, and turbidity are also found to be affected by non-equilibrium. This suggests that processes that are affected on their part by pH and EC, for example, the stability of colloidal solutions, should then be controlled by non-equilibrium. For prolonged no-flow conditions, one would expect that the increase of EC due to the dissolution of carbonates should result in the destabilization of colloids, thus affecting also any contaminants associated with these colloids. This is indeed found for turbidity and for the particle-associated PAH in the retentate. However, the release of particles (0.7–200 µm) was independent of pH and EC. We suggest that in our materials the dominant release processes for particles are the first-flush, the particle detachment due to hydraulic mobilization and, to a lesser extent, the rate-limited release.
For the low flow velocity, 33% of the total PAH export is found in the retentate (two-thirds in the filtrate), while for the high flow velocity the amount of particle-associated PAH increases to 42% of the total PAH. The comparison of measured concentrations in this fraction and aqueous equilibrium concentrations calculated according to Raoult's law shows that dissolution of PAH from NAPL seems to be of minor importance. This, however, might be different in groundwater environments with low lateral flow velocities, where the prolonged residence might be high enough to more closely achieve the dissolution equilibrium.
The PAH in the filtrate (fraction <0.7 µm) are mobilized under rate-limited conditions and show strong correlation to the fraction of DOC released under rate-limited conditions. This suggests that in the filtrate PAH seem to be closely connected with the DOC. One possible explanation would be that the DOC itself is part of the NAPL phase which is released in form of small fragments or droplet. This is also corroborated by the similarity of the PAH patterns of the filtrate and the retentate.
The particle-associated PAH account for up to 42% of the total exported PAH in our gravelly soil material. This transport process should be more thoroughly considered in risk assessment at contaminated sites. For the particle-associated PAH, we conclude that the dominant release processes are once again the first flush, the hydraulic mobilization and, to a lesser extent, the rate-limited release of PAH bearing NAPL fragments or droplets.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPEXECUTIVE SUMMARYABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  K. U. Totsche, S. Jann, and I. Kogel-Knabner Single Event-Driven Export of Polycyclic Aromatic Hydrocarbons and Suspended Matter from Coal Tar-Contaminated Soil Vadose Zone J., April 9, 2007; 6(2): 233 - 243. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||
