Introduction

Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous organic contaminants that are synthesised during incomplete combustion processes. PAHs tend to adsorb onto solid particles (Abdel-Shafy and Mansour 2016; Moyo et al. 2014), namely soils and sediments acting as the main sink for these contaminants (Cerniglia 1992; Nam et al. 2008; Srogi 2007; Wilson and Jones 1993). Once accumulated in soils, PAH transference to biota and humans can pose serious threats (Lasota and Błońska 2018; Yakovleva et al. 2016; Zhang et al. 2015). Soil remediation techniques are therefore a key in the prevention of PAH transference. However, PAHs are not easy to remove from contaminated soil due to their hydrophobic nature. Regular remediation treatments usually encounter difficulties in PAH accessibility and mobility (Gan et al. 2009; Lima et al. 2012b; Ortega-Calvo et al. 2013; Rivas 2006; Wilson and Jones 1993). Recent developments have been made regarding the usage of clay in aiding PAH degradation via bioremediation (Biswas et al. 2015) and the use of rhizosphere phytoremediation (Bisht et al. 2015), all with proven in situ limitations (Lima et al. 2017). In fact, in situ remediation of clayey layers remains challenging because of their low hydraulic conductivity that hampers flushing of these layers.

In the beginning of the 1990s, electroremediation has been proposed as a suitable method for the in situ clean-up of low-permeable clayey soils and sediments from contaminants (Acar et al. 1995; Acar and Alshawabkeh 1993; Probstein and Hicks 1993). While the principle of electroosmosis to remove hydrophobic and strongly adsorbed contaminants like PAHs from soil is still under research (Ammami et al. 2015; Gan et al. 2009; Lima et al. 2017; Pamukcu et al. 2016), in situ soil remediation efficiency can be improved by the use of hydrophobic and hydrophilic surfactants, which enhance desorption of hydrophobic organic chemicals by improvement of the mass transfer from the solid into the aqueous phase by accumulating the chemicals in so-called micelles and decreasing the interfacial tension (Li and Chen 2002; Zhu and Zhou 2008). Although the feasibility of surfactant-enhanced electroremediation of clayey materials has been shown in numerous laboratory studies (Khodadoust et al. 2005; Lima et al. 2011; Saichek and Reddy 2003), only one study has been published until today, where the successful remediation of a contaminated field site at the pilot-scale has been reported (Lima et al. 2012a). In that study, a soil volume of roughly 1 m3 at 3.7–4.2 m below ground surface was treated for approximately 4 months at a field site of a former asphalt production plant, where a tar oil contamination was found even down to a depth of 12–13 m below ground surface. Treatment details can be found in the study of Lima et al. (2012a, b). At the end of the experiment, it remained unclear whether the observed PAH heterogeneous distribution pattern was caused by the electroremediation treatment or was an initial condition. In addition, how the presence of the surfactant affected desorption and subsequent transport of PAHs could also not be demonstrated, although recent studies have shown that the use of Tween 80 surfactant has great potential for PAH remediation (Befkadu and Chen 2018; Cheng et al. 2018).

Using soil retrieved from the former asphalt production plant in Olst, The Netherlands, this study aims to address the following questions:

  1. 1.

    Is the PAH distribution pattern in the soil after the electroremediation treatment caused by this treatment or does it also reflect the heterogeneity of the soil itself?

  2. 2.

    To which extent the surfactant desorption and subsequent bioavailability of the PAHs are affected?

The first question was addressed by analysing randomly selected soil samples from different locations of the field site before the electroremediation treatment with respect to the amount and composition of organic matter and PAH concentrations. The second issue was studied in batch desorption experiment with selected contaminated soil samples with and without 1% Tween 80 in the solution phase. These data facilitate our understanding of the processes that occurred at the field site during the remediation treatment.

Materials and methods

Soil samples

Soil samples were taken from the site of a former asphalt production plant in Olst, The Netherlands, located at the river Ijssel (Fig. 1a). Asphalt production was ongoing for around 100 years and ceased in 1983. A contamination by tar oil was found even down to a depth of 12–13 m below ground surface, where a clay layer, called the “eemklei” layer, is situated. This clay was deposited during the Late Pleistocene and is comparably rich in organic material due to intercalation of peat. It is believed that the contamination of this deep layer is caused by long-term leaching during the years of production. In a first remediation attempt in the 1990s, the upper 6 m of the soil was partly excavated and replaced by sand. Although this sand was supposed to be clean, tar oil was later found in this layer as well, which is maybe caused by mixing during the activities at the site. Due to the extensive contamination, no further use of this site is possible until today, and remediation is still in progress.

Fig. 1
figure 1

Soil sampling locations and instrumentation. a Asphalt production plant location, Olst, The Netherlands, and sample locations. b, c Soil core sampling equipment. d An illustration of a soil sample

Full size image

Prior to the remediation treatment (Lima et al. 2012a), soils were sampled from different locations (Fig. 1a) and from different depths (Table 1) down to the eemklei layer at this site in order to be able to assess the heterogeneity of the soil with respect to organic matter and tar oil content, PAH concentrations and the assumed contact time of the PAHs with the soil. Four soils were taken from clay layers of various depths (soils A–D), and a fifth soil (soil E) was sampled from the upper sandy layer containing tar oil.

Table 1 Characterisation of the soil samples
Full size table

All soils were stored at 4 °C until further use.

Soil characterisation

The soils were characterised with respect to organic matter content and composition and PAH content. Prior to these analyses, aliquots of all samples were air-dried and the pH was determined potentiometrically with a glass electrode in 1 M KCl at a solid/solution ratio of 1:2.5.

Total C and N contents were measured in duplicate by means of dry combustion, using an elemental analyser (Euro EA 3000; HEKAtech, Wegberg, Germany). Detection limits for both C and N were 0.02%, respectively.

The composition of the organic matter was assessed by 13C nuclear magnetic resonance (NMR) spectroscopy with a DSX 200 spectrometer (Bruker BioSpin, Rheinstetten, Germany), operating with a cross-polarisation magic angle spinning (CP MAS) technique with a spinning speed of 6.8 kHz at the resonance frequency of 13C (50.32 MHz). Acquisition parameters included a contact time of 1 ms and a pulse delay of 0.4 s. Line broadening of 75 Hz was applied. Depending on the C content of the sample, a minimum of 24,110 scans was accumulated for each sample. Chemical shifts were calibrated relative to tetramethylsilane (0 ppm). The contributions of the various C groups were determined by integration of their signal intensity by means of the instrument software. Chemical shift regions were used as follows: alkyl C 0–45 ppm, O/N alkyl C 45–110 ppm, aryl C 110–160 ppm and carboxyl C 160–220 ppm (Knicker and Lüdemann 1995).

The contents of 16 PAHs that were identified by the US Environmental Protection Agency as priority pollutants (16 EPA PAHs) were analysed by gas chromatography/mass spectrometry (GC/MS) after solid-phase extraction as described below.

Desorption experiment

The effect of the surfactant on the desorption of PAHs was studied in batch desorption experiments on two different soils. Soil E was selected because the sand layer with its tar oil contamination differs significantly from all other soils of this site because of its history and composition and because it represents the filled upper layer of this site. From the other soils, which are all more clayey and display the heterogeneity of the subsoil, a representative sample was chosen that exhibited an intermediate C and PAH content.

The experimental set-up of the batch desorption experiments is based on the study of Heister et al. (2013). Approximately 5 g of the wet sample material was filled into a standard grade–regenerated cellulose membrane dialysis tubing with a molecular weight cut-off of 1000 Da (Spectra/Por 6; Spectrum Laboratories, Inc., Los Angeles, CA). The membrane was stored in 0.5% sodium azide solution at 4 °C to avoid microbial alteration. The filled tubings were sealed with copper wire at both ends and placed into brown borosilicate glass bottles (1 l) with Teflon-lined screw caps. The bottles were then either filled with 1000 ml deionised water or a 1% Tween 80 solution. Tween 80 (Merck, Darmstadt, Germany) was used as received. Both solutions were spiked with sodium azide at a concentration of 0.5 mmol/l to prevent microbial activity that could have possibly been introduced during the preparation. Respectively, two bottles of each of the two soils were filled with deionised water and two with the surfactant solution. In addition, two blanks were prepared: one blank of deionised water, consisting of a dialysis membrane filled with deionised water and placed into a bottle of deionised water, and one blank of surfactant solution, consisting of a membrane filled with 1% Tween 80 solution in a bottle of 1% Tween 80 solution. Accordingly, the experiment comprised a total of ten bottles. All bottles were kept in the dark at a constant temperature of 20 °C during the course of the experiment.

Liquid-phase samples (approximately 90 ml) were taken directly at the beginning of the experiment and after 1 day, 4 days, 7 days, 32 days and 176 days. These sampling days were selected because PAH concentrations in the liquid phase increase exponentially, as determined in previous studies (Heister et al. 2013). Around 80 ml was used for PAH analysis as described below, while the remaining 10 ml served to measure electrical conductivity, pH and dissolved organic carbon (DOC). Electrical conductivity was measured with a LF 300 conductivity meter equipped with a TetraCon 325 electrode and pH with a pH meter 340 and a SenTix 81 pH electrode (WTW, Weilheim, Germany), respectively. DOC was determined as non-purgeable organic carbon with a TOC analyser (5050A; Shimadzu, Kyoto, Japan) after acidification with 2 N HCl. According to the concept of the equilibrium solution, it is assumed that the soil samples became equilibrated with the surrounding solution, until the sampled solution is finally equal to the pore water (Heister et al. 2013).

PAH extraction and analysis

PAH concentrations of the soils were determined by solid-phase extraction based on the method developed by Hartmann (1996). The procedure contains a saponification step with methanolic KOH in an ultrasonic bath, followed by partitioning of the PAHs into hexane and clean-up of the extracts by Al2O3/silica gel columns with subsequent elution and volume reduction.

The PAH concentrations of the liquid phases collected from the desorption experiments were determined by liquid-phase extraction based on the method of Chladek and Marano (1984). The samples were extracted by columns of Bakerbond® Octadecyl (C18) material (J.T. Baker, Philipsburg, NJ) directly after sampling, followed by elution and volume reduction. Thereafter, they were subjected to an additional purification step by means of Al2O3/silica gel columns to remove the surfactant (Lima et al. 2012a). In order to avoid artefacts due to sample treatment, this purification step was applied to all liquid samples, irrespective of whether they contained the surfactant or not.

Analysis of the 16 EPA PAHs was conducted by GC/MS with a GC8000 gas chromatograph (Thermo Fisons, Milan, Italy) supplied with a fused silica column (VF-35ms, with a length of 30 m, inner diameter of 0.25 mm, film thickness of 0.25 μm; Varian, Middelburg, The Netherlands) and a MD800 mass spectrometer (Thermo Fisons, Manchester, UK). Quantification was done by the addition of a mixture of eight deuterated PAHs to the samples prior to extraction (Supelco, Bellefonte, PA, and Cambridge Isotope Laboratories, Andover, MA). Recoveries of the internal standard were quantified by the addition of deuterated perylene-D12 (Supelco, Bellefonte, PA) after extraction. The linear range of the calibration line was between 0.01 and 1 ppm.

Results and discussion

Assessment of soil heterogeneity

The pH values of the soils, which range between pH 2.3 and pH 6.9 (Table 1), are all below neutral, indicating that no free carbonates are present, and thus, the total C content determined by elemental analysis is equal to the organic carbon (OC) content. The extreme low pH values of soils A and B can be explained by the presence of pyrite that became oxidised after sampling, since the groundwater table at that site is normally higher than 2 m below ground surface, resulting in anoxic conditions in situ (Lima et al. 2011). Due to the intercalation with peat, the clayey subsoil samples (soils A–D) are rather rich in OC. The organic material seems to be heterogeneously distributed, which is reflected in the heterogeneous OC content. The lowest OC content with 7.86 mg OC/g is found in the sand (soil E), which is significantly lower than the OC contents of the other soils (Table 1).

Solid-state 13C NMR spectroscopy revealed that not only the amount of organic matter but also its composition varies between the soil samples. Soils B and C can be considered rather similar regarding C group types and intensity (Supplementary Material, Figure S1), since both are dominated by O/N-alkyl species and contain 28–29% aromatic C (Table 1). Soils A and D contain more aromatic C with 36% and 45%, respectively. Soil E consists predominantly of aromatic C, which is inferred from the high signal intensity of 71% in the aryl C region (Table 1). As a rule, high signal intensities in the aryl C region can point to the presence of tar oil (Kipling and Waldron 1976), which consists of aromatic C, although also higher plant-derived material like lignin causes signals in this region. Another characteristic peak for lignin is at 72 ppm in the NMR spectra (Kögel-Knabner 2002), not well pronounced in soils D and E and noticeably reduced compared to the peak in the aryl C region in the spectrum of soil A (Supplementary Material, Figure S1). Therefore, it can be concluded that soil E consists almost exclusively of tar oil, soils A and D contain increasing amounts of tar oil and the organic material of soils B and C consists mainly of peat.

Accordingly, the organic material of the five soil samples can be characterised as follows:

  • Soil A has a high content of organic material, consisting of a mixture of inherent peat and tar oil, originating from the former asphalt production.

  • Soil B contains a low content of organic material that is composed solely of peat.

  • Soil C exhibits a high content of organic material, consisting exclusively of peat.

  • Soil D has a low content of organic material that is significantly influenced by the tar oil contamination.

  • Soil E contains very little organic material compared to the other soils, consisting solely of tar oil.

Regarding PAH concentrations (Table 2), it can been seen that soils C and A have the highest concentrations, while soil D exhibits an intermediate level of contamination and soil B seems to be only slightly affected by PAH contamination. Interestingly, soil E shows 1023 μg/g of total PAHs, only an intermediate level of PAH contamination. This is unexpected since soil E presents a tar oil dominance over the total organic phase, and tar oil is known to be associated with high levels of PAHs (e.g. Abdel-Shafy and Mansour 2016). PAHs are hydrophobic organic pollutants and tend to accumulate in organic material (De Paolis and Kukkonen 1997; Rivas 2006) with only reports of naphthalene or phenanthrene (2–3-ring PAHs) having the tendency to volatilise (Cabrerizo et al. 2011). Thus, it was expected that with increasing OC concentration, PAH concentrations would increase as well. Such a relationship between organic matter content and PAH concentration was found, e.g. in the sediments of marine coastal lagoons (Giacalone et al. 2004). In the study of Giacalone et al. (2004), the authors also observed a heterogeneous PAH distribution throughout the studied soil samples, but the total amounts of the 16 EPA PAHs correlated very well with the organic matter content. Contrary to that study, we did not observe such a correlation. PAH concentrations did not increase with increasing OC contents (Fig. 2a), which is in line with observations by Liu et al. (2010) about the PAH concentrations in the surface soils of the Shanghai area. There, the spots with the highest PAH contamination were found in the urban and petrochemical industrial areas, and thus, it was concluded that the level of PAH contamination was predominantly influenced by the input of anthropogenic materials from traffic, petrochemical industry and coal combustion (Liu et al. 2010). On the other hand, in the coastal lagoons studied by Giacalone et al. (2004), there was no evidence for hot spots of anthropogenic pollution, although a contamination by tar oil or creosote was expected at the sites of mussel farms. Lastly, Trellu et al. (2017) found PAHs associated to the carbonaceous particulate materials. Consequently, these findings imply that in the present study, the PAH concentrations should be correlated with the amount of tar oil, as tar oil is the type of organic material that introduced the PAHs into the soil. From the characterisation of the organic material by 13C NMR spectroscopy, it is inferred that high signal intensities in the aryl C region (110–160 ppm) are indicative of high concentrations of aromatic C. Therefore, the signal intensity of this C group can act as a proxy for the tar oil contamination. However, no correlation between the percentage of aromatic C and the PAH concentration was observed (Fig. 2b). Since soil contamination without the addition of an organic matter phase that is high in aromatic C is very unlikely, it is assumed that the PAHs were already gone—at least partly—from the tar oil phase and, most probably, were leached through the soil. This could have been caused by tar oil weathering itself, and the mobilised PAHs could have been sorbed to the natural organic matter present in the soil, e.g. peat.

Table 2 Solid-phase PAH concentrations of the soils determined by GC/MS after solid-phase extraction
Full size table
Fig. 2
figure 2

Correlation between PAH concentrations (sum of 16 EPA PAHs) and organic carbon contents (a) and between PAH concentrations and the percentages of aromatic C derived from the integration of the signal intensity of the aryl C region (110–160 ppm) as a proxy for the presence of tar oil (b)

Full size image

Consequently, we can conclude that the distribution of organic material, tar oil and PAHs was due to natural soil weathering processes, before any remediation efforts had started. This means that it cannot be concluded that the heterogeneous distribution patterns observed by Lima et al. (2012a, b) are solely due to the remediation treatment. In fact, PAH distribution patterns observed by the authors after site remediation are rather a superimposition of the treatment on the already PAH heterogeneity of the contaminated site. Since surfactants were used in this remediation effort, in tandem with electrokinetics (Lima et al. 2012a), we subsequently study the effect of the surfactant on PAH desorption in a representative sample for this site. Soil D was selected due to its intermediate content of both organic matter and PAHs, in spite of exhibiting a comparably high percentage of aromatic C compared to the other subsoil types investigated.

Effect of the surfactant on the desorption of PAHs

In the field experiment described by Lima et al. (2012a, b), the non-ionic surfactant Tween 80 was used in order to enhance desorption of PAHs and subsequent removal in the electrolyte solutions at the electrodes. To test the effect and importance of the surfactant on PAH mobilisation, batch desorption experiments with and without Tween 80 were conducted with selected soils D and E. The differences in organic matter composition between these two soils are assumed to cause differences in PAH desorption because PAHs have a high affinity to aromatic C moieties (De Paolis and Kukkonen 1997; Perminova et al. 1999) and should be more strongly bound to the tar oil phase.

As expected, the two soils exhibited differences in desorption behaviour during the course of the experiment. In the batches containing deionised water, the electrical conductivity increased from an initial value ranging between 50 and 65 μS/cm to 138 μS/cm for soil E and 215 μS/cm for soil D, respectively, while it did not exceed 60 μS/cm in the corresponding blank. Similarly, there was an increase in DOC concentrations to 2.9 ppm for soil E and 3.3 ppm for soil D, respectively (blank 2.0 ppm), and an increase in pH from around 6.3 to 7.3 for soil E and 7.7 for soil D, while the pH remained constant in the blank (Supplementary Material, Figures S2–S4). The observed increases are comparable to another desorption batch experiment of a similar set-up (Heister et al. 2013) and recent findings (Yang et al. 2014) and can be explained by the equilibration of the soil with the surrounding aqueous phase. The asymptotic behaviour of the electrical conductivity and the constant values of DOC and pH at the later stages of the experiment indicate that this equilibration had been obtained. The higher values for soil D are explained by the fact that more substances were desorbed from this soil compared to the sandy soil E.

Also, desorption of PAHs increased over time, but for soil D, the sum of the 16 EPA PAHs did not exceed 0.5 μg/l (Supplementary Material, Figure S5), and for soil E, the total concentration increased to a maximum of 22 μg/l on day 7 and then decreased again to 0.3 μg/l at day 176 (Fig. 3). Because the total PAH concentration of soil E exceeded the concentration of soil D by almost a factor of 4 (Table 2), it is likely that higher amounts of PAHs will become desorbed from soil E. However, soil E exhibited a disproportional increase in PAH aqueous-phase concentrations, exceeding this factor of 4. This demonstrates that contrary to the assumption that PAHs are more strongly bound in the tar oil that dominated the organic matter of soil E, PAH sorption is stronger in soil D. It is assumed that this increased sorption in soil D is caused by the longer contact times of the PAHs with the organic material, resulting in ageing (Semple et al. 2004). It is supposed that the PAHs become strongly sorbed and sequestered in the microporous structure of the organic material (Rivas 2006). Subsequently, the amount and composition of the organic material and contact time were responsible for the ageing and caused the reduced desorption of the PAHs from soil D.

Fig. 3
figure 3

Development of the PAH concentration (sum of 16 EPA PAHs) in the aqueous phase in the desorption experiment of soil E in water. The values are the average of two replicates and are corrected for the values of the blanks

Full size image

The observed decrease in PAH aqueous-phase concentration of soil E after day 7 (Fig. 3) indicates that microbial degradation of the desorbed PAHs occurred, as the set-up was a closed system. This desorbed fraction was readily available for microorganisms present (Alexander 2000; Semple et al. 2003), since the added sodium azide acts as a respiratory poison and is not effective against all potential microorganisms present in the soils. Consequently, the data clearly demonstrate the importance of desorption for the bioavailability of chemicals in soils and sediments.

In the batches containing the 1% Tween 80 solution, desorption was significantly increased for both soils. The electrical conductivity rose from an initial value ranging between 60 mS/cm and 65 μS/cm to 195 μS/cm for soil E and 360 μS/cm for soil D, respectively. Enhanced solubilisation was also detected in the blank with an increase of up to 136 μS/cm, indicating that the surfactant was also able to solubilise substances from the dialysis membrane (Supplementary Material, Figure S6). DOC concentrations were not measured in these samples because the presence of the surfactant resulted in such high DOC concentrations that differences between the various samples were too small to be detected. The pH values increased from around 6.0 to 7.3 for soil D but decreased from 6.1 to 5.7 for soil E and from 6.1 to 4.0 for the blank, respectively (Supplementary Material, Figure S7). Obviously, this decrease in pH was caused by the surfactant and could be buffered by the clayey soil D. In another laboratory study on biodegradation of PAHs in the presence of surfactants, Bautista et al. (2009) observed an increase of pH from around pH 7.0 to nearly 9.0 in a 1% Tween 80 solution with different PAH-degrading bacteria over 31 days and explained this increase by the formation of intermediate products during degradation.

In the present study, enhanced desorption in the presence of the surfactant was demonstrated, but a positive effect on biodegradation as observed by Bautista et al. (2009) was not detected. PAH aqueous-phase concentrations in the presence of Tween 80 increased to almost 73 μg/l (sum of 16 EPA PAHs) for soil D (Fig. 4) and to approximately 1600 μg/l for soil E (Fig. 5), respectively. Since the decrease in the batches without surfactant was explained by biodegradation of the PAHs, Tween 80 may have prevented PAH degradation here. Moreover, due to the inhibition of PAH biodegradation, no intermediate products were formed which could raise the pH, which is in agreement with the findings of Bautista et al. (2009).

Fig. 4
figure 4

Development of the PAH concentration (sum of 16 EPA PAHs) in the aqueous phase in the desorption experiment of soil D in 1% Tween 80 solution. The values are the average of two replicates and are corrected for the values of the blanks

Full size image
Fig. 5
figure 5

Development of the PAH concentration (sum of 16 EPA PAHs) in the aqueous phase in the desorption experiment of soil E in 1% Tween 80 solution. The values are the average of two replicates and are corrected for the values of the blanks

Full size image

The inhibitory effect of a surfactant on the bioavailability of PAHs has been observed before (Laha and Luthy 1991, 1992; Volkering et al. 1997). It is explained by the toxicity effects of the surfactant on the microorganisms, entrapment of contaminants inside micelles to which microorganisms cannot attach to and interactions with the surfactant molecules on the membranes of the microorganisms. Toxicity effects of a 1% Tween 80 solution is not likely because non-ionic surfactants like Tween 80 are less active against bacteria than ionic surfactants and therefore less toxic (Volkering et al. 1997). For the complete entrapment of PAH molecules inside surfactant micelles, the critical micelle concentration (CMC), i.e. the concentration of a surfactant in a solution at which micelle formation starts, has to be achieved or exceeded. As surfactant molecules also adsorb to soil particles, the CMC is higher in a soil suspension than in pure water (Laha et al. 2009). The amount of sorption depends, among others, on the soil organic matter content and the mineral composition (Laha et al. 2009) and therefore varies between different soils. The CMC of Tween 80 in water is 0.01 mM (Cheng and Wong 2006; Laha et al. 2009) and is 720 times exceeded in a 1% Tween 80 solution. Accordingly, it is inferred that the CMC in the batch desorption experiment was significantly exceeded, although some sorption to soil particles probably occurred. Subsequently, this resulted in complete entrapment of desorbed PAH molecules inside micelles and lowered their bioavailability. Moreover, it is also conceivable that microorganisms became covered with surfactant molecules because surfactant molecules also interact with lipid components of the cell membranes, enzymes and other proteins. This attachment is assumed to be a reversible binding without disrupting cell membrane structures (Laha and Luthy 1992), but it hinders the attachment of the microorganisms to organic substrates. In the present study, it cannot be distinguished between these two processes, but it is very likely that both play a role. Consequently, at such high surfactant concentrations as in the present study, the surfactant significantly enhances desorption of organic contaminants but also causes an inhibition of their biodegradation by microorganisms.

Implications for the surfactant-enhanced soil remediation

After a successful pilot-scale field experiment demonstrating that PAHs can be removed from clayey soil by electroosmosis, Lima et al. (2012a, b) detected not only increased PAH concentrations in the electrolytes but also a heterogeneous distribution of the PAHs throughout the soil. However, they could not ascribe this distribution pattern to the redistribution of PAHs due to the treatment, as destructive sampling was only done afterwards. The data of the current study show that such a heterogeneous distribution was already very likely before the remediation treatment because in samples taken from different locations at the field site before, neither a correlation between OC content and PAH concentration nor a correlation between tar oil content and PAH concentration was observed. This implies that PAHs are not solely present within the tar oil phase, but that they also moved alone through the soil and sorbed to inherent organic matter like peat. It is assumed that this can be caused by weathering of the tar oil, since the asphalt production and subsequent contamination at that site were ongoing for around 100 years. Accordingly, for assessing the remediation efficiency, different states of the presence of PAH have to be considered, namely bound in the tar oil or sorbed to soil organic matter. It is supposed that the PAHs in these different states have various mobilities and thus various bioavailabilities and, consequently, remediation efficiency will also differ.

The success of the remediation field experiment, as seen in noticeable outflow of PAHs in the electrolyte solutions after a minimum of 50 days, was ascribed to the effect of the surfactant on desorption of PAHs (Lima et al. 2012a, b), since higher removal rates in the presence of the surfactant were already indicated in a previous laboratory electroosmosis experiment (Lima et al. 2011). The data of the present study support this assumption because they prove the considerable increase in PAH solubilisation (about 72 times for soil E and even 146 times for soil D). It is supposed that significantly less or even no removal of PAHs could have been detected without the use of the surfactant. Accordingly, the use of the surfactant appears to be very essential for the success of the remediation.

Electroosmosis in tandem with bioremediation for PAH treatment has been suggested (Secord et al. 2016). However, it is inferred that the desorbed PAHs will not become degraded by microorganisms, as they are entrapped in surfactant micelles. These micelles are persistent and mobile in the soil. As a consequence for the design of a remediation set-up, particular care should be taken to ensure complete recovery of the surfactant from the soil. The use of a readily degradable biosurfactant as proposed by Mulligan et al. (2001) could be an alternative in order to avoid subsequent contamination by the surfactant. However, one should keep in mind that the degradation of the surfactant molecules will release entrapped PAHs into the soil solution. Overall, for modelling the remediation process, the presence of PAHs inside surfactant micelles has to be considered.

Conclusions

The heterogeneous distribution of PAHs in the subsoil after a successful electroremediation treatment at a field site with a long-term tar oil contamination (Lima et al. 2012a, b) cannot solely be ascribed to the redistribution of PAHs during the treatment. The data of the present study show that the PAHs that were likely introduced into the soil with the tar oil phase became partly released from the tar oil and sorbed to inherent organic matter. It is proposed that this is caused by the long contact times and the weathering of the tar oil phase.

The use of the non-ionic surfactant Tween 80 in the anolyte solution appears to be inevitable for the success of the remediation treatment, as significantly enhanced desorption in the presence of the surfactant was observed in batch desorption experiments. Desorption efficiency was even higher in a soil with a rather short contact time, indicating that due to ageing, desorption efficiency decreases with increasing contact time. Moreover, the reduced bioavailability of desorbed PAHs points to the fact that the desorbed PAHs are entrapped in surfactant micelles, which reduces bioavailability significantly. Consequently, it is inferred that the desorbed PAHs are present in micelles and therefore not bioavailable, which has to be taken into account for the design of similar electroremediation treatments. This will have consequences for the transport of the desorbed PAHs and the necessity to ensure complete recovery of the surfactant together with the PAHs from the soil.