WU Liang, SONG Dandan, YAN Lele, LIANG Shengkang, *, YANG Yanqun, PENG Changsheng SHANG Yujun, WANG Xiuli, and DONG Xueliang

Simultaneous Desorption of Polycyclic Aromatic Hydrocarbons and Heavy Metals from Contaminated Soils by Rhamnolipid Biosurfactants

WU Liang1),2), #, SONG Dandan3), #, YAN Lele4), 5), LIANG Shengkang4), 5), *, YANG Yanqun4), 5), PENG Changsheng1), SHANG Yujun3), WANG Xiuli3), and DONG Xueliang3)

1)College of Environmental Science and Engineering,Ocean University of China, Qingdao 266100, China 2) China Offshore Environmental Service Limited, Tianjin 300452, China 3) High and New Technology Research Center of Henan Academy of Sciences, Zhengzhou 450002, China 4) Key Laboratory of Marine Chemistry Theory and Technology, Ministry of Education, Qingdao 266100, China 5) College of Chemistry and Chemical Engineering,Ocean University of China, Qingdao 266100, China

Hydrophobic organic compounds (HOCs) and heavy metals (HMs) are toxins that usually coexist in natural environments. Due to the differences in their properties, it remains challenging to simultaneously remove HMs and HOCs. In this study, the removal of phenanthrene (Phe) and lead (Pb) from co-contaminated soils by single rhamnolipid (RL) and mixed RL-sophorolipid (SL) biosurfactants were evaluatedsoil column experiments. Biosurfactant micelle sizes were determined by dynamic light scattering, and the mechanisms of micelle solubilization were studied. The effects of biosurfactant concentrations, pH, washing agent salinity and the ageing time of polluted soils on Phe and Pb desorption efficiencies were also assessed. The substantial removal of Phe and Pb using mixed RL-SL systems, when molar fractions of RLs were 0.7, was attributed to large mixed micelle formation and lower sorption losses of these systems. The optimal pH value was 6.0, while Phe desorption was favoured at high RLs and low ionic strengths. However, the RLs concentration and ionic strength had no obvious influence on Pb removal.In addition, both Phe and Pb desorption decreased with increased ageing of the polluted soils. Combined RL-SL biosurfactants can be effective for simultaneously removing HOCs and HMs from polluted soils.

rhamnolipid; desorption; polycyclic aromatic hydrocarbons; heavy metals

1 Introduction

The co-contamination of persistent organic pollutants and heavy metals (HMs) in soils is a widespread environmental problem as a result of numerous industrial activities and rapid urbanization. In recent years, surfactant enhanced remediation (SER) was suggested as a promising method for removing hydrophobic organic contaminants (HOCs) and HMs from soils, and has attracted great attention (Song., 2008; Yuan., 2010; Rivero- Huguet and Marshall, 2011; Wen and Marshall, 2011; Zhao., 2016). Over the years, surfactants used in SER have usually been synthesizedchemical methods. The application of these surfactants has been limited by the sources of the raw materials, prices and the propertiesofproducts (Rouse., 1993; Zhao., 2015). How-ever, it is more important that the synthetic surfactants are toxic and easily cause secondary pollution (Yuan, 2010). Compared with synthetic surfactants, biosurfactants, which are produced by microorganisms or extracted from plants, possess lower critical micelle concentrations (CMC), higher solubilizing capacities, lower toxicities and are more favourable for biodegradation and biocompatibility (Mao, 2015; Tmáková, 2015). Rha- mnolipids (RLs) and sophorolipids (SLs) are two kinds of representative biosurfactants; both of them consist of carbohydrates and long-chain aliphatic acids or aliphatic esters, of which the structures are shown in Fig.1 (Banat, 2010). The structures of RLs and SLs result in distinctive physicochemical properties and self-assembly behaviours, such as high surface activities (Nguyen, 2008; Daverey and Pakshirajan, 2009), the ability to complex cationic HMs (Mulligan, 2001; Chen, 2017) and tendencies to form larger micelles, which can increase the solubility capacity for HOCs (Kitamoto, 2009; Pornsunthorntawee, 2009; Wan, 2011). The removal of various HOCs or HMs from soils by RL and SL biosurfactants has been extensively studied (Wang and Mulligan, 2009; Wan, 2015; Rangarajan and Narayanan, 2018). In our previous study (Song, 2016),it showed that binary RL-SL mixed systems exhibited greater solubilisation capacities for polycyclic aromatic hydrocarbons (PAHs) than did individual glycolipids. Moreover, the solubility capacities of the SL-RL mixed biosurfactants were enhanced in the presence of a trace amounts of HMs. As a consequence, using mixed RL-SL biosurfactants as washing agents (WAs) to remove HOCs and HMs from soils simultaneously could achieve high washing efficiencies with lower biosurfactant dosages and reduce the environmental risks. However, there was little report on this approach.

When RL-SL mixed biosurfactants are used as WAs, the distribution of RLs and SLs between the solid and aqueous phases will be influenced by each other. Consequentially, the desorption of HOCs and HMs could be reflected. In this study, the desorption of phenanthrene (Phe) and Pb by RLs and mixed RL-SL systems were investigated. Desorption efficiencies depend, to a certain extent, on the adsorption characteristics of biosurfactants on soil matrix, and therefore the adsorption behaviors of RLs on soil matrix were also investigated. In addition, factors that might affect desorption efficiency, such as the concentrations of RL, pH, salinity and ageing time of polluted soils, were also investigated. The experimental results can provide important complementary information for the use of RLs enhanced-flushing systems to simultaneously remove HOCs and HMs.

Fig.1 Chemical structures of rhamnolipids (RLs) and sophorolipids (SLs): mono- and dirhamnolipids, lactonic and acidic sophorolipids (Banat et al., 2010).

2 Materials and Methods

2.1 Materials

Phenanthrene (>99% purity) was chosen as model PAHs, and Pb (in nitrate form, > 99%purity) was selected as the representative of HMs. The RLs were produced by theO-2-2 and SLs were produced byO-13-1 strains, respectively, in our laboratory. The physical and chemical properties of RLs and SLs used in this study were reported in our previous studies (Wang, 2005; Song, 2011).

2.2 Preparation of Phe and Pb Contaminated Soil

Uncontaminated soil was collected (0-20cm) from Qingdao, China. The soil was dried at room temperature and passed through a 100 mesh sieve. The main physical and textural characteristics of the tested soil are given in Table 1. To produce spiked soil, Phe and Pb were dissolved in acetone and water, respectively, and a measured amount of soil was added slowly and stirred. Following the complete evaporation of the solvent in a ventilation hood, the contaminated soil was stored in sealed glass bottles at room temperature (25±1℃) for a week. The resulting freshly amended soils were in final concentrations of 400.0mgPhekg?1of soil and 278.7mgPbkg?1of soil, respectively.

Table 1 Properties of the soil used in this study

2.3 Desorption of Phe and Pb from Contaminated Soil Samples

Acryliccolumns (2.5cm in depth and 20cm in length) were used in the experiments. The columns were packed with soil in incremental steps and uniform bulk density was achieved. After packing, the columns were saturated with electrolyte solution (0.01molL?1NaNO3). Rhamnolipids were dissolved in an aqueous solution of sodium azide (0.02%) and the pH was adjusted by dilute nitric acid and a sodium hydroxide solution. A peristaltic pump was used to pump RL solution into the columns at a constant flow rate of 0.4mLmin?1until the effluent concentrations of both Phe and Pbno longer varied. Each test was performed in triplicate. The desorption percentages of Phe and Pb were calculated from the difference between their initial and resident concentrations in the soil. The factors effecting the desorption of Phe and Pb from contaminated soils (.., SL biosurfactants, pH, ionic strength and ageing time of polluted soils) were also investigated. The pH in the range 4.0-9.0 of the background solution was adjusted by addition of HCl or NaOH solution. The effect of ionic strength was tested by adding different concentrations of NaCl (10, 20, 40, 80and 100mmolL?1) to the RL solution. The ageing times of polluted soils investigated were 10, 40, 60, 90 and 120d. All the measurements were reproducible, and the mea- surement error was < 5%.

2.4 Isothermal Adsorption Curves of RLs and SLs on Soils

The experimental procedures involved in constructing isothermal adsorption curves were similar to those presented in section 2.3, but the soil samples were uncontaminated. After elution, the supernatants (1.0mL) were mixed with 5mL of an anthrone reagent, and the mixture was colored for 10min in the boiling water bath, then removing to cool to room temperature. The concentrations of RLs and SLs were determined using a spectrophotometer (Model UV-2550; Shimadzu, Japan) at 650 and 628nm, respectively. The concentrations of RLs and SLs were fitted to the Langmuir isothermal adsorption equation (Eq. (1)) (Liu, 1992; Pazos, 2010):

where0is the maximum adsorption,is the capacity factor, which refers to the adsorption capacity of the soil organic matter at a particular concentration,qis the adsorption capacity of the organic matter under balanced conditions (mgg?1), andCis the equilibrium concentration of organic matter in solution (mgL?1).

2.5 Micelle Size Studies

Gadelle(1995) reported that the solubilisation capability is an increasing function of the size of surfactant micelles. Therefore, the micelle sizes of the RLs, SLs and mixed RL-SL systems under different experiment conditions were determined by dynamic light scattering (DLS) measurements. The instrument type used was a Malvern Panalytical Zetasizer Nano ZS90 (UK). The light source was a He-Ne laser with a 4 mW power and 633nm wavelength. The scattering angle () was set as 90?. The studied surfactant solutions were filtered through Millipore membranes (average pore size of 0.45μm) before measurement. And the measurement in thermostatic sample chamber was maintained at 298.5K by thermoelectric Peltier module.

2.6 Analytical Methods

2.6.1 Phe analysis

Phenanthrene concentration adsorbed to soils was determined using a Soxhlet extraction procedure according to the Environmental Protection Agency (EPA) of the United States of America method 3540C (US-EPA, 1986). The Soxhlet extraction was analysedgas chromatography (GC) using an Agilent Technologies 6890N (USA) equipped with a model 5975B mass selective detector (MSD;.., through GC-mass spectrometry (MS)). Phenanthrene in the effluent was extracted with dichloromethane, and extractant was properly diluted, then their concentrations were analysed by GC-MS.

2.6.2 Pb analysis

Lead concentration in the soil was determined according to US-EPA test method 3050 (US-EPA, 1986). The Pb in the effluent was directly tested using an atomic absorption spectrophotometer (Thermo-Fisher Scientific M6, USA), following US-EPA method 7420 (US-EPA, 1986).

3 Results and Discussion

3.1 Adsorption of RLs and SLs on Soils

The accumulative adsorption quantities of RLs and SLs increased dramatically and then remained nearly unchanged with increased washing time (Fig.2). Therefore, RLs and SLs reached adsorption equilibrium when washing time continued to 32h and 36h, and the accumulative adsorption quantities of RLs and SLs were 148mg and 205mg, respectively. The adsorption of SLs was slightly greater than that of RLs when the equilibrium concentration of RLs and SLs were above 1mgmL?1. This may be due to the negative surface charges of soil and the anionic biosurfactant RLs producing electrostatic repulsion from these surfaces. Thus, in desorption experiments, the con- centration of SLs will be lower than that of RLs in the eluent after reaching adsorption equilibrium. The sorption loss of biosurfactants to soils decreased the effective con- centrations of biosurfactants, which then decreased the desorption efficiency of Phe and Pb from soils.

Fig.2 Curve of the adsorption rates of RLs and SLs on soils.

3.2 Desorption of Phe and Pb from Contaminated Soils

3.2.1 Effects of rhamnolipid concentrations

The desorption of Phe increased gradually with increasing RL concentration up to 1.0gL?1(Fig.3a). With respect to the micelle microstructure, the micelle diameter increased distinctly (from 180nm to 210nm) with the increased of RL concentrations from 0.5gL?1to 1.0gL?1(Fig.3b), indicating that large micelle sizes favoured the micellar core solubilisation of Phe (Gadelle, 1995). When the concentration of RLs further increased from 1.0 gL?1to 3.0gL?1, the removal efficiency of Phe increased slowly and by a small amount; meanwhile, the same changes in micelle sizes were observed.

At a concentration of 1.0gL?1of RLs, desorption of Pb reached a plateau at a removal efficiency of 15.1%. The interfacial tension between solid surfaces and HMs decreased with the addition of RLs, therefore increasing the mobility of HMs (Mulligan, 2001). However, when the concentration of RLs was more than 1.0gL?1, the increasing rate of desorption of Pb became low. This is possibly due to excess quantities of RLs approaching its maximum dissociate capacity of Pb from soil. Additionally, the removal efficiencies of HMs were strongly related to their speciation in soils, and fraction bound to organic matter and residual fraction were more difficult to remove (Wan, 2015). According to these results, an RL concentration of 1.0gL?1is suggested for the soil washing of Phe and Pb.

Fig.3 Effects of RL concentrations on (a) phenanthrene (Phe) and Pb desorption from soils and (b) micelle size distribution.

3.2.2 Effects of SLs in binary RL-SL mixed systems

Fig.4 showed the desorption of Phe and Pb in single and different RL-SL mixed binary systems. It can be seen that the mixed RL-SL systems were more effective for the desorption of Phe and Pb from soils than either the single RLs or SLs. It was further observed that the desorption percentage increased as the molar fraction of RLs in mixed systems increased before reaching a maximum atRLs=0.7 (Fig.4a). The desorption percentages of Phe with mixed RL-SL systems (RLs=0.7) were higher than those by single RL or SL systems, at the same total concentration (1000mgL?1) of SLs, RLs, and RL-SL systems. Similar results were also reported by Lu(2009); in their study, the desorption percentages of oil contaminated-soils by anion-nonionic surfactants were higher than those by single anion or nonionic surfactants. It can be concluded that substantial synergy occurs in the Phe and Pb desorption of mixed RL-SL systems, which is mainly attributable to the presence of SLs decreasing the adsorption of RLs onto soils. The reason may be that the SLs added to anionic RLs solution decreased the electrostatic self-repulsion of RLs and increased the attractive ion-dipole interaction between hydrophilic groups of the both SLs and RLs, thereby decreasing the adsorption of RLs onto soils. (Song., 2016).The isothermal adsorption curves of single and RL-SL mixed systems generated by fitting a Langmuir model (Eq. (1)) are shown in Fig.4b, and the corresponding fitting parameter values are listed in Table 2. It can be seen that adsorption of RLs on soils was greatly reduced by adding SLs. The Langmuir isotherm constant (max) values were calculated as 7.44 mgg?1for RLs, 10.3mgg?1for SLs and 2.92mgg?1for RL-SL mixed systems (RLs=0.7). The total sorption loss of RL-SL mixed systems was lower than that of single system. According to one previous study (Wan, 2015), while a small amount of SLs were added to the RLs solution, electrostatic attractive interactions occurred between RLs and SLs in the mixed micelles. When the RL molar fractions were 0.7 and 0.9, the mixed micelles were more stable than that of other mixed systems. As the negative surface charges of the mixed surfactant micellar shell increased, the polarity of mixed micellar cores decreased (Garamus, 2003; Penfold, 2005). Thus, the adsorption losses of RLs onto the soil columns in mixed surfactant solutions reduced in the presence of nonionic SLs.

The active concentrations of mixed RL-SL systems in aqueous solutions were higher than those of individual RLs or SLs. Meanwhile, the mixed micelles formed in mixed RL-SL systems plays a synergistic role for the transfer of Phe and Pb from the soil solid phase to the liquid phase. As shown in Fig.4c, larger micelles were formed in mixed RL-SL systems. Most of the mixed RL- SL systems had larger micelle sizes than the single RLs and SLs, which indicates synergy in the mixed micelles. In terms of the microstructure, the addition of a small amount of SLs can enlarge micelle diameter, while the micelle diameter decreased after SL concentrations exceeded a certain value. Compared with the results of solubilisation behaviour in the water solubilisation of PAHs, in which the maximum also occurred when larger micelles were formed (Song, 2013, 2016), it can be concluded that RL-SL mixed systems are expected to form larger and incompact vesicular structures because of the long hydrophobic chains linked to the hydrophilic carbohydrate-based head group of RLs and SLs, thereby enhancing micellar core solubilisation. Thus, the performance of mixed RL-SL biosurfactants in removing Phe and Pb from contaminated soils will be better than those of individual RLs or SLs. As a result of the loss decrease of mixed RL-SL biosurfactants due to adsorption, the active concentrations of biosurfactants in aqueous solutions increased. Therefore, lower doses of biosur- factants will be needed in soil remediation.

Fig.4 Effects of different mixed RL-SL systems on (a) the desorption percentages of Phe and Pb, (b) sorption isotherms of RLs, SLs and RL-SL mixed systems on soils and (c) micelle size distribution. αRLs represents the molar fraction of RLs in mixed RL-SL systems.

Table 2 Langmuir isotherm constants for RLs, SLs and RL-SL mixed systems on soils

3.2.3 Effects of pH

Rhamnolipids have been reported to exhibit pH-dependent behaviours. Its self-assembled behavior and aggregation morphology are all governed by pH (Ishigami, 1987; Baccile, 2012). Meanwhile, the desorption capabilities of biosurfactants for contaminants are strongly dependent on their micelle properties. Data presented in Fig.5a suggest that the desorption efficiencies of Phe and Pb were highly influenced by pH values. The desorption efficiencies of Phe and Pb increased with increasing pH from 4.0 to 7.0; while the substantial decrease in the adsorption of RLs on soils (Fig.5b), by which the effective concentration of RLs increased and the desorption of Phe and Pb from soils was facilitated. The micelle sizes of RLs increased with the increase of pH from 4.0 to 6.0 (Fig.5c). The desorption of Phe and Pb was maximized (to 66.1% and 17.4%, respectively) when the solution pH increased to 6.0, which is close to the value of the dissociation constant (pa) for RLs (Ishigami, 1987). RL head groups became more negatively charged with the increase of pH from 6.0 to 8.0, and a larger size head formed due to the increase in charge repulsion, thereby facilitating morphology change from large lamellae to vesicles, and finally to micelles (Baccile, 2012).

As the pH increased from 6.0 to 8.0, the desorption of Phe gradually decreased. Meanwhile, the precipitation of anionic surfactant ions (RLs) with divalent cations (.., Ca2+and Mg2+) should be substantial and result in RL loss (Yang, 2006), decreasing the removal efficiency of Pb. The results show that the pH of eluent not only affected the micelle morphology of RLs and the solubilisation capability of RLs for Phe, but it also affected the state of RLs and the complexation with HM ions. Therefore, the enhancement of desorption capabilities for PAHs and HM ions can obtain by adjusting the pH of the biosurfactant solution. Furthermore, it is essential to also take into account of environmental concern of pH for in situ applications. Therefore, a pH of 6.0 is suggested.

Fig.5 Effects of pH on (a) desorption percentages of Phe and Pb, (b) RL adsorption on soils and (c) micelle size distribution.

3.2.4 Effects of ionic strength

Phenanthrene desorption decreased with increasing NaCl concentrations (Fig.6a). Because of the electrolyte reduced the electrostatic repulsion between RLs and soils, the adsorption of RLs on soils greatly increased as the ionic strength increased (Fig.6b). Compared with the increase in the micelle sizes of RLs with increasing ionic strength (Fig.6c), the effect of the adsorption loss of RLs on soils could predominate. The effective concentration of RLs for the solubilisation of Phe greatly decreased, and thus the solubilisation capabilities of RLs for Phe also decreased. Additionally, the solubility of HOCs might decrease in a solution with a high ionic strength due to the salt-out effect (Brunk, 1996). This phenomenon has been reported for different soil in RLs and different organic acid system because increased ionic strength causes decreased Phe desorption (An, 2011). Although increased ionic strength caused the increase of RL adsorption loss on soils and the decrease of RL concentrations in solution, there was no substantial influence on the desorption of Pb from soils. This result is similar to that of RL concentrations on the desorption of Pb discussed in Section 3.2.1. The effect of RL concentration on the desorption of Pb was not significant.

3.2.5 Effects of the ageing time of polluted soils

Ageing appears to have had significant impact on the removal of Phe and Pb by RL (Fig.7). With the increased ageing time of polluted soils, these two kinds of pollutants was gradually adsorbed into soil micropores. With the increase of ageing time, the resistance of chemicals to desorption, volatilization, biodegradation, and extraction, increased (Loehr and Webster, 1996). It was reported the amounts of remobilized benzo[a]pyrene decreased significantly with aging (Umeh, 2018). Meanwhile, the proportions of Pb associated with the most weakly bound fraction tended to decrease, with corresponding increases in the other five more strongly binding fractions during the soil ageing (Jalali and Khanlari, 2008). Moreover, organic matter in soils or mineral matrices combined with the increasingly closeadsorption, leading to the substantially reduced removal efficiency of the two types of pollutants.

Fig.6 Effects of ionic strength on (a) the desorption percentages of Phe and Pb, (b) RL adsorption on soils and (c) micelle size distribution.

Fig.7 Effect of the ageing time of polluted soils on desorption percentages of Phe and Pb.

4 Conclusions

The substantial removals of Phe and Pb from soils were achieved when RL concentrations were above 1.0gL?1. Synergistic effects on Phe and Pb with the RL-SL mixed systems were promoted when the molar fraction of RLs was 0.7, and the maximal desorptions of Phe and Pb were achieved, 42.0% and 16.0%, respectively. This was attributed to the lower sorption loss and the formation of large mixed micelles of RL-SL mixed systems in the presence of SLs in RL solutions. Therefore, compared with individual anionic RLs, using anionic-nonionic mixed biosurfactants (RL-SL), higher removal efficiencies for Phe and Pb from soils can be obtained with lower biosurfactant doses.

The extents of Phe and Pb desorption were more notable at a pH of 6.0, and the increased ageing time of polluted soils decreased desorption of Phe and Pb. Phenanthrene desorption was favoured at high RLs and low ionic strengths. The efficiency of Pb washing was unaffected by RL concentration. Meanwhile, ionic strength conferred no substantial effect on the desorption of Pb from soil. Moreover, unlike most of the synthetic chemical surfactants, the use of biosurfactants is environmentally friendly, having little or no negative effect on the environment. Thus, mixed glycolipid biosurfactants have great potential in biosurfactant-enhanced flushing remediation of co- contaminated soils with PAHs and HMs.

Acknowledgements

This work was supported by the National Natural Science Fund Projects of China (Nos. 41371314 and 51202229), the Key Research & Development Project of Shandong Province (No. 2017GHY15117), the Major Focus Project of Henan Academy of Sciences (No. 19ZD08001), and the Fundamental Research Funds for the Central Universities (No. 18JK02025).

An, C. J., Huang, G. H., Wei, J., and Yu, H., 2011. Effect of short-chain organic acids on the enhanced desorption of phenanthrene by rhamnolipid biosurfactant in soil-water environment., 45 (17): 5501-5510.

Baccile, N., Babonneau, F., Jestin, J., Pehau-Arnaudet, G., and Van Bogaert, I., 2012. Unusual, pH-induced, self-assembly of sophorolipid biosurfactants., 6 (6): 4763-4776.

Banat, I. M., Franzetti, A., Gandolfi, I., Bestetti, G., Martinotti, M. G., Fracchia, L.,Smyth, T. G., and Marchant, R., 2010. Microbial biosurfactants production, applications and future potential., 87 (2): 427-444.

Brunk, B. K., Jirka, G. H., and Lion, L. W., 1996. Effects of salinity changes and the formation of dissolved organic matter coatings on the sorption of phenanthrene: Implications for pollutant trapping in estuaries., 31 (1): 119-125.

Chen, W., Qu, Y., Xu, Z., He, F., Chen, Z., Huang, S., and Li, Y., 2017. Heavy metal (Cu, Cd, Pb, Cr) washing from river sedi- ment using biosurfactant rhamnolipid., 24 (19): 16344-16350.

Daverey, A., and Pakshirajan, K., 2009. Production, characterization, and properties of sophorolipids from the yeastusing a low-cost fermentative medium., 158 (3): 663-674.

Gadelle, F., Koros, W. J., and Schechter, R. S., 1995. Solubilization isotherms of aromatic solutes in surfactant aggregates., 170 (1): 57-64.

Garamus, V. M., 2003. Formation of mixed micelles in salt-free aqueous solutions of sodium dodecyl sulfate and C12E6., 19 (18): 7214-7218.

Ishigami, Y., Gama, Y., Nagahora, H., Yamaguchi, M., Nakahara, H., and Kamata, T., 1987. The pH-sensitive conversion of molecular aggregates of rhamnolipid biosurfactant., 16 (5): 763-766.

Jalali, M., and Khanlari, Z. V., 2008. Effect of aging process on the fractionation of heavy metals in some calcareous soils of Iran., 143 (1-2): 26-40.

Kitamoto, D., Morita, T., Fukuoka, T., Konishi, M. A., and Imura, T., 2009. Self-assembling properties of glycolipid biosurfactants and their potential applications., 14 (5): 315-328.

Liu, Z., Edwards, D. A., and Luthy, R. G., 1992. Sorption of non- ionic surfactants onto soil., 26 (10): 1337- 1345.

Loehr, R. C., and Webster, M. T., 1996. Behavior of fresh. aged chemicals in soil., 5 (4): 361-383.

Lu, Y., Ma, X. D., Sun, H. W., and Zhang, W., 2009. Surfactant enhanced washing of heavily oil-contaminated soil., 3 (8): 1483-1487.

Mao, X., Jiang, R., Xiao, W., and Yu, J., 2015. Use of surfactants for the remediation of contaminated soils: A review., 285: 419-435.

Mulligan, C. N., Yong, R. N., and Gibbs, B. F., 2001. Heavy metal removal from sediments by biosurfactants., 85 (1-2): 111-125.

Nguyen, T. T., Youssef, N. H., McInerney, M. J., and Sabatini, D. A., 2008. Rhamnolipid biosurfactant mixtures for environmental remediation., 42 (6-7): 1735-1743.

Pazos, M., Rosales, E., Alcántara, T., Gómez, J., and Sanromán, M. A., 2010. Decontamination of soils containing PAHs by electroremediation: A review., 177 (1-3): 1-11.

Penfold, J., Tucker, I., Thomas, R. K., Staples, E., and Schuermann, R., 2005. Structure of mixed anionic/nonionic surfactant micelles: Experimental observations relating to the role of headgroup electrostatic and steric effects and the effects of added electrolyte., 109 (21): 10760-10770.

Pornsunthorntawee, O., Chavadej, S., and Rujiravanit, R., 2009. Solution properties and vesicle formation of rhamnolipid biosurfactants produced bySP4., 72 (1): 6-15.

Rangarajan, V., and Narayanan, M., 2018. Biosurfactants in soil bioremediation. In:. Springer, Singapore, 193-204.

Rivero-Huguet, M., and Marshall, W. D., 2011. Scaling up a treatment to simultaneously remove persistent organic pollutants and heavy metals from contaminated soils., 83 (5): 668-673.

Rouse, J. D., Sabatini, D. A., and Harwell, J. H., 1993. Minimizing surfactant losses using twin-head anionic surfactants in subsurface remediation., 27 (10): 2072-2078.

Song, D., Li, Y., Liang, S., and Wang, J., 2013. Micelle Behaviors of Sophorolipid/Rhamnolipid Binary Mixed Biosurfactant Systems., 436: 201-206.

Song, D., Liang, S., and Wang, J., 2011. Structure characterization and physic-chemical properties of sophorolipid biosurfactants., 30: 1474-1479 (in Chinese with English abstract).

Song, D., Liang, S., Yan, L., Shang, Y., and Wang, X., 2016. Solubilization of polycyclic aromatic hydrocarbons by single and binary mixed rhamnolipid-sophorolipid biosurfactants., 45 (4): 1405-1412.

Song, S., Zhu, L., and Zhou, W., 2008. Simultaneous removal of phenanthrene and cadmium from contaminated soils by saponin, a plant-derived biosurfactant., 156 (3): 1368-1370.

Tmáková, L., Sekretár, S., and Schmidt, ?., 2016. Plant-derived surfactants as an alternative to synthetic surfactants: Surface and antioxidant activities., 70 (2): 188-196.

Umeh, A. C., Duan, L., Naidu, R., and Semple, K. T., 2018. Time-dependent remobilization of nonextractable Benzo[a] pyrene residues in contrasting soils: Effects of aging, spiked concentration, and soil properties., 52 (21): 12295-12305.

United States Environmental Protection Agency (US-EPA), 1986.,-846. 3rd ed. Office of Solid Waste and Emergency Response, Washington D C.

Wan, J., Chai, L., Lu, X., Lin, Y., and Zhang, S., 2011. Remediation of hexachlorobenzene contaminated soils by rhamnolipid enhanced soil washing coupled with activated carbon selective adsorption., 189 (1- 2): 458-464.

Wan, J., Meng, D., Long, T., Ying, R., Ye, M., Zhang, S., Li, Q., Zhou, Y., and Lin, Y., 2015. Simultaneous removal of lindane, lead and cadmium from soils by rhamnolipids combined with citric acid., 10 (6): 1-14.

Wang, S., and Mulligan, C. N., 2009. Rhamnolipid biosurfactant-enhanced soil flushing for the removal of arsenic and heavy metals from mine tailings., 44 (3): 296-301.

Wang, X., Gong, L., Liang, S., Han, X., Zhu, C., and Li, Y., 2005. Algicidal activity of rhamnolipid biosurfactants produced by., 4 (2): 433-443.

Wen, Y., and Marshall, W. D., 2011. Simultaneous mobilization of trace elements and polycyclic aromatic hydrocarbon (PAH) compounds from soil with a nonionic surfactant and [S, S]- EDDS in admixture: Metals., 197: 361-368.

Yang, K., Zhu, L., and Xing, B., 2006. Enhanced soil washing of phenanthrene by mixed solutions of TX100 and SDBS., 40 (13): 4274-4280.

Yuan, S., Wu, X., Wan, J., Long, H., Lu, X., Wu, X., and Chen, J., 2010. Enhanced washing of HCB and Zn from aged sediments by TX-100 and EDTA mixed solutions., 156 (3-4): 119-125.

Zhao, B., Che, H., Wang, H., and Xu, J., 2016. Column flushing of phenanthrene and copper (II) co-contaminants from sandy soil using Tween 80 and citric acid., 25 (1): 50-63.

Zhao, S., Huang, G., An, C., Wei, J., and Yao, Y., 2015. Enhancement of soil retention for phenanthrene in binary cationic gemini and nonionic surfactant mixtures: Characterizing two-step adsorption and partition processes through experimental and modeling approaches., 286: 144-151.

# These authors contributed equally to the work.

. Tel: 0086-532-66786355

E-mail: liangsk@ouc.edu.cn

June 20, 2019;

November 29, 2019;

March 2, 2020

(Edited by Ji Dechun)