RESEARCH ARTICLE

Influence of soil and hydrocarbon properties on the solventextraction of high-concentration weathered petroleumfrom contaminated soils

Hong Sui & Zhengtao Hua & Xingang Li & Hong Li &Guozhong Wu

Received: 7 October 2013 /Accepted: 29 December 2013# Springer-Verlag Berlin Heidelberg 2014

Abstract Petroleum ether was used to extract petroleumhydrocarbons from soils collected from six oil fields withdifferent history of exploratory and contamination. It wascapable of fast removing 76–94 % of the total petroleumhydrocarbons including 25 alkanes (C11–C35) and 16 USEPA priority polycyclic aromatic hydrocarbons from soils atroom temperature. The partial least squares analysis indicatedthat the solvent extraction efficiencies were positively corre-lated with soil organic matter, cation exchange capacity, mois-ture, pH, and sand content of soils, while negative effects wereobserved in the properties reflecting the molecular size (e.g.,molecular weight and number of carbon atoms) and hydro-phobicity (e.g., water solubility, octanol–water partition coef-ficient, soil organic carbon partition coefficient) of hydrocar-bons. The high concentration of weathered crude oil at theorder of 105 mg kg−1 in this study was demonstrated adversefor solvent extraction by providing an obvious nonaqueousphase liquid phase for hydrocarbon sinking and increasing the

sequestration of soluble hydrocarbons in the insoluble oilfractions during weathering. A full picture of the mass distri-bution and transport mechanism of petroleum contaminants insoils will ultimately require a variety of studies to gain insightsinto the dynamic interactions between environmental indica-tor hydrocarbons and their host oil matrix.

Keywords Soil remediation . Solvent extraction . Alkanes .

PAHs . NAPL . Petroleum ether

Introduction

Spills or leaks of storage tanks during oil exploitation, dischargeoperations, and oil processing process have resulted in a con-siderable number of oil contaminated sites (Wehrer and Totsche2009). The issue of land contamination is particularly salient inChina and is being exacerbated driven by the rapid developmentof petroleum chemical industry. Most of the existing oil fields inChina have been exploited for more than 10 years and theresidual oil in some soils exceeded 105 mg kg−1 (Wu 2012).This is one order of magnitude higher than the ecological riskthreshold of 1.0×104 mg kg−1 (Saterbak et al. 1999), whichsuggested that such kinds of soils were not good candidates forless cost-intensive bioremediation methods due to the unfavor-able conditions for microbial degradation in the presence ofhigh-concentration nonaqueous phase liquid (NAPL). By con-trast, solvent extraction is amore suitable technology for cleanupof crude oil contamination, which is also relatively expensivecompared with bioremediation technologies. For example, thecost for solvent extraction is up to fivefold higher than that forbiopiles or bioaugmentation. Therefore, considerable effortshave been devoted to develop individual or mixture solventsto improve the oil recovery efficiency during the last decades(Avila-Chavez and Trejo 2010; Latawiec and Reid 2010;Viglianti et al. 2006; Wu et al. 2013a).

Responsible editor: Ester Heath

Electronic supplementary material The online version of this article(doi:10.1007/s11356-014-2511-x) contains supplementary material,which is available to authorized users.

H. Sui : Z. Hua :X. Li :H. LiSchool of Chemical Engineering and Technology, Tianjin University,Tianjin 300072, China

H. Sui :X. Li :H. LiCollaborative Innovation Center of Chemical Scienceand Engineering, Tianjin 300072, China

H. Sui :X. Li :H. LiNational Engineering Research Center for Distillation Technology,Tianjin 300072, China

G. Wu (*)Division of Ocean Science and Technology, Graduate School atShenzhen, Tsinghua University, Shenzhen 518055, Chinae-mail: w.guozhong@sz.tsinghua.edu.cn

Environ Sci Pollut ResDOI 10.1007/s11356-014-2511-x

Majority of these works focused on alkanes and polycyclicaromatic hydrocarbons (PAHs) in crude oil, because some ofthese oil fractions have carcinogenic, mutagenic, and toxiceffects on receptors. Actually, crude oil is an extremely com-plex mixture of hydrocarbons. After release into the soil, thefate and transport of crude oil include volatilization, oxidation,dissolution, biotransformation, and sequestration (Semple et al.2001). The evaporation of the light fractions leaves behind themedium and heavy portion of the oil. This remaining mixtureis defined as weathered crude oil which is now the focus of thecleanup efforts. To date, the relatively recalcitrant fractions ofcrude oil such as asphaltenes and resins were less incorporatedin the current models for assessing human health risks fromcontaminated soils due to their low bioavailability. However,they might interact with the aliphatic and aromatic fractions byaggregation, sequestration, and precipitation, which influencethe mass transport and distribution of oil fractions that areultimately responsible for the overall toxicity potential risksand cleanup efficiency (He et al. 2013; Li et al. 2012b; Wuet al. 2013b). These interactions were expected to be morepronounced in high-concentration weathered crude oil-associated soils. However, they were often ignored as thepetroleum concentration in most of previous studies was rela-tively low (<5×103 mg kg−1) (Wu et al. 2013a). Particularlyaiming at high-concentration weathered oil contaminated soils,we conceptualized a treatment process including solvent ex-traction, solvent recycling, and biodegradation of the residualoil in the extracted soils (Li et al. 2012a; Wu et al. 2011,2013a). A further study is required to gain insights into theselectivity of solvent extraction on the individual indicatorhydrocarbon compounds as these studies were focused on theoverall total petroleum hydrocarbon (TPH) or hydrocarbongroups which would potentially overestimate the environmen-tal risks (Wu et al. 2013c). Indeed, there is a lack of statisticalcorrelation between solvent extraction and the properties ofspecific hydrocarbons and the content of crude oil in such soils.

Additionally, the weathering processes, a phenomena oftime-dependent decline in the bioavailability of organic con-taminants (Alexander 2000), tends to change the critical factorsinfluencing solvent extraction by (a) increasing the hydropho-bicity of crude oil via shifting the chemical composition to-wards recalcitrant and asphaltenic fractions (Brassington et al.2007; Reid et al. 2000) and (b) enhancing the sorption andsequestration by modifying soil properties such as pH, soilorganic matter (SOM) both in amount and in nature, and soilinorganic constituents with particular reference to pore size andstructure (Alexander 1995; Pignatello and Xing 1996).Although these important qualitative and quantitative differ-ences between weathered and non-weathered petroleum hy-drocarbons were widely acknowledged, it was not comprehen-sively understood before remediation projects were performed(Brassington et al. 2010). This resulted in repeated remediationattempts, which required additional resources and energy but

had additional negative environmental consequences withoutachieving the treatment objectives. Compared with the artifi-cially spiked contaminated soils, the interactions between oiland soil were more complex in the authentic soils with a longhistory of oil contamination. Therefore, knowledge on thecorrelation between solvent extraction and soil characteristicswas highly demanded.

In this study, we used petroleum ether to extract 25 aliphat-ic hydrocarbons and 16 US EPA priority PAHs in authenticsoils from six heavily contaminated oil fields. Petroleum etheris a group of light hydrocarbon mixtures obtained from petro-leum refineries as a portion of the distillate, which consistsmainly of hexane and pentane. Our recent study demonstratedthat a mixture of hexane and pentane was effective for remov-ing petroleum oil from contaminated soil (Wu et al. 2013a).Therefore, we attempted to use petroleum ether instead ofhexane and pentane in the present study due to its much lowercost. Many factors would influence the solvent extractionefficiency such as temperature, extraction time, the ratio ofsolvent to soil, and the physicochemical properties of soils andhydrocarbons. The specific objectives of this study were todetermine the efficiency of petroleum ether on the recovery ofoverall TPH and individual hydrocarbons, correlate the sol-vent extraction with the physicochemical properties of soilsand hydrocarbons, and evaluate the main mass transfer limi-tation process for the cleanup of high-concentration weatheredcrude oil contaminated soils using solvent extraction.

Materials and methods

Chemicals and soils

Acetone, n-hexane, petroleum ether, tetrachloromethane(CCl4), and anhydrous sodium sulfate were purchased fromTianjin Jiangtian Technologies, China. All the solvents wereanalytical grade except that the n-hexane was HPLC grade.

Duplicate contaminated soil samples were obtained fromsix oil fields in China including Daqing oil field (S1), Liaoheoil field (S2), Dagang oil field (S3), Shengli oil field (S4),Huabei oil field (S5), and Jianghan oil field (S6). The locationof the six oil fields is shown in Fig. SM-1 in SupplementaryMaterials (SM). It was estimated that the soils had years ofcontamination age as all these oil fields have been exploitedfor more than 10 years. Soil samples were collected at 5–10 cm depth, which appeared brown and black with strongsmell of oil. The soils were air dried at room temperature,homogenized, and passed through a 2-mm sieve to removelarge vegetable roots and rocks (>0.42 mm) and kept at 4 °Cbefore use.

Soil pH, moisture, and SOM were determined using themethods described by Wu et al. (2013a). Briefly, pH wasmeasured by 1 mol L−1 NaCl on a pH meter. Moisture was

Environ Sci Pollut Res

determined gravimetrically after drying soil samples at 105 °Cfor 8 h. SOM was measured by placing 2 g of samples in thecrucible, weighed, and heated in a furnace at 500 °C for 2 h toallow the organic matter to burn away. After removal from theoven, the crucibles were allowed to cool before being weighedagain. The weight after baking was subtracted from the initialweight to give the weight of organic matter in each sample.Particle size distribution was determined using standard meth-od (ISO 2010). Initial concentration of crude oil in soils wasmeasured using gravimetric method. Briefly, soil samples(2.5 g) and CCl4 (10 mL) were added into a centrifuge tube.After ultrasonication for 20 min, the mixture was centrifugedfor 5 min at 2,000 rpm. The supernatant was collected forfiltration and the residual soil was extracted again using CCl4(10 mL). After repeating this procedure twice, the solvent inthe extracts was allowed to evaporate in a rotary evaporator.The residual soil was oven dried at 105 °C for 2 h followed byweighing.

Solvent extraction

All glassware for extraction was acid washed, soaked in 10 %nitric acid bath overnight, rinsed with deionized water andethanol followed by hexane, and oven dried at 40 °C beforeuse. Petroleum ether (80 mL) was added into a 100-mLconical flask containing 10 g of contaminated samples, whichwas magnetically agitated for 20 min. The supernatant wasremoved after filtration while the residual soil was collectedfor petroleum analysis. The ratio of solvent to soil (8:1, v/w)was inspired from our preliminary experiments which indicat-ed that the TPH removal increased from 50 to 85 % with theratio of solvent to soil increasing from 4:1 to 8:1 (Fig. SM-2).A higher ratio than 8:1 was not likely to obviously increase thesolvent extraction efficiency, but would increase the cost inpractice.

Petroleum hydrocarbon analysis

The petroleum hydrocarbons in soils were determined usingthe method previously described by Risdon et al. (2008).Briefly, soils (5 g) were chemically dried with 5 g anhydroussodium sulfate in 40-mL Teflon centrifuge tubes. Acetone(4 mL) was added and sonicated for 2 min at 20 °C.Acetone (6 mL) and hexane (10 mL) were added to thesamples and sonicated for 10 min, followed by manuallyshaking to mix the solvent and soil. This step was repeatedtwice followed by centrifugation for 5 min at 750 rpm. Afterpassing the supernatant through a filter column fitted with aglass receiver tube, a sequential step series, including resus-pension of the samples in 10 mL of acetone/hexane (1:1),sonication for 15 min at 20 °C, centrifugation for 5 min at750 rpm, and decantation into a filter column, was repeatedtwice. The final extract volume was adjusted to 40 mL with a

mixture solvent of acetone and hexane (1:1) and homogenizedby manual shaking before analysis.

Total petroleum hydrocarbons were identified and quanti-fied using an Agilent 7890A gas chromatograph coupled to anAgilent 5975C mass spectrometer operated at 70 eV in posi-tive ion mode. TPH was separated on a fused silica capillarycolumn HP-5MS (30 m×0.25 mm internal diameter) withhelium as a carrier gas at a flow rate of 1 mL min−1.Splitless injection with a sample volume of 1 μL and theinjector temperature of 280 °C was applied. The oven temper-ature was increased from 80 to 300 °C at 5 °C min−1 and heldat this temperature for 15 min. TheMSwas operated using thefull scan mode and each compound was quantified by inte-grating the peak at specific m/z.

External calibrations were performed using the C8–C40

hydrocarbon window defining standards and PAH mix stan-dards. The target petroleum compounds were 25 n-alkanes(C11–C35) and the 16 PAHs including naphthalene (Nap),acenaphthylene (Acp), acenaphthene (Ace), fluorene (Flu),phenanthrene (Phen), anthracene (Ant), fluoranthene (Flt),pyrene (Pyr), benzo[a]anthracene (BaA), chrysene (Chr),benzo[b]fluoranthene (BbF), benzo(k)fluoranthene (BkF),benzo[a]pyrene (BaP), indeno[1,2,3-cd]pyrene (Ind),dibenzo[a,h]anthracene (DbA), and benzo[ghi]perylene(BgP). Deuterated alkane standards included decane-d22,nonadecane-d40, and triacontane-d62. Deuterated PAH stan-dards included naphthalene-d8, anthracene-d10, chrysene-d12,and perylene-d12 each at a concentration of 2,000 μg mL−1 inDCM. The deuterated standards were used as internal stan-dards for the analysis. The physicochemical properties of thetarget aliphatic and aromatic hydrocarbons were compiledfrom Mackay et al. (1997), Mackay (2001), and TPHCWG(1999), which are listed in Table SM-1.

Statistical analysis

In order to gain insights into the underlying structure of thesolvent extraction data, partial least squares (PLS) analysiswas performed using statistics software Simca-p. PLS is awidely used multivariate regression method which has beenelaborated in handbooks (Vinzi 2010) and research papers(Jonsson et al. 2007; Navarro-Villoslada et al. 2004). In thisstudy, three PLS models were computed to identify the influ-ence of soil properties (model 1), alkane properties (model 2),and PAH properties (model 3) on the solvent extraction effi-ciency. In model 1, the extractability of ∑25alkanes and∑16PAHs was both set as Y-variables. In the remainingmodels, the mean values for the 25 alkanes or the 16 PAHsin the six soils were used as Y-variable, respectively. Auto-scaling, mean centering, and log transformation were per-formed to the data before PLS calculation in order to obtaina normal distribution. The R2Y and Q2Y parameters were cal-culated to assess the goodness of fit and predictability of the

Environ Sci Pollut Res

models, respectively. The relative importance of each influ-ence factor on the solvent extraction was evaluated by thevariable influence on the projection parameter (VIP) which isthe weighted sum of squares of the PLS loading weights. Ahigh VIP value (>1) indicates an influential factor that willcontribute strongly to the solvent extraction efficiency. Theconfidence level was set at 95% during the statistical analysis.

Results

Characterization of soils and petroleum hydrocarbons

Table 1 indicated significant difference in the particle sizedistribution of the six soils, which were characterized as clay,sandy clay loam, silty clay, silt loam, silt clay loam, and sandyloam, respectively. The pH of the soils varied from slightlyalkaline to strongly alkaline (pH value 7.2–8.9) as describedby previous studies (Masakorala et al. 2013; Wu et al. 2013a).The largest difference in the physicochemical properties wasobserved between S1 and S6, especially in terms of moisture,cation exchange capacity (CEC), and SOM. For instance, theSOM in S1 was almost fivefold higher than that in S6(Table 1).

Overall, the initial TPH concentration in S1, S2, and S6was one order of magnitude higher than that in the remainingthree soils (Table 2). The majority of initial TPH was alkanesirrespective of the soil types, while PAHs accounted for lessthan 10 % of the TPH. Similar distribution patterns of alkanefractions were found in S1, S2, S5, and S6. In these four soils,medium molecular weight fractions predominated except thatparticularly high concentration of C35 was observed in S1, S2,and S5 which accounted for up to 10 % of the total alkanes(Fig. 1). By contrast, a shape of exponential increase was

observed in the composition of aliphatic compounds in S3and S4.With the carbon number of alkanes increasing from 11to 35, the concentration increased from 1.2 to 63.3 mg kg−1 inS3 and from 1.2 to 41.9 mg kg−1 in S4, respectively. Mainlythree-ring (Acp, Ace, Flu, Phen, and Ant) and four-ring (Flt,Pyr, BaA, and Chr) PAHs predominated the aromatic fractionsespecially in S1 and S2 where the sum of these two groupscontributed to 70 and 80 %, respectively, of the total PAHs(Fig. 2).

TPH removal by solvent extraction

Results demonstrated that TPH removal efficiencies from thesix oil field samples ranged from 76 to 94 % (Table 2). Therewas little difference in the contribution of aliphatic and aro-matic fractions to the overall TPH after solvent extraction.However, the distribution of individual alkanes and PAHs inthe residual soils was obviously changed (Figs. 1 and 2).

A general trend was observed that the concentration of theremaining alkanes increased with the number of carbon num-bers. For example, the C11–15 fractions accounted for less than6% of∑Alkanes in the extracted soils, while the percentage ofC26–30 and C30–35 in the total alkanes ranged from 20.0 to30.0 % and from 29.9 to 53.4 %, respectively. Figure 2 alsodemonstrated that the largest removal rate for aromatic frac-tions was noted for the three-ring PAHs irrespective of soiltypes, while the six-ring PAHs were the most recalcitrantfraction to be extracted by petroleum ether.

Multivariate evaluation of solvent extraction

Figure 3 showed excitatory effects of pH, moisture, SOM,CEC, and sand fractions on solvent extraction efficiency,while the remaining physicochemical properties of soils

Table 1 Physicochemical properties of soils

Soil samples

S1 S2 S3 S4 S5 S6

Location Daqing oil field(Heilongjiang)

Liaohe oil field(Liaoning)

Dagang oil field(Tianjin)

Shengli oil field(Shandong)

Huabei oil field(Hebei)

Jianghan oil field(Hubei)

Sand (%) 18 56 2 28 8 66

Silt (%) 34 20 40 52 56 18

Clay (%) 48 24 58 20 36 16

pH 8.9 8.5 8.4 7.2 8.4 8.0

Moisture (%) 2.6 4.6 2.8 2.4 2.5 0.8

CEC (cmol kg−1) 21.4 13.2 19.4 6. 6 11.9 5.8

SOM (%) 9.1 2.3 2.9 6.4 4.3 1.9

Crude oil (mg kg−1)a 2.6×105 1.1×105 2.9×105 2.3×105 4.2×104 1.9×105

SOM soil organic matter, CEC cation exchange capacitya Crude oil including asphaltenes, resins, aromatics, and saturates

Environ Sci Pollut Res

showed inhibitory effects. All the oil properties selected in thisstudy were negatively correlated with the removal rate of bothPAHs and alkanes using solvent extraction method. Watersolubility was characterized as the parameter most stronglycorrelated with solvent extraction with the correlation coeffi-cient of 0.84 and 0.86 for alkanes and PAHs, respectively.

The PLS loading plots are shown in Fig. 4. The factorloading is the Pearson correlation between a factor and avariable. The coordinates of variables along each axis repre-sent the strength of relationship between that variable andeach factor. In each PLS model, three principle components

were identified. The PLS calculation for the soil propertiesyielded a cumulative R2Y of 93.0 % and a Q2Y of 24.2 %,respectively (Fig. 4a). The moisture, CEC, pH value, and thepercentage of sand fractions were overall important variables(VIP>1) for the extraction of alkanes and PAHs. The watersolubility and log Koc were characterized as the important oilproperties, while the carbon number and molecular weightseemed less important for oil removal (Fig. 4b, c). The cumu-lative R2Y and Q2Y in the PLS models for alkane propertieswere 67.1 and 54.4 %, respectively. For PAH properties, thecumulative R2Yand Q2Ywere 56.2 and 15.2 %, respectively.

Table 2 Composition changes in the petroleum hydrocarbons after solvent extraction (mean ± standard error, n=2)

Soil ∑Alkanes ∑PAHs TPH

Initial(mg kg−1)

Treated(mg kg−1)

Removal(%)

Initial(mg kg−1)

Treated(mg kg−1)

Removal(%)

Initial(mg kg−1)

Treated(mg kg−1)

Removal(%)

S1 12,415.7±335.3 781.2±19.4 93.7 80.8±9.3 9.2±0.8 88.7 12,496.5±326.0 790.4±20.2 93.7

S2 8,013.5±70.2 730.6±120.9 90.9 97.7±20.3 9.4±2.3 90.4 8,111.2±59.9 740.0±118.5 90.9

S3 514.4±6.5 125.6±3.0 75.6 55.5±3.9 10.6±1.6 80.9 569.8±2.7 136.2±4.6 76.1

S4 500.8±30.2 118.8±0.9 76.3 28.9±1.2 4.8±0.6 83.5 529.7±31.3 123.6±1.4 76.7

S5 936.9±0.3 164.2±0.3 82.5 14.5±3.2 2.7±3.2 81.3 951.5±3.2 166.9±3.2 82.5

S6 9,565.9±486.4 1,483.6±98.3 84.5 19.3±1.3 5.1±0.1 73.4 9,585.1±485.0 1,488.8±98.4 84.5

0

200

400

600

800

1000 Initial soil Treated soil

S2S1

0

200

400

600

0

10

20

30

40

50

60

S5

S4S3

0

10

20

30

40

C11

C12

C13

C14

C15

C16

C17

C18

C19

C20

C21

C22

C23

C24

C25

C26

C27

C28

C29

C30

C31

C32

C33

C34

C35

0

20

40

60

80

100

120S6C

once

ntra

tion

of a

lkan

es (

mg

kg-1)

C11

C12

C13

C14

C15

C16

C17

C18

C19

C20

C21

C22

C23

C24

C25

C26

C27

C28

C29

C30

C31

C32

C33

C34

C35

0

200

400

600

Fig. 1 Changes in thedistribution of individual alkanesafter solvent extraction (Errorbars represent stand errors, n=2)

Environ Sci Pollut Res

Discussion

All the soil samples used in this study were seriously contam-inated for years, while this study clearly demonstrated that thepetroleum ether was a highly effective solvent for removing

high-concentration weathered petroleum from contaminatedsoils. During the last decades, a great number of aqueous ornonaqueous solvents have been developed for petroleum oilremoval from contaminated soil, which was well documentedin a most recent review paper (Lau et al. 2014). Generally, theoil extraction efficiency reported so far was between 75 and99 % (Khodadoust et al. 2005; Risdon et al. 2008; Silva et al.2005). This study proposed an effective solvent for fast re-moving up to 94 % of TPH within 20 min at 25 °C, while theextraction time ranged from 1 to 48 h and the temperatureranged from 70 to 100 °C in previous studies (Ahn et al. 2008;Han et al. 2009; Khodadoust et al. 2005). More importantly, itremained unclear whether the solvent extraction method waseffective when high-concentration weathered crude oil waspresent in the soil, because most of the oil concentrationpreviously reported was less than 5×103 mg kg−1 (Wu et al.2013a). The mass transfer resistance for the release of TPHfrom soil to the solvent would be highly increased after theweathering of the high-concentration crude oil (up to 3×105 mg kg−1 as shown in Table 1). In our recent studies, wedemonstrated the high effectiveness of hexane–acetone mix-ture (Li et al. 2012a) or hexane–pentane mixture (Wu et al.2013a) for treating such contaminated soils. This study sug-gested that petroleum ether could be used as an alternative

0

1

2

3

15

30

45

0

2

4

10

20

30

Con

cent

ratio

n of

PA

Hs

(mg

kg-1)

Initial soil Treated soil

0

2

4

8

12

16

20

S1

0.0

0.2

0.4

0.6

4

8

12

Nap

Acp

Ace Flu

Phe

n

Ant Flt

Pyr

BaA Chr

BbF

BkF

BaP In

d

DbA

BgP

0

1

2

3

4

Nap

Acp

Ace Flu

Phe

n

Ant Flt

Pyr

BaA Chr

BbF

BkF

BaP In

d

DbA

BgP

0

1

2

3S6S5

S4S3

S2Fig. 2 Changes in thedistribution of individual PAHsafter solvent extraction (Errorbars represent stand errors, n=2)

Sand

Silt

Clay

pH

Moisture

CEC

SOM

Oil

CN

MW

WS

log Kow

log Koc

-1.2 -0.8 -0.4 0.0 0.4 0.8 1.2

Correlation coefficient

PAHAlkane

Soi

l pro

pert

ies

Oil

prop

ertie

s

Fig. 3 Correlation coefficient between solvent extraction efficiency andthe properties of soils and hydrocarbons. CEC cation exchange capacity,SOM soil organic matter, CN carbon number, MWmolecular weight, WSwater solubility

Environ Sci Pollut Res

because (a) the effects of secondary pollution can be mini-mized without inducing a new kind of organic solvent aspetroleum ether itself is a distillation fraction of petroleumand (b) the boiling point of petroleum ether is low (40–80 °C);therefore, it is easy for recycling and reusing by separating itfrom the extracted oil using distillation techniques.

Before performing real remediation of oil contaminatedsoil via solvent extraction using petroleum ether, soil charac-terization is a crucial step since the deviation of soil propertiesmight result in different distribution and transports of oilcontaminants which would in turn influence the status of soils.The Pearson correlation coefficient was computed in order toidentify possible interactions between soil properties(Table 3). The CEC was significantly correlated with clay(p<0.01) and pH (p<0.05). Insignificant correlation was ob-served between the initial concentration of heavy crude oil andother soil properties (p>0.05). Several studies reported thesignificant dependence of TPH level on the physicochemicalcharacteristics of soils (Bauder et al. 2005; Brohon et al. 2001;Masakorala et al. 2013). Particularly, Masakorala et al. (2013)demonstrated a highly significant (p<0.001) positive

correlation between TPH and the SOM (R2=0.895) and pH(R2=0.846) of the soils in Dagang oil field. However, suchphenomenon was not clearly observed in this study. Forexample, the lowest concentration of initial TPH was notedin S4, but the corresponding SOM content was the secondhighest among the six soils. TPH in S6 was one order ofmagnitude higher than that in S3, S4, and S5, but the leastSOM was noted in S6.

These different findings may be attributed to the origin ofSOM and the various degree of weathering. The soil samplesused byMasakorala et al. (2013) were collected from the sameoil field and the largest distance between each two samplingpoints was less than 150 m. By contrast, the soils used in thisstudy originated from six different oil fields distributed acrossnorth China (Fig. SM-1). It is believed that SOM from differ-ent sources differs in composition, functionalities, and confor-mations which also depend on the age and other environmen-tal factors (Niederer et al. 2007; Wen et al. 2007). For exam-ple, if xenobiotic organics represent the main fraction ofSOM, it may inhibit the carbon cycle and restrain the accu-mulation of natural organic matter in TPH-contaminated soil

Fig. 4 PLS loading plot of solvent extraction variables. The PLS predictability (Q2Y) for a soil properties, b alkane properties, and cPAH properties was24.2, 54.4, and 15.2 %, respectively. (Red: important X-variables with VIP>1; blue: less important X-variables with VIP<1; black: Y-variable)

Environ Sci Pollut Res

(Valsecchi et al. 1995). Moreover, the conformation and sur-face polarity of SOM can change during weathering processwhen it is bound to soil minerals (Feng et al. 2006; Sun et al.2013). This means that the overall TPH sorption was influ-enced by SOM amount, SOM type, and weathering degree.There should be not much difference in the SOM types inMasakorala et al. (2013) as the soil samples they used hadsimilar history of exploration and contamination. Therefore,the observed correlation between TPH and SOM amountshould be more pronounced than that in the present study.Caution therefore should be taken before drawing a finalconclusion on the relationship between SOM and the TPHpresent in the soil.

Nevertheless, a positive correlation was noted between theextractability of petroleum hydrocarbons and the SOM andpH of soils (Fig. 3). It is not surprising to find a higher solventextraction efficiency at higher pH values, because the crudeoil would be strongly attached to the soil surface at low pHvalues (Painter et al. 2010). The alkaline environment couldpotentially weaken the sorption of oil by changing the surfacestructure of soils which was favorable for the oil recovery bysolvent extraction (Ji and Zhou 2007). By contrast, the report-ed influences of SOM on the solubility and mobility of oilhave long been controversial. Considerable works demon-strated that the presence of SOM served a major compartmentfor the sink of hydrophobic pollutants making them lessextractable after sequestration during weathering process(Jonsson et al. 2010; Yang et al. 2010a, b). On the other hand,it could also increased the desorption rate by changing the zetapotential of porous particles (Pelley and Tufenkji 2008), de-creasing the binding sites of the solid surface due to compet-itive adsorption (Flores-Céspedes et al. 2006; Nelson et al.2000), or reducing the surface tension by forming micelleslike a biosurfactant (Montoneri et al. 2009; Quagliotto et al.2006). The results obtained from this study seem to beconsistent with the latter process. Our results furtherindicated that the influence of SOM on the extractionof alkanes was more pronounced than PAHs as shownby the higher correlation coefficient obtained for PAHs(0.44) than for alkanes (0.17) (Fig. 3).

In addition to desorption from SOM, the other two keyprocesses controlling the recovery of petroleum hydrocarbonsby solvent extraction are (a) dissolution in the solvent mole-cules in the oil layer and diffusion back into the solvent bodyand (b) breaking through the porous network structure formedby the insoluble oil fractions such as asphaltenes on the soilsurface (Li et al. 2012b). The former process depended on theconcentration gradient, which suggested that a higher oilcontent should benefit for solvent extraction. However, thesolvent extraction efficiency was negatively correlated withthe initial oil content (Fig. 3). Therefore, it was inferred thatdissolution and diffusion were not likely the mass transfercontrolling processes during oil extraction in this study. Thisfinding highlighted the complex interactions between hydro-carbons and their host matrix (i.e., oil phase) which was lessrepresented within risk assessment for hydrocarbon-contaminated sites (Pollard et al. 2008). Early studies sug-gested that a threshold of 103 mg kg−1 soil is sufficient for theoil phase to function as a distinct partitioning phase separatedfrom SOM (Boyd and Sun 1990; Sun and Boyd 1991). Thecrude oil concentration in this study was two order of magni-tude higher than this value (Table 1), which suggested that anexplicit NAPL compartment existed. Consequently, the se-questration of soluble petroleum compounds inside theasphaltene networks which was insoluble in petroleum etherwould potentially become the predominant mass transfer ratecontrolling process. This was further supported by the stron-ger correlation with oil content observed for alkanes than forPAHs because of their higher hydrophobicity resulting in astronger binding to oil phase (Fig. 3). Therefore, understand-ing the behavior of petroleum hydrocarbons in the residualNAPL phase is essential for improving solvent extractionefficiency.

Although the release of petroleum hydrocarbons fromNAPL into the water phase is often negligible (Coulon et al.2010; Pollard et al. 2008; Zemanek et al. 1997), the moisturecontent was positively correlated with the solvent extractionespecially for PAHs (Fig. 3). Similar results were reported byShu and Lai (2001) because a high ratio of moisture to soilcontaminants would promote the contaminant’s desorption

Table 3 Correlation matrix ofsoil properties

*p<0.05; **p<0.01

Sand Silt Clay pH Moisture CEC SOM Oil

Sand 1.000

Silt −0.788* 1.000

Clay −0.812* 0.279 1.000

pH −0.208 −0.302 0.610 1.000

Moisture −0.100 −0.042 0.196 0.334 1.000

CEC −0.595 0.021 0.908** 0.795* 0.413 1.000

SOM −0.442 0.420 0.291 0.101 −0.024 0.388 1.000

Oil −0.207 −0.103 0.420 −0.087 −0.232 0.369 0.336 1.000

Environ Sci Pollut Res

(Thibaud et al. 1993). However, Gong et al. (2005) found thatthe removal rate increased by 15.5 % when moisture contentdecreased from 17.1 to 0.3 %. This was attributed to the factthat the contact area between organic solvent and contaminantwould decrease when increasing the moisture ratio, which isadverse for solvent extraction. Moreover, the discrete water-filled soil pores would form when large quantity of watermolecules takes up the micropores in the soils, which wouldreduce the capillarity effect that drove hydrocarbons to thesolvent phase through the micro channels (Wu et al. 2011).Therefore, it is speculative that the effects of moisture on thesolvent extraction efficiency would potentially change frompositive to negative when it increased to a critical value. Thepositive effects of moisture predominated in this study as themoisture content was less than 5 % (Table 1), which wassmaller than the critical value (if existed) of approximately10 % reported by Lian et al. (2009).

The higher solvent extraction efficiency observed in thesoils with larger particle sizes was consistent with previousstudies (Freeman and Harris 1995), because the release andtransport of petroleum were influenced by the detachment ofparticles associated with hydrocarbons (Bauder et al. 2005)and the solvent could penetrate the particles more thoroughlyin coarse sands than in clay or silt.

Conclusions

The petroleum ether was demonstrated to be an effectivesolvent for removing high concentration of weathered petro-leum from contaminated soil with TPH removal up to 94 %.Little changes were found in the contribution of total alkaneand total PAH to the overall TPH after solvent extraction, butthe profile of individual hydrocarbons was obviouslychanged. This highlighted the need to target specific indicatorcompounds other than TPH for a more realistic assessment ofremediation. The dependence of initial TPH on the soil prop-erties such as SOM, pH, moisture, and CEC was not observedin this study, but these characteristics were positively corre-lated with the removal rate of both alkanes and PAHs bysolvent extraction. All the oil properties selected in this studywere negatively correlated with solvent extraction, while thewater solubility was characterized as the most significantparameter. The high concentration of crude oil in soils wasadverse for solvent extraction, which highlighted thedemand for focusing on the NAPL and residual oilphase when performing solvent extraction for hydrocarbon-contaminated soils, because the predominant mass trans-fer controlling process was the sequestration of solublefractions inside the porous network structure formed bythe insoluble fractions such as asphaltenes in the weath-ered crude oil.

Acknowledgments This work was financially supported by NationalNatural Science Foundation of China (No. 21307069, No. 21306129 andNo. 41201497), Municipal Natural Science Foundation of Tianjin(No. 12JCQNJC05300), Guangdong Natural Science Foundation(No. S2013040012413), and China Postdoctoral Science Foundation(No. 2013M530641).

References

Ahn CK, Kim YM,Woo SH, Park JM (2008) Soil washing using variousnonionic surfactants and their recovery by selective adsorption withactivated carbon. J Hazard Mater 154:153–160

Alexander M (1995) How toxic are toxic chemicals in soil? Environ SciTechnol 29:2713–2717

Alexander M (2000) Aging, bioavailability, and overestimation of riskfrom environmental pollutants. Environ Sci Technol 34:4259–4265

Avila-Chavez MA, Trejo A (2010) Remediation of soils contaminatedwith total petroleum hydrocarbons and polycyclic aromatic hydro-carbons: extraction with supercritical ethane. Ind Eng Chem Res 49:3342–3348

Bauder T, Barbarick K, Ippolito J, Shanahan J, Ayers P (2005) Soilproperties affecting wheat yields following drilling-fluid applica-tion. J Environ Qual 34:1687–1696

Boyd SA, Sun S (1990) Residual petroleum and polychlorobiphenyl oilsas sorptive phases for organic contaminants in soils. Environ SciTechnol 24:142–144

Brassington KJ, Hough RL, Paton GI, Semple KT, Risdon GC, CrossleyJ, Hay I, Askari K, Pollard SJT (2007) Weathered hydrocarbonwastes: a risk management primer. Crit Rev Environ Sci Technol37:199–232

Brassington K, Pollard S, Coulon F (2010) Handbook of hydrocarbonand lipid microbiology, Weathered hydrocarbon biotransformation:implications for bioremediation, analysis, and risk assessment.Springer, Heidelberg, pp 2487–2499

Brohon B, Delolme C, Gourdon R (2001) Complementarity of bioassaysand microbial activity measurements for the evaluation ofhydrocarbon-contaminated soils quality. Soil Biol Biochem 33:883–891

Coulon F, Whelan MJ, Paton GI, Semple KT, Villa R, Pollard SJT (2010)Multimedia fate of petroleum hydrocarbons in the soil: oil matrix ofconstructed biopiles. Chemosphere 81:1454–1462

Feng X, Simpson AJ, Simpson MJ (2006) Investigating the role ofmineral-bound humic acid in phenanthrene sorption. Environ SciTechnol 40:3260–3266

Flores-Céspedes F, Fernández-Pérez M, Villafranca-Sánchez M,González-Pradas E (2006) Cosorption study of organic pollutantsand dissolved organic matter in a soil. Environ Pollut 142:449–456

Freeman HM,Harris EF (1995) Hazardouswaste remediation: innovativetreatment technologies. CRC, Boca Raton

Gong Z, Wilke BM, Alef K, Li P (2005) Influence of soil moisture onsunflower oil extraction of polycyclic aromatic hydrocarbons from amanufactured gas plant soil. Sci Total Environ 343:51–59

Han M, Ji GD, Ni JR (2009) Washing of field weathered crude oilcontaminated soil with an environmentally compatible surfactant,alkyl polyglucoside. Chemosphere 76:579–586

He L, Li X, Wu G, Lin F, Sui H (2013) Distribution of saturates,aromatics, resins and asphaltenes fractions in the bituminous layerof Athabasca oil sands. Energ Fuel 27:4677–4683

ISO (2010) BS ISO 11277:2009: Determination of particle size distribu-tion in mineral soil material—method by sieving and sedimentation

Environ Sci Pollut Res

Ji G, Zhou G (2007) Remediation of soil contaminated with oil pollutantsby ex-situ chemical washing. Acta Sci Nat—Univ Pekin (Chin) 43:863–871

Jonsson S, Persson Y, Frankki S, van Bavel B, Lundstedt S, Haglund P,Tysklind M (2007) Degradation of polycyclic aromatic hydrocar-bons (PAHs) in contaminated soils by Fenton’s reagent: a multivar-iate evaluation of the importance of soil characteristics and PAHproperties. J Hazard Mater 149:86–96

Jonsson S, Lind H, Lundstedt S, Haglund P, Tysklind M (2010) Dioxinremoval from contaminated soils by ethanol washing. J HazardMater 179:393–399

Khodadoust AP, Reddy KR, Maturi K (2005) Effect of different extrac-tion agents on metal and organic contaminant removal from a fieldsoil. J Hazard Mater 117:15–24

Latawiec AE, Reid BJ (2010) Sequential extraction of polycyclic aromatichydrocarbons using subcritical water. Chemosphere 78:1042–1048

Lau EV, Gan S, Ng HK, Poh PE (2014) Extraction agents for the removalof polycyclic aromatic hydrocarbons (PAHs) from soil in soil wash-ing technologies. Environ Pollut 184:640–649

Li X, Du Y,WuG, Li Z, Li H, Sui H (2012a) Solvent extraction for heavycrude oil removal from contaminated soils. Chemosphere 88:245–249

Li X, He L, Wu G, Sun W, Li H, Sui H (2012b) Operational parameters,evaluation methods, and fundamental mechanisms: aspects of non-aqueous extraction of bitumen from oil sands. Energ Fuel 26:3553–3563

Lian J, Shi S, Guo M, Li X (2009) Influence of soil properties on thesolvent extraction of oil from contaminated soil using n-hexane.Chem Ind Eng Process (Chin) 28:530–532

Mackay D (2001) Multimedia environmental models: the fugacity ap-proach. Lewis, Chelsea

Mackay D, Shiu WY, Ma KC (1997) Illustrated handbook of physical-chemical properties and environmental fate for organic chemicals:pesticide chemicals, 5. CRC, Boca Raton

Masakorala K, Yao J, Chandankere R, Liu H, Liu W, Cai M, Choi MM(2013) A combined approach of physicochemical and biologicalmethods for the characterization of petroleum hydrocarbon-contaminated soil. Environ Sci Pollut Res. doi:10.1007/s11356-013-1923-3

Montoneri E, Boffa V, Savarino P, Tambone F, Adani F, Micheletti L,Gianotti C, Chiono R (2009) Use of biosurfactants from urbanwastes compost in textile dyeing and soil remediation. WasteManage 29:383–389

Navarro-Villoslada F, Vicente BS, Moreno-Bondi MC (2004)Application of multivariate analysis to the screening of molecularlyimprinted polymers for bisphenol A. Anal Chim Acta 504:149–162

Nelson S, FarmerW, Letey J,Williams C (2000) Stability and mobility ofnapropamide complexed with dissolved organic matter in soil col-umns. J Environ Qual 29:1856–1862

Niederer C, Schwarzenbach RP, Goss K-U (2007) Elucidating differencesin the sorption properties of 10 humic and fulvic acids for polar andnonpolar organic chemicals. Environ Sci Technol 41:6711–6717

Painter P,Williams P, LupinskyA (2010) Recovery of bitumen fromUtahtar sands using ionic liquids. Energ Fuel 24:5081–5088

Pelley AJ, Tufenkji N (2008) Effect of particle size and natural organicmatter on the migration of nano- and microscale latex particles insaturated porous media. J Colloid Interface Sci 321:74–83

Pignatello JJ, Xing B (1996) Mechanisms of slow sorption of organicchemicals to natural particles. Environ Sci Technol 30:1–11

Pollard SJT, Hough RL, Kim KH, Bellarby J, Paton G, Semple KT,Coulon F (2008) Fugacity modelling to predict the distribution oforganic contaminants in the soil: oil matrix of constructed biopiles.Chemosphere 71:1432–1439

Quagliotto P, Montoneri E, Tambone F, Adani F, Gobetto R, Viscardi G(2006) Chemicals from wastes: compost-derived humic acid-likematter as surfactant. Environ Sci Technol 40:1686–1692

Reid BJ, Jones KC, Semple KT (2000) Bioavailability of persis-tent organic pollutants in soils and sediments—a perspectiveon mechanisms, consequences and assessment. Environ Pollut108:103–112

Risdon GC, Pollard SJT, Brassington KJ, McEwan JN, Paton GI, SempleKT, Coulon F (2008) Development of an analytical procedure forweathered hydrocarbon contaminated soils within a UK risk-basedframework. Anal Chem 80:7090–7096

Saterbak A, Toy RJ, Wong DC, McMain BJ, Williams MP, Dorn PB,Brzuzy LP, Chai EY, Salanitro JP (1999) Ecotoxicological andanalytical assessment of hydrocarbon‐contaminated soils and appli-cation to ecological risk assessment. Environ Toxicol Chem 18:1591–1607

Semple KT, Reid BJ, Fermor TR (2001) Impact of composting strategieson the treatment of soils contaminated with organic pollutants.Environ Pollut 112:269–283

Shu YY, Lai TL (2001) Effect of moisture on the extraction efficiency ofpolycyclic aromatic hydrocarbons from soils under atmosphericpressure by focused microwave-assisted extraction. J ChromatogrA 927:131–141

Silva A, Delerue-Matos C, Fiuza A (2005) Use of solvent extraction toremediate soils contaminated with hydrocarbons. J Hazard Mater124:224–229

Sun S, Boyd SA (1991) Sorption of polychlorobiphenyl (PCB)congeners by residual PCB-oil phases in soils. J EnvironQual 20:557–561

Sun K, Jin J, Kang M, Zhang Z, Pan Z, Wang Z, Wu F, Xing B (2013)Isolation and characterization of different organic matter fractionsfrom a same soil source and their phenanthrene sorption. EnvironSci Technol 47:5138–5145

Thibaud C, Erkey C, Akgerman A (1993) Investigation of theeffect of moisture on the sorption and desorption of chloro-benzene and toluene from soil. Environ Sci Technol 27:2373–2380

TPHCWG (1999) Total Petroleum Hydrocarbon Criteria Working GroupSeries Volume 5: Human health risk-based evaluation of petroleumrelease sites: implementing the working group approach. AmherstScientific, Amherst

Valsecchi G, Gigliotti C, Farini A (1995) Microbial biomass, activity, andorganic matter accumulation in soils contaminated with heavymetals. Biol Fertil Soils 20:253–259

Viglianti C, Hanna K, De Brauer C, Germain P (2006) Removal ofpolycyclic aromatic hydrocarbons from aged-contaminated soilusing cyclodextrins: experimental study. Environ Pollut 140:427–435

Vinzi VE (2010) Handbook of partial least squares: concepts, methodsand applications. Springer, Heidelberg

Wehrer M, Totsche KU (2009) Difference in PAH release processes fromtar-oil contaminated soil materials with similar contamination histo-ry. Chem Erde-Geochem 69:109–124

Wen B, Zhang J, Zhang S, Shan X, Khan SU, Xing B (2007)Phenanthrene sorption to soil humic acid and different humin frac-tions. Environ Sci Technol 41:3165–3171

Wu G (2012) Insights into sustainable environmental remediation ap-proaches and the fate and transport of petroleum hydrocarbons insoils. Dissertation, Tianjin University

Wu G, Li X, Coulon F, Li H, Lian J, Sui H (2011) Recycling of solventused in a solvent extraction of petroleum hydrocarbons contaminat-ed soil. J Hazard Mater 186:533–539

Wu G, Coulon F, Yang Y, Li H, Sui H (2013a) Combining solventextraction and bioremediation for removing weathered petroleumfrom contaminated soil. Pedosphere 23:455–463

Wu G, He L, Chen D (2013b) Sorption and distribution of asphaltene,resin, aromatic and saturate fractions of heavy crude oil on quartzsurface: molecular dynamic simulation. Chemosphere 92:1465–1471

Environ Sci Pollut Res

Wu G, Kechavarzi C, Li X, Sui H, Pollard SJT, Coulon F (2013c)Influence of mature compost amendment on total and bioavailablepolycyclic aromatic hydrocarbons in contaminated soils.Chemosphere 90:2240–2246

Yang Y, Tao S, Zhang N, Zhang DY, Li XQ (2010a) The effect of soilorganic matter on fate of polycyclic aromatic hydrocarbons in soil: amicrocosm study. Environ Pollut 158:1768–1774

Yang Y, Zhang N, XueM, Tao S (2010b) Impact of soil organic matter onthe distribution of polycyclic aromatic hydrocarbons (PAHs) insoils. Environ Pollut 158:2170–2174

Zemanek MG, Pollard SJT, Kenefick SL, Hrudey SE (1997) Multi-phasepartitioning and co-solvent effects for polynuclear aromatic hydro-carbons (PAH) in authentic petroleum- and creosote-contaminatedsoils. Environ Pollut 98:239–252

Environ Sci Pollut Res