Approaches to the remediation of a polychlorinated biphenyl (PCB)

contaminated soil–a laboratory study

Qixiang Wu and William D. Marshall*

Department of Food Science and Agricultural Chemistry, Macdonald Campus of McGillUniversity, 21111 Lakeshore Road, Ste-Anne-de-Bellevue, Quebec, Canada H9X 3V9.E-mail: Marshall@macdonald.mcgill.ca; Fax: (514)-398-7977; Tel: (514)-398-7921

Received 9th February 2001, Accepted 18th April 2001First published as an Advance Article on the web 9th May 2001

A soil that had been historically contaminated with Aroclor 1242, 1248, 1254 and 1260 was decontaminated by

two surfactant-mediated cleaning procedures that had been chosen to mimic ex-situ washing and in-situ soil

flushing processes. A preliminary screening selected four surfactants (from 17 commercial formulations) for

their ability to mobilise PCBs from the soil while suffering minimal losses to the supercritical carbon dioxide

(scCO2) that was used in a separate back-extraction procedure. The mobilisation was enhanced, with minimal

foam formation, by the presence of 17% (v/v) IBMK in the surfactant suspension. Each of the four surfactants,

at 1, 3, or 5% (v/v) concentration, was evaluated by (i) 15 successive 10 min sonication–filtrations and (ii)

continuous soil column flushing during 20 h. Each filtrate from (i) and samples, taken at hourly intervals, from

(ii) were analysed for their PCB and surfactant content. Both extraction procedures mobilised PCBs efficiently

when extended for longer periods and were modelled accurately as the sum of a constant and single-term

exponential increase to a maximum. The predicted number of replicate stages required to mobilise 50% of the

toxicants (t50) varied from 7 to 3 for sonication–washing of the soil (10 g) or from 6.8 to 2.8 h for column

flushing of 30 g soil and decreased as the concentration of surfactant in the aqueous phase was increased. The

combined PCB-laden aqueous suspensions were then back-extracted efficiently with scCO2 and the eluate was

dechlorinated quantitatively as it traversed a short, heated column of silver–iron bimetallic mixture.

Introduction

Proposals for the restoration of soils that have been pollutedwith polychlorinated biphenyl (PCB) compounds haveincluded incineration, solidification/vitrification,1 phyto-remediation,2 bioremediation,3 and electrokinetic4 approaches.These strategies, however, have been applied as treatments inthe field only infrequently because of costs, environmentalconstraints and efficacy.5 Other more efficient methods fortreating PCB contaminated soils continue to be proposed,optimised and evaluated.6

In recent years, the use of surfactants has received increasedattention as a means of enhancing the removal of organicpollutants from contaminated soils,7–9 especially for thosesituations when conventional ‘‘remove and treat’’ protocolshave proven to be prohibitively costly.10,11 Surfactants areattractive for the mobilisation/mass transfer of organiccontaminants because of their reduced acute toxicity andtheir favourable rates for environmental degradation toinnocuous products. Aqueous surfactant suspensions havebeen considered as more environmentally friendly thancompeting organic-solvent-based extraction systems.12,13

Surfactant-enhanced soil remediation procedures includeboth soil washing (ex situ) and soil flushing (in situ).14 Theseprocesses become economically feasible only if they are of lowcost. A further processing stage that is able to decontaminatethe soil extract can help to increase its remediation value,15 andoften combinations of different techniques must be used duringthe processes.16 An attractive combination has involvedsurfactant recovery and re-use.17

Other approaches to the removal of hydrophobic contami-nants with volatile organic solvents or surfactants have beenproposed and tested.11,18 These separation methods haveinclude: air-stripping or vacuum-stripping within packedcolumns, hollow fiber membrane contactors,18 liquid–liquid

extraction within a hollow fiber membrane contactor19 andpervaporation of surfactant solution.20 Although their highremoval efficiencies have encouraged researchers to evaluatethese processes in the field, often they have not been sufficientlyefficient, they have been environmentally damaging or theyhave been viewed as being too expensive. More information ofthe effects of surfactants on the extraction operation and novelseparation techniques remain to be identified/optimised.Promising in this respect are the reports by Hawthorne andco-workers21,22 who mobilised PCBs from field-contaminatedsoil with pressurised sub-critical water and the report of Waiand co-workers23 who observed the virtually quantitativedechlorination of PCBs during extraction with sub-criticalwater in the presence of zero valent iron.

Supercritical carbon dioxide (scCO2) based separationspossess several advantages over conventional extractionswith organic solvents; including cost, a lack of appreciabletoxicity of the solvent, increased diffusivities and penetrationrates through particulate media, high mobilisation efficienciesfor relative non-polar analytes and environmental friendli-ness.24 We were anxious to evaluate novel methods/proceduresthat might permit continuous cleaning of contaminated soil.For particulate matrices, extractions with scCO2 are inherentlybatch processes in which the extractor is charged with soil,pressurised, extracted with scCO2 then depressurised and thesoil is replaced with fresh substrate before the cycle can berepeated. Conventionally, pressure within the extractor ismaintained with a capillary restrictor (y50 mm diameter) thatis easily fouled with particles. We chose to minimise foulingwith a two-stage process in which PCBs would be mobilisedfrom the soil into an aqueous surfactant emulsion and thenstripped from the water medium with scCO2. It seemedprobable that surfactants could be identified that wouldpromote the mobilisation from soil but would be relativelyinsoluble in the scCO2 and it was anticipated that the cleaned

DOI: 10.1039/b101323h J. Environ. Monit., 2001, 3, 281–287 281

This journal is # The Royal Society of Chemistry 2001

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online / Journal Homepage / Table of Contents for this issue

surfactant suspension could be recycled back to the soil tomobilise more toxicant. It also seemed to be inefficient tosimply transfer the toxicants from one medium to another (soilto scCO2) so that an on line detoxification stage was envisagedfor the process. The effluent from the scCO2 extractor would bedechlorinated by passage through a short column of zero-valent bimetallic mixture at elevated temperature underconditions that had been demonstrated to dechlorinate acontinuous stream of these substrates virtually quantitatively.24

The current report focuses on the use of surfactants to mobilisePCBs from the soil and the use of scCO2 to back-extract thesetoxicants from the surfactant suspension. Once in scCO2,dechlorination over Ag0/Fe0 was evaluated.

Materials and methods

Chemicals, solvents and surfactants

Magnesium granules (12–50 mesh, 99.8% purity), potassiumhexachloropalladate (K2PdCl6), cobaltous nitrate[Co(NO3)2?6H2O], and ammonium thiocyanate (NH4SCN)were purchased from Alfa Aesar, Ward Hill, MA,USA. Hexane, ethylene glycol dimethyl ether, isobutylmethylketone (IBMK, 99z%), methanol, methylene chloride(CH2Cl2), propan-2-ol (all HPLC grade), and glass woolwere purchased from Fisher Scientific, Ottawa, ON. Biphenyl,surfactants, antifoam A, and antifoam B were purchased fromSigma-Aldrich, Oakville, ON. All chemicals (ACS reagentgrade or better), solvents and surfactants were used as received.

The surfactants selected for study included anionic (AerosolOT, sodium dodecyl sulfate) and nonionic chemicals (Brij 30,Brij 56, Brij 78, Brij 97, Brij 98, Triton X-100, Triton X-405,Triton X-705, Triton CF54, Triton DF16, Tween 20, Tween 40,Tween 60, Tween 80, Tween 85).

Soil samples and pre-treatment

The soil had been contaminated with PCB compounds formore than 30 years. The pH of the contaminated soil was 7.4and the organic matter content was 4.8%. The properties of thissoil have been described previously.25 The total PCB contentwas determined to be 30.3¡6.6 mg g21 and involved Aroclors1242, 1248, 1254, and 1260. Post sampling, the soil was airdried, mixed thoroughly, passed through a 2 mm sieve (10mesh), and stored at 4 ‡C to await further treatment.

PCBs extraction/mobilisation

In preliminary trials, 17 surfactants were screened to mobilisePCBs from an aqueous PCB–surfactant mixture into super-critical CO2 (scCO2). The aqueous suspension (25 mL)consisting of surfactant (3% v/v), Aroclor 1242 (1% v/v) anddistilled water was supplemented with 5 mL IBMK, and theresulting suspension (30 mL) was well mixed and placed in acylindrical supercritical fluid (SCF) extraction vessel[320 mm615 mm inner diameter (id)]. The vessel was wrappedwith a heating tape (Swagelok, Montreal) and heated to theoperating temperature (30–80 ‡C) of the experiment. ScCO2

provided by a scCO2 pump (Newport Scientific Inc., Jessup,MD USA), was then added to the base of the extraction vesseland permitted to percolate up through the surfactant suspen-sion and to accumulate within the headspace above thesuspension. The fluid in the headspace was replaced con-tinuously with fresh solvent from below and pressure within thesystem was maintained with a short section of flexible silicatubing (y25 cm650 mm id) that terminated with the stainless-steel transfer line and acted as a capillary restrictor. Extractoreluate was trapped in 25 mL hexane contained in a 50 mLtest-tube. Once suitable surfactants had been identified,scCO2 extractions (second stage) were performed on

PCB–surfactant mixtures (in the absence of soil) at varioustemperatures.

The third stage consisted of evaluations of the optimisedsurfactant-extraction time–extraction temperature combina-tions to mobilise PCBs from the contaminated soil. Bothsonication–soil washing and continuous flushing of packed soilcolumns were evaluated. The sonication–washing treatmentwas performed by placing the air-dried soil sample (10 g) into afritted glass funnel (30 cm640 cm, Fisher Scientific), togetherwith surfactant and the suspension was sonicated [Polytronhomogenizing unit (N.Y. 11590, Brinkmann Instruments,Mississauga, ON) for 10 min prior to filtration under agentle vacuum. Separate soil washes were performed with 1,3, 5% (v/v) aqueous surfactant suspension and each extract wasanalysed for PCB and surfactant content. Each sonication–vacuum filtration procedure was repeated 14 times with thesame soil sample but with fresh surfactant suspension eachtime.

An empty HPLC column (30 cm61 cm id Swagelok,Montreal, QC) was used to perform the soil column flushingexperiments. Soil (30 g) was packed into the column andterminated with a coarse filter at both column ends. Thecolumn was mounted horizontally and flushed for 20 h(1.0 mL min21) with [1, 3 or 5% (v/v)] surfactant suspensiondelivered from an HPLC pump (model 100A, Altex Scientific,Berkley, CA). The effluent was analysed at hourly intervals forPCB and surfactant content.

Finally, the cumulative soil extracts were back-extractedwith scCO2 under the optimised mobilisation conditions torecover the PCBs from the surfactant suspension. PCBs, thathad been extracted into scCO2, were dechlorinated with zerovalent bimetallic mixture (Ag0/Fe0) and the cleaned surfactantsuspension in the retentate was then re-used for the soil PCBextraction.

PCB analysis

The total content of PCBs in soil was determined by gaschromatography equipped with flame ionisation detection(GC-FID) or a Hewlett-Packard model 5890 gas chromato-graph equipped with a model 5971 mass selective detector (GC-MS). Both methods have been described previously.24,25 Inbrief, the PCB concentration in soil extracts was determined byconversion of PCB residues to biphenyl. Magnesium flakes(1 g, 12–50 mesh), potassium hexachloropalladate (K2PdCl6),2 mg, propan-2-ol (2 mL), and 2 mL soil extract werecombined in a 50 mL flask. The mixture was vortex mixedfor 1 min repeatedly at approximately 2 h intervals. After 10 h(5 cycles), the biphenyl was extracted into hexane (1 mL) andquantified as described previously. The estimation of PCBcontent was based on calibrations using Aroclor 1260.

Surfactant analysis

The Standard Analysis Method, APHA,26 was followed todetermine the surfactant concentration in both soil and in themicelle suspension post-treatment. Because anionic surfactantsproved to be unsuitable, surfactant analyses were performedonly for nonionic surfactants. A standard calibration curve wasgenerated for each nonionic surfactant. Micellar suspension(1 mL) was diluted (0–10 fold) to obtain a suitable workingconcentration. Diluted surfactant suspension (2 mL) containedin a 100 mL flask, was evaporated to dryness on a water bath.The residues were triturated with methylene chloride (10 mL)and the resulting extract was added to a 125 mL separatoryfunnel containing 5 mL cobaltous thiocyanate solution. After60 s vigorous shaking and phase separation, the lower aqueouslayer was transferred, via a funnel containing a plug of glasswool, to a 5 mL spectrophotometer cell. The absorbance, at620 nm, was determined vs. a blank of CH2Cl2.

282 J. Environ. Monit., 2001, 3, 281–287

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online

Results and discussion

Surfactant screening

Seventeen surfactants were screened in this study. Eachsurfactant and its relevant properties is listed in Table 1. Thefactors that influenced selection for further study included anability to increase the mobilisation of Aroclor 1242 into scCO2,

a lack of appreciable foaming during the extraction process andefficient recovery of surfactant in the aqueous retentate.Surfactant suspensions that could be extracted with scCO2

during 30 min, without the addition of antifoaming reagentand/or co-solvent(s) and with only minimal losses to the mobilephase were to be selected. Aqueous mixtures of PCB–surfactantsuspension could be extracted with scCO2 if IBMK, 5 mL, wasadded to the suspension coupled, in exceptional case (cited

below), with supplementing the mobile phase scCO2 withIBMK (0.4 or 0.2 mL min21 added prior to its entry into theextraction vessel).

PCB mobilisation from aqueous surfactant suspension

To optimise the extraction conditions, suspensions of 1% (v/v)Aroclor 1242 with each of four selected surfactants (3% v/v Brij97, Triton CF54, Triton DF16 or Tween 85) and 5 mL IBMKwere extracted with scCO2 (1.42 MPa and 30–80 ‡C) for either30 or 60 min. For suspensions of PCB and Brij 97 that had beenextracted for 30 min, mobilisation efficiencies increased withincreased operating temperature ($ symbol, Fig. 1A). How-ever, the content of Brij 97 in the retentate also decreasedsubstantially with increased operating temperature. As sum-marised in Table 2, the quantity of mobilised PCBs was only

Table 1 Chemical and physical characteristics of surfactants selected for the PCB extraction study

Trade name Chemical structure HLBa Viscosity/cP Molecular weight

Anionic—SDS Sodium dodecyl sulfate 1.12 288b

Aerosol OT Dioctyl sulfosuccinate 0.95 445c

Nonionic—Brij 35 POE(10) lauryl ether 16.9 1.21 1198cd

Brij 56 POE(10) cetyl ether 12.9 683Brij 78 POE(20) stearyl ether 15.3 1152c

Brij 97 POE(10) lauryl ether 12.4 1.28Brij 98 POE(20) oleyl ether 15.3 1.38 1150c

Triton X-100 POE(10) isooctylphenyl ether 13.5 624c

Triton X-114 POE(8) isooctylphenyl ether 12.4Triton X-405 POE(40)-(aromatic ring)6–(CH)8 17.9 1966Triton CF54 Alkylaryl polyether 13.6e

Triton DF16 POE linear alcoholTween 20 POE(20) sorbitan monolaurate 16.9 1.19 1266cd

Tween 40 POE(20) sorbitan monopalmitate 15.6 1.18Tween 60 POE(20) sorbitan monostearate 14.9 1312d

Tween 80 POE(20) sorbitan monoleate 15 1.19 1310d

Tween 85 POE(20) sorbitan monolaurate 11.0 1840b

aHLB: hydrophilic–lipophilic balance. bRef. 29. cRef. 11. dRef. 17. eRef. 28.

Fig. 1 PCB mobilisation efficiency (%) as a function of operating temperature and surfactant remaining with the aqueous phase (%) for 30 min ($)or 60 min (&) of scCO2 extraction, at 1.42 MPa, from suspensions containing (A) Brij 97, (B) Triton CF54, (C) Triton DF16 or (D) Tween 85.

J. Environ. Monit., 2001, 3, 281–287 283

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online

15% for 30 min of extraction at 30 ‡C, however, the content ofBrij 97 in the retentate remained virtually unchanged (99.9%)for extraction at this temperature. The PCB mobilisationincreased to 80% when the extraction was conducted at 80 ‡C;however, the remaining content of Brij 97 was reduced to 58%.Linear regression analysis (Table 3) of three factors (PCBrecovery, extraction temperature, and surfactant content thatremained with the aqueous retentate) indicated that thequantity of PCB and the quantity of Brij 97 surfactant thatwas mobilised were both related linearly to the operatingtemperature. Increased operating temperatures enhanced PCBextraction efficiencies but reduced the content of surfactantthat remained with the aqueous suspension.

The PCB mobilisation efficiency reached 99% if theextraction was continued for 60 min at 80 ‡C (Table 2; &

symbol, Fig. 1A). The quantity of Brij 97 that remained in thetreated suspension did not change substantially when 30 minmobilisation (58% retention) was increased to 60 min (54%retention) for extraction at 80 ‡C. The three-factor linearregression model (Table 3) again indicated that the PCBextraction efficiency was related linearly to extraction tem-perature and increased substantially with increases in thisparameter. For the Brij 97 surfactant, the optimised PCBmobilisation conditions included supplemental IBMK(0.4 mL min21) that was merged with the scCO2 stream priorto entry into the extraction vessel.

Extraction efficiencies from the 3% (v/v) suspensions ofTriton CF54, Triton DF16 or Tween 85, were qualitativelysimilar. Recoveries of Aroclor 1242 from these media arerecorded in Fig. 1B, C and D for 30 ($ symbol) or 60 min ofextraction (& symbol). Recoveries at the extraction tempera-ture extremes (30 or 80 ‡C) during 30 or 60 min extraction arealso summarised in Table 2. Mobilisation efficiencies increasedwith both longer extraction time (60 vs. 30 min) and withincreased temperature of extraction so that the mobilisationbecame virtually quantitative above 60 ‡C for the extendedextraction time. For all four surfactant suspensions, increasedPCB mobilisation efficiency was accompanied by increasedlosses of surfactant. The optimised linear regression models(Table 3) indicated that the PCB extraction efficiency waspositively related to the operating temperature, but that thecontent of surfactant in the retentate was negatively related tothe temperature. For mobilisations from the Triton CF54suspensions, it was necessary to supplement the scCO2 mobilephase with 0.2 mL min21 IBMK. But it was not necessary tosupplement the mobile phase during extractions of either theTriton DF16 or Tween 85 suspensions. Extractions could beeffected smoothly if 5 mL IBMK was added to the surfactantsuspension (25 mL) prior to extraction. In summary, optimal

surfactant performance was in the order Tween 85wTritonDF16wTriton CF54wBrij 97.

PCB extraction/mobilisation from soil

Since surfactants, at concentrations less than their criticalmicelle concentration (cmc),8 have been reported to promotethe sorption of PCBs to soil particles during extraction,11 thesame surfactants, at 1, 3, or 5% (v/v), were evaluated for PCBmobilisation from the contaminated soil. Both ex-situ sonica-tion–washing and in-situ flushing of a packed soil column wereevaluated. To optimise the sonication treatment with the soil–surfactant suspensions, Brij 97 was selected to define a suitablesonication time. PCB mobilisation efficiencies from 10 g soil,increased moderately with increased sonication times (Fig. 2).Although 15 min of sonication mobilised more PCB than the10 min treatment that, in turn, was more efficient than the5 min treatment, appreciable foam formation was observedwith the later washes. Therefore, 10 min of sonication wasemployed in subsequent trials. A single ‘‘exponential rise to amaximum’’ term provided an expression that modelled theexperimental data accurately. The cumulative percentage ofPCBs mobilised was modelled as the sum of a constant (Y0)plus an exponential increase term that took the formA(12e2lx) where A is a coefficient, l is a ‘‘time’’ constant

Table 2 Mobilised PCB (%) and surfactant content (%) in the retentate post-extraction with scCO2 at 30 ‡C or 80 ‡C for 30 or 60 min

Brij 97 Triton CF54 Triton DF16 Tween 85

Extraction time/min 30 30 60 60 30 30 60 60 30 30 60 60 30 30 60 60Temperature/‡C 30 80 30 80 30 80 30 80 30 80 30 80 30 80 30 80PCBs mobilised (%) 15 80 19 99 45 92 55 91 88 100 92 100 34 96 60 100Remaining surfactant (%) 100 58 100 54 92 30 74 20 52 12 40 7 80 34 72 30

Table 3 Linear regression models for three factors consisting of PCB mobilisation efficiency, percentage of initial surfactant remaining in theretentate and processing temperature for 30 or 60 min of extractiona

Surfactant 30 min of extraction R2 60 min of extraction R2

Brij 97 % PCBs mobilised~3.5Sz0.7T26.4 0.998 % PCBs mobilised~1.4Sz2.9T268.8 0.996Triton CF54 % PCBs mobilised~0.9Sz2.1Tz79.0 0.999 % PCBs mobilised~21.2Sz0.9Tz14.0 0.996Triton DF16 % PCBs mobilised~0.4Sz0.6Tz67.5 0.996 % PCBs mobilised~0.3Sz0.7Tz82.0 0.994Tween 85 % PCBs mobilised~22.9Sz4.9Tz143.3 0.982 % PCBs mobilised~12.4S213Tz130.8 0.989aS~percentage of initial surfactant content remaining in the treated solutions, T~extraction temperature.

Fig. 2 Cumulative extraction efficiency from soil (10 g) in consecutive10 mL washes with Brij 97 aqueous suspension (3% v/v) followingsonication for 5 (+), 10 ($) or 15 min (,). The three-parameter bestfit curves were generated with a model consisting of a single exponentialincrease to a maximum.

284 J. Environ. Monit., 2001, 3, 281–287

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online

and x represents the number of successive sonication–washings. The optimised models indicated that t50, thenumber of sonication–washings required to mobilise one-halfof the extractable fraction of total PCB soil burden, was 4.08,2.68 and 2.20 for 5, 10 and 15 min treatments, respectively. Theinclusion of additional exponential terms in the model did notincrease the goodness of fit to the experimental data. Thus,there was no evidence for the existence of separate fractions ofsoil PCBs that could be differentiated based on their rates ofmobilisation. This contrasts with an earlier report27 thatdescribed the scCO2 fractionation of soil PCBs from ‘‘rapidlydesorbing, moderately, slowly, and very slowly desorbing’’sites. If present, the differences in the fractions were masked byincluding the IBMK within the aqueous slurry.

As illustrated in Fig. 3A, 87%, 92%, or 95% of the PCBs wasextracted from the soil (10 g) with 15 consecutive sonication–washing treatments using 1%, 3% or 5% (v/v) Brij 97suspension, respectively. An increased concentration ofsurfactant in the aqueous suspension enhanced PCB extraction,however, recoveries from 3 or 5% suspension were notsignificantly different from each other. A single exponentialincrease to a maximum modelled the experimental dataaccurately for all three concentrations and indicated t50

values of 6.93, 2.98 and 2.89 sonication–washes for the 1%,3%, and 5% surfactant solutions, respectively. The standarderrors of estimate (a parameter of goodness of fit) weresatisfyingly low (1.94, 1.59 and 3.57, respectively). The meanPCB content remaining in the soil post sonication–washing(¡1 standard deviation, s) was 2.4¡0.6, 1.6¡0.3, and1.5¡0.4 mg g21 (dry weight basis) for the 1, 3, and 5% (v/v)surfactant suspensions, respectively (Table 4). The loss of Brij97 surfactant from the soil suspension was appreciable for theearly washes (Table 5) but losses became negligible in the laterfractions indicating that sorption to soil particles was theprinciple route of loss. Once the soil had become saturated withsurfactant, further losses were negligible. The net percent loss(by difference) to the soil (Table 6) was 3.6¡3.3, 8.4¡4.5 or10.4¡5.6 for the 5%, 3% or 1% Brij 97 treatments, respectively.

The results of packed column soil flushing trials (Fig. 3A,open symbols) were rather similar to the observations in thesoil washing experiments. Most of PCBs (w95%) had beenextracted/mobilised after 20 h (1200 mL) of continued flushingwith 5% (v/v) Brij 97 suspension. Decreasing the concentrationof surfactant in the carrier reduced the extraction efficiency.For 1% Brij 97 suspension, 85% of the PCB soil burden hadbeen mobilised after 20 h of flushing [# symbol, Fig. 3A] and

Fig. 3 Cumulative PCB extraction efficiencies for mobilisations from 10 g soil by sonication (10 min)–washing with 1% ($), 3% (r) or 5% (&) or bycontinuous flushing of 30 g soil with 1% (#), 3% (e) or 5% (%) aqueous suspension of A, Brij 97; B, Triton CF54; C, Triton DF16; or D, Tween 85surfactant.

Table 4 Residual PCB concentrations (mg g21 dry weight¡1s based on 3 separate determinations) in soil post 15 consecutive sonication–washingsor continuous flushing during 20 h

Brij 97 Triton CF54 Triton DF16 Tween 85

1% 3% 5% 1% 3% 5% 1% 3% 5% 1% 3% 5%

Treatment—Sonication–washing 2.4¡0.6 1.6¡0.3 1.5¡0.4 1.0¡0.2 0.3¡0.1 0.2¡0.1 1.5¡0.2 0.2¡0.1 0.2¡0.1 4.8¡0.5 — —Column flushing 2.7¡0.4 2.1¡0.5 1.5¡0.3 1.2¡0.3 0.8¡0.2 0.4¡0.2 2.8¡0.1 0.9¡0.2 2.1¡0.2 2.0¡0.3 1.8¡0.3 2.9¡0.4

J. Environ. Monit., 2001, 3, 281–287 285

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online

differences in extraction efficiencies between the 3% (e symbol)and 5% (% symbol) Brij 97 suspensions were not significant.The mean PCB contents in the soil post-flushing (¡1s) were2.7¡0.4, 2.1¡0.5, and 1.5¡0.3 mg g21 for the 1%, 3%, and 5%aqueous suspensions, respectively (Table 4). The residualquantities of surfactant sorbed to the soil post-flushing weresimilar to the quantities observed for the sonication–washingexperiments. Quantities of surfactant in the extract post-processing did not differ greatly (Table 6) from the quantitiespresent in the suspension prior to processing.

The PCB extractions from soil by sonication–washing withTriton CF54 were reasonably efficient. As illustrated inFig. 3B, the quantities of mobilised PCB were 80.1%, 91.1%and 93.5% for aqueous suspensions of 1% ($ symbol), 3%(r symbol) and 5% (& symbol), respectively. The singleexponential rise to a maximum also accurately modelled theexperimental observations with the CF54 surfactant. Values fort50 of 5.74, 5.37 and 2.94 sonication–washes and 3.95, 2.85 and2.92 h of flushing were predicted for the 1%, 3%, and 5%suspensions. The quantities of PCBs in the soil post-treatmentwith 1%, 3% or 5% Triton CF54 were 1.0¡0.2, 0.3¡0.1, and0.2¡0.1 mg g21, respectively (Table 4). Results for soil columnflushing with this surfactant were comparable to extractions bysonication–washing (Fig. 3B, open symbols). The PCB extrac-tion was enhanced when the concentration of CF54 wasincreased from 1% to 3% (v/v) but a further increase from 3% to5% did not significantly increase the mobilisation further. Theamount of PCB that remained in the treated soil was 1.2¡0.3,0.2¡0.2, or 0.4¡0.2 mg g21 for flushing with 1%, 3%, or 5%Triton CF54 suspension (Table 4).

PCB sonication–washing treatments with Triton DF16proved to be the most efficient among the four surfactants(Fig. 3C). The efficiency of PCB purging from the soil was 89%,95%, and 97% for 1%, 3% and 5% suspensions, respectively. Ananalysis of variance (ANOVA) indicated that differences inefficiency among the treatments were not significant. The PCBcontents remaining with the soil post-washing were 1.5¡0.2,0.2¡0.1, and 0.2¡0.1 mg g21 for the 1%, 3% or 5% DF16

suspensions. Soil flushing treatments with Triton DF16provided analogous results (Fig. 3C, open symbols). Thequantities of mobilised PCBs were 89%, 95%, or 97% withsuspensions of 1%, 3% or 5%, respectively. The residual PCBcontents in the treated soil were 2.8¡0.4, 0.9¡0.2, and2.1¡0.3 mg g after flushing with 1%, 3%, and 5% DF16suspension, respectively.

Unlike the other surfactants, Tween 85 proved to be lesssuitable for soil sonication–washings. The Tween 85 suspensiondispersed soil particles very strongly so that, once sonicated, itbecame difficult to filter materials from the treated slurry. Thisbehaviour was evident even after the initial treatment with the1% (v/v) suspension. The more treatments performed, the moredifficult it became to filter the suspension, and increasedconcentration of this surfactant only exacerbated the problem.In consequence, sonication–washing with the 3% and the 5%Tween suspension were not completed. Extracted PCBsamounted to 81% for 1% suspension (Table 6) and the residualPCB content was 4.8¡0.5 mg g21 (Table 4). This residual PCBcontent was higher than for similar treatments with the othersurfactants.

The concentration of aqueous Tween 85 solutions had less ofan effect on the packed soil flushing treatments, although theappreciable soil particle dispersion required higher pressure toforce the mobile phase through the column. The quantity ofPCB extracted from the soil was 85%, 89% and 90% for theflushing trials with 1%, 3% or 5% suspension. The residual PCBcontents were 2.8¡0.4, 1.8¡0.3, and 2.9¡0.4 mg g21 forsuspensions containing 1%, 3%, or 5% surfactant. The fractionof surfactant that remained with the soil was also relativelygreater than with similar concentrations of other surfactantsfor both the flushing and washing treatments (Table 6).

For the soil cleaning trials, the mean standard error of estimate(a measure of the goodness of fit of the mathematial model to thedata) for all 22 trials (1.998; range, 1.1908–3.5736) did not varyappreciably as the quantity of surfactant in the suspension(2.5255 for 1% vs. 2.54528 for 3% vs. 2.3903 for the 5% sonication;and 1.905 for 1% vs. 1.803 for 3% vs. 1.800 for 5% flushing).

Table 5 Surfactant content (%) remaining in the extract post soil sonication–washing (10 g) or continuous column flushing (30 g)

Brij 97 Triton CF54 Triton DF16 Tween 85

1% 3% 5% 1% 3% 5% 1% 3% 5% 1% 3% 5%

Sonication–washing treatment—1 68.6¡6.9 76.1¡7.1 79.4¡4.3 62.1¡6.3 74.6¡5.6 80.1¡4.7 63.4¡5.4 70.6¡5.1 78.6¡4.8 60.3¡3.62 80.2¡4.2 89.3¡3.9 92.8¡4.0 83.8¡4.8 87.5¡3.9 94.8¡3.2 78.6¡4.1 86¡3.8 95.2¡3.1 72.5¡4.23 90.1¡2.1 93.5¡2.7 98.5¡1.9 91.8¡3.0 98.2¡2.5 99.1¡2.2 89.8¡3.0 98¡2.3 99.5¡2.2 80.9¡2.74 95.6¡1.6 98.0¡1.7 100¡2.0 97.9¡2.3 99¡1.7 100¡1.3 98.9¡2.1 100¡1.5 100¡1.4 89.3¡4.85 99.0¡2.0 100¡1.1 100¡1.8 100¡1.2 100¡1.3 100¡1.0 100¡2.0 100¡1.4 100¡1.2 95.5¡4.6

Column flushing/min—10 32.4¡13.9 38.6¡10.1 43.2¡8.9 37.8¡9.9 42.4¡5.8 45.7¡6.5 39.4¡8.6 40.5¡6.7 47.1¡6.6 43.5¡6.7 47.6¡6.4 48.2¡7.120 47.2¡9.8 53.2¡8.9 60.4¡5.5 49.7¡6.5 53.9¡4.5 57.8¡4.1 47.5¡5.8 52.8¡5.4 60.2¡4.5 50.9¡6.8 58.5¡4.5 66.5¡6.130 66.3¡7.4 69.7¡5.5 76.8¡3.5 64.9¡4.4 68.9¡2.3 71.9¡3.3 68.1¡4.6 65.7¡3.7 73.2¡4.5 71.4¡4.5 73.6¡3.3 77.7¡4.640 75.7¡5.7 80.7¡4.6 89.8¡3.5 79.9¡2.9 83.2¡3.9 87.4¡2.8 78.8¡3.8 85.6¡3.6 88.9¡3.1 83.5¡4.6 88.8¡4.7 90.9¡3.650 82.6¡5.4 89.6¡3.6 95.6¡2.7 88.5¡2.4 93.6¡2.9 96.5¡3.7 89.2¡3.5 92.6¡3.2 97.4¡3.4 92.2¡3.8 96.8¡3.9 98.1¡2.460 98.9¡2.6 99.9¡1.6 99.9¡1.0 100¡1.6 100¡0.5 100¡1.0 100¡1.5 100¡1.7 100¡2.1 100¡1.4 100¡2.0 100¡1.7

Table 6 Surfactant content (%) remaining with the aqueous phase post soil flushing, soil sonication–washing or remediation of the eluate by back-extraction with scCO2

Brij 97 Triton CF54 Triton DF16 Tween 85

1% 3% 5% 1% 3% 5% 1% 3% 5% 1% 3% 5%

Treatment—Sonication–

washing89.6¡5.6 91.6¡4.5 96.4¡3.3 90.1¡4.4 95.6¡3.6 98.1¡2.7 91.4¡4.6 96.6¡3.1 98.6¡3.8 80.6¡6.9

Soil columnflushing

90.6¡4.3 93.4¡3.2 97.6¡3.6 89.3¡3.6 94.9¡3.7 98.8¡3.1 92.3¡3.2 95.7¡2.6 97.7¡2.5 85.6¡6.6 90.8¡5.8 92.3¡5.5

ScCO2

extraction70.2¡4.0 76.3¡6.1 78.8¡4.6 60.9¡4.5 66.6¡5.2 75.3¡4.6 20.3¡5.6 23.6¡4.5 31.6¡4.7 45.3¡5.6 48.6¡4.6 58.8¡3.8

286 J. Environ. Monit., 2001, 3, 281–287

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online

PCB separation from soil extract and surfactant regeneration

PCBs were mobilised from the soil cleaning emulsion by back-extraction with scCO2 during 30 min at 50 ‡C. The resultsindicated that w99% of PCB (data not shown) were mobilisedfrom the soil extracts and 23–78% of surfactant (Table 6)remained in the aqueous suspension post-treatment. The cleanedsurfactant suspension could then have been re-used to mobilisemore PCB from the contaminated soil. Based on the PCBextraction efficiencies, the suitability of surfactants for eithersoil sonication–washing or column flushing treatments were inthe order: Triton DF16wTriton CF54wBrij 97wTween 85,whereas increased surfactant losses were observed in the order:Tween 85wBrij 97wTriton CF54wTriton DF16.

PCB dechlorination

Once PCBs had been mobilised into scCO2, they weredechlorinated (on line) to biphenyl by passing the extractoreffluent through a reactor column of zero valent silver–iron(Ag0/Fe0) bimetallic mixture under anoxic conditions.24

Analyses by GC-MS did not detect any chlorinated productsthat were eluted from the analytical capillary column within thewindow for PCB compounds. There was no evidence thatorganic matter or surfactant that had been mobilised from theaqueous surfactant–soil suspension affected the dechlorinationefficiency, which proved to be both rapid and apparentlyquantitative.

In total, the results have demonstrated that PCBs can beremoved from the soil by extraction with dilute suspensions (1–5% v/v) of aqueous surfactant. All four commercial nonionicformulations were approximately equally efficient when usedwith procedures designed to mimic either ex-situ washing or in-situ soil flushing. The removal efficiency from the soil waspredicted accurately with a simple exponential expression thatincluded the number of successive treatments and a ‘‘time-constant’’ (l) that was characteristic of the washing process.The number of successive washes required to extract 50% of thePCBs (t50) varied as the concentration of surfactant in theemulsion for both washing and flushing experiments. Equili-bration with Triton CF54 was typical; 5.74 and 2.94 washeswere predicted to be required for the t50 with 1% and 5%surfactant emulsions, respectively. The mathematical modelalso accurately predicted the course of the flushing process withthe soil column. In this case, the PCB concentration waspredicted to be reduced two-fold after 3.95 (237 mL) or 2.92 h(y175 mL) of flushing. Comparable values for sonication–washing and for flushing were observed with similar concen-trations of the other surfactants. Despite differences in thequantity of soil treated (10 g for sonication–washing vs. 30 g forsoil flushing) comparable mobilisation efficiencies wereachieved with both procedures.

There were no apparent differences in the avidity with whichseparate fractions of the total PCB burden interacted with thesoil. Although three of the surfactant formulations did notchange the physical properties of the soil, one formulation wasobserved to disperse the soil agglomerates appreciably andrequired increased pressure to force the aqueous phase throughthe soil. Nonetheless, large volumes of surfactant suspension

were required to mobilise the PCBs efficiently. The challengeremains to improve the efficiency of this stage of the overallprocess. The PCBs in the cumulative aqueous extracts wereback-extracted efficiently (w99%) into scCO2–IBMK. Lossesof surfactant during this stage were appreciable but thedechlorination was virtually quantitative.

References

1 Analytical of Chemistry of PCBs, ed. M. D. Erickson, CRC press/Lewis Publishers, Boca Raton, FL, 2nd edn., 1997, 688 pp.

2 E. K. Dzantor, T. Chekol and L. R. Vough, J. Environ. Sci.Health, Part A, 2000, A35, 1645.

3 A. C. Alder, M. M. Haggblom, S. R. Oppenheimer andL. Y. Young, Environ. Sci. Technol., 1993, 27, 530.

4 S.-O. Kim, S. H. Moon and K.-W. Kim. , Environ. Technol., 1999,21, 417.

5 J. W. W. Kovalick and J. Kingscott, Hazard. Waste Hazard.Mater., 1996, 13, editorial/opinion.

6 A. S. Abdul and T. L. Gibson, Environ. Sci. Technol., 1991, 25,665.

7 A. S. Abdul, T. L. Gibson and D. N. Rai, Ground Water, 1990, 28,920.

8 S. B. Burchfield, D. J. Wilson and A. N. Clarke, Sep. Sci. Technol.,1994, 29, 47.

9 C. Huang, J. E. Van Benschote, T. C. Healy and M. E. Ryan,J. Soil Contam., 1997, 6, 537.

10 V. Weerasooriya, S. L. Yeh, G. A. Pope and W. H. Wade, ACSSymp. Ser., 2000, 740, 23.

11 J.-S. Kim and K. Lee, J. Environ. Sci. Health, Part A, 2000, 35,915.

12 S. Deshpande, B. J. Shiau, D. Wade, D. A. Sabatini andJ. H. Harwell, Water Res., 1999, 33, 351.

13 M. J. Beck, C. A. Layton, J. P. Easter, G. S. Sayler, J. Barton andM. Reeves, Environ. Sci. Res., 1997, 54, 73.

14 J. F. Scamehorn and J. H. Harwell, ACS Symp. Ser., 2000, 740, 1.15 D. N. Sabatini and H. J. Robert, IAHS Publ., 1998, 250, 361.16 F.-H. Haegel, F. Dierkes, S. Kowalski, K. Monig, M. J. Schwuger,

G. Subklew and P. Thiele, ACS Symp. Ser., 2000, 740, 35.17 M. Zhou and R. D. Rhue, Environ. Sci. Technol., 2000, 34, 1985.18 B. Kitiyanan, J. H. O’Haver, J. H. Harwell and D. A. Sabatini,

ACS Symp. Ser., 2000, 740, 76.19 C. C. Ang and A. S. Abdul, Proc. Annu. Meet.-Air Waste Manage.

Assoc., 1995, 15, 95.20 L. M. Vane, E. L. Giroux, F. R. Alvarez and L. Hitchens, ACS

Symp. Ser., 2000, 740, 57.21 Y. Yang, S. Bowadt, S. B. Hawthorne and D. J. Miller, Anal.

Chem., 1995, 67, 4571.22 S. B. Hawthorne, C. B. Grabanski, K. J. Hageman and D. J. Miller,

J. Chromatogr., A, 1998, 814, 151.23 H. J. Yak, B. W. Wenclawiak, I. F. Cheng, J. G. Doyle and

C. M. Wai, Environ. Sci. Technol., 1999, 33, 1307.24 Q. Wu, W. D. Marshall and A. Majid, Green Chem., 2000, 2, 127.25 Q. Wu and W. D. Marshall, Int. J. Environ. Anal. Chem., 2001,

79, 1.26 A. E. Dreebberg, L. S. Clescer and A. D. Eaton, Standard

Methods, American Public Health Association, Washington, DC,18th edn., 1992, p. 5.36–5.41.

27 E. Bjorklund, S. Bowadt, L. Mathiasson and S. B. Hawthorne,Environ. Sci. Technol., 1999, 33, 2193.

28 A. Eckelhoff and A. V. Hirner, Prog. Colloid Polym. Sci., 1998,111, 189.

29 D. Myers, Surfactant Science and Technology, VCH, New York,1988.

J. Environ. Monit., 2001, 3, 281–287 287

Publ

ishe

d on

09

May

200

1. D

ownl

oade

d by

Uni

vers

ity o

f Y

ork

on 2

2/10

/201

4 14

:18:

27.

View Article Online