Environmental Impacts ofRemediation of aTrichloroethene-Contaminated Site:Life Cycle Assessment ofRemediation AlternativesG I T T E L E M M I N G , * , †

M I C H A E L Z . H A U S C H I L D , ⊥

J U L I E C H A M B O N , † P H I L I P J . B I N N I N G , †

C E C I L E B U L L E , ‡ M A N U E L E M A R G N I , †

A N D P O U L L . B J E R G †

Department of Environmental Engineering, TechnicalUniversity of Denmark (DTU), DK-2800 Lyngby, Denmark,

Department of Management Engineering, Technical Universityof Denmark (DTU), DK-2800 Kgs. Lyngby, Denmark, and

The Interuniversity Research Centre for the Life Cycle ofProducts, Processes and Services (CIRAIG), Ecole Polytechniquede Montreal., P.O. Box 6079, Montreal,Quebec H3C 3A7, Canada

Received June 14, 2010. Revised manuscript receivedOctober 20, 2010. Accepted October 25, 2010.

The environmental impacts of remediation of a chloroethene-contaminated site were evaluated using life cycle assessment(LCA). The compared remediation options are (i) in situbioremediation by enhanced reductive dechlorination (ERD),(ii) in situ thermal desorption (ISTD), and (iii) excavation of thecontaminated soil followed by off-site treatment and disposal.The results showed that choosing the ERD option will reduce thelife-cycle impacts of remediation remarkably compared tochoosing either ISTD or excavation, which are more energy-demanding. In addition to the secondary impacts of remediation,this study includes assessment of local toxic impacts (theprimary impact) related to the on-site contaminant leaching togroundwater and subsequent human exposure via drinkingwater. The primary human toxic impacts were high for ERD dueto the formation and leaching of chlorinated degradationproducts, especially vinyl chloride during remediation. However,the secondary human toxic impacts of ISTD and excavationare likely to be even higher, particularly due to upstream impactsfrom steel production. The newly launched model, USEtox,was applied for characterization of primary and secondary toxicimpacts and combined with a site-dependent fate model ofthe leaching of chlorinated ethenes from the fractured clay tillsite.

Introduction

Chlorinated ethenes, such as perchloroethene (PCE) andtrichloroethene (TCE), are among the most frequent con-taminants found in soil and groundwater due to their

extensive and widespread use as cleaning agents and metaldegreasers (1). In the 2004 U.S. National Priority Listchlorinated ethenes were by far the most common group oforganic contaminants at sites prioritized for remediation (2).Remediation methods for chlorinated ethenes can be eitherex situ methods, where contaminated soil/groundwater isexcavated/pumped to the surface and treated on- or off-site,or they can be in situ methods that remove contaminantsvia mass transfer or mass removal by targeting them at theiractual location in the subsurface.

Life cycle assessment (LCA) is an ISO standardized andwidely used method for environmental assessment ofproducts and services. It has also been applied in the fieldof soil and groundwater remediation to compare theenvironmental impacts of remediation alternatives as re-ported in two recent literature reviews (3, 4). Existing studieshave, however, focused mainly on ex situ remediation andcontaminants such as metals, PAHs, and hydrocarbons (3).LCA studies addressing chlorinated solvent remediation (5, 6)focus on the comparison of groundwater plume remediationtechniques e.g. in situ permeable reactive barriers andconventional pump-and-treat systems, whereas in situmethods for source zone remediation of chlorinated etheneshave not yet been a focus of published LCA studies.

Environmental impacts from remediation can be dividedinto primary and secondary impacts (see e.g. refs 7 and 8).Primary impacts are the local toxic impacts from the residualsite contamination, whereas secondary impacts are impactson the local, regional, and global scale generated by theremediation activities. In addition, a study introduced theterm tertiary impacts to describe impacts associated withthe postremediation fate of a brownfield (9), but these arenot considered here. Existing studies that include primaryimpacts use generic characterization factors that do not takethe site-specific contaminant fate and exposure into con-sideration and furthermore focus on impacts in surface wateror soil (7-9). Although highly relevant for remedial actions,primary impacts in groundwater have not been targeted inexisting studies. This may be because of the fact that thegroundwater compartment is not included as emission orimpact compartment in the applied models for life cycleinventory (LCI) and life cycle impact assessment (LCIA).

The aim of this study is to use LCA for a comparison ofprimary and secondary environmental impacts of threealternative technologies for remediating a TCE-contaminatedsource zone. Site-generic characterization factors for toxicemissions do not adequately represent the fate of chlorinatedethenes at contaminated sites because they disregard deepersoil layers and groundwater causing the main part of thecontamination to end up in the atmosphere (10). Further-more, they do not include the formation of metabolites duringbiodegradation of chlorinated ethenes, of which particularlyvinyl chloride is problematic due to its toxic and carcinogeniceffects (10). Therefore a site-dependent assessment is usedhere, taking into account the site-specificity of the fate of thecontaminant including formation of metabolites. In addition,site-dependent exposure parameters are used for calculationof exposure concentrations and the exposed number ofpeople. In this case the primary impacts are related to humanexposure via ingestion of groundwater due to on-site leachingof contaminants and are evaluated using a site-specificleaching model. Primary ecotoxic impacts in groundwaterare neglected, and no discharge to surface water is includedbecause the groundwater plume is assumed to be fullyabstracted by the downstream drinking water well. Thecompared remediation technologies are as follows: (i)

* Corresponding author phone: (+45) 4525 1595; fax: (+45) 45932850; e-mail: gile@env.dtu.dk.

† DTU Environment.⊥ DTU Management Engineering.‡ CIRAIG.

Environ. Sci. Technol. 2010, 44, 9163–9169

10.1021/es102007s 2010 American Chemical Society VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9163

Published on Web 11/05/2010

enhanced reductive dechlorination (ERD) which is an in situbioremediation method employing bioaugmentation andbiostimulation; (ii) in situ thermal desorption (ISTD) whichis a physical remediation method deploying thermal con-ductive heating of the contaminated soil; and (iii) excavationof the contaminated soil followed by off-site treatment andfinal disposal in a landfill.

Based on the LCA results the environmental impacts ofthe three remediation alternatives are compared. Environ-mental hotspots of each technology are identified andimprovement options suggested.

Materials and MethodsCase study. A TCE-contaminated site in central Copenhagenis situated within the catchment of a water supply that extractsdrinking water for approximately 44,000 people located2000 m down gradient from the site. The site was used as ametal shop, boilermaker shop, woodcutting shop, etc. until1986. A conceptual sketch of the contaminated site is shownin Figure 1. The contamination is located mainly in a fracturedclay till and partly in a sand layer embedded in the clay till.The contaminated source zone extends vertically from 3 to8 m below ground surface and has a horizontal cross-sectionalarea of 140 m2. The estimated mass of TCE in the ap-proximately 700 m3 of contaminated soil is 40 kg. The claytill overlies a regional limestone aquifer used for drinkingwater extraction.

The goal of the LCA is to compare three options forremediating the contaminated source zone. In addition tothe three remediation scenarios the assessment includes ano action scenario, in which no active remediation takes

places and monitoring of the naturally occurring degradationis the only activity included. The no action scenario isincluded as a reference scenario only and cannot be seen asa viable management scenario comparable to the threeremediation scenarios.

The functional unit provided by all options compared inthis study is the treatment of the 700 m3 of contaminated soilwith a 98% removal of the contaminant mass within thisvolume. This remediation target was chosen as it reducesthe mass discharge to the groundwater sufficiently to ensurethat the Danish groundwater quality criterion (1 µg/L) forchlorinated ethenes is not exceeded. The timeframe for theLCI is unrestricted in order to capture both short- and long-term impacts.

The scope of the LCA is to include all important activitieson site and off site (see overview in Table 1 and a morecomplete listing in the Supporting Information (SI) TableS1) of each scenario, i.e. covering raw materials acquisition,materials production, use stages, and end-of-life processes.Primary data regarding energy and material consumptionfor the different remedial activities is collected from the actualconsultants, contractors, or producers likely to undertakethe work. Inventory data for the background system (pro-duction of steel, electricity, plastic, asphalt, lorry transport,etc.) were based on average technology data from theecoinvent life cycle unit process database version 2.0 (11).These generic data are supported by a collection of specificdata for remediation-related processes for which no data arefound in ecoinvent or other general LCA databases (produc-tion of bioculture and activated carbon, laboratory analysesof soil and groundwater). Electricity used in the operation

FIGURE 1. Conceptual model of the contaminated site and placement of the microbial degradation zones used in the numericalmodel of natural (no action) and enhanced reductive dechlorination (ERD). The sand layer embedded in the clay till is neglected inthe numerical modeling of the site because of its limited extent. Model results of mass discharge of trichloroethene (TCE),cis-dichloroethene (cis-DCE), and vinyl chloride (VC) to the regional aquifer is shown together with the accumulated emissions ofeach chlorinated ethene to the aquifer.

9164 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010

phases is assumed produced as the average Danish electricitygrid mix including imports (11). All steel (unalloyed steeland stainless steel) is modeled as secondary steel producedfrom scrap, and a recycling rate of 90% is assumed for allsteel products. Based on experience from contractors, heaters,and temperature sensors for ISTD are assumed to be reuseddirectly on four projects each, and the sheet pile wall for theexcavation scenario is reused directly on two projects beforethe steel is recycled. For all scenarios, including the long-term ones (ERD and NoA) current technology is assumed forthe entire period. Additional assumptions and the appliedecoinvent processes for the background system are sum-marized in Tables S2-S4 in the SI. A number of sensitivityscenarios have been analyzed in order to test the importanceof main input parameters to each of the remediation methods.The substrate amount (molasses), the electricity source, andthe soil transportation distance are the parameters tested inthe sensitivity analysis of the ERD, ISTD, and excavationscenario, respectively.

The life cycle impact assessment method applied isEDIP2003 (12) for the categories global warming, ozoneformation, acidification, and eutrophication. Respiratory im-pacts associated with particulate matter (PM2.5-10 µm, PM<2.5 µm)and inorganics (NOx, SO2, and NH3) are assessed usingcharacterization factors from Humbert et al. (13), and the newlydeveloped scientific consensus model, USEtox (version 1.0,January 2010), is used for human toxicity and ecotoxicity (14).The USEtox model addresses ecotoxic impacts in the freshwatercompartment only as the modeling of ecotoxic impacts in themarine and terrestrial compartments is considered too im-mature at present. All USEtox characterization factors providedfor metal emissions are classified as interim as they disregardcentral fate aspects like speciation (14).

The LCA results are normalized against the backgroundimpacts from society by expressing them as person equiva-lents. The applied normalization references represent theannual impacts of an average European/world citizen in 2004for regional/global impacts (listed in the Supporting Infor-mation (SI) C.2).

System description. No action (NoA) is a baseline scenarioreflecting the situation with no active remediation but onlyannual monitoring of groundwater samples. The on-siteleaching of contaminants to groundwater is modeled usinga dynamic discrete fracture model of the fractured clay system(presented below). Natural attenuation of TCE via anaerobicreductive dechlorination is assumed to take place at a lowrate as observed at the site.

System description. Enhanced reductive dechlorination(ERD) aims to enhance microbial degradation via theanaerobic reductive dechlorination pathway, which sequen-tially degrades trichloroethene (TCE) to cis-dichloroethene(cis-DCE), vinyl chloride (VC), and finally to ethene/ethane(15). The enhancement is done via the injection of a bioculturecontaining specific degrader organisms (Dehalococcoides)together with a fermentable substrate - in this case organicsugar cane molasses from Brazil, which is a byproduct fromcane sugar production. The substrate generates hydrogenduring degradation, which acts as an electron donor in thereaction with the chlorinated ethene. The substrate is mixedwith the bioculture and injected at 56 points using direct-push Geoprobe technology. Injection is done at every 25 cmdepth in the contaminated clay till to ensure a gooddistribution in the subsurface. The total necessary amountof substrate required to react with all electron acceptors wasestimated using a safety factor of 2 (see details in the SI A.3).Safety factors are recommended in design guidelines due touncertainties in the amounts of sediment bound electronacceptors in the source zone and the microbial processes.Potential methane generation due to the anaerobic degrada-tion of the substrate was estimated at 0.1 kg methane per kgof molasses (SI A.4) and is included as an emission to air inthe inventory. All generated methane is assumed to escapeto the atmosphere via high-permeability soil features orunderground infrastructure. Monitoring activities includebiannual sampling of groundwater (source zone and regionalaquifer) during the first 2 years after each injection andannually in the remaining period. Sampling of intact soilcores is carried out after 5 years and again after termination

TABLE 1. Overview of Main Activities Included in Each Scenario

no action (NoA)enhanced reductivedechlorination (ERD)

in situ thermaldesorption (ISTD)

excavation withoff-site treatment

and disposal (EXC)

monitoring of groundwater(person transportation, watersampling and analysis)

installation of additionalmonitoring wells

installation of heater andextraction wells, heaters,collection pipes and top cover

installation of sheet pile wall

pumping and injection ofmolasses and bioculture

heating of soil excavation and backfilling

monitoring of soiland groundwater

ventilation of soiland pumping of water

soil transportation

transportation of materials,equipment and people

activated carbon treatment off-site soil treatment

monitoring of soil soil disposal

transportation of materials,equipment and people

monitoring of soil

removal of sheet pile walland asphalting

transportation of materials,equipment and people

timeframe: 1200 years timeframe: 38 years timeframe: ∼3 months timeframe: ∼1 month at site

VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9165

of the treatment in order to assess the mass removal in theclay till matrix of the source zone.

System description. In situ thermal desorption (ISTD)involves the heating of the contaminated clay till by electri-cally powered heating elements submersed in heater wells(16). This causes a fast transfer of the chlorinated ethenes tothe vapor phase, which is then extracted using vacuumextraction wells and treated on-site by adsorption to activatedcarbon. The wells consist of an outer steel pipe lined witha stainless steel pipe, into which the stainless steel heatersare submersed. The 20 wells are placed in gravel packstogether with an extraction filter, and the bottom is sealedwith concrete. The total number of heaters was estimatedassuming that each heater covers an area with a 1.5 m radius.The asphalt paving at the site is retained and is covered byan insulating layer of expanded polystyrene. An energy useof 400 kWh of electricity per m3 of soil heated was assumedfor this site as the water inflow to the treated zone is relativelylow. Additional electricity (6%) is used for ventilation andpumping. The heaters deliver a maximum of 1.2 kW per meterof heating element. With a contaminated soil depth of 5 m,the heating period is approximately 3 months. Treatmenttermination is confirmed by soil sampling at 8 sites.

System Description. Excavation with ex situ soil treat-ment and disposal (EXC) initially requires removal of thepavement from the site and installation of a sheet pile wallof steel to support the excavation. The soil is excavated downto 8 m below surface. The top 3 m of soil is assumeduncontaminated and is left at the site. After excavation,completion is confirmation by soil sampling at 8 sites, andthe pit is backfilled with clean soil. The sheet pile is removed,and the pavement reconstructed. The excavated soil istransported 100 km to a soil treatment facility, where the soilis placed in piles and turned regularly until soil samplescomply with soil standards for landfill disposal. The soil isthen transported 50 km to the final disposal site.

Primary impacts due to leaching of contaminants togroundwater in the no action scenario and the enhancedreductive dechlorination scenario were estimated using anumerical model of multicomponent contaminant transportand microbial degradation in a low-permeability fracturedmedia developed by Chambon et al. (17) (see model detailsand inputs in the SI A.3). The model simulates advectivetransport of dissolved contaminant through fractures andback-diffusion of contaminant from the clay matrix to thefracture. The transport model is combined with a Monod-type degradation model for simulation of the microbial-driven sequential dechlorination of TCE via cis-DCE and VCto ethene. The modeled fracture network and the locationof the microbial degradation zones for the two scenarios areillustrated in Figure 1.

Based on the modeled concentration at the fracture outletto the aquifer, the temporal change of mass discharge (g/year) of each chlorinated ethene to the aquifer is calculatedand shown in Figure 1. Compared to NoA, ERD causes asignificant decrease in the accumulated leached masses ofTCE and cis-DCE; however, the leached mass of VC, whichis the most toxic degradation product, is the almost equal forthe two scenarios (see Figure 1). The model-predictedtimeframes for NoA and ERD are 1200 years and 38 years,respectively, before the remedial target is met (98% massremoval in source zone). The mass discharge results are usedto estimate the exposure concentrations in downstreamextracted groundwater and resulting human toxic impacts(see the SI B). The calculation assumes that all mass iscaptured in the supply well and that no further degradationor purification takes place after the contaminants enter theaquifer. Primary impacts are disregarded for the thermalremediation scenario and the excavation scenario as theseprovide a fast (<3 months) removal and compliance with thecleanup target.

Results and DiscussionThe life cycle impacts are listed in characterized form in

the SI (Table S11), and the normalized results expressed asperson equivalents are presented in Figure 2, including bothprimary and secondary impacts. ISTD is seen to have thehighest global warming potential of the compared alterna-tives, whereas the excavation option has the largest impactscore in most other impact categories. Remediation usingERD is a low-impact alternative compared to both ISTD andexcavation. Compared to the no action scenario ERD hascomparable or lower impacts in all categories except for ozoneformation and eutrophication. It can be noted from Figure2 that the normalized toxic impacts are generally much higherthan the normalized nontoxic impacts. Reliable normaliza-tion references are difficult to construct for toxicity-relatedimpacts due to the large number of substances involved, thedata gaps for toxic releases in inventories used as a basis forthe establishment of normalization references, and the lackof characterization factors for some of the releases (18, 19).The normalization references may therefore be underesti-mated, and a better estimation could bring the toxic impactscores more on par with the rest of the environmental impactscores in Figure 2.

Consumption of nonrenewable resources (fossil energycarriers and metals) was also included in the assessmentand is reported in the SI. The results show that the resourceconsumption is highest for EXC due to the large consumptionof scarce metals.Environmental hotspots of each remediation system andimprovement options. Figure 3 shows the result of a

FIGURE 2. Normalized impacts in person equivalents (PE) for each scenario. NoA: no action, ERD: enhanced reductivedechlorination, ISTD: in situ thermal desorption, EXC: excavation with off-site treatment and disposal. Note expanded scale onright-hand figure.

9166 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010

contribution analysis performed to reveal the most importantcontributing activities for each of the analyzed systems.

No action impacts are mainly due to person transportationfor sampling but also stem from laboratory analyses andsampling. Human toxic impacts are mainly caused by theon-site leaching of TCE and degradation products to thegroundwater. It is important to note that the VC formationaccounts for 86-87% of the primary human toxic impactdue to its high carcinogenic and noncarcinogenic toxicity.

Enhanced reductive dechlorination. The main contributorto global warming and ozone formation is the formation ofmethane during anaerobic degradation of the substrate(molasses) in the subsurface. Production and transportationof bioculture and molasses, on-site pumping and injection,and production of the monitoring wells also contributenotably to impacts. In contrast to ISTD and excavation,monitoring activities are important contributors to impacts

in ERD and should not be excluded. The dominant cause ofhuman toxic impacts is the generation of VC (as for NoA),which is responsible for 97-98% of the primary toxic impact.The main potential improvement of this technology is theminimization of the amounts of bioculture and substrateadded to the subsurface. Furthermore, substitution to locallyproduced substrates and bioculture would remove thecontributions from transport, which causes up to 55% ofimpacts associated with the bioculture and up to 84% of theimpacts from molasses (in the impact categories acidificationand respiratory inorganics).

In situ thermal desorption. The impacts of the ISTDmethod are primarily due to the operational energy demandfor the electrical heating of the soil. In addition, steel andstainless steel for wells, pipes, and heaters give largecontributions, especially to the toxic impacts. In combination,electricity for heating and use of steel generate between 90

FIGURE 3. The percent contribution of different subparts of the assessed remediation systems to each impact category.

VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9167

and 99% of impacts in all categories. Improvement of thistechnology will therefore mainly be the reduction of these;however, the potential may be limited as it is already assumedthat heating elements and temperature sensors are reuseddirectly four times. A sheet pile wall around the heated areawas not necessary at this site due to a very low groundwaterflow in the upper aquifer, but it would contribute considerablyto impacts if applied; as discussed for excavation below.

Excavation with off-site treatment and disposal. Thetransportation of soil, excavation and backfilling, and thesheet pile wall are the main contributors to all nontoxicimpacts. Steel use for the sheet piles is the dominant causeof the elevated toxic impacts. The environmental impacts ofthe sheet pile wall would be even higher if it was not assumedto be reused directly on two other projects. An importantimprovement potential for this technology is the reductionof transport distances for treatment and disposal. In addition,steel use for the sheet pile wall should be minimized andreuse rates kept high. The environmental impact would begreatly increased if the steel wall is not reused or if it is leftin ground. The study site is located in an urban area withspace limitations and excavation without a sheet pile wall istherefore not possible.

Primary and secondary toxic impacts. The LCA resultsunderlined that primary toxic impacts caused by release ofchlorinated ethenes on-site during ERD and NoA areimportant and should not be disregarded. The assessmentof primary toxic impacts due to leaching of chlorinatedethenes was based on detailed numerical modeling of fateand transport including metabolite formation and site-specific exposure parameters. As USEtox does not includemetabolite formation and uses generic fate and exposuredata, the resulting primary toxic impacts using this site-dependent procedure are higher than if e.g. USEtox char-acterization factors for freshwater emissions were applieddirectly. To accord with USEtox assumptions, no removal ofcontaminants in drinking water was assumed at the watersupply. Furthermore, no further degradation of TCE and itsmetabolites was assumed during the transport to the drinkingwater supply well. Thus, the calculated primary impacts maybe seen as worst case estimates.

Despite the high primary toxic impacts related to NoAand ERD, the secondary toxic impacts of ISTD and EXC wereeven higher. The human toxic impacts from these methodsare dominated by metal emissions, particularly hexavalentchromium (CrVI) leaching from landfill sites for disposal ofsteel slags (residual from steel production) and discardedpreserved wood poles (from electricity distribution network).For instance, CrVI alone is responsible for 99% of thecarcinogenic toxicity of ISTD and EXC. As USEtox charac-terization factors for metals are reported as interim and not

final recommended values, it should be stressed that thecalculated secondary toxic impacts of ISTD and EXC canonly be taken as guidance. A recent comparison of currentlyavailable LCIA methods including USEtox (20) showed thatLCIA models show poor or no agreement in characterizationof metal toxicity. USEtox, however, was found to representa harmonization of the four LCIA methods that provided thebasis for its development.

Sensitivity scenarios. The sensitivity of the environmentalimpact scores to some of the main parameter choicesidentified above is illustrated for the three remediationsystems in Figure 4. The results shows that the soil trans-portation distance in the EXC system must be increased atleast by a factor of 3 (>450 km) before the global warmingpotential of EXC exceeds that of ISTD. At the same time,however, all other impacts largely exceed those of ISTD. Theelectricity used on-site is the main contributor to impactsfor the ISTD option, and it was assumed to be generated asthe average Danish electricity grid mix (based on 38% coal,21% natural gas, 19% renewables, 3% oil and 19% importfrom Sweden, Germany, and Norway). It could be arguedthat the increase in electricity demand due the remediationhave to be met by an increase in the Danish electricityproduction on the short-term. Therefore, the marginalprovider of electricity should be used, which for Denmarkis most likely to be a coal-based power plant (this is in LCAreferred to as a consequential approach to the modeling). Asillustrated in Figure 4, the change to marginal electricity highlyimpacts the ISTD result and leads to large increases in globalwarming potential, ozone formation, acidification, andrespiratory impacts. However, the impact scores in thecategories ozone formation, terrestrial and aquatic eutrophi-cation, ecotoxicity, and human toxicity are still lower thanfor EXC.

For the ERD scenario, the substrate demand is one of thecritical parameters for the assessment. With a 3 times increasein substrate demand (and a corresponding increase inbioculture, injection energy, and site visits), the ozoneformation of ERD exceeds that of ISTD and the globalwarming potential approaches that of EXC (see Figure 4).However, in most remaining impact categories, the impactscores of ERD are still lower than those of ISTD and EXC.

Use of LCA in evaluation of remediation alternatives. Itshould be stressed that life cycle assessments of remediationis very site-specific and that the result of this analysis cannotbe directly transferred to other sites as transport distances,electricity source, need for sheet pile wall, etc. will varybetween sites. Furthermore, differences in the definition ofthe functional unit and the setting of boundaries make itdifficult to compare results across studies. However, previousstudies also noted the important contribution to environ-

FIGURE 4. Sensitivity analysis of main parameters. The substrate demand for ERD is increased by a factor of 3, electricity generationis changed to the marginal producer (coal-based) for ISTD, and the soil transportation distances are varied for EXC with a factor of0.5, 2, and 3, respectively.

9168 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 23, 2010

mental impacts from steel (5, 21) and transport of excavatedsoil (6, 8). This study made a consistent life-cycle comparisonof primary and secondary impacts of remediation using LCA,which can be used for decision-support for remedy selectionat the site. The site-dependent assessment of primary impactsillustrated the importance of including metabolite formationas vinyl chloride contributed significantly to human toxicimpacts of the bioremediation scenario. The analyzed casestudy showed that the bioremediation scheme (ERD) isenvironmentally superior to ISTD and excavation. However,the long timeframe and larger uncertainty in the efficiencyof the ERD option is definitely a drawback compared to theother remediation options. In addition to the assessment ofenvironmental impacts a holistic remedy selection processshould also include aspects such as remediation cost andsocial impacts associated with current and future use of thesite.

AcknowledgmentsThe authors wish to acknowledge Alexis Laurent, DTUManagement Engineering, for importing the characterizationfactors for USEtox and particulate matter to SimaPro. DTUfunded the Ph.D. fellowship and financial support for researchvisit to CIRAIG was given by Otto Moensted Foundation.The Capital Region of Copenhagen provided site data for thecase study. Furthermore, we wish to acknowledge allcompanies that provided data for the life cycle inventory.

Supporting Information AvailableAdditional information about the life cycle inventory, lifecycle impact assessment method, and results. The materialis available free of charge via the Internet at http://pubs.acs.org.

Literature Cited(1) Stroo, H. F.; Unger, M.; Ward, C. H.; Kavanaugh, M. C.; Vogel,

C.; Leeson, A.; Marqusee, J. A.; Smith, B. P. Remediatingchlorinated solvent source zones. Environ. Sci. Technol. 2003,37 (11), 224A–230A.

(2) US EPA. Cleanup of the Nation’s Waste Sites: Markets andTechnology trends; EPA 542-R-04-015; U.S. EnvironmentalProtection Agency, U.S. Government Printing Office: Wash-ington, DC, 2004.

(3) Lemming, G.; Hauschild, M. Z.; Bjerg, P. L. Life cycle assessmentof soil and groundwater remediation technologies: literaturereview. Int. J. Life Cycle Assess. 2010, 15 (1), 115–127.

(4) Morais, S. A.; Delerue-Matos, C. A perspective on LCA applicationin site remediation services: Critical review of challenges. J.Hazard. Mater. 2010, 175 (1-3), 12–22.

(5) Higgins, M. R.; Olson, T. M. Life-Cycle Case Study Comparisonof Permeable Reactive Barrier versus Pump-and-Treat Reme-diation. Environ. Sci. Technol. 2009, 43 (24), 9432–9438.

(6) ScanRail Consult; HOH Water Technology A/S; NIRAS Consult-ing Engineers and Planners A/S; Revisorsamvirket/Pannell KerrForster. Environmental/Economic Evaluation and Optimisingof Contaminated Sites Remediation; EU LIFE Project no. 96ENV/DK/0016; Copenhagen, Denmark, 2000.

(7) Cadotte, M.; Deschenes, L.; Samson, R. Selection of a reme-diation scenario for a diesel-contaminated site using LCA. Int.J. Life Cycle Assess. 2007, 12 (4), 239–251.

(8) Toffoletto, L.; Deschenes, L.; Samson, R. LCA of ex-situbioremediation of diesel-contaminated soil. Int. J. Life CycleAssess. 2005, 10 (6), 406–416.

(9) Lesage, P.; Ekvall, T.; Deschenes, L.; Samson, R. Environmentalassessment of Brownfield rehabilitation using two different lifecycle inventory models - Part 2: Case study. Int. J. Life CycleAssess. 2007, 12 (7), 497–513.

(10) Lemming, G. Environmental assessment of contaminated siteremediation in a life cycle perspective. Ph.D. Thesis, TechnicalUniversity of Denmark, Department of Environmental Engi-neering, September 2010.

(11) Frischknecht, R.; Jungbluth, N.; Althaus, H.-J.; Doka, G.; Dones,R.; Heck, T.; Hellweg, S.; Hischier, R.; Nemecek, T.; Rebitzer, G.;Spielmann, M.; Wernet, G. Overview and Methodology; eco-invent report No. 1. 2007; Swiss Centre for Life Cycle Inventories,Dubendorf, 2007.

(12) Hauschild, M. Z.; Potting J. Spatial differentiation in life cycleimpact assessment - the EDIP-2003 methodology; Environ-mental news No. 80 2005; Environmental Protection Agency,Danish Ministry of the Environment, 2005.

(13) Humbert, S.; Marshall, J.; Shaked, S.; Nishioka, Y.; Preiss, P.;Spadaro, J.; Horvath, A.; Jolliet, O. Intake fractions and char-acterization factors for particulate matter: review and recom-mendations (manuscript). 2009.

(14) Rosenbaum, R. K.; Bachmann, T. M.; Gold, L. S.; Huijbregts,M. A. J.; Jolliet, O.; Juraske, R.; Koehler, A.; Larsen, H. F.; MacLeod,M.; Margni, M.; McKone, T. E.; Payet, J.; Schuhmacher, M.; vande Meent, D.; Hauschild, M. Z. USEtox-the UNEP-SETAC toxicitymodel: recommended characterisation factors for humantoxicity and freshwater ecotoxicity in life cycle impact assess-ment. Int. J. Life Cycle Assess. 2008, 13 (7), 532–546.

(15) Scheutz, C.; Durant, N.; Dennis, P.; Hansen, M. H.; Jørgensen,T.; Jakobsen, R.; Cox, E.; Bjerg, P. L. Concurrent EtheneGeneration and Growth of Dehalococcoides Containing VinylChloride Reductive Dehalogenase Genes During an EnhancedReductive Dechlorination Field Demonstration. Environ. Sci.Technol. 2008, 42 (24), 9302–9309.

(16) Heron, G.; Parker, K.; Galligan, J.; Holmes, T. C. Thermaltreatment of eight CVOC source zones to near nondetectconcentrations. Ground Water Monit. Rem. 2009, 29 (3), 56–65.

(17) Chambon, J. C.; Broholm, M. M.; Binning, P. J.; Bjerg, P. L.Modeling multi-component transport and enhanced anaerobicdechlorination processes in a single fracture-clay matrix system.J. Contam. Hydrol. 2010, 112 (1-4), 77–90.

(18) Heijungs, R.; Guinee, J.; Kleijn, R.; Rovers, V. Bias in normaliza-tion: Causes, consequences, detection and remedies. Int. J. LifeCycle Assess. 2007, 12 (4), 211–216.

(19) Sleeswijk, A. W.; van Oers, L. F. C. M.; Guinee, J. B.; Struijs, J.;Huijbregts, M. A. J. Normalisation in product life cycle assess-ment: An LCA of the global and European economic systemsin the year 2000. Sci. Total Environ. 2008, 390 (1), 227–240.

(20) Pizzol, M.; Christensen, P.; Schmidt, J.; Thomsen, M. Impactsof “metals” on human health: a comparison between ninedifferent methodologies for Life Cycle Impact Assessment. J.Cleaner Prod. 2010, doi:10.1016/j.jclepro.2010.05.007 (In press).

(21) Bayer, P.; Finkel, M. Life cycle assessment of active and passivegroundwater remediation technologies. J. Contam. Hydrol. 2006,83 (3), 171–199.

ES102007S

VOL. 44, NO. 23, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 9169