lable at ScienceDirect

Journal of Environmental Radioactivity 102 (2011) 193e199

Contents lists avai

Journal of Environmental Radioactivity

journal homepage: www.elsevier .com/locate/ jenvrad

Using radon-222 as indicator for the evaluation of the efficiency of groundwaterremediation by in situ air sparging

Michael Schubert a,*, Axel Schmidt a, Kai Müller b, Holger Weiss a

aUFZ e Helmholtz-Centre for Environmental Research, Permoserstr. 15, 04318 Leipzig, GermanybGICON GmbH, Tiergartenstr. 48, 01219 Dresden, Germany

a r t i c l e i n f o

Article history:Received 10 September 2010Received in revised form15 November 2010Accepted 17 November 2010Available online 10 December 2010

Keywords:Air spargingRadonTracerGroundwater remediationVOC

* Corresponding author. Tel.: þ49 341 235 1410; faE-mail address: michael.schubert@ufz.de (M. Schu

0265-931X/$ e see front matter � 2010 Elsevier Ltd.doi:10.1016/j.jenvrad.2010.11.012

a b s t r a c t

A common approach for remediation of groundwater contamination with volatile organic compounds(VOCs) is contaminant stripping by means of in situ air sparging (IAS). For VOC stripping, pressurized airis injected into the contaminated groundwater volume, followed by the extraction of the contaminant-loaded exhaust gas from the vadose soil zone and its immediate on-site treatment. Progress assessmentof such remediation measure necessitates information (i) on the spatial range of the IAS influence and (ii)on temporal variations of the IAS efficiency. In the present study it was shown that the naturallyoccurring noble gas radon can be used as suitable environmental tracer for achieving the related spatialand temporal information. Due to the distinct water/air partitioning behaviour of radon and due to itsstraightforward on-site detectability, the radon distribution pattern in the groundwater can be used asappropriate measure for assessing the progression of an IAS measure as a function of space and time. Thepresented paper discusses both the theoretical background of the approach and the results of an IAStreatment accomplished at a VOC contaminated site lasting six months, during which radon was appliedas efficiency indicator.

� 2010 Elsevier Ltd. All rights reserved.

1. Introduction

1.1. Motivation

A major environmental problem at sites (formerly) occupied bylarge-scale chemical industry is groundwater contamination withvolatile organic compounds (VOCs). In particular chemical plantsthat have been in operation for several decades are likely to have, atsome point in time, caused subsurface contamination due toa hidden leakage or a major single accident. An example for thelatter, which is frequently encountered in Central Europe, isextensive VOC contamination of the subsurface caused by air raidsor other combat operations executed during the SecondWorldWar.

A suitable clean-up measure for VOC contamination of thegroundwater is remediation by in situ air sparging (“IAS”). In short,a contaminant-freepressurizedgaseousmedium(air) is injected intoor below the contaminated aquifer domain (Miller, 1996; Bass et al.,2000). Within the physical range of the air sparging, i.e. within therange in which the air bubbles rise through the contaminatedgroundwater, the volatile contaminantspartition fromthewater into

x: þ49 341 235 1443.bert).

All rights reserved.

the injected air, and are thus stripped from the groundwater. Thestripped VOCs are carried into the overlying vadose soil zone, wherestrong air pumps that are installed in gas extractionwells collect theexhaust gas and pump it to on-site treatment facilities, such asactivated carbon filters or catalytic oxidation systems (Fig. 1).

When applying IAS as remediation technology, informationregarding the physical range of air sparging around the gas injectionwell and its efficiency in terms of mass transport from the water tothe gas is mandatory (Selker et al., 2007). In order to continuouslyobtain the related data during an on-going remediation measure,groundwater samples and/or samples of the exhaust gas can beanalyzed directly for VOCs (Johnston et al., 2002). However, therequired chemical (laboratory) analysis is timeconsuming, costlyandelaborate. An alternative or supplementary approach is an indirectestimation of both, IAS range and efficiency, by using artificial ornaturally occurring partitioning tracers. In previous studies artificialgas tracers such as SF6 (Johnson et al., 2001) or helium (Berkey et al.,2003) have been suggested for approaching the problem. However,laboratory analysis for SF6 or helium is not trivial either.

The study presented here discusses the application of radon-222(222Rn, hereafter referred to as radon) as naturally occurring andeasily detectable IAS efficiency indicator. At the study site, which ischaracterized by severe groundwater contamination with VOCs, insitu air sparging was applied as remediation tool. The results of the

Fig. 1. Schematic sketch of the principle of using radon as IAS efficiency indicator; notto scale.

M. Schubert et al. / Journal of Environmental Radioactivity 102 (2011) 193e199194

study prove the general applicability of radon as tracer for theevaluation of the effectiveness of forced water / gas transfer ofVOCs for remediation purposes.

1.2. Radon as aqueous tracer e general remarks

In modern hydrology radon is used as environmental tracer orindicator for various applications. Outstanding examples are thelocalization and evaluation of non-aqueous phase liquid contami-nation in soils and aquifers (Schubert et al., 2001, 2007), theinvestigation of groundwater/surface water interaction processes(Burnett and Dulaiova, 2003; Schubert et al., 2008; Schmidt et al.,2009, 2010; Schubert et al., 2010b), or the estimation of ground-water flow velocities (Cook et al., 1999; Hamada, 2000; Schubertet al., 2010a).

As for some of the mentioned applications, the suitability ofradon as indicator for IAS efficiency assessment measures arisesfrom its distinct water/air partitioning behaviour. At temperaturesthat are typical for groundwater (about 10 �C in Central Europe), thewater/air partitioning coefficient of radon amounts to approximately0.35 (Clever, 1979). That significant affinity of radon to the gaseousphase leads to its strong tendency to be stripped from the ground-water into the air that is sparged through it. The resulting decrease ofthe radon concentration in the affected groundwater volume allowsusing the local radon concentration pattern in the groundwater asindicator for the evaluation of the influential range and the efficiencyof the IAS measure. The schematic sketch shown in Fig. 1 illustratesthe principle of using radon as IAS efficiency indicator.

If compared to other partitioning tracers that are applicable forthe evaluation of the progress of an IAS measure, one outstandingadvantage of radon is its ubiquitous occurrence in groundwaterwhich makes redundant the injection of artificial compounds intothe aquifer. Radon is part of the uranium-238 decay chain andrepresents its only gaseous component. Radon is radioactive andexhibits a half-life of 3.83 days. As daughter product of the ubiqui-tously-occurring radium nuclide radium-226, radon is omnipresent

in all groundwater, appearing as dissolved gas with concentrationsranging from about 1 up to over 50 Bq L�1.

A second advantage of using radon for IAS efficiency assessment,in comparison to other potentially applicable indicators, is that themass transfer direction (water / gas) is the same as that which isrelevant for VOC removal from the water. Tracer applications thatare based on the opposite transfer direction (gas / water), as forinstance the use of oxygen brought into the aquifer with theinjected air, are likely to show transfer kinetics that are significantlydifferent (less effective) from that of the VOC mass transfer(Rutherford and Johnson, 1996).

Regarding the use of oxygen as IAS efficiency indicator, it has alsoto be considered disadvantageous that the oxygen concentration inthe contaminated (potentially anoxic) groundwater might beinfluencedbyoxygen consumingprocesses such as instantaneous Fe(II) oxidation (Stumm and Morgan, 1996), solid phase sulphideoxidation (JohnstonandDesvignes, 2003), or IAS-stimulated aerobicbiodegradation (Adams andReddy, 2003). Thus, a third advantage ofradon is its noble gas configuration leading to an inert behaviour inthe groundwater. While other potentially applicable trace compo-nents are subject to sorption/desorption, precipitation/remobiliza-tion, species interaction, or biodegradation, the mobility of radon ingroundwater is virtually independent from the chemical/mineral-ogical/biological environment present. Its concentration in thegroundwater only depends on the radon emanation rate from theaquifer matrix (constant in time), the geohydraulic transport byadvection/dispersion and diffusion (steady state in the consideredtime scales), and the efficiency of the IAS process, i.e. the variableparameter that is to be evaluated.

A fourth and final advantage of radon that shall be pointed outhere is its fast and straightforward detectability on-site even at lowconcentrations. Hence, on-site determination and evaluation of theresults allow immediate decisions concerning the remediationprocess management and a reduction of the number of relatedtime-consuming laboratory experiments.

1.3. In situ air sparging e general remarks

In order to discuss observations made during the study in anappropriate context, somegeneral aspects concerning IAS remediationmeasures are addressed in the following section. IAS-induced air flowthrough heterogeneous porous media with small effective particlesizes (i.e. a scenario comparable to theactual situationat the studysite)does preferentially occur through highly permeable zones of theaquifer, e.g. along channel-like flowpaths. In the case of amulti-phasesystem (gas e liquid) these zones are characterized by low capillaryforces (Ahlfeld et al., 1994; Geistlinger et al., 2006), i.e. they are occu-pied by groundwater that can easily be moved by hydromechanicforces (“mobile groundwater”). On the contrary, aquifer zones that arecharacterized by a significantly lower permeability are hardly pene-trated by the injected air due to higher capillary forces. The ground-water in these zones is virtually stagnant, i.e. it cannot be moved bynatural hydromechanic forces (“immobile groundwater”). Thus, thesezones are barely directly affected by the remediationmeasure (Ahlfeldet al., 1994; Leeson et al., 2002; Geistlinger et al., 2006; Pohlert et al.,2008). Whereas contaminant transfer from mobile groundwater tothe injected air is governed mainly by advection and dispersion withdiffusion only needed for short mass transfer lengths, mass transferfrom immobile groundwater into the injected air is mainly governedby diffusion over longer distances (Suthersan, 1999). Thus, because ofthe intense gas/water interaction in highly permeable aquifer zones(due to high gas saturation and large water/gas interface), volatilecompounds (such as VOCs and radon) are quickly stripped from thegroundwater. In low permeability zones however contaminantremoval from the enclosed volumes of immobile groundwater is only

Fig. 2. Typical pattern for the contaminant concentration in the exhaust gas recordedduring an IAS measure.

M. Schubert et al. / Journal of Environmental Radioactivity 102 (2011) 193e199 195

possible via aqueous diffusion from the immobile into the mobilegroundwater from where the contaminants can easily be stripped.That limiting aqueous diffusion step occurs at very lowmass transportrates and results in a general process progression typical for airsparging in heterogeneous aquifer domains as it will be described inthe following:

At first a rapid decrease in VOC concentration in the ground-water is observed. Subsequently, the concentration lessens moreslowly but still gradually until contaminant removal from themobile groundwater and diffusive contaminant replacement fromzones of immobile groundwater reaches a diffusion-controlledsteady state. The inverse concentration pattern is recorded in theexhaust gas. A concentration peak immediately after the start of theIAS measure is followed by a steady concentration decrease untila steady state is reached. Fig. 2 displays a typical example for thatbehaviour as it was recorded during the study discussed below. Theplot shows benzene concentrations in the exhaust gas detected ina soil gas extraction well located next to water sampling well SW2

Fig. 3. Setup of wells at the study site; AS ¼ air injection well, SW ¼ groundwatersampling well.

(Fig. 3). Soil gas extraction and analysis started immediately afterthe onset of the air sparging procedure.

Another factor that is influential on the IAS efficiency isa changing permeability of the aquifer matrix. On the one hand,a process that hampers the water/ air mass transfer und thus theremediation efficiency is a decrease in permeability of formerlyfreely-accessible aquifer volumes due to pore clogging by ferrous orcalcite precipitates or biomass (Baveye et al., 1998). On the otherhand, air sparging can also cause an increase in aquifer perme-ability. Whereas preferential air pathways through the poroussystem develop without changing the aquifer structure if the airinjection rate is below a certain aquifer specific threshold (Tsaiet al., 2006), pneumatic fracturing of the aquifer matrix mayoccur if the aquifer structure around the injection point is notcapable of transferring the injected air, resulting in an increase inpermeability of an aquifer section. Such pneumatic fracturingnormally occurs either immediately after the start of an air injec-tion measure or immediately after a significant increase of the airflow rate. Pneumatic fracturing is supposed to lessen thewater / gas mass transfer due to a more channelled air flow anda resulting smaller water/gas interface.

2. Site description

The study was carried out at the site of a former hydrogenationplant located in Central Germany close to the town of Zeitz, about40 km south to the city of Leipzig. The plant was founded in 1938with the main goal of gasoline and lubricant production for sup-porting the Germanwarfare during the SecondWorldWar. However,production ended abruptly in 1944/1945when the plant was heavilyattacked by air strikes, which resulted in the destruction of severalcentral production facilities. Among the damaged facilities werestorage tanks and pipelines, which led to considerable spills of liquidhydrocarbons, resulting in large-scale subsurface contamination(soil and groundwater). The estimated volume of light non-aqueousphase liquids (LNAPL) that were spilled due to damaged pipelinesand storage tanks is about 2500 m3 (Schirmer et al., 2006).

After the end of the war the plant was refurbished. However,due to the low technical standards available at the time, subsurfacecontamination continued, benzene being the main contaminant.Benzene production at the site lasted from 1963 until the finalshutdown of the plant in 1990, when all facilities were demolishedand removed. During these years contaminationwas mainly causedby leaking pipes (corrosion) paired with a lack of technical securitymeasures (Gödeke, 2004; Schirmer et al., 2006). Due to the spillhistory and the geohydraulic situation at the site large quantities ofLNAPL are now present as residual and dissolved phase within thesaturated aquifer. Because of its relatively high solubility and theamounts spilled, benzene is the main contaminant dissolved in thegroundwater with concentrations of up to 1000 mg L�1.

Of the two aquifers that can be locally distinguished the upperone is of interest within the presented study. It consists of hetero-geneous Quaternary and Tertiary sand and gravel deposits underlainby a lignite and clay unit. The current depth of the groundwater tableis approx. 8 m below ground. It shows hardly any gradient resultingin a low groundwater flow velocity of only a few mm per day.However, if groundwater flow is forced by pumping, preferentialflow paths are likely within the aquifer.

3. On-site activities

For the pilot study one exemplary air injection well (labelled inthe figures below as “AS” for “air sparging”) was chosen. In thevicinity of the AS well the saturated aquifer has a thickness of about4.5 m. The actual point of air injection was located approx. 0.5 m

M. Schubert et al. / Journal of Environmental Radioactivity 102 (2011) 193e199196

above the underlying aquitard, thus leaving a sparging influencedaquifer thickness of approx. 4 m.

The injection well was surrounded by seven 5 cm groundwatersamplingwells (SW1e SW7) located in distances between about 0.7and 3.5 m from the air injection well. The entire setup is shown inFig. 3 (the shaded sector in theupper part of the sketch illustrates theedge of a formermajor NAPL storage tank). All seven samplingwellsare fully screened within the saturated aquifer allowing ground-water sampling from the aquifer section that is reached by theinjected air.

Inhomogeneities in the mineralogical aquifer composition resultin spatially variable radium-226 concentrations in the aquifermatrixand thus in spatially variable radonbackgroundconcentrations in thegroundwater (Fig. 4). The inhomogeneities were taken into consid-eration by using well-specific background-normalized radonconcentration rather than absolute radon values for data evaluation.Thus the temporal changes of the radon concentrations detected ineach of the sampling wells, i.e. the temporal changes of the radonconcentration patterns in the groundwater that can solely be attrib-uted to the air sparing process, could be analyzed, thereby allowingan efficiency assessment of the IAS measure for each of the wells.

The IAS measure was carried out continuously for six months as“pulsed air sparging” with 10 min of air injection followed by an80-min pause. An air injection rate of 165 L/min, i.e. 1.65 m3 of airper pulse, was chosen. Assuming a (theoretically) cone shapedsparging zone with a radius of roughly 5 m (Müller, 2009), aninjection depth of 4 m, an aquifer porosity of 0.35 (representativefor the sand and gravel deposits of the aquifer) and an average air-filled pore volume in the sparging affected zone during sparging ofabout 5% (Müller, 2009), the gas-filled pore volume in the spargingzone is about 1.8 m3. That implies that each sparging pulsedisplaces roughly the complete air volume in the sparging zone.

Keeping inmind the kinetics of water/ gasmass transfer in thesparging zone, the continuous radon emanation from the aquifermatrix and the relatively long radon half-life of 3.8 days, the sixmonth IAS procedure as a whole can, in the given context, bereferred to as “continuous”, although the air injection was actuallypulsed as described above.

Water sampling and on-site radon detection were carried out inseven campaigns. The first set of data was collected immediatelybefore the in situ air sparging was put into action (“before”), the lastone was collected three months after the IAS procedure wasfinished (“after”). Both, “before” and the “after” data sets representthe uninfluenced well-specific radon background concentration ofthe groundwater in each of the wells. During the six months IASprocedure, five campaigns for on-site radon detection were carriedout (t1 e t5). In order to cover a remediation period long enough toobserve potentially occurring significant changes, approximatelyfour weeks were allowed between each two consecutive sampling

Fig. 4. Spatial radon distribution patterns around the air injection well (AS) before the

campaigns. Due to the applied radon detection equipment, fast andstraightforward radon sampling and on-site detectionwas possible.In order to guarantee compatibility of the results, all samplingcampaigns (t1 e t5) were scheduled in a way that the samples weretaken within the same short timeslot (about 20 min) between twoair injection pulses.

For the purpose of representative water sampling, all samplesduring thewholefield test periodwere taken in an identicalmanner.Approximately 100 L of groundwater were pumped from each wellat a pump rate of 7.5 L min�1 before the actual groundwater samplewas taken. It wasmade sure that the standardwater parameters pH,temperature and electrical conductivity had reached stable values inthe water pump stream before sampling. The water samples forradon measurements were collected from a depth of 2 m below thegroundwater tablewith a pump rateof 2 L/min in 6-LHDPEcanistersandanalyzed for their radon concentration immediatelyon-site. Themethodical approach for radon on-site detection is based on theprinciple of liquid-gas-membrane-extraction as described bySchmidt et al. (2008). Radon concentrations were detected bymeans of the radonmonitor RAD7 (Durridge Inc., BedfordMA/USA).The related measurement uncertainties (2s) are mainly based oncounting statistics. Considering the chosen sampling procedure andthe used measuring equipment the precision of the radon analysescan be reported as 10%.

Samples for detection of benzene concentrations were filled on-site in gas-tight GC-vials and stored until further processing at 4 �C.In the laboratory concentrations were detected by an automatedheadspace-gas chromatograph (Varian GC 3800, PALO ALTO)equipped with a CP SIL 5 CB capillary column (Varian, Germany)and a flame ionisation detector. For calibration, diluted standards ofbenzene prepared from stock solutions were treated in the sameway as the actual samples (Herrmann, 2009). Analysis of thediluted standard solutions showed an analytical error of themethod of 3%.

4. Results and discussion

All radon concentrations detected at the seven sampling wellsbefore, during and after the IAS measure are summarized in Table 1.The concentrations detected “before” show the uninfluenced well-specific radon background concentration in the groundwater. Thevalues range between 6.3 Bq L�1 (SW6) and 9.3 Bq L�1 (SW7). Thatrange of background concentrations can, as discussed above, beattributed to small-scale lithological heterogeneities (resulting ina heterogeneous radium-226 distribution), which are due to thefluvial genesis of the aquifer system. However, since relativechanges in the radon concentrations are of interest rather thanabsolute concentrations, the spatially slightly varying radonconcentrations do not complicate the final data evaluation.

IAS procedure (radon background, 4A) and during the IAS procedure (4A and B).

Table 1Radon concentrations detected in groundwater sampling wells SW1 e SW7 [Bq/l].

Before t1 t2 t3 t4 t5 After

SW1 8.83 8.62 6.63 5.88 6.84 6.59 8.37SW2 7.06 5.79 4.82 4.23 4.14 5.50 6.45SW3 8.18 5.23 3.88 6.05 4.20 5.71 9.54SW4 8.16 4.73 4.94 5.03 4.61 n.d. 8.28SW5 7.45 4.39 3.53 5.10 5.21 4.86 6.76SW6 6.28 4.27 4.35 4.61 5.09 4.80 6.69SW7 9.32 7.18 6.17 6.98 n.d. 6.36 8.18

n.d. ¼ not detected.

Fig. 5. Radon concentration versus benzene concentration in the groundwater duringthe five campaigns t1 e t5 averaged for each campaign over all seven sampling wells.

M. Schubert et al. / Journal of Environmental Radioactivity 102 (2011) 193e199 197

The set of data collected three months “after” the IAS procedurewas finished also reflects the uninfluenced well-specific radonbackground, i.e. a situation similar to that represented by the“before” data set. If the 10% uncertainty of the radon detectionmethod is taken into consideration, it can be stated that the“before” and the “after” data sets show, as expected, similar valuesfor each of the wells.

In Fig. 4A, the radon background distribution pattern at the siteis plotted based on the average values of the “before” and “after”measurements. The data reflect a radon background distributionwhich indicates a linear geologic structure running NW-SE.However, interpolation of the sparse data for creation of a radondistribution map might exaggerate the confidence in how the datareflect the subsurface situation. Thus the plot in Fig. 4A, showingabsolute concentrations, can be referred to as only approximatelyreflecting the actual situation on-site.

Compared to the background concentrations shown in Fig. 4A,the concentration patterns detected during the air spargingprocedure are significantly different (Table 1). The generallyreduced concentrations indicate that all sampled groundwaterwells are (individually to a different degree) affected by airsparging. Thus, the general appropriateness of the chosen clean-upapproach for VOC removal from the groundwater could be shownfor the site. At the same time, the data shown in Fig. 4B and Cillustrate that the relative decrease in radon concentration in theindividual wells (DRn given in % of the respective background value)and hence the effectiveness of the stripping procedure in therespective aquifer sections varies significantly in both ways,spatially between the wells as well as temporally between thesampling campaigns. Both effects are discussed in the following.

Fig. 6. Radon and benzene concentrations relative to their respective backgrounddetected in SW4 before (bf ¼ 100% background), during (t1 e t5), and after (af) the airsparging measure.

4.1. Spatial variations in sparging efficiency

As an example for the spatial variability of the effectiveness ofthe stripping procedure the DRn distribution pattern detectedduring the sampling campaign t2 is plotted (Fig. 4B). The lowestrelative changes in radon concentration were observed in thesouth-eastern part of the study area, the highest in the north andnorth-western part. Thus, the data indicate the highest efficiency ofair sparging during sampling campaign t2 around the northern andnorth-western wells. The lowest reduction in radon concentrationwas detected in well SW1, indicating only limited access of theinjected air to that part of the aquifer domain. This observationwassupported by an assessment of the microbiological activity in thewell, applying the approach of the Most Probable Number (MPN),an estimate of microbial density per unit volume of water sample.The lowest MPN of aerobic microbes for all sampling wells wasobserved in SW1 (a maximum count of 1.5 � 105 per mL comparedwith maximum counts in the other wells ranging from 2.7 � 105 to4.6 � 105 per mL), also indicating limited oxygen access to the well.

Fig. 4C illustrates the DRn distribution pattern during the entireperiod of air sparging averaged for all complete samplingcampaigns t1 e t5. The resulting picture is generally similar to the

“snap shot” shown in Fig. 4B. Both plots show a heterogeneousdistribution of the air sparging impact around the injection well“AS”. That spatial variability of the impact was, as discussed above,due to the inhomogeneous permeability distribution in the aquifer.For instance, the data recorded at well SW7 indicate a comparablypoor efficiency of the air sparging despite the fact that SW7was thesampling well closest to the injection well.

A general conclusion that can be drawn from the results dis-played in Fig. 4 is that the efficiency of the air sparging measure islimited, i.e. that the accessibility of the saturated zone for theinjected air varies spatially significantly. The heterogeneous DRndistribution pattern reveals as well that a circularly distributedsparging cone around the injection well, as it could be expectedtheoretically (Fig. 1), has not developed.

4.2. Temporal variations in sparging efficiency

Shown in Fig. 5 is the reduction of radon concentration versusthe reduction of the benzene concentration in the groundwater asa result of the remediation measure detected during the five

Fig. 7. Temporal variability of the radon concentration and thus the air sparging efficiency for all seven wells.

M. Schubert et al. / Journal of Environmental Radioactivity 102 (2011) 193e199198

campaigns t1 e t5 averaged for each campaign over all sevensampling wells. The background value (“before”) is plotted as 100%for both, radon and benzene. The positive correlation between thetwo parameters is obvious. However, since radon is constantlyproduced at a constant rate in the whole aquifer volume, itsconcentration did not as steadily decline as that of benzene, which isnot dissolved (i.e. mobilized) with a constant rate from its hetero-geneously distributed residual source. Although residual organicphase (NAPL) trapped within pockets of “immobile groundwater”acts as a long term source, the benzene dissolution rate is expectedto be limited and decreasing with time. As discussed above, in thelow permeability source zones no groundwater migration applies,but the rate of aqueous diffusion limits the rate of benzene disso-lution. The gentler the concentration gradient becomes with time,the lower becomes the diffusion rate and hence the contaminantinput into the mobile water phase. The decelerated benzene disso-lution from the scattered residual NAPL source on the one hand andthe constant and ubiquitous radon production in the aquifer on theother hand result in the non-linear relationship between the twoparameters displayed in Fig. 5 (the dashed graph is not a calculatedfit of the data points, but a line sketched in order to highlight thatcorrelation).

Fig. 6 shows the concentrations of radon and benzene measuredduring and after (“af”) the air sparging measure in relation to theirrespective background values detected before (“bf” ¼ 100% back-ground) the sparging of sampling well SW4 (the t5 radon samplecould not be analyzed due to a technical obstacle on-site). The SW4data were chosen for display because they depict the typicalbehaviour of radon and benzene as result of air sparging underideal conditions. On the one hand the radon concentration quicklydrops to a value of about 60% of the background at which it comesto a steady state equilibrium held in balance by radon stripping(sink term) and radon production (source term). On the other hand,after a quick drop, a gradual and continuous decrease in benzeneconcentration was observed, which is due to the fast removal ofbenzene from the air sparged aquifer zone in the beginning of theoperation followed by a continuous, diffusion-controlled removalof benzene from the surrounding low-permeable zones that are notdirectly affected by the IAS measure. Both, the steadiness of theradon equilibrium concentration and the gradually decreasingbenzene concentration indicate a good efficiency and a temporalconsistency of the air sparging procedure at sampling well SW4.

In Fig. 7 the temporal variability of the air sparging efficiency isillustrated for all seven wells. As mentioned above, radon concen-trations relative to their respective background values are shown in

order to enable direct comparability between the data setscollected from the individual wells.

Again, the data show that the air sparging process was mostconstant over time at sampling well SW4. In the neighbouring wellsSW3 and SW5, the average radon concentrations decreased toapproximately the same level, dropping to 58% and 64%, respectively.However, the efficiency of the air sparging observed in these twowells is not as consistent with time as in SW4. The concentrationsmeasured varied with time (evenwith regard to the 10% uncertaintyof the radon detection method). That was in particular the case forSW3 where the radon concentration oscillated around the meanvalue ranging between 3.9 and 5.7 Bq L�1. In SW5 a temporal trendcan only vaguely be seen: the radon concentrationswere low to start,with the first two values at 62 and 50% of the background, respec-tively. However, subsequently the influence of the stripping seemedto lessen, leading to significantly higher radon concentrations. Acomparable trendwasevenmoreobvious insamplingwell SW6.Herethe initial radon concentration during sparging was 65% of thebackground with the stripping efficiency lessening afterwards,resulting in a gradual rise of the radon concentration. As mentionedabove, a reason for a decreasing efficiency can be a sparging inducedchange of porosity within the aquifer, which is likely to alter airaccessibility and flow paths through the aquifer.

Sampling wells SW1 and SW2 were located south of the airinjection well, i.e. opposite to the wells SW3 e SW6 discussedabove. The radon concentrations detected in the two wells showedtemporal patterns, with the lowest concentrations in the middle ofthe remediation measure (Fig. 7). In SW1 and SW2, the stripping ofradon (and hence of benzene) was somewhat delayed and did notreach the intensity observed in wells SW3 to SW6, which wasattributed to the lower permeability and to the resulting reducedaccessibility of that part of the aquifer to the injected air (Fig. 4Band C). The data show clearly an alteration of the impact of airsparging with time, which was probably due to temporary block-ages of the injected air flow resulting from clogging of migrationpathways in the aquifer. As shown, such obstruction can be a shortterm incidence (e.g. SW3) or steadily increasing (e.g. SW6).

5. Conclusions

If volatile organic compounds are to be removed from contami-nated groundwater, in situ air sparging is a common remediationapproach. For assessing andpredicting the success of such the clean-up effort, the efficiency and the physical range of influence of the airsparging around the air injection well have to be known. In the

M. Schubert et al. / Journal of Environmental Radioactivity 102 (2011) 193e199 199

present study, it was shown that the concentration of the naturallyoccurring radioactivenoble gas radon ingroundwater canbeappliedas a useful and easily obtainable indicator for attaining relatedinformation. Comparable to dissolved VOCs, radon is removed fromthegroundwater if air is sparged through it. Thus, the decrease of theradon concentration in the groundwater compared to its undis-turbed background level can be used as at least a semi-quantitativemeasure for the efficiency of VOC removal from the groundwater byair sparging. In the present pilot study, it was shown that the effi-ciencyof air sparging around anair injectionwellmayshowapatchypattern and may also vary with time. Both lateral and temporalvariability are due to a spatially variable aquifer permeability and/ordue to a temporally clogging of pore space thereby blocking waterand air migration pathways and slowing down aqueous diffusion.

Acknowledgements

The authors want to thank Carsten Vogt and Steffi Herrmann forbenzene analysis and providing the MPNs as well as Nadine Zim-mer and Katja Klemm for technical assistance in the field and thelaboratory. Furthermore we acknowledge the support by ZSGZeitzer Standortgesellschaft mbH throughout the project. Parts ofthe work was done within the project BEOQUE (“Evaluation andoptimization of Source Zone Remediation Approaches”) funded bythe German Federal Ministry of Education and Research (supportcode 02WN0651).

References

Adams, J.A., Reddy,K.R., 2003. Extentofbenzenebiodegradation in saturatedsoil columnduring air sparging. GroundWater Monitoring and Remediation 23 (3), 85e94.

Ahlfeld, D.V., Dahmani, A., Ji, W., 1994. A conceptual model of field behavior of airsparging and its implications for application. Ground Water Monitoring andRemediation 14 (4), 132e139.

Bass, D.H., Hastings, N.A., Brown, R.A., 2000. Performance of air sparging systems:a review of case studies. Journal of Hazardous Materials 72, 101e119.

Baveye, P., Vandevivere, P., Hoyle, B.L., DeLeo, P.C., de Lozada, D.S., 1998. Environ-mental impact and mechanisms of the biological clogging of saturated soils andaquifer materials. Critical Reviews in Environmental Science and Technology289 (2), 123e191.

Berkey, J.S., Lachmar, T.E., Doucette, W.J., Dupont, R.R., 2003. Tracer studies forevaluation of in situ air sparging and in-well aeration system performance ata gasoline-contaminated site. Journal of Hazardous Materials 98, 127e144.

Burnett, W.C., Dulaiova, H., 2003. Estimating the dynamics of groundwater inputinto the coastal zone via continuous radon-222 measurements. Journal ofEnvironmental Radioactivity 69, 21e35.

Clever, H.L. (Ed.), 1979. Krypton, Xenon and Radon e Gas Solubilities. IUPAC.Solubility Data Series, vol. 2. Pergamon Press, Oxford/UK.

Cook, P.G., Love, A.J., Dighton, J.C., 1999. Inferring ground water flow in fracturedrock from dissolved radon. Ground Water 37 (4), 606e610.

Gödeke, S., 2004. Evaluierung und Modellierung des Natural Attenuation Potentialsam Industriestandort Zeitz. PhD-Thesis, Eberhard-Karls-Universität Tübingen,Tübingen/Germany.

Geistlinger, H., Krauss, G., Lazik, D., Luckner, L., 2006. Direct gas injection intosaturated glass beads: transition from incoherent to coherent gas flow pattern.Water Resources Research 42. doi:10.1029/2005WR004451.

Hamada, H., 2000. Estimation of groundwater flow rate using the decay of 222Rn ina well. Journal of Environmental Radioactivity 47, 1e13.

Herrmann, S., 2009. Molecular biological, physiological and isotope chemicalanalyses of BTEX (bio)degradation processes within a contaminated aquifer.PhD-Thesis, TU Bergakademie Freiberg, Freiberg/Germany.

Johnson, R.L., Johnson, P.C., Amerson, I.L., Johnson, T.L., Bruce, C.L., Leeson, A.,Vogel, C.M., 2001. Diagnostic tools for integrated in situ air sparging pilot tests.Bioremediation Journal 5 (4), 238e298.

Johnston, C.D., Desvignes, A., 2003. Evidence for biodegradation and volatilisationof dissolved petroleum hydrocarbons during in situ air sparging in large labo-ratory columns. Water, Air, and Soil Pollution 3/3, 25e33.

Johnston, C.D., Rayner, J.L., Briegel, D., 2002. Effectiveness of in situ air sparging forremoving NAPL gasoline from a sandy aquifer near Perth, Western Australia.Journal of Contaminant Hydrology 59, 87e111.

Leeson, A., Johnson, P.C., Johnson, R.L., Vogel, C.M., Hinchee, R.E.,Marley,M., Peargin, T.,Bruce, C.L., Amerson, I.L., Coonfare, C.T., Gillespie, R.D., McWhorter, D.B., 2002. AirSparging Design Paradigm Battelle Memorial Institute Project Report, Columbus/OH, USA.

Müller, K., 2009. Bewertung und Optimierung von Quellensanierungsansätzen imHinblick auf NA und ENA in der Abstromfahne des Standortes HydrierwerkZeitz e Stoffbilanzierung des Air Sparging-Versuchs Final Report BMBF-Project"BEOQUE".

Miller, R.R., 1996. Air Sparging. Technology Overview Report GWRTAG O-Series TO-96-04. Ground-Water Remediation Technologies Analysis Center, Pittsburgh/PA,USA.

Pohlert, M., Martienssen, M., Geistlinger, H., Weiß, H. Trabitzsch, R., 2008. Gas-tracer-Versuche zur Abschätzung der Raumwirkung bei Direktgasinjektionenzur Grundwassersanierung. Proceedings Berg- und Hüttenmännischer TagTUBA Freiberg, TU Bergakademie Freiberg, Freiberg/Germany.

Rutherford, K.W., Johnson, P.C., 1996. Effects of process control changes on aquiferoxygenation rates during in situ air sparging in homogeneous aquifers. GroundWater Monitoring and Remediation Fall 1996, 132e141.

Schirmer, M., Dahmke, A., Dietrich, P., Dietze, M., Gödeke, S., Richnow, H.H.,Schirmer, K., Weiß, H., Teutsch, G., 2006. Natural attenuation research at thecontaminated megasite Zeitz. Journal of Hydrology 328, 393e407.

Schmidt, A., Schlüter, M., Melles, M., Schubert, M., 2008. Continuous and discreteon-site detection of radon-222 in ground- and surface waters by means of anextraction module. Applied Radiation and Isotopes 66, 1939e1944.

Schmidt, A., Stringer, C.E., Haferkorn, U., Schubert, M., 2009. Quantification ofgroundwater discharge into lakes using radon-222 as naturally occurring tracer.Environmental Geology 56, 855e863.

Schmidt, A., Gibson, J.J., Santos, I.R., Schubert, M., Tattrie, K., Weiß, H., 2010. Therelevance of groundwater discharge to the overall water budget of two typicalBoreal lakes in Alberta/Canada estimated from a radonmass balance. Hydrologyand Earth System Sciences 14, 79e89.

Schubert, M., Freyer, K., Treutler, H.C., Weiß, H., 2001. Using soil gas radon as anindicator for ground contamination by non-aqueous phase-liquids. Journal ofSoils and Sediments 1, 217e222.

Schubert, M., Paschke, A., Lau, S., Geyer, W., Knöller, K., 2007. Radon as a naturallyoccurring tracer for the assessment of residual NAPL contamination of aquifers.Environmental Pollution 145 (3), 920e927.

Schubert, M., Schmidt, A., Lopez, A., Balcázar, M., Paschke, A., 2008. In situ deter-mination of radon in surface water bodies by means of a hydrophobicmembrane tubing. Radiation Measurements 43, 111e120.

Schubert M., Brüggemann L., Schirmer M., Knoeller K, 2010a. Radon monitoring inwells as tool for the estimation of the groundwater flow rate. Water RecoursesResearch, submitted for publication.

Schubert M., Knoeller K., Einsiedl F., 2010b. Using radon for the investigation ofsubmarine groundwater discharge into Kinvarra Bay, Ireland e a case study.Journal of Hydrology, submitted for publication.

Selker, J.S., Niemet, M., McDuffie, N.G., Gorelick, S.M., Parlange, J.Y., 2007. The localgeometry of gas injection into saturated homogeneous porous media. Transportin Porous Media 68, 107e127.

Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry: Chemical Equilibria and Rates inNatural Waters, third ed. John Wiley& Son, New York.

Suthersan, S.S., 1999. In: Situ Air Sparging - Remediation Engineering: DesignConcepts. CRC Press, Boca Raton/FL, USA.

Tsai, Y.J., Kuo, Y.C., Chen, T.C., Chou, F.C., 2006. Estimating the change of porosity inthe saturated zone during air sparging. Journal of Environmental Sciences eChina 18, 675e679.