este

r

ent

for7 S

Abstract

1. Introduction

colloids from drinking water supplies (e.g., [42,81,114]).Microorganisms can also travel attached to abiotic particles

(e.g., [83,24,55,61]). In addition, certainmicroorganisms can

of the uid velocity eld and the tortuosity of the pathsthrough the porous media. Dispersion can be more impor-tant for colloids than for solutes, since it can lead to earlierbreakthrough of the colloids, as presented below. In addi-tion, random interactions among molecules and/or parti-cles result in Brownian movements [117] that diuse the

* Corresponding author.E-mail address: arturo.keller@gmail.com (A.A. Keller).

Advances in Water Resources 30The transport of biocolloids (e.g., viruses, bacteria, sporesand other microorganisms) through saturated and unsatu-rated porous media is of signicant interest, from the per-spective of protection of groundwater supplies fromcontamination (e.g., [87,53,128,85,104,99]), assessment ofrisk from pathogens in groundwater (e.g., [9,40,49,51,54,111,143,1,28,124,14,88,108,95,130]), natural and enhancedbioremediation (e.g., [106,96,141,3,38,33,37,2,68]) and forthe design of better water treatment systems to remove bio-

also facilitate the transport of metals and other chemicals(e.g., [72,35,132,145]). Thus, it is important to understandthe transport of colloids in general, and that of biocolloidsin specic.

Biocolloids are aected by many of the physical andchemical processes that inuence solute transport, i.e.,advection, diusion, dispersion and adsorption (Fig. 1).Advection is the motion of the biocolloids along the trajec-tories of the uid streamlines. This mechanism can createdispersion of the biocolloids because of the heterogeneityField and column studies of biocolloid transport in porous media have yielded a large body of information, used to design treatmentsystems, protect water supplies and assess the risk of pathogen contamination. However, the inherent black-box approach of theselarger scales has resulted in generalizations that sometimes prove inaccurate. Over the past 1015 years, pore scale visualization tech-niques have improved substantially, allowing the study of biocolloid transport in saturated and unsaturated porous media at a level thatprovides a very clear understanding of the processes that govern biocolloid movement. For example, it is now understood that the reduc-tion in pathways for biocolloids as a function of their size leads to earlier breakthrough. Interception of biocolloids by the porous mediaused to be considered independent of uid ow velocity, but recent work indicates that there is a relationship between them. The exis-tence of almost stagnant pore water regions within a porous medium can lead to storage of biocolloids, but this process is strongly col-loid-size dependent, since larger biocolloids are focused along the central streamlines in the owing uid. Interfaces, such as the airwaterinterface, the soilwater interface and the soilwaterair interface, play a major role in attachment and detachment, with signicantimplications for risk assessment and system design. Important research questions related to the pore-scale factors that control attach-ment and detachment are key to furthering our understanding of the transport of biocolloids in porous media. 2006 Elsevier Ltd. All rights reserved.

Keywords: Colloid; Sorption; Desorption; Filtration; Straining; GroundwaterA review of visualization techniquat the pore scale under satura

Arturo A. Kelle

Bren School of Environmental Science and Managem

Received 15 April 2005; received in revisedAvailable online0309-1708/$ - see front matter 2006 Elsevier Ltd. All rights reserved.doi:10.1016/j.advwatres.2006.05.013of biocolloid transport processesd and unsaturated conditions

*, Maria Auset

, University of California, Santa Barbara, CA, USA

m 8 November 2005; accepted 5 May 2006eptember 2006

www.elsevier.com/locate/advwatres

(2007) 13921407

Deposition

WaNomenclature

A Hamaker constant (J)ap particle radius (m)As soil porosity function ()AWI airwater interface ()D1 colloid bulk diusion coecient (m

2/s)dg eective grain diameter (m)dp particle eective diameter (m)g gravitational acceleration (m/s2)kB Boltzmanns constant (J/K)NA Hamaker number ()NG gravity number ()NPe Peclet number ()

Grain

A.A. Keller, M. Auset / Advances inbiocolloids. The biocolloids can attach to the soilwaterinterface (SWI), the airwater interface (AWI), or the triplecontact of soilwaterair (SWA). Attachment/adsorptionto these interfaces can be reversible or essentially irrevers-ible under certain conditions, and is perhaps the most com-plex process, given the large number of colloid and grainsurface characteristics that determine the probability ofattachment, and the inuence of the dissolved chemicalspecies in the aqueous solution on attachment and detach-ment. In addition to those four processes, colloids are sub-ject to removal by physical mechanisms, such as straining,interception, diusion to the wall and gravitational deposi-tion. These physical processes are precursors to attachment(Fig. 2).

At the level of the individual biocolloid, there are pro-cesses that can result in the formation of clusters of biocol-loids, either attached to an interface or mobile within theaqueous phase. Clusters can also be initiated via biologicalprocesses, to form biolms (e.g., [23,76,129,21,94,126,7]).Individual or clustered biocolloids can break o from thelm, releasing them into the owing aqueous medium(e.g., [18,140,127]).

Biological processes such as growth, death, predation,parasitism and other processes can result in the increase

Advectionalongcentral

streamline

Diffusion into

stagnant region

Straining

Adsorption

Flow direction

Fig. 1. Schematic of pore scale processes under saturated ow.NR Reynolds number ()NvdW van der Waals number ()SWA soilwaterair interface ()SWI soilwater interface ()T temperature (K)T/C pore throat to colloid diameter (m/m)U pore water velocity (m/s)g collision eciency ()hm soil matrix porosity ()lw viscosity of aqueous solution (kg/m s)qp particle density (kg/m

3)qw density of aqueous solution (kg/m

3)

Flow direction(vertical)

Gravitational

ter Resources 30 (2007) 13921407 1393or removal of mobile or attached microorganisms (e.g.,[30,36,48,91]). Many of these biological processes are alsoinuenced by physical and chemicals conditions, and thechanges in these conditions. Although these processes areextremely important, they are outside the scope of thismanuscript, which will focus on the transport of biocol-loids through porous media.

Conventional methods to investigate biocolloid trans-port through saturated and unsaturated porous mediaoften include column and eld studies (e.g.,[58,67,86,30,42,53,114,130]). These experiments are gener-ally limited to the evaluation of euent breakthroughcurves and destructive sampling at the end of the experi-mentation that represent some average behavior of biocol-loids. Some studies focus on the collection ofbiogeochemical parameters that can monitor the biologicalprocess. Unfortunately, direct observations of the internalprocesses occurring are not possible, and mechanisms thatcontrol biocolloid transport are therefore poorly under-stood. A useful method to investigate pore scale processesimplicates the use of micromodels.

In recent years, micromodels have been increasinglyemployed to study the fate and transport of colloids and

Diffusionoff central

streamlines

Interception

Fig. 2. Schematic of processes that lead to attachment.

Table 1Micromodel and ow cell studies of biocolloid transport

Conditions Reference Material Pattern Dimensions Key ndings

Saturatedporous medium

[75] Etched glass Homogeneous periodic network Pore depth 80 lm, pore width360 lm

Dispersion of E. coli anddetermination of dispersioncoecient

[16] PDMS on glass Homogeneous network ofsquares

2 2 lm square arrays spaced1 lm apart

Particle deposition (adsorption)inheterogeneously charged surfaces

[8] Etched silicon Homogenous network of circles,300 lm diameter

Pore depth 50 lm, pore space173 lm, pore throat 35 lm

Transport along streamlines andattachment

[121,122] Etched silicon Realistic sand pore network Pore depth 15 lm, porediameters: 2.430 lm, porethroat 110 lm

Pathway a function of colloid size,higher dispersion for small colloids

[4] PDMS Homogeneous network ofsquares

Pore depth 12 lm, pore throats10 and 20 lm

Inuence of colloidal size on colloidaldispersion

Saturated porousmedium withbiolm

[123] Poly(methyl methacrylate)(PMMA) and glass Parallel plate ow cell 5.5 3.8 0.06 cm (l w h)Adhesionof

biocolloids tosolid substrata

[31] Etched silicon Network of squares; simulationof a ne homogeneous sand;porosity 37%

Pore depth 200 lm, meanchannel width 75 lm grain sizes(0.5 mm), pore sizes (50200 lm)

Rerouting of ow due to biomassgrowth

[32] Etched silicon Network of squares, channelwidth randomly distributed

Pore depth 200 lm, channelwidth 75 and 123 lm

Conductivity decreases correlatedwith biolm growth Microorganisms strongly attachingto surfaces and to each other are themost eective at reducingpermeability Continuous, rather than periodic,disinfection is recommended

[73] Etched glass Homogeneous triangular lattice Pore bodies 300 lm, pore throats30100 lm

Biomass accumulation causespermeability reduction Existence of a critical shear stress

[125] Etched glass Homogeneous triangular lattice Pore bodies 300 lm, pore throats30130 lm

Exopolymer production by bacterialeads to biomass plug and pressuredrop increase

[90] Etched silicon Homogenous network of circles 1 cm 1 cm packed array of300 lm diameter silicon postsseparated by 35 lm pore throats15 lm deep

Biomass growth changes water owpaths

[78] Steel and glass Flow cells packed with quartzsand

8 3 54 mm Colloidbiolm interactions haveimplications for colloid transport andremobilizationSolution low ionic strength (I)remobilizes attached bacterialbiomassBiomass and clay colloidsremobilized by deplecting I orincreasing ow rate

1394A.A.Keller,

M.Auset

/Advances

inWater

Reso

urces

30(2007)13921407

Conditions

Reference

Material

Pattern

Dimensions

Key

ndings

[11]

Glass

Sandstonerock

Pore

depth

34.9lm

Biolm

developmentand

accumulationin

leadingfacesof

obstructions

Un-saturated

porousmedium

[138,139]

Etched

glass

Hexagonal,quadrilateraland

heterogeneousnetworks

Pore

sizesof20400

lm1250

pore

bodieseach

AWIisan

additionalsorbentphase

forcolloids

[46]

poly(m

ethyl

methacrylate)

(PMMA)

Parallelplateowcham

ber

7650.6mm

(lwh)

AWIdetaches

particles

from

collectorsurface

[121,122]

Etched

silicon

Realisticsandpore

network

Pore

depth

15lm,pore

diameters:2.430

lm,pore

throat

110

lm

Colloidsattached

toAWIform

acluster

withthedissolutionofair

bubble

[25]

Sem

i-translucentsilica

sand

Inltrationcham

ber

264.80.5cm

0.430.60mm

graindiameter

Colloidaltrappingat

SWAInterface

[5]

PDMS

Realisticsandpore

network

Pore

diameters:3060lm

Rem

obilizationofbiocolloidsby

interm

ittentunsaturatedow

A.A. Keller, M. Auset / Advances in Waspecically biocolloids at the pore scale (Table 1). Micro-models are transparent physical models of porous media,with a pore size in the range of 10100 lm, etched in glass(e.g., [138,75]), silicon wafers (e.g., [69,121,122,19,8]), orpolymer substrates (e.g., [4]) like the ones presented inFig. 3. Some recent studies have used silica particles asthe porous media in three dimensions, visualizing the topsurface (e.g., [25]). In addition, ow cells have also beenused to study physical processes such as attachment,detachment and mass transfer rates (e.g., [59,123,80,82]).Recent work by Sherwood et al., Olson et al., [118,97] usingmagnetic resonance imaging has also served to betterunderstand biocolloid transport at small scales. The mainpurpose of these microscale experiments has been to visual-ize biocolloid transport processes at the dimensions of apore or collection of pores, validating or negating hypoth-esis that have been put forward with regards to processesthat had not been actually observed; a secondary objectivehas been to quantify the importance of these processes.Although the use of micromodels has increased, there arestill many questions that need to be answered with regardsto attachment and detachment from interfaces, and the roleof physical, chemical and biological heterogeneity in suchprocesses.

In this paper, we review the most recent ndings on bio-colloid migration and immobilization at the pore scaleusing micromodels. The experimental details can be foundin the original papers, so only the most relevant conditionsare discussed in this manuscript. We begin by examiningthe processes that aect biocolloid advection and disper-sion under saturated conditions. We then explore the roleof interfaces on biocolloid retention in saturated and unsat-urated porous media. We conclude with recommendationsfor future research.

2. Biocolloid transport processes in saturated porous media

2.1. Advection, diusion, dispersion

Imposing a pressure gradient across a porous mediumrapidly generates a stable ow eld with dened stream-lines. Even fairly signicant changes in pressure gradienthave minimal inuence on the streamlines that dene thepathways within the medium, although these changes cer-tainly aect the rate of transport. Colloids and solutesundergo advective transport moving with the pore-water,whose velocity is governed by the hydraulic pressure gradi-ent, porosity, and permeability distribution [44]. Solutionof the NavierStokes equations at the pore scale (e.g.,[121]) indicates that even for fairly complex geometries,the local velocity prole is nearly parabolic [26,8], withthe faster streamlines in the center of the pore throats,and slower streamlines along the solidwater interface(Fig. 4). In the complex geometries of natural porousmedia, there are many regions which are almost stagnant

ter Resources 30 (2007) 13921407 1395(darker blue regions in Fig. 4), while only a few pathwaysexhibit signicant ow (lighter blue to yellow to red regions

Water Resources 30 (2007) 139214071396 A.A. Keller, M. Auset / Advances inin Fig. 4). The pressure gradient is from right to left in thissimulation that solves the NavierStokes equations for arealistic pore space. Thus, solutes or colloids that begintheir transport near the central streamlines are advectedat a considerably higher rate than those along the SWI.

Fig. 3. Scanning electron micrographs of PDMS and silicon wafer micromodels. Typical pore size 10100 lm, pore throats 320 lm.

Fig. 4. Solution of NavierStokes equation for a complex pore spacegeometry using FEMLAB. Flow is from right to left, and in the laminarregime.Since diusion due to Brownian motion is inversely pro-portional to the mass of the molecule or particle, soluteshave a much higher probability of transferring amongstreamlines than colloids. Even the smallest colloidsobserved to date (MS-2 viruses, about 50 nm in diameter)exhibit very low transfer among streamlines within thelength of a typical micromodel (a few mm). At largerscales, with increasing transport time, transfer amongstreamlines will eventually occur, slowing down some ofthe faster colloids and speeding up some of the slower ones.However, low diusion tends to focus mobile colloidsalong the certain streamlines; slower colloids near theSWI have a higher probability of depositing onto theSWI by a number of processes. Larger colloids are forcedto remain near the central streamlines, while the smallercolloids can sample a wider range of streamlines (Fig. 5).The schematic shows two sizes of colloids (2 and 7 lm indiameter) in two dierent pore throats (10 and 20 lm indiameter) and the range of streamlines they can travelthrough as indicated by the black rectangles. For a smallerpore throat to colloid diameter ratio (T/C ratio), the col-loid is severely constrained to the central streamlines.

Under controlled conditions, Auset and Keller [4]showed that some colloids follow these streamlines evenFig. 5. Schematic of possible distribution of small (2 lm) and largecolloids (7 lm) within pore throats of dierent diameters (10 and 20 lm).

Waalong sharp turns into perpendicular pore throats at theend of pore bodies. Smaller colloids can easily follow alongthe pore walls, making many detours along their path,while the larger colloids tend to stay on the central stream-lines and in general have fewer detours (Fig. 6). For thesame size of colloids, travel through narrower pore throatsresults in shorter average residence time and a narrower

Fig. 6. Schematic of dierential colloid transport along streamlinescalculated using the solution to the NavierStokes equation for a simplepore geometry.

A.A. Keller, M. Auset / Advances indistribution of residence times, relative to a wider pore net-work (Fig. 7a,b). Travel through a more complex network,closer to real pore spaces, results in longer average resi-dence time and a broader distribution than those of simplepore networks (Fig. 7c). Colloid residence time is also afunction of the pressure gradient (Fig. 8); large gradientsresult in wider dierences in residence time between col-loids of dierent sizes, while small gradients tend to reducethe dierences. Torquato [131] also discusses the eect ofheterogeneity on colloid dispersion.

For complex pore geometries such as that shown inFig. 4, the dierence in colloid size has increasing impor-tance. Smaller colloids sample many of the pathways avail-able to them, traveling though both narrow and wide porethroats, and are thus more likely to move into regionswhere ow is almost stagnant (Fig. 4). Larger colloidsare excluded from many regions and pathways, in partbecause they remain in the central streamlines, as shownby Sirivithayapakorn and Keller [121]. This dierentialbehavior can have a signicant eect on the average resi-dence time of dierent colloid sizes, since the larger colloidscan travel at signicant faster velocities through the porousmedium compared to the smaller colloids. At the porescale, this phenomenon can result in colloid velocities thatare 1.53 times greater than the average water velocity(Fig. 9). This eect has been designated as a velocityter Resources 30 (2007) 13921407 1397enhancement (e.g., [52,4,70]). Mathematically, it has beenproposed that this could be handled as a retardation factorless than unity or a lower eective porosity [43]. Due to col-loid removal processes the magnitude of this eectdecreases with travel distance, as shown by Keller et al.[70], but can nevertheless result in earlier breakthrough ofcolloids moving through a porous medium, as seen in lar-ger scale studies (Table 2).

An important result from these studies is that dispersiv-ity, which is generally considered an intrinsic property ofthe porous medium [10], is a function of colloid size [4];it may be more appropriate to denominate it apparent dis-persivity when discussing colloid transport. The eect had

Fig. 7. Experimentally measured residence time distributions for 2 lmcolloids in dierent pore geometries, with a pressure gradient of 500 Paacross the micromodel (visualization method as presented in [4]).

Wa1398 A.A. Keller, M. Auset / Advances inbeen observed at larger scales. For example, Shonnardet al.,Pang et al. [119,98], analyzing earlier breakthroughof microbes relative to a tracer, assigned a lower dispersiv-ity for microbes than for solutes. They noted dierences indispersion that led to faster breakthrough, although theywere unable to pinpoint the mechanism that caused thesedierences. Sinton et al. [120] reported reductions in thedispersivity when modeling migration of dierent sized

Fig. 8. Comparison between geometries at dierent pressure gradients. Mean rmodel (circles), 20 lm-channel model (squares), zig-zag model (triangle). (a) 1

Fig. 9. Ratio of ensemble mean velocity and the straight path meanvelocity for four colloid sizes, at the highest-pressure gradient (1500 Pa) ineach pore geometry.esidence time as a function of colloidal diameter. Ten micrometer-channel500 Pa; (b) 1000 Pa; (c) 500 Pa and (d) 100 Pa.

ter Resources 30 (2007) 13921407microorganisms in an alluvial gravel aquifer. Schulze-Mak-uch et al. [116] also found variable longitudinal dispersivi-ties between bromide and MS2 virus in a model aquifer andshowed that vertical dispersion of MS2 is actually less thanthat of bromide. The micromodel studies have provided thevisual explanation for these macroscale observations.

2.2. Exclusion

A number of colloid exclusion processes have been dis-cussed in the literature (e.g., [43,44,13,12]). The most evi-dent exclusion process occurs when the colloid diameteris larger or equal to the pore throat to be entered, resultingin either exclusion (the colloid does not enter the downgra-dient pore space) or straining, with attachment of the col-loid to the SWI. A more subtle exclusion process wasobserved in the micromodel experiments conducted by Sir-ivithayapakorn and Keller [121,122] which revealed thatthe pore T/C ratio threshold for entering a pore throatwas about 1.5, due to the hydrodynamics of the system.Since colloids are focused towards the central streamlines,they rarely enter small pore throats. In these studies, morethan 100 colloids were tracked through various pores, andthe T/C threshold seemed to hold for various sizes andtypes of colloids, including viruses [121,122]. The pressuregradient was seen not to have a signicant eect on theT/C threshold. Although the exact T/C ratio thresholdwas not determined, one can use this value to consider thatbiocolloids larger than about 15 lm can be excluded from

Table 2Column and eld studies of colloid velocity enhancement

Reference Travel distance Medium, particle size Colloids, size (lm) Velocity enhancement Velocity (m/d)

Colloid transport through laboratory columns

[153] 110 cm Particles, 18, 40, 58 lm Microspheres,1, 2, 3, 5, 7, 10 lm 1.031.09 3.27

[150] 30 cm Quartz powder, 30 lm Microspheres 0.04 lm 1.06 0.1440.17 lm 1.110.31 lm 1.13

[147] 60 cm Column sediments, 0.51 mm 0.2 lm 1.9 1.40.7 lm 1.71.3 lm 1.6

[50] 46 cm Soil aggregates 12 mm Microspheres 0.11 lm 1.4 10

[52] 10 cm Coarse sand, 1.42.4 mm Medium sand, 0.40.5 mm Cryptosporidium parvum oocysts, 4.55.5 lm 11.38 0.7Fine sand, 0.180.25 mm 7

[29] 40 cm 50 cm Sand sediments Comamonas sp., 0.6 1.1 lm 1.11.551.8 0.5

[149] 120 cm Crushed int gravel,1.53 mm Aeolian quartz silt, 260 lm 0.751.08 10.4432

[116] 109 cm Sieved play sand Phage MS2 = 0.024 lm 0.88 (pH 6.1) 2301.03 (pH 7.5)1.14 (pH 8.1)

[70] 60 cm Medium sand Microspheres, 3 and 0.05 1.051.09 1.4Phage MS2 = 0.025 lm 1.111.14 14

Field studies of colloid transport

[152] Aquifer Escherichia coli 1.161.2

[148] 0.57 m Sand aquifer 1.001.3 0.431.081.62 m Fulvic acid, 1 nm 1.12.3 0.360.624

Polystyrene sulphonate, 20 nm 1.041.11 0.61.31.01.4 0.431.08

[146] 6.9 m downgradient Sandy aquifer, 0.5 mm Carboxylated microspheres 0.23 lm 1.4 0.330.53 lm 1.40.91 lm 1.41.35 lm 1.1

[151] 385 m Alluvial gravel aquifer Fecal coliforms 1.29 160F-RNA coliphages 1.88Escherichia coli,J6-2 1.05Phage MS2 1.25

[98] 61.63 m Alluvial gravel aquifer Bacillus subtilis endospores 1.16 64

[120] 1218 m Alluvial gravel aquifer Escherichia coli, 1.56 lm 1.3 and 2 94Endospores, 0.81.5 lm 1.22 94Phage MS2, 0.026 lm 1.21 72

[144] 0.30 m Colluvial aquifer,silt to gravel size particles Microsphere, 0.98 lm 1.810.55 m 1.51.81 m 1.1

A.A.Keller,

M.Auset

/Advances

inWater

Reso

urces

30(2007)13921407

1399

most small pore throats on the order of a few lm [121,122].A third exclusion process can occur for higher pressure gra-dients, since the colloids will tend to by-pass relatively stag-nant regions, traveling along the central streamlines. Inaddition, larger colloids are excluded from some of thestreamlines near the pore body and pore throat walls[121,122]. Finally, biocolloids may have surface chargesthat result in repulsion from the grain surfaces, thusexcluding them from certain pore regions (e.g., [113]).

The size of the microbe had previously been observed tobe an important factor in bacterial transport in porousmedia (e.g., [41,39,29]). Variation on the macroscopictransport behavior of dierent sized biocolloids can benow explained by mechanisms that occur at the scale ofpores and pore networks. All four exclusion processesresult in selectively faster transport of larger colloids, rela-tive to smaller colloids.

stant (m/s2), lw = viscosity of aqueous solution (kg/m s),U = pore water velocity (m/s), dp = particle eective diam-eter (m), dg = eective grain diameter (m), As = soil specicconstant related to hm = soil matrix porosity (),kB = Boltzmanns constant (J/K), and T = temperature(K). Recently, Tufenkji and Elimelech [133] have proposedthe following rened correlation based on experimentalevidence:

g 2:4A1=3S N0:081R N0:715Pe N 0:052vdW 0:55ASN 1:675R N 0:125A 0:22N0:24R N 1:11G N 0:053vdW 3

where NR dpdg ; NPe UdpD1

; N vdW AkBT ;

NA A12plwa2pU

; NG 29

a2pqp qfglU

;

D1 = colloid bulk diusion coecient (m2/s), A = Hamaker

21 20

1400 A.A. Keller, M. Auset / Advances in Water Resources 30 (2007) 139214072.3. Collision with SWI

From a theoretical perspective, colloids are thought toreach the SWI based on three mechanisms: interception,diusion and gravitational deposition. The theoreticalframework was put forward by Yao et al. [142] and Rajag-opalan and Tien [102], and has since been rened by severalauthors, in particular by Rajagopalan et al., Ryan andElimelech [103,109]. Based on this theoretical approach,the probability of a collision can be estimated from:

g 0:897As

3p kBT

lwdpdgU

2=3 32As

dpdg

2 qp qwg

18lwUdp

1As 21 p5=2 3p 3p5 2p6; p 1 hm1=3 2

where qp = particle density (kg/m3), qw = density of aque-

ous solution (kg/m3), g = gravitational acceleration con-Fig. 10. Estimate of relative importance of interception, diusion and gravitavelocities, for colloids of (a) 50 nm; (b) 1.0 lm; (c) 2.5 lm and (d) 5.0 lm.constant (3 10 4 10 J), and ap = particle radius(m). The three terms correspond to interception, diusionand gravitational deposition.

Using general values for biocolloids, such as a particledensity of 1050 kg/m3 (ref) and particle size ranging from50 nm to 5 lm, the relative importance of these three pro-cesses as a function of velocity can be estimated using Eq.(3) (Fig. 10), considering a porosity of 30% and an eectivegrain diameter of 100 lm (ne sand). The range of owvelocities corresponds to a few cm/d to about 100 m/d,which is the range of interest for transport in porous media.From Eq. (1), interception is considered to be mostly afunction of the relative size ratio between the colloid andthe grains of the porous medium, as well as the porosityvia As, independent of U. Interception is expected to bestrongly inuenced by matrix porosity, particularly asporosity decreases below 10%. However, the empirical evi-dence used to parameterize Eq. (3) indicates that intercep-tion is in fact a function of ow velocity, decreasing withtional deposition, and total collision probability (Eq. (3)) at dierent ow

increasing velocity. This was recently observed in micro-model studies by Baumann and Werth [8]. These experi-ments show that at high ow velocities interception is lessprobable, since the colloids follow along the streamlinesand are generally diverted from the grain surfaces.

For small biocolloids such as viruses and microorgan-isms up to about 1 lm, interception is thought to be negli-gible, while diusion dominates over gravitationaldeposition at all ow velocities of interest (Fig. 10a,b).From the micromodel studies and calculation of the veloc-ity eld within a complex pore network, there are regionsof stagnant water which are shielded from the main owdirection by the grains, are in crevasses or dead end pores,or along the walls of wide pore bodies (Fig. 11). Small bio-colloids are likely to accumulate initially in these regions,since they are more likely to be traveling along thesestreamlines and can more easily diuse into stagnantregions. For larger biocolloids, interception should domi-nate, followed by diusion (Fig. 10d). Gravitational depo-sition becomes important only for ow velocities less than1 106 m/s, or on the order of mm/day, since biocolloidsthat are almost buoyancy neutral.

2.4. Attachment

stability [62]. Recently, work by Tufenkji and Elimelech,Redman et al. [134,105] suggests that additional aspectsneed to be considered. Grain surface composition andcharge have been shown to be important (e.g.,[135,89,56]), as well as the biocolloid surface proteinsand other charged chemical species (e.g., [74,136]). Basedon theoretical calculations, Baumann and Werth [8] esti-mated that the probability of attachment for their colloidsis in the range of 104106. Schijven et al. [115] reportedvalues for a of 0.000270.0014 for MS2 viruses in dunesand. Keller et al. [70] reported values of 0.0080.0026for MS2 viruses in medium sand at ow velocities of1.414 m/day. For Cryptosporidium, Harter et al. [52]reported values from 0.37 to 1.1 in sand. In Sedimentcores, Dong et al. [29] measured values of 0.0030.025for Comamonas sp. Most of the experimental evidence isfrom column studies, leaving this as an area of openresearch at the microscale.

A number of studies have addressed the mechanisms ofbiocolloid attachment to the SWI and/or a growing bio-lm. A biolm may include cells as well as exopolymericsubstances that serve as a substrate modier for a numberof reasons (e.g., [22,17,47]). Dierent bacterial strains mayexhibit dierential attachment (e.g., [6]). Surface physico-

A.A. Keller, M. Auset / Advances in Water Resources 30 (2007) 13921407 1401Once the biocolloid collides with the SWI, the proba-bility of attaching to the surface, also denominated theattachment eciency, a, is thought to be controlled byelectrostatic and van der Waals interactions [84]. Theseinteractions have been estimated using DerjaguinLan-dauVerweyOverbeek (DLVO) theory of biocolloidalFig. 11. Images of collision via interception and diusion into dead-end pores fwithin a PDMS micromodel. Clusters of colloids form even at very low ionicchemical properties inuence the ability of biocolloids toform biolms (e.g., [15,79,63,92]). These biolms caninduce changes in hydrodynamic properties that inuencethe transport of subsequent biocolloids (e.g., [107]).Detachment of biocolloids from these biolms is a sourcefor downgradient sites, and may be inuenced by a varietyof processes including ow velocity and associated shearor 5 lm latex microspheres at an average velocity of 143 lm/s = 12 m/day,strength (deionized water) (visualization method as presented in [4]).

stress, chemical conditions or biolm thickness (e.g.,[57,45,140,60]).

Particleparticle interactions may lead to attachmentand clustering (Fig. 11e,f, Fig. 12). Attached biocolloidscan also form large clusters up to a certain thickness

matrix, and will also inuence the pathways of subsequentcolloids, modifying the dispersivity of the matrix. Clustersof colloids can also form at the AWI (discussed in thenext section), which upon release from the interface canlead to a collision with the SWI and subsequentattachment.

3. Biocolloid transport processes in unsaturated porous

media

Flow and transport mechanisms in the unsaturated zonebecome more complicated than those in the saturated zonebecause of the presence of the AWI, ow discontinuitiesand wetting history [93,110]. Investigations at larger scaleshave shown that volumetric moisture content and porewater velocity play a key role in biocolloid transport inthe vadose zone (e.g., [101]). Biocolloid sorption at theAWI has been recognized as an important process for sev-eral years [138,139,100].

Pore scale studies in unsaturated conditions [139,122]have shown that, like the SWI, the AWI serves as collectorof biocolloids (Fig. 13). Some of the colloids might also betrapped at the triple junction, the SWA. These interfaces(AWI and SWA) are therefore important barriers for bio-colloid transport. Colloids can interact with the AWI

Fig. 12. Clustering of colloids that results in signicant modication ofpermeability and colloid transport pathways (visualization method aspresented in [4]).

1402 A.A. Keller, M. Auset / Advances in Water Resources 30 (2007) 13921407(i.e., biolms, aggregates and laments, biowebs), untilsome of the cells in the interior become starved of a par-ticular chemical (e.g., electron acceptor, nutrients), lead-ing to rupture of the biolm and subsequent release(biosloughing) (e.g., [32]). These clusters can result in sig-nicant modication of the permeability of the porousFig. 13. Sequence showing the imbibition process that displaces the air phase, eEventually the air bubble dissolves, leading to the formation of clusters of colbreak up (visualization method as presented in [4]).through the same collision processes described before.However, in part due to the hydrophobicity of the AWIand the proliferation of these interfaces as the porousmatrix drains, the probability of attachment increasessignicantly.

ventually leading to a detached air bubble with several colloids at the AWI.

loids. The clusters can then be transported through the pore space, or can

WaThe earlier work visualizing colloid sorption at the AWIwas done under steady pore water ow [138], and it led tothe conclusion that colloids were sorbed irreversibly at theAWI. Increasing air saturation increased retention at theAWI [139]. These observations were supported by resultsof experiments on mass balance of breakthrough colloidconcentrations in sand columns [112,6466,77] where moreparticles were retained at lower water contents. Calcula-tions by Sirivithayapakorn and Keller [122] using DLVOtheory and evaluating the electrostatic and capillary forcesindicated that colloids, including MS2 viruses, should beheld almost irreversible at the AWI, once they cross overthe energy barrier for attachment. The energy barrierincreases with particle size, but is on the scale of 115 nmmeasured from the AWI into the bulk solution. Thus, col-loids can migrate slowly very near the AWI along stream-lines perpendicular to the interface and not be capturedunless they cross the energy barrier due to some mechanism(diusion or interception). On the other hand, water owaround entrapped air bubbles decreases substantially, sincethe dimensions of the water lms through which water owis on the order of a few lm2 at most [69].

As with attachment to the SWI, biocolloid attachmentto the AWI is a function of ionic strength and the surfaceproperties of the biocolloid, such as hydrophobicity andsurface charge [138]. Increases in ionic strength will reducethe magnitude of the repulsive energy barrier between thenegatively charged airwater interface and the biocolloids,leading to progressively more favorable conditions forattachment and faster rates of airwater interface capture.

Wan and Tokunaga [137] introduced an additionalmechanism of colloid immobilization in partially saturatedporous media. They used the concept of lm straining tosuggest that the transport of suspended colloids could beretarded due to physical restrictions imposed when thethickness of water lms is smaller than the diameter ofthe colloids. Wan and Tokunaga [137] estimated that theselms should be on the order of 2040 nm, which is consid-erably thinner than a 1 lm Escherichia coli, but may notcompletely immobilize a 25 nm virus. Chu et al. [20] esti-mated a lm thickness in the range of 1521 nm for dier-ent soils, at a water content of 0.170.29 cm3 cm3. In theircolumn studies, Keller et al. [70] estimated a thickness forthe water lms of 3060 nm in a medium sand and averagewater content of 0.110.18 cm3 cm3.

According to Wan and Tokunaga [137], colloid reten-tion by lm straining depends on the existence of pendularring discontinuity, on the ratio of biocolloid size to lmwidth and on ow velocity. A pendular ring is dened aswater retained by capillarity around the adjacent grains.The possibility of pendular ring discontinuity augmentsas the capillary pressure decreases [27]. When the biocol-loid diameter is smaller than lm thickness, strainingremains ineective. When the biocolloid diameter is similaror bigger than lm thickness than surface tension forces

A.A. Keller, M. Auset / Advances inretain biocolloids against grain surfaces. Crist et al. [25]provided visual evidence that biocolloid retention can alsooccur via trapping at the solidwaterair (SWA) interface.These thin water lms serve as storage locations for biocol-loids under unsaturated conditions, but may also serve toplace the biocolloid in direct contact with the SWI if thewater lm thickness decreases even more.

Transient ow, generated by rainfall and snowmeltevents interspersed between dry periods or due to articialaquifer recharge or other anthropogenic actions, can pro-mote very rapid biocolloid mobilization (e.g., [34]). Undertransient conditions it has been observed that the move-ment of biocolloids is aected by the movement of air bub-bles and AWI (e.g., [46,45,71]). Sirivithayapakorn andKeller, Auset et al. [122,5] observed in micromodels howinltration events can mobilize the AWI, thicken the waterlms where colloids are immobilized, dissolve some of thegas phase and mobilize air bubbles (Fig. 13). First, theAWI is displaced as the water re-imbibes into the porousmedia. Colloids trapped in stagnant water regions are ableto remobilize. At some point, an air bubble breaks o fromthe main air phase. Colloids which were attached to theAWI remain attached until the AWI disappears. Eventu-ally, these rewetting processes lead to the remobilizationof all colloids trapped at the AWI or in thin water lms.

Depending on colloid surface properties, the colloidsmay tend to cluster even in solution. However, in manycases surface charges are similar, creating an electrostaticbarrier that reduces the likelihood of clustering, as calcu-lated by Sirivithayapakorn and Keller [122] for MS2viruses and latex microspheres in a weak ionic solution.Nevertheless, colloid clusters can form at the AWI as thesize of the AWI shrinks, as observed in Fig. 13gh. Theseclusters might be stable enough to travel as a single body,or they might break up while traveling (Fig. 13i). Theseobservations suggest that coagulation at the AWI mayincrease the overall ltration for biocolloids transportedthrough the vadose zone.

Whether all colloids on the AWI are actually at theSWA interface is an open question. In most micromodelvisualizations, the pore space being observed reects a thin(1050 lm) wedge between the top and bottom surfaces(see for example the diagram in [25]). Colloids whichappear to be at the AWI could in fact be at the SWA. Cer-tainly, some of the colloids observed in these experimentsare at the SWA, as suggested by Crist et al. [25]. Even asthe imbibing water front displaces the air phase in Figs.13bg, some of the colloids remain in place, suggestingattachment to the SWI at the same time that the colloidswere in contact with the AWI. However, in other sequences(e.g., [5,122]; and Fig. 13), some colloids are seen to rapidlytravel through the porous media as soon as the AWI disap-pears, suggesting no attachment to the SWI.

Rewetting processes and intermittent wetting and dryingevents thus can result in signicant mobilization of biocol-loids that had been considered irreversibly retained at theAWI. Colloid remobilization appears to be a strong function

ter Resources 30 (2007) 13921407 1403of particle size [71]. Although intermittent ltration providessignicant pathogen removal capacity, it is important to

unanswered. The conditions that result in attachment to

ter and the UCSB Academic Senate. M. Auset thank the

Wapostdoctoral fellowship support from the Secretara de Es-tado de Educacion y Universidades (Spain). Jose Saletaperformed the ESEM at MEIAF/UCSB (NSF 9977772).

References

[1] Adelman DD, Stansbury J, Tabidian MA. A risk/cost analysis tomanage viral contamination of groundwater. Water Sci Technol1998;38(12):16.

[2] Adriano DC, Wenzel WW, Vangronsveld J, Bolan NS. Role ofassisted natural remediation in environmental cleanup. Geoderma2004;122(24):12142.

[3] Allard A-S, Neilson AH. Bioremediation of organic waste sites: athe SWI need to be better understood at this scale. Surfaceheterogeneity needs to be characterized, so that we canmake better predictions of the probability of attachmentto a particular surface given information on the biocolloid,the attachment surface and the aqueous solution composi-tion. Detachment from the surface has also not been stud-ied in any detail at this scale, whether it is directly from thesurface or from a growing biolm. We need to understandthe eect of changes in physical, chemical and/or biologicalconditions that result in detachment of biocolloids. Biolog-ical processes that inuence surface characteristics and het-erogeneity are also targets for future research at this scale.Pore scale observations can help to validate our under-standing of the mechanisms that govern the macroscopicequations used to simulate the fate and transport of biocol-loids in porous media.

Acknowledgements

The authors also wish to acknowledge partial fundingfrom US EPA Exploratory Research Grant NumberR826268 and US EPA Grant Number R827133, as wellas from the University of California Water Resources Cen-take into consideration the potential for biocolloid remobi-lization over time.

These pore-scale mechanisms in the unsaturated zoneplay a signicant role in the macroscopic transport of bio-colloids; biocolloids can be signicantly retarded in theirtransport through the porous media due to the interactionwith the AWI and SWA interface, but they can also bereleased from these interfaces to continue their path. Inaddition, sorption of biocolloids at the AWI, SWA inter-face and thin water lms can result in increased probabilityof sorption onto the SWI.

4. Future research directions

Although signicant advances have been made in ourunderstanding of biocolloid fate and transport in saturatedand unsaturated porous media with pore scale visualiza-tions, there are a number of important questions still left

1404 A.A. Keller, M. Auset / Advances incritical review of microbiological aspects. Int Biodeterior Biodegr1997;39(4):25385.[4] Auset M, Keller AA. Pore-scale processes that control dispersion ofbiocolloids in saturated porous media. Water Resour Res2004;40(3):W03503. doi:10.1029/2003WR002800.

[5] Auset M, Keller AA, Brissaud F, Lazarova V. Intermittent ltrationof bacteria and colloids at pore and column scales. Water ResourRes 2005;41(9):W09408. doi:10.1029/2004WR003611.

[6] Bakker DP, Postmus BR, Busscher HJ, Van der Mei HC. Bacterialstrains isolated from dierent niches can exhibit dierent patterns ofadhesion to substrata. Appl Environ Microbiol 2004;70:375860.

[7] Banning N, Toze S, Mee BJ. Persistence of biolm-associatedEscherichia coli and Pseudomonas aeruginosa in groundwater andtreated euent in a laboratory model system. Microbiology2003;149(1):4755.

[8] Baumann T, Werth CJ. Visualization and modeling of polystyrolcolloid transport in a silicon micromodel. Vadose Zone J2004;3:43443.

[9] Bitton G, Gerba CP. Groundwater pollution microbiology: theemerging issue. New York (NY): Wiley; 1984. pp. 713.

[10] BouwerH.Groundwater hydrology. NewYork: McGraw-Hill; 1978.[11] Dunsmore Braden C, Bass Catherine J, Lappin-ScottHilary M. A

novel approach to investigate biolm accumulation and bacterialtransport in porous matrices. Environ Microbiol 2004;6(2):1837.doi:10.1046/j.1462-2920.2003.00546.x.

[12] Bradford SA, Simunek J, Bettahar M, Van Genuchten MTh, YatesSR. Modeling colloid attachment, straining, and exclusion insaturated porous media. Environ Sci Technol 2003;37(10):224250.

[13] Bradford SA, Bettahar M. Straining, attachment, and detachment ofcryptosporidium oocysts in saturated porous media. J Environ Qual2005;34:46978.

[14] Bruins MR, Kapil S, Oehme FW. Pseudomonas pickettii: a commonsoil and groundwater aerobic bacteria with pathogenic and biodeg-radation properties. Ecotox Environ Safe 2000;47(2):10511.

[15] Chavant PB, Martinie T, Meylheuc M-N, Bellon-Fontaine, HebraudM. Listeria monocytogenes LO28: surface physicochemical proper-ties and ability to form biolms at dierent temperatures and growthphases. Appl Environ Microbiol 2002;68:72837.

[16] Chen JY, Klemic JF, Elimelech M. Micropatterning microscopiccharge heterogeneity on at surfaces for studying the interactionbetween colloidal particles and heterogeneously charged surfaces.Nano Lett 2002;2(4):3936.

[17] Cheung HY, Sun SQ, Sreedhar B, Ching WM, Tanner PA.Alterations in extracellular substances during the biolm develop-ment of Pseudomonas aeruginosa on aluminum plates. J ApplMicrobiol 2000;89:1006.

[18] Choi YC, Morgenroth E. Monitoring biolm detachment underdynamic changes in shear stress using laser-based particle sizeanalysis and mass fractionation. Biolm Monitoring. Water SciTechnol 2003;47(5):6976.

[19] Chomsurin C, Werth CJ. Analysis of pore-scale nonaqueous phaseliquid dissolution in etched silicon pore networks. Water Resour Res2003;39:126575.

[20] Chu YJ, Jin Y, Baumann T, Yates MV. Eect of soil properties onsaturated and unsaturated virus transport through columns. JEnviron Qual 2003;32:201725.

[21] Cooksey KE, Wigglesworth-Cooksey B. Adhesion of bacteria anddiatoms to surfaces in the sea: a review. Aquat Microb Ecol1995;9(1):8796.

[22] Costerton JW, Lewandowski Z, Caldwell DE, Korber DR, Lappin-Scott HM. Microbial biolms. Annu Rev Microbiol 1995;49:71145.

[23] Costerton JW. Introduction to biolm. Int J Antimicrob Agent1999;11:21721.

[24] Craig D, Falloweld H, Cromar N. Enumeration of fecal coliformsfrom recreational coastal sites: evaluation of techniques for theseparation of bacteria from sediments. J Appl Microbiol2002;93(4):55765.

[25] Crist JT, McCarthy JF, Zevi Y, Baveye P, Throop JA, Steenhuis TS.

ter Resources 30 (2007) 13921407Pore-scale visualization of biocolloid transport and retention inpartly saturated porous media. Vadose Zone J 2004;3:44450.

Wa[26] de Marsily G. Quantitative hydrogeology. San Diego, CA,USA: Academic Press; 1986. p. 440.

[27] DeNovio NM, Saiers JE, Ryan JN. Colloid movement in unsatu-rated porous media. Recent advances and future directions. VadoseZone J 2004;3:33851.

[28] Detay M, Alessandrello E, Come P, Groom I. Groundwater contam-ination and pollution in Micronesia. J Hydrol 1989;112(12):14970.

[29] Dong HL, Onstott TC, Deaun MF, Fuller ME, Scheibe TD,Streger SH, et al. Relative dominance of physical versus chemicaleects on the transport of adhesion-decient bacteria in intactcores from South Oyster Virginia. Environ Sci Technol 2002;36:891900.

[30] Dowd SE, Pillai SD. Survival and transport of selected bacterialpathogens and indicator viruses under sandy aquifer conditions.Environ Sci Health, Pt. A: Environ Sci Eng Toxic Hazard SubstControl 1997;32A(8):224558.

[31] Dupin HJ, McCarty PL. Mesoscale and microscale observations ofbiological growth in a silicon pore imaging element. Environ SciTechnol 1999;33(8):12306.

[32] Dupin HJ, McCarty PL. Impact of colony morphologies anddisinfection on biological clogging in porous media. Environ SciTechnol 2000;34(8):151320.

[33] Ebihara T, Bishop PL. Eect of acetate on biolms utilized in PAHbioremediation. Environ Eng Sci 2002;19(5):30519.

[34] El-Farhan YH, DeNovio NM, Herman JS, Hornberger GM.Mobilization and transport of soil particles during inltrationexperiments in and agricultural eld, Shenandoah Valley, Virginia.Environ Sci Technol 2000;34:35559.

[35] Ferguson AD, Deisenhofer J. Metal import through microbialmembranes. Cell 2004;116(1):1524.

[36] Filip Z, Kaddu-Mulindwa D, Milde G. Survival of some pathogenicand facultative pathogenic bacteria in groundwater. Water SciTechnol 1988;20(3):22731.

[37] Fiorenza S, Rifai HS. Review of MTBE biodegradation andbioremediation. Bioremediation J 2003;7(1):135.

[38] Foght J, April T, Biggar K, Aislabie J. Bioremediation of DDT-contaminated soils: a review. Bioremediation J 2001(5):22546.

[39] Fontes DE, Mills AL, Hornberger GM, Herman JS. Physical andchemical factors inuencing transport of microorganisms throughporous media. Appl Environ Microbiol 1991;57:247381.

[40] Frankenberger WT. Fate of wastewater constituents in soil andgroundwater: pathogens. In: Irrigation with reclaimed municipalwastewater a guidance manual. Chelsea, Michigan: Lewis Pub-lishers; 1985. p. 425.

[41] Gannon JT, Manilal VB, Alexander M. Relationship between cellsurface properties and transport of bacteria through soil. ApplEnviron Microbiol 1991;57:1903.

[42] Gerba CP, Goyal SM. Pathogen removal from wastewater duringgroundwater recharge articial recharge of groundwater. Boston,Massachusetts: Butterworth Publishers; 1985. pp. 317.

[43] Ginn TR. A travel time approach to exclusion on transport inporous media. Water Resour Res 2002;38(4). doi:10.1029/2001WR000865.

[44] Ginn TR, Wood BD, Nelson KE, Scheibe TD, Murphy EM,Clement TP. Processes in microbial transport in the naturalsubsurface. Adv Water Resour 2002;25(812):101742.

[45] Gomez-Suarez C, Busscher HJ, Van der Mei HC. Analysis ofbacterial detachment from substratum surfaces by the passage ofairliquid interfaces. Appl Environ Microbiol 2001;67:25317.

[46] Gomez-Suarez C, Noordmans J, van der Mei HC, Busscher HJ.Removal of biocolloidal particles from quartz collector surfaces asstimulated by the passage of liquidair interfaces. Langmuir1999;15:51237.

[47] Gomez-Suarez C, Pasma J, van der Borden AJ, Wingender J,Flemming HC, Busscher HJ, et al. Inuence of extracellularpolymeric substances on deposition and redeposition of Pseudo-

A.A. Keller, M. Auset / Advances inmonas aeruginosa to surfaces. Microbiology-SGM 2002;148:11619.[48] Gordon C, Toze S. Inuence of groundwater characteristics on thesurvival of enteric viruses. J Appl Microbiol 2003;95(3):53644.

[49] Goyal SM, Amundson DA, Robinson RA, Gerba CP. Viruses anddrug resistant bacteria in groundwater of southeastern Minnesota. JMinn Acad Sci 1989;55(1):5862.

[50] Grolimund D, Elimelech M, Borkovec M, Barmettler K, Kretzsch-mar R, Sticher H. Transport of in situ mobilized biocolloidalparticles in packed soil columns. Environ Sci Technol 1998;32(22):35629.

[51] Hagedorn C, McCoy EL, Rahe TM. The potential for groundwatercontamination from septic euents. J Environ Qual 1981;10(1):18.

[52] Harter T, Wagner S, Atwill ER. Biocolloid transport and ltrationof Cryptosporidium parvum in sandy soils and aquifer sediments.Environ Sci Technol 2000;34(1):6270.

[53] Harvey RW, Kinner NE, Bunn A, MacDonald D, Metge D.Transport behavior of groundwater protozoa and protozoan-sizedmicrospheres in sandy aquifer sediments. Appl Environ Microbiol1995;61(1):20917.

[54] Herbold-Paschke K, Straub U, Hahn T, Teutsch G, Botzenhart K.Behaviour of pathogenic bacteria, phages and viruses in groundwa-ter during transport and adsorption. Water Sci Technol 1991;24(2):3014.

[55] Hiney M, Stanley C, Martin L, Cooney A, Smith P. The inuenceof sediment type on the survival of colony forming ability ofAeromonas salmonicida in laboratory microcosms. Aquaculture2002;212(14):119.

[56] Hoek EMV, Bhattacharjee S, Elimelech M. Eect of membranesurface roughness of colloid-membrane DLVO interactions. Lang-muir 2003;19:483647.

[57] Horn H, Rei H, Morgenroth E. Simulation of growth anddetachment in biolm systems under dened hydrodynamic condi-tions. Biotechnol Bioeng 2003;81(5):60717.

[58] Hornberger GM, Mills AL, Herman JS. Bacterial transport inporous media: evaluation of a model using laboratory observations.Water Resour Res 1992;28:91538.

[59] Huang CT, Peretti SW, Bryers JD. Use of ow cell reactors toquantify biolm formation kinetics. Biotechnol Tech 1992;6(3):1938.

[60] Hunt SM, Werner EM, Huang B, Hamilton MA, Stewart PS.Hypothesis for the role of nutrient starvation in biolm detachment.Appl Environ Microbiol 2004;70(12):741825.

[61] Indest KJ, Betts K, Furey JS, Fredrickson HL, Hinton VR.Evaluation of a real-time Taqman registered PCR method forassessment of pathogenic coliform contamination in sediment.Aquat Ecosyst Health Manage 2004;7(3):41524.

[62] Israelachvili JN. Intermolecular and surface forces. 2nd ed. NewYork: Academic Press; 1992.

[63] Ista LK, Callow ME, Finlay JA, Coleman SE, Nolasco AC, SimonsRH, et al. Eect of substratum surface chemistry and surface energyon attachment of marine bacteria and algal spores. Appl EnvironMicrobiol 2004;70:41517.

[64] Jewett DG, Logan BE, Arnold RG, Bales RC. Transport ofPseudomonas uorescens strain P17 through quartz sand columns asa function of water content. J Contam Hydrol 1999;36:7389.

[65] Jewett DG, Hilbert TA, Logan BE, Arnold RG, Bales RC. Bacterialtransport in laboratory columns and lters: Inuence of ionicstrength and pH on collision eciency. Water Res 1995;29(7):167380.

[66] Jin Y, Chu YJ, Li YS. Virus removal and transport in saturated andunsaturated sand columns. J Contam Hydrol 2000;43.

[67] Johnson PR, Sun N, Elimelech M. Biocolloid transport in geo-chemically heterogeneous porous media: Modeling and measure-ments. Environ Sci Technol 1996;30:328493.

[68] Kalogerakis N, Alvarez P, Psillakis E. Special issue: recent advancesin bioremediation. Environ Int 2005;31(2):147.

[69] Keller AA, Blunt MJ, Roberts PV. Micromodel observation of the

ter Resources 30 (2007) 13921407 1405role of oil layers in three-phase ow. Transport Porous Med1997;26(3):27797.

Wa[70] Keller AA, Sirivithayapakorn S, Chrysikopoulos CV. Early break-through of colloids and bacteriophage MS2 in a water-saturatedsand column. Water Resour Res 2004;40(8). Art. No. W08304.

[71] Keller AA, Sirivithayapakorn S. Transport of colloids in unsatu-rated porous media: explaining large-scale behavior based on pore-scale mechanisms. Water Resour Res 2004;40(12). Art. No. W12403.

[72] Kersting AB, Efurd DW, Finnegan DL, Rokop DJ, Smith DK,Thompson JL. Migration of plutonium in groundwater at theNevada Test Site. Nature 1999;397:569.

[73] Kim DS, Fogler HS. Biomass evolution in porous media and itseects on permeability under starvation conditions. BiotechnolBioeng 2000;69(1):4756.

[74] Kuznar ZA, Elimelech M. Role of surface proteins in the depositionkinetics of Cryptosporidium parvum oocysts. Langmuir 2005;21:7106.

[75] Lanning LM, Ford RM. Glass micromodel study of bacterialdispersion in spatially periodic porous networks. Biotechnol Bioeng2002;78:55666.

[76] Laspidou CS, Rittman BE. A unied theory for extracellularpolymeric substances, soluble microbial products, and active andinert biomass. Water Res 2002;36:271120.

[77] Lenhart JJ, Saiers JE. Transport of silica biocolloids throughunsaturated porous media: Experimental results and model com-parisons. Environ Sci Technol 2002;36:76977.

[78] Leon-Morales CF, Leis AP, Strathmann M, Flemming HC. Inter-actions between laponite and microbial biolms in porous media:implications for colloid transport and biolm stability. Water Res2004;38(16):361426.

[79] Liu Y, Yang S-F, Li Y, Xu H, Qin L, Tay J-H. The inuence of celland substratum surface hydrophobicities on microbial attachment. JBiotechnol 2004;110:2516.

[80] Lopez LAG, Veiga MC, Nogueira R, Aparicio A, Melo LF. Atechnique using a membrane ow cell to determine average masstransfer coecients and tortuosity factors in biolms. BiolmMonitoring. Water Sci Technol 2003;47(5):617.

[81] Macler BA, Merkle JC. Current knowledge on groundwatermicrobial pathogens and their control. Hydrogeol J 2000;8(1):2940.

[82] McClaine JW, Ford RM. Characterizing the adhesion of motile andnonmotile Escherichia coli to a glass surface using a parallel-plateow chamber. Biotechnol Bioeng 2002;78(2):17989.

[83] Magarinos B, Romalde JL, Barja JL, Toranzo AE. Evidence of adormant but infective state of the sh pathogen Pasteurella piscicidain seawater and sediment. Appl Environ Microbiol 1994;60(1):1806.

[84] Marshall KC, Stout R, Mitchell RJ. Gen Microbiol 1971;68:337.[85] Matthess G, Pekdeger A, Schroeter J. Persistence and transport of

bacteria and viruses in groundwater a conceptual evaluation. JContam Hydrol 1988;2(2):17188.

[86] McGechan MB, Jarvis NJ, Hooda PS, Vinten AJA. Parameteriza-tion of the MACRO model to represent leaching of biocolloidallyattached inorganic phosphorus following slurry spreading. Soil UseManage 2002;18:617.

[87] Moe CL, Cogger CG, Sobsey MD. Viral and bacterial contamina-tion of groundwater by on-site wastewater treatment systems insandy coastal soils. in: Proceedings of the second, internationalconference on ground water quality research. Houston (TX):National Center for Ground Water Res; 1985. p 1324.

[88] Morris BL, Foster SSD. Cryptosporidium contamination hazardassessment and risk management for British groundwater sources.Water Sci Technol 2000;41(7):6777.

[89] Mylon SE, Chen KL, Elimelech M. Inuence of natural organicmatter and ionic composition on the kinetics and structure ofhematite colloid aggregation: implications to iron depletion inestuaries. Langmuir 2004;20:90006.

[90] Nambi IM, Werth CJ, Sanford RA, Valocchi AJ. Pore-scale analysisof anaerobic halorespiring bacterial growth along the transverse

1406 A.A. Keller, M. Auset / Advances inmixing zone of an etched silicon pore network. Environ Sci Technol2003;37(24):561724.[91] Nasser AM, Tchorch Y, Fattal B. Comparative Survival of E. coli,F+Bacteriophages, HAV and Poliovirus 1 in wastewater andgroundwater. Water Sci Technol 1993;27(3/4):4017.

[92] Nedwell DB, Gray TRG. Soils and sediments as matrices formicrobial growth. In: Fletcher M, Gray TRG, Jones JG, editors.Ecology of Microbial Communities, vol. 41. New York: CambridgeUniversity Press; 1987. p. 2154.

[93] Nielsen DR, Genuchten MTH, Biggar JW. Water ow and solutetransport processes in the unsaturated zone. Water Resour Res1986;22(9):89S108S.

[94] Nivens DE, Palmer Jr RJ, White DC. Continuous nondestructivemonitoring of microbial biolms: a review of analytical techniques. JInd Microbiol 1995;15(4):26376.

[95] Nola M, Njine T, Sikati VF, Djuikom E. Distribution of Pseudo-monas aeruginosa and Aeromonas hydrophila in groundwater inequatorial zone in Cameroon and relationships with some environ-mental chemical factors. Rev Sci Eau/J Water Sci 2001;14(1):3553.

[96] Norris RD, Hinchee RE, Brown R, McCarty PL, Semprini L,Wilson JT, Kampbell DH, Reinhard M, Bouwer EJ, Borden RC,Vogel TM, Thomas JM, Ward CH. In situ bioremediation of groundwater and geological material: a review of technologies. Report No.EPA/SR-93/124 United States Environmental Protection Agency,Cincinnati, OH 45268, USA; 1993.

[97] Olson MS, Ford RM, Smith JA, Fernandez EJ. Quantication ofbacterial chemotaxis in porous media using magnetic resonanceimaging. Environ Sci Technol 2004;38(14):386470.

[98] Pang LP, Close M, Noonan M. Rhodamine WT and bacillus subtilistransport through an alluvial gravel aquifer. Ground Water1998;36(1):11222.

[99] Pitt R, Clark S, Field R. Groundwater contamination potential fromstormwater inltration practices. Urban Water 1999;1(3):21736.

[100] Powelson DK, Mills AL. Bacterial enrichment at the gaswaterinterface of a laboratory apparatus. Appl Environ Microbiol1996;62:25937.

[101] Powelson DK, Simpson JR, Gerba CP. Virus transport and survivalin saturated and unsaturated ow through soil columns. J EnvironQual 1990;19:396401.

[102] Rajagopalan R, Tien C. Trajectory analysis of deep-bed ltrationwith the sphere-in-cell porous media model. AIChE J 1976;22(3):52333.

[103] Rajagopalan R, Tien C, Pfeer R, Tardos G. Letter to the editor.AIChE J 1982;28(5):8712.

[104] Redman JA, Grant SB, Olson TM, Estes MK. Pathogen ltration,heterogeneity, and the potable reuse of wastewater. Environ SciTechnol 2001;35(9):1798805.

[105] Redman JA, Walker SL, Elimelech M. Bacterial adhesion andtransport in porous media: role of the secondary energy minimum.Environ Sci Technol 2004;38:177785.

[106] Rittmann BE, Valocchi AJ, Seagren E, Ray C, Wrenn B. Criticalreview of in situ bioremediation. Topical report GRI-92/0322. GasResearch Inst., Chicago, IL and Department of Energy, Morgan-town, WV, Morgantown Energy Technology Center; 1992. p. 185.

[107] Rockhold ML, Yarwood RR, Niemet MR, Bottomley PJ, Selker JS.Considerations for modeling bacterial-induced changes in hydraulicproperties of variably saturated porous media. Adv Water Resour2002;25:47795.

[108] Rose JB, Vasconcelos GJ, Harris SI, Klonicki PT, Sturbaum GD,Moulton-Hancock C. Giardia and Cryptosporidium occurrence ingroundwater. J Am Water Works Assoc 2000;92(9):11723.

[109] Ryan JN, Elimelech M. Colloid mobilization and transport ingroundwater. Colloid Surface A-Physicochem Eng Aspects 1996;107:156.

[110] Saiers JE, Lenhart JJ. Biocolloid mobilization and transport withinunsaturated porous media under transient-ow conditions. WaterResour Res 1996;39(1):1019. doi:10.1029/2002WR001370.

[111] Scandura JE, Sobsey MD. Viral and bacterial contamination of

ter Resources 30 (2007) 13921407groundwater from on-site sewage treatment systems. Water SciTechnol 1997;38(12):1416.

A.A. Keller, M. Auset / Advances in Wa[112] Schafer A, Ustohal P, Harms H, Stauer F, Dracos T, Zehnder AJB.Transport of bacteria in unsaturated porous media. J ContamHydrol 1998;33:14969.

[113] Scheibe TD, Wood BD. A particle-based model of size or anionexclusion with application to microbial transport in porous media.Water Resour Res 2003;39(4). doi:10.1029/2001WR001223.

[114] Schijven JF, Hassanizadeh SM. Virus removal by soil passage ateld scale and groundwater protection of sandy aquifers. Secondworld water congress: environmental monitoring, contaminants andpathogens. Water Sci Technol 2001;46(3):1239.

[115] Schijven JF, Hoogenboezem W, Hassanizadeh SM. Modelingremoval of bacteriophages MS2 and PRD1 by dune rechargeat Castricum, Netherlands. Water Resour Res 1999;35(4):110112.

[116] Schulze-Makuch D, Guan H, Pillai SD. Eects of pH and geologicalmedium on bacteriophage MS2 transport in a model aquifer.Geomicrobiol J 2003;20(1):7384.

[117] Schwarzenbach RP, Gschwend PM, Imboden DM. Environmentalorganic chemistry. New York: WileyInterscience; 1993.

[118] Sherwood JL, Sung JC, Ford RM, Fernandez EJ, Maneval JE,Smith JA. Analysis of bacterial random motility in a porous mediumusing magnetic resonance imaging and immunomagnetic labeling.Environ Sci Technol 2003;37(4):7815.

[119] Shonnard DR, Taylor RT, Hanna ML, Boro CO, Duba AG.Injection-attachment of methylosinus-trichosporium ob3b in a 2-imensional miniature sand-lled aquifer simulator. Water ResourRes 1994;30(1):2535.

[120] Sinton LW, Noonan MJ, Finlay RK, Pang L, Close ME. Transportand attenuation of bacteria and bacteriophages in an alluvial gravelaquifer. NZ J Mar Freshwater Res 2000;34(1):17586.

[121] Sirivithayapakorn S, Keller AA. Transport of colloids in saturatedporous media: a pore-scale observation of the size exclusion eectand colloid acceleration. Water Resour Res 2003;39(4):1109.doi:10.1029/2002WR001583.

[122] Sirivithayapakorn S, Keller AA. Transport of biocolloids inunsaturated porous media: a pore-scale observation of processesduring the dissolution of airwater interface. Water Resour Res2003;39(12):1346. doi:10.1029/2003WR002487.

[123] Sjollema J, Busscher HJ, Weerkamp AH. Real-time enumeration ofadhering microorganisms in a parallel plate ow cell usingautomated image analysis. J Microbiol Meth 1989;9(2):738.

[124] Snowdon JA, Coliver DO. Coliphages as indicators of humanenteric viruses in groundwater. Crit Rev Environ Contr 1989;19(3):23149.

[125] Stewart TL, Fogler HS. Biomass plug development and propagationin porous media. Biotechnol Bioeng 2001;72(3):35363.

[126] Stewart PS. A review of experimental measurements of eectivediusive permeabilities and eective diusion coecients in biolms.Biotechnol Bioeng 1998;59(3):26172.

[127] Stoodley P, Cargo R, Rupp CJ, Wilson S, Klapper I. Biolmmaterial properties as related to shear-induced deformation anddetachment phenomena. J Ind Microbiol Biotechnol 2002;29(6):3617.

[128] Surampalli RY, Lin KL, Banerji SK, Sievers DM. Impact of long-term land application of biosolids on groundwater quality andsurface soils. J Environ Syst 1997;26(3):30524.

[129] Sutherland IW. The biolm matrix: an immobilized but dynamicmicrobial environment. Trend Microbiol 2001;9:2227.

[130] Taylor R, Cronin A, Pedley S, Barker J, Atkinson T. Theimplications of groundwater velocity variations on microbial trans-port and wellhead protectionreview of eld evidence. FEMSMicrobiol Ecol 2004;49(1):1726.

[131] Torquato S. Random heterogeneous materials: microstructure andmacroscopic properties. New York (NY), USA: Springer; 2002.

[132] Tsuruta T. Cell-associated adsorption of thorium or uranium fromaqueous system using various microorganisms. Water, Air, Soil

Pollut 2004;159(1):3547.[133] Tufenkji N, Elimelech M. Correlation equation for predicting single-collector eciency in physicochemical ltration in saturated porousmedia. Environ Sci Technol 2004;38:52936.

[134] Tufenkji N, Elimelech M. Deviation from classical colloid ltrationtheory in the presence of repulsive DLVO interactions. Langmuir2004;20:1081828.

[135] Tufenkji N, Elimelech M. Breakdown of colloid ltration theory:role of the secondary energy minimum and surface charge hetero-geneities. Langmuir 2005;21:84152.

[136] Walker SL, Redman JA, Elimelech M. Role of cell surfacelipopolysaccharides (LPS) in Escherichia coli K12 adhesion andtransport. Langmuir 2004;20:773646.

[137] Wan JM, Tokunaga TK. Film straining of colloids in unsaturatedporous media: conceptual model and experimental testing. EnvironSci Technol 1997;31(8):241320.

[138] Wan JM, Wilson JL. Of the role of the gaswater interface on thefate and transport of biocolloids in porous-media. Water ResourRes 1994;30:1123.

[139] Wan JM, Wilson JL. Colloid transport in unsaturated porous-media. Water Resour Res 1994;30:85764.

[140] Wilson S, Hamilton MA, Hamilton GC, Schumann MR, StoodleyP. Statistical quantication of detachment rates and size distribu-tions of cell clumps from wild-type (PAO1) and cell signaling mutant(JP1) Pseudomonas aeruginosa biolms. Appl Environ Microbiol2004;70(10):584752.

[141] Wilson SC, Jones KC. Bioremediation of soil contaminated withpolynuclear aromatic hydrocarbons (PAHs): a review. EnvironPollut 1993;81(3):22949.

[142] Yao K, Habibian MT, OMelia CR. Water and wastewaterltration: concepts and applications. Environ Sci Technol1971;5(11):110512.

[143] Yates MV, Yates SR. Virus survival and transport in groundwater.Water Sci Technol 1988;20(11/12):3017.

[144] Zhang P, Johnson WP, Piana MJ, Fuller CC, Naftz DL. Potentialartifacts in interpretation of dierential breakthrough of biocolloidsand dissolved tracers in the context of transport in a zero-valent ironpermeable reactive barrier. Ground Water 2001;39(6):83140.

[145] Zouboulis AI, Loukidou MX, Matis KA. Biosorption of toxicmetals from aqueous solutions by bacteria strains isolated frommetal-polluted soils. Process Biochem 2004;39(8):90916.

[146] Harvey RW, George LH, Smith RL, LeBlanc DR. Transport ofmicrospheres and indigenous bacteria through a sandy aquifer:Results of natural and forced-gradient tracer experiments. EnvironSci Technol 1989;23:516.

[147] Harvey RW, Kinner NE, MacDonald D, Metge DW, Bunn A. Roleof physical heterogeneity in the interpretation of small-scale labo-ratory and eld observations of microorganism, microsphere, andbromide transport through aquifer sediments. Water Resour Res1993;29:271321.

[148] Higgo JJW, Williams GM, Harrison I, Warwick P, Gardiner MP,Longworth G. Colloid transport in a glacial sand aquifer laboratory and eld studies. Colloid Surface A 1993;73:179200.

[149] Massei N, Lacroix M, Wang HQ, Dupont JP. Transport ofparticulate material and dissolved tracer in a highly permeableporous medium: comparison of the transfer parameters. J ContamHydrol 2002;57:2139.

[150] Nagasaki S, Tanaka S, Suzuki A. Fast transport of colloidalparticles through quartz-packed columns. J Nucl Sci Technol1993;30:113644.

[151] Sinton LW, Finlay RK, Pang L, Scott DM. Transport of bacteriaand bacteriophages in irrigated euent into and through an alluvialgravel aquifer. Water Air Soil Poll 1997;98:1742.

[152] Sinton LW, Close ME. Groundwater Tracing Experiments. Publi-cation No. 2 of the Hydrology Centre, Ministry of Works andDevelopment, Christchurch, New Zealand. 1983. p. 38.

[153] Small H. Hydrodynamic Chromatography Technique for size

ter Resources 30 (2007) 13921407 1407analysis of colloidal particles. J Colloid Interf Sci 1974;48:14761.

A review of visualization techniques of biocolloid transport processes at the pore scale under saturated and unsaturated conditionsIntroductionBiocolloid transport processes in saturated porous mediaAdvection, diffusion, dispersionExclusionCollision with SWIAttachment

Biocolloid transport processes in unsaturated porous mediaFuture research directionsAcknowledgementsReferences