ARTICLE IN PRESS

Available at www.sciencedirect.com

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 2 0 5 9 – 2 0 6 8

0043-1354/$ - see frodoi:10.1016/j.watres

�Corresponding auE-mail address:

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

Influence of biofilms on the movement of colloids in porousmedia. Implications for colloid facilitated transport insubsurface environments

Carlos Felipe Leon Morales�, Martin Strathmann, Hans-Curt Flemming

FB Chemie—Biofilm Centre, University of Duisburg-Essen, Geibelstrasse 41, D-47057 Duisburg, Germany

a r t i c l e i n f o

Article history:

Received 1 May 2006

Received in revised form

11 January 2007

Accepted 20 February 2007

Available online 9 April 2007

Keywords:

Biofilms

Colloids

Porous matrix

Transport

Retention

nt matter & 2007 Elsevie.2007.02.024

thor. Tel.: +49 203 379 3801felipe@uni-duisburg.de (C

a b s t r a c t

Colloid transport through porous media can be influenced by the presence of biofilms.

Sterile and non-sterile sand columns were investigated using Laponite RD as model colloid

and a highly mucoid strain of Pseudomonas aeruginosa as model biofilm former. Laponite RD

was marked specifically by fluorescent complexes with rhodamine 6G. Breakthrough

curves (BTCs) were used as parameters for determination of colloid transport character-

istics. In the sterile columns, the colloid was mobile (collision efficiencies from 0.05 to 0.08)

both after the presence of Na+ and Ca2+ ions followed by deionised water influent. In the

biofilm-grown column, the same treatment did not result in colloid retention in the case of

Na+ exposure, but in altered or enhanced colloid transport. In the case of Ca2+ ions

exposure, colloid retention increased with biofilm age. After 3 weeks, almost complete

retention was observed. Similar observations were made in columns packed with material

from slow sand filtration units. These data reveal the complex interactions between

biofilms, cations and colloid transport. Changes in the electrolyte composition of water

percolating the subsurface can frequently occur and will result in different colloid transport

characteristics with regard to the dominating species of ions and the relative abundance of

microbial biofilms. This has to be considered when modelling colloid transport through the

subsurface.

& 2007 Elsevier Ltd. All rights reserved.

1. Introduction

Scientific interest in the way colloids (particles with an

average diameter between 10�9 and 10�6 m) are transported

through the intricate pores of subsurface environments has

been inspired by two facts affecting human health: colloids

can be contaminants by themselves (bacteria, viruses,

organics) and colloids can be carriers of contaminants

(Kretzschmar et al., 1999). The awareness that colloids can

act under certain conditions as contaminant carriers has

motivated a great number of studies on colloidal transport in

the subsurface throughout the last decade (Grolimund et al.,

r Ltd. All rights reserved.

, +49 203 379 3801 3796; fa.F. Leon Morales).

1996, 1998; Kim and Corapcioglu, 1997; Corapcioglu et al.,

1999; Tatalovich et al., 2000; Cherrey et al., 2003).

Many colloid transport experiments have been carried out

under well-defined physicochemical conditions in which the

systems are composed of only the collectors (sand grains,

glass spheres, polystyrene spheres, etc.), the studied colloidal

particles (latex colloids, bacteria, viruses, metal complexes,

etc.) and the defined electrochemical conditions of both the

liquid flowing phase (presence of different ion species,

different pH conditions, etc.) and the solid surfaces. Data

obtained from these well-defined experiments have been

used for the development of colloid transport prediction

x: +49 203 379 1941

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 2 0 5 9 – 2 0 6 82060

theories such as the colloid filtration theory (CFT) (Yao et al.,

1971). The equations comprising the CFT only describe colloid

transport under limited conditions and hence are of little use

when trying to predict colloid transport under more complex

real-life situations.

The solution chemistry in many natural subsurface envir-

onments is dominated by the relative contributions made by

Ca2+ ions and Na+ ions. Calcium in groundwater comes from

the decomposition of rocks and mainly from the dissolution

of carbonate minerals, both of which represent important

players in groundwater chemical equilibria. Sodium has also

several sources including rock weathering, contaminant

plumes, and salt-water infiltration in near-coast environ-

ments. Several well-controlled studies have investigated

calcium and sodium influence on colloid transport and

aggregation (Chen and Kojouharov, 1998; Kretzschmar and

Sticher, 1998; Davis et al., 2001). Colloid stability and transport

has been generally observed to be very sensitive to the

presence of Ca2+ ions, resulting in colloid retention or

aggregation even at relatively low Ca2+ concentrations.

Subsurface biofilm formation is a well-observed phenom-

enon. Biofilm accumulation in sediments can be considered

as a highly complex and dynamic organic matter accumula-

tion. The porous nature of many subsurface environments

facilitates microbial deposition and colonization (Bouwer et

al., 2000). High specific surface areas result in high collision

rates for migrating particles and filtration of suspended

nutrients. The production of extracellular polymeric sub-

stances (EPS) (Wingender et al., 1999) and continued microbial

growth results in the consolidation of biofilms in the subsur-

face (Bouwer et al., 2000). Biofilms in many sedimentary

environments including polluted rivers are normally quanti-

fied in terms of their constituents (microorganisms and EPS

among others) (Battin et al., 2001; Brummer et al., 2000;

Orvain et al., 2003). In the case of surface or close to the

surface biofilms, where microalgae such as diatoms are

dominant, chlorophyill becomes another commonly mea-

sured biofilm parameter (Battin and Sengschmitt, 1999; Battin

et al., 2003). Interestingly, calcium ions have been found to be

important EPS structural components (Flemming et al., 2000).

The presence of calcium ions contribute to the architecture

and viscoelastic properties of biofilms (Koerstgens et al.,

2001). The presence of Ca2+ ions can therefore drastically

increase biofilm stability and alter detachment rates in these

environments.

Direct physical contact between the subsurface biofilm

mass and the mobile colloidal particle may result in reten-

tion. Many potential sorption sites have been identified in

biofilms (Flemming, 1995). This is not however, the only

possible mechanism of biofilm influence. Losses of perme-

ability and changes of the subsurface hydraulic conditions

due to biofilm growth (Vandevivere and Baveye, 1992) will

result in a complex indirect influence on colloid and colloid-

bound contaminant transport. The effect of biofilm structure

and complex mass transfer processes, including diffusion

and convection, on colloid transport, however, remains

unclear.

The significant role of calcium and sodium in many

environmental systems and the ubiquity and complexity of

biofilms raises important questions about their influence on

colloid transport. This is especially true in contaminated

environments where biofilm presence is more evident. Large-

scale predictions of colloid and colloid-bound contaminant

transport, especially in contaminated environments, should

therefore include the influence of biofilms.

Typically, the study and modelling of biofilm formation in

the subsurface has focused on bioremediation (biobarrier

technologies) or bioclogging, an important process for filtra-

tion and for example, oil recovery. Surprisingly few reported

investigations have been conducted on the details of subsur-

face biofilm influence on colloidal or colloid-bound contami-

nant transport.

Breakthrough patterns, evidenced as breakthrough curves

(BTCs), as well as attachment or collision efficiencies are

important colloid transport parameters when assessing

colloid mobility through porous materials. In sand column

experiments, these parameters can be obtained with relative

ease if the detection system is able to discriminate well

between colloidal particles and other material potentially

present in the column effluent. This is especially true in

complex experimental settings, such as those with high

organic matter content or as in the case of this study, the

presence of biofilms. Providing the fact that this is achieved,

the deposition kinetics behaviour (assessment of colloidal

transport) of the mobile colloidal particles can be obtained by

using experimental methods such as the pulse technique

(Grolimund et al., 2001). The deposition rate constant, kd, and

collision efficiency, a, can be calculated from these BTCs using

the normalized suspended particle concentration ðC=C0Þ as

previously explained (Leon Morales et al., 2004).

A reduction in the saturated hydraulic conductivity can be

used as an indirect indicator of biofilm development. For

example, the presence of bacterial extracellular polymers has

been found to be responsible for a reduction in saturated

hydraulic conductivity in sand columns (Vandevivere and

Baveye, 1992). This is therefore an indication of an active

biofilm community. In saturated conditions, Darcy’s law can

be used to measure the hydraulic conductivity of the medium

at any stage of biofilm development:

K ¼QðdlÞAðdhÞ

, (1)

where Q is discharge (V T�1), K is hydraulic conductivity

(LT�1), A is the cross-sectional area of the sand column (L2),

dh is the difference in hydraulic head and dl is the distance

over which head is lost.

The main aim of the work described in this paper is to

determine the influence of biofilms on colloidal movement

under changing electrochemical conditions.

2. Materials and methods

2.1. Porous media and colloids

Quartz sand F34 with an average diameter of 0.2 mm was

used. The sand was washed several times with deionised

water followed by an acid–base treatment (0.2 M NaOH and

0.2 M HCl) prior to each experiment. Between the acid and

base steps deionised water was applied until neutral pH.

ARTICLE IN PRESS

Table 1 – Hydraulic characteristics of sand columns usedfor colloid transport studies

Flow rate (mL h�1) 50

Average linear velocity (cm min�1) 1.1270.08

Bulk density (g cm�3) 1.36470.001

Porosity 0.4870.04

Pore volume (mL) 7.470.5

Hydrodynamic disp. coefficient (cm2 min�1) 7.4� 10�275� 10�3

Pe 15272� 10�6

Hydraulic conductivity (cm s�1) 3.1� 10�274� 10�3

WAT E R R E S E A R C H 41 (2007) 2059– 2068 2061

Laponite RD an artificial, clay was used for colloid transport

studies in the quartz sand-packed columns, while in columns

packed with aged filter sand, the remobilization of in situ

colloids was observed under controlled electrochemical

conditions. Aged filter sand was taken from an active slow

sand filter at a drinking water purification plant. This material

consisted of gravel, sand, and native mixed biofilm popula-

tion and natural colloids.

The laponite RD colloid resembles in structure, the one of

the hectorite type of clays. In deionised water and close to

neutral pH, the clay will be present as highly and homo-

geneously dispersed colloidal disks (30 nm diameter�1–2 nm

thick). For the colloid transport studies, laponite suspensions

were used at 200 mg L�1, suspended in deionised water

(approx. 40 min stirring). Additionally, the colloidal dispersion

was stained with the fluorochrome rhodamine 6G (Fluka,

Switzerland) at a concentration of 5�10�6 M according to the

protocol described in Tapia Estevez et al. (1993). The

laponite–rhodamine complexes were quantified in column

effluents by fluorescence (lex,max 480 nm, lemm,max. 551 nm).

To obtain relative fluorescence, the same suspension used for

colloidal pulse injections was used to calibrate the fluorom-

eter at 100% fluorescence. After pulse injections, fractions

collected from column effluents were measured against this

calibration.

2.2. Column setup, main experimental setup andphysicochemical conditions

Borosilicate glass columns with a diameter of 1.4 and 10 cm

length were used for the transport and remobilization

experiments. The columns were packed with quartz sand

under saturated conditions, as described in Deshpande and

Shonnard (1999). For the packing procedure, constant water

saturation was maintained during packing by filling the

column with water and then step-wise introducing wet sand

through a funnel. For the aged filter sand columns, water

from the same slow sand filter, was used for this purpose.

Column parameters such as pore volumes, porosity, col-

umn volume and density were obtained from deionised

water-packed columns by relating column dimensions with

measured weights of wet and dried porous material. For these

calculations a sand density of 2.65 g mL�1 and water density

of 1 g mL�1 were assumed. For the calculation of the Peclet

number, Pe, the following equation was used: Pe ¼ vxL/DL.

Here, the flow distance, L, was chosen as the reference length

(length of the sand column), vx, is the average linear velocity

and DL is the longitudinal hydrodynamic dispersion coeffi-

cient (Fetter, 1998). Colloid transport parameters such as

deposition rate constants, normalized effluent concentra-

tions, particles average velocity and collision efficiencies were

obtained from colloid BTCs, as described earlier. Packed

column parameters can be observed in Table 1.

For both the quartz sand-packed and aged filter sand-

packed columns, Tygon tubing (Novodirect, Kiel, Germany)

was used to connect the columns to peristaltic pumps

(Ismatec SA, Switzerland), to a flow cell for online UV/VIS

measurements and to standard fraction collectors.

Previous results (Leon Morales et al., 2004) showed that at

low ionic strength conditions, laponite RD is highly mobile

through sand columns. As the salt concentration rises, the

laponite particles are destabilized and aggregation as well as

retention occurs. Also from the previous study, it was

observed that with the dominance of monovalent cations,

detachment of biofilm cells will occur after a drastic decrease

in ionic strength. It was therefore necessary to gradually

lower the influent’s ionic strength to a point where transport

of an injected colloidal pulse could occur. This was done in a

series of steps that were consistent for all columns. After the

biofilm growth period of 1, 2 or 3 weeks (as described in

Section 2.4); the nutrient medium as influent, was replaced by

a high ionic strength (70 mM) salt solution (CaCl2 or NaCl).

This was done in order to: (i) eliminate as many suspended

particles as possible before the colloid pulse experiments and

(ii) expose the system to either monovalent or divalent

cations. After 20 pore volumes of high ionic strength

monovalent or divalent cation influent, a constant and low

plateau (UV/VIS) of eluted particles was observed in all

conditions used. The ionic strength in the influent was then

halved, and run for three pore volumes, followed by deionised

water as column influent for two further pore volumes.

Following the injection of the colloidal pulse (0.6 mL), the

BTCs of laponite RD were recorded by measuring fluorescence

of rhodamine 6G–laponite RD complexes with time. Transport

patterns were extracted from these BTCs as previously

described. Sterile columns were run at the same described

conditions, as controls.

After packing, aged filter sand columns (colloid remobiliza-

tion experiments) were left stagnant overnight before intro-

ducing the high ionic strength solutions described earlier. The

ionic strength of the slow sand filter influent water in situ was

around seven times lower. After around 20 pore volumes,

when the suspended colloid concentration was low and

stable (as judged by the optical signal), the ionic strength of

the influent was decreased step-wise (five pore volume steps)

until deionised water was allowed to flow through the

columns. The remobilization patterns of natural colloid

particles were obtained by monitoring the effluent of these

columns using UV/VIS spectrometry. Calcium chloride and

sodium chloride were used with the aim of keeping constant

Cl� as a counter ion.

All described experiments were run at least in triplicate and

mean values are presented. Differences of these means are

considered significant at a 95% confidence level. Pearson

correlation coefficients, r, were considered only significant for

po0.05 (t-test distributions).

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 2 0 5 9 – 2 0 6 82062

2.3. Hydraulic conductivity measurements

The saturated hydraulic conductivity of sterile and biofilm

containing columns was measured using a constant head

permeameter. The permeameter consisted of a constant head

reservoir connected to the bottom of a sand column from

which the discharge was measured by quantifying the volume

of liquid flowing over a period of time. Depending on the

magnitude of the discharge, the change in hydraulic head,

delta h, could be altered simply by changing the height of the

influent reservoir above the discharge point. Darcy’s law was

then used (as depicted in the theoretical background section)

to calculate hydraulic conductivity.

2.4. Bacterial inoculum and nutrient media

Pseudomonas aeruginosa SG81 was used as a model biofilm

forming microorganism. P. aeruginosa SG81 is a mucoid, Gram-

negative, motile bacterium which was originally isolated from

a technical water system (Grobe et al., 1995). In agar plates,

this microorganism produces slimy and shiny colonies. This

‘‘slime’’ is composed of EPS including polysaccharides and

proteins (Wingender et al., 2001). This strain especially

produces copious amounts of the acidic polysaccharide

alginate. Biofilm developing conditions inside sand columns

have not yet been characterized for this organism. The

organism was maintained and pre-grown on Pseudomonas

Isolation Agar (DifcoTM) which by the addition of glycerine

stimulates alginate production. Fresh cultures were then

transferred and grown in Tryptic Soy Broth (TSB) (Merck,

Darmstadt). The columns were inoculated with one pore

volume of logarithmic phase Pseudomonas liquid cultures. The

inoculated columns were left stagnant overnight, then they

were constantly fed with modified alginate promoting med-

ium APM50 consisting of 50 mM Sodium gluconate, 1 mM

KNO3, 0.1 mM MgSO4 � 7H2O, 0.05% yeast extract and 0.2 M

NaCl.

2.5. Effluent cell concentrations and bacterial massdetermination

Bacterial cell detection in the column effluent was carried out

either online by measuring optical density at 240 nm or offline

by the determination of total cell counts (TCC) using Thoma

cell chambers in collected fractions. For the optical measure-

ments, bacterial numbers were calibrated from known

bacterial numbers suspended in the same buffer or nutrient

media as the one being injected to the column. Optical

density was measured using a UV/VIS spectrophotometer

(Cary 50, Varian Inc.). Column effluent fractions were

obtained using a fraction collector (Pharmacia FRAC-100).

Dry cell mass determination was carried out in accordance

with DIN EN 12880. The pellet dry weight was divided by the

total number of cells, obtaining the single bacterial dry weight

used for the calculations.

2.6. Biofilm growth determination and analyses

Column biofilm content was characterized by determining

cell numbers, and EPS contents. EPS were represented by total

extracellular carbohydrates and total extracellular proteins.

Biofilm quantification was done before and after the colloid

transport experiments. For biofilm quantification before the

colloid transport experiments, separate columns were

packed, inoculated and fed as described. After 1, 2 and 3

weeks of biofilm growth, the nutrient influent was replaced by

the 70 mM Na+ solution. The column material was extruded

and analysed, before the ionic strength reduction steps.

Biofilm content from the aged filter sand material was

determined by EPS content only.

EPS were extracted from the sand matrices by mechanical

shear force (stomachers 400 circulator) in the presence of a

cation exchange resin (DOWEX 50�8, Fluka ref.nr. 44445) in a

modified version of the procedure described by Frolund et al.

(1996). For this study, the initial settling step was not

necessary because the column material was transferred from

the column directly to the circulator without excess of fluid.

TCC were obtained directly from the previously extracted

material by taking aliquots of the well-mixed supernatants,

diluting if necessary and counting in Thoma chambers, using

phase contrast microscopy. The biochemical EPS analysis was

conducted in accordance with the protocol described in

Wingender et al. (2001) for total EPS protein and carbohydrate

fraction. All analyses were carried out with analytical grade

chemicals.

3. Results

3.1. Biofilm formation inside sand columns

Changes in sand column hydraulic conditions provide

indirect evidence of both biofilm formation and its possible

influence on colloid transport. Darcy’s law was used to obtain

the saturated hydraulic conductivity of sand columns at

different times after inoculation with P. aeruginosa SG81. This

parameter was measured constantly during the biofilm

growth period. Results indicate that after an instability

period, the measured hydraulic conductivity decreased stea-

dily until a constant lower plateau was reached for the

remainder of the observation period. This plateau was

observed after approximately 7 days of constant high nutrient

load influent (Fig. 1).

In non-inoculated columns a reduction in saturated hy-

draulic conductivity was observed after the first hours of

column packing. The measured hydraulic conductivity re-

mained more or less stable at a higher value than obtained

with the biofilm containing columns (Fig. 1).

Depicted in Fig. 2 are EPS concentrations and cell numbers

from biofilms after 1, 2 and 3 weeks of constant nutrient

influent. The cell/carbohydrate ratio (cell mass in mg was

obtained as described in Section 2.5) increased with time,

from a value of approximately 5 after 1 week of growth to

approximately 8 after 3 weeks of biofilm growth. Although the

cell/protein ratio was relatively high for all weeks, it

decreased with time, from approximately 19 in the first week,

to around 14 in the third week. The carbohydrate/protein ratio

decreased from around 4 in the first week, to a value of

approximately 2 in the third week of growth.

ARTICLE IN PRESS

Fig. 2 – Biofilm components (cells and EPS) in sand columns

before colloid transport experiments. Cell numbers are

represented in the right y-axis.

Fig. 3 – Laponite RD breakthrough through columns in

which Ca2+ ions were predominant before the

establishment of low ionic strength conditions by replacing

the salt influent with deionised water after 1, 2 or 3 weeks of

biofilm growth (means7SE; n ¼ 3). Also plotted, are sterile

controls, at the same electrochemical conditions

(means7SE; n ¼ 9).

Fig. 1 – Variation in saturated hydraulic conductivity with

time for media (sand F34, 0.2 mm diameter) sustaining the

growth of P. aeruginosa SG81 biofilms. The columns were

constantly fed with high nutrient load (50 mM Na-D-

gluconate+0.1 M NaCl).

WAT E R R E S E A R C H 41 (2007) 2059– 2068 2063

3.2. Influence of biofilms on inorganic colloid transportthrough sand columns

In sterile columns, the laponite RD colloid was transported

through both Na+ and Ca2+-treated columns using deionised

water as the background solution. However, a significant

increase in collision efficiency, calculated from the BTCs

obtained (Figs. 3 and 5), was observed in the Ca2+ columns

(Table 2) compared with the Na+ columns. For each of the

treatments, Ca2+ and Na+, there was incomplete colloid

retention in the sterile columns.

In non-sterile columns, biofilms developed increasingly

with time after constant nutrient influent. Two types of

effects on colloid transport were observed depending on the

dominant ionic species present during the ionic strength

reduction steps. The BTCs obtained from Na+-treated col-

umns showed no retention of the rhodamine 6G–stained

laponite RD. When compared with BTCs obtained in the

sterile counterparts, there is a significant increase in the area

under the curves after 2 and 3 weeks of biofilm growth.

Different degrees of tailing were also evident (Fig. 5).

In contrast, in Ca2+-treated columns, it was found that

colloid retention (measured as collection of breakthrough

patterns and collision efficiencies) was proportional to biofilm

age (Table 2, Fig. 3). EPS analyses indicate that colloid

retention also increased with an increase of the EPS content,

especially proteins remaining in the sand columns after the

colloid transport experiments (Fig. 6 compared to Fig. 3). At

the established confidence level, the correlation between

carbohydrates and colloid retention (0.917) was not signifi-

cant. The correlation between proteins and colloid retention

was significant (0.999) as was the correlation between cell

concentration and colloid retention (0.998).

In terms of the ratio of EPS components remaining in sand

columns after the colloid transport experiments, in the Na+-

treated columns, cell/carbohydrate, cell/protein and carbohy-

drate/protein ratios remained constant even after the longest

biofilm growing period of 3 weeks (Fig. 6A). In contrast, in

Ca2+-treated columns, the cell/carbohydrate ratio increased

from 2 to 3, the cell/protein ratio decreased from 16 to 3 and

the carbohydrate/protein ratio decreased from 7 to 1 (Fig. 6B).

The ratio of total EPS plus cells remaining in Ca2+-treated

columns divided by total EPS plus cells remaining in Na+-

treated columns was 1.5 for the first week, 1.9 for the second

week and 2.2 for the third week.

A pulse of laponite as described earlier, was injected

into aged filter sand-packed columns. The BTCs recorded

from these experiments, as well as, total cell numbers, are

plotted in Fig. 4. In agreement with results from the quartz

sand-packed columns, colloid retention is evident in Ca2+

exposed columns (Fig. 5). The EPS data obtained from aged

filter sand columns after these pulse experiments can be seen

in Fig. 6A and B.

ARTICLE IN PRESS

Table 2 – Transport parameters obtained from the colloid transport experiments

Week Cation C/C0 kd (s�1) Colloid travel time (s) Collision efficiency

1 Ca2+ 0.57a 8.2� 10�4a 688a 0.1a

Na+ 0.79a 4.6� 10�4a 516a 0.06a

2 Ca2+ 0.29a 1.8� 10�3a 689a 0.22a

Na+ 0.99a 1.3� 10�5a 516a 0.002a

3 Ca2+ 0.02a 7.5� 10�3b 532b 0.9b

Na+ 0.90a 2.1� 10�4a 492a 0.03a

Sterile columns Ca2+ 0.65a 6.8� 10�4b 654a 0.08b

Na+ 0.79a 4.4� 10�4b 547a 0.05b

Standard error less or equal to: (a) 10% and (b) 20%.

Fig. 5 – Laponite RD breakthrough in columns in which Na+

ions were predominant before the establishment of low

ionic strength conditions by replacing the salt influent with

deionised water after 1, 2 or 3 weeks of biofilm growth

(means7SE; n ¼ 3). Also plotted, are sterile controls at the

same electrochemical conditions (means7SE; n ¼ 9).

Fig. 4 – Breakthrough patterns of laponite RD when injected

in columns packed with aged filter sand at the same

conditions as established for the quartz sand systems.

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 2 0 5 9 – 2 0 6 82064

3.3. Remobilization of natural colloids including bacteriafrom aged filter sand-packed columns

Figs. 7 and 8 show the remobilization patterns of natural colloids

after Na+ and Ca2+ influence, respectively, from columns packed

with aged filter sand. After a period of stabilization, the columns

influent was changed as described in Section 2. In the case of Na+

columns, colloid remobilization occurred after every step-

decrease in ionic strength (Fig. 7). 40-6-diamidino-2-phenylindole

(DAPI—a DNA binding stain)-stained particles seen in the

micrographs show different morphologies and cell sizes con-

sistent with both bacteria and microalgae (diatoms). Increments

in particle concentrations in the effluents of these columns

coincide with the UV/VIS peaks observed (Fig. 7).

In the case of Ca2+ columns no remobilization of microorgan-

isms and other particles was observed during the treatment as

compared with Na+ columns. UV/VIS peaks were absent in

these columns and no increase in DAPI-stained particles was

observed during the ionic strength reduction steps. Remobiliza-

tion begins only after several pore volumes of deionised water

influent (Fig. 8). Particle size distributions (data not shown)

indicated that larger particles were mainly present in the first

fractions (at higher ionic strength) and smaller particles were

present in the middle or last fractions (lower ionic strength).

4. Discussion

4.1. Colloid transport through sand columns and theinteraction between ionic composition and biofilm presence

In biofilm containing columns, the exposure to Ca2+ or Na+

ions prior to decreasing ionic strength to very low levels had a

distinct impact on colloid transport as compared with sterile

columns.

ARTICLE IN PRESS

Fig. 6 – Biofilm components (cells and EPS) remaining in sand columns after colloid transport experiments. Note the steady

increase on EPS protein content with time. (A) After experiment and Na+ exposure. (B) After experiment and Ca2+ exposure.

Cells numbers are represented in the right y-axes.

Fig. 7 – Remobilization patterns of natural colloids from

aged filter sand-packed columns after step-wise decrease

in Na+ ions concentration. The micrographs show

morphology features observed by fluorescence of DAPI

stained particles from column effluents.

Fig. 8 – Remobilization patterns of natural colloids from

aged filter sand-packed columns after step-wise decrease

in Ca2+ ions concentration. Note no remobilization even at

low ionic strengths as compared with Na+-treated columns.

WAT E R R E S E A R C H 41 (2007) 2059– 2068 2065

In sterile columns (biofilm absence experiments), it was

observed that a two pore volume deionised water influent

resulted in considerable transport of laponite RD through

both Ca2+ and Na+-exposed columns. This probably means

that in both cases, the concentration of cations inside the

sand column was not high enough to cause colloid destabi-

lization and retention.

The influence of the ionic composition on the transport of

both natural and artificial colloids has been investigated

previously (Grolimund et al., 1998; Davis et al., 2001). It has

been observed that Ca2+ ions, in agreement with the DLVO

theory, decrease the stability of colloids much more effec-

tively than Na+ ions (Derjaguin and Landau, 1941). This has

been demonstrated for instance by comparing the electro-

phoretic mobility of colloids as a function of Na+ and Ca2+

concentrations. Ca2+ counterions result much more effective

in lowering particle electrokinetic charge than Na+ counter-

ions. In terms of retention in porous media, this can be

observed for example when comparing colloid deposition

rates in experiments with either type of cation (Grolimund

et al., 1998). In those experiments, substantial particle

deposition occurs at a much lower concentration of Ca2+

than Na+. Remobilization patterns observed in this study

confirm this sodium–calcium effect. This was observed not

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 2 0 5 9 – 2 0 6 82066

only in artificial model colloids such as laponite RD (data not

shown) but also in natural colloids (Figs. 7 and 8) such as the

ones remobilized from a column packed with aged filter sand.

Considering the heterogeneity of these natural colloids it is

evident that this type of ionic influence is not restricted to

certain specific colloids.

In contrast, colloid retention was found to be proportional

to biofilm presence and age when columns were subjected to

Ca2+ exposure. Ca2+ is known to directly influence the

mechanic and viscoelastic properties of biofilms (Koerstgens

et al., 2001). Ca2+ is important for the stability of the biofilm

matrix acting as a bridging and stabilization factor for the EPS

matrix components (Flemming et al., 2000). A decrease in

biofilm detachment is therefore expected when lowering of

ionic strength occurs in the presence of Ca2+ ions. This was

the case for older than 1 week biofilms (Fig. 3) and aged filter

sand (Figs. 4 and 8). Biofilm stability in this case is due to the

accumulation of Ca2+ ions and slower ion exchange pro-

cesses, resulting in colloid sorption and retention. On the

contrary, Na+ ions with a less important structural role (Mayer

et al., 1999) did not influence biofilm detachment processes.

This resulted in the observed alteration of colloid break-

through patterns (see Fig. 5). Alteration of laponite RD

breakthrough patterns after the presence of Na+ and at low

ionic strength, had been previously observed in other studies

(Leon Morales et al., 2004). The increased cell concentrations

in column effluents during colloid elution in those studies,

suggested a co-elution effect between the colloid and biofilm

cells. In the present study, biofilm influence on colloid

transport was further characterized. Here, biofilm age, EPS

components remaining in the columns before and after the

colloid transport studies, as well as, the type of cation

(monovalent or divalent) dominating the influent solution

were all taken into account. The use of fluorescence detection

of laponite–rhodamine complexes allowed a more accurate

determination of colloid transport patterns in the presence of

biofilms. These improvements not only in the detection

system but also on biofilm quantification, provided a clearer

Fig. 9 – Representation of the combined effect between the pre

transport of colloidal particles. The term collector refers to the s

deposited.

view of the interplay between biofilm formation and ionic

type and strength and their influence on colloid transport in

porous media.

A schematic representation of the influence of ionic

composition in the presence of biofilms as suggested from

the findings of this study is presented in Fig. 9. The thickness

of the dotted arrows represents the intensity of particle

mobilization. The combined calcium–biofilm influence on

colloid transport is notorious (shadowed square in biofilm

presence section) at electrochemical conditions in which

injected colloids normally would be mobile (shadowed square

in biofilm absence section).

4.2. Impact of biofilm architecture and EPS composition oncolloid transport

The influence of biofilm architecture on how substances and

particles are retained and transformed has been reported in

the literature (Okabe et al., 1998; Stoodley et al., 1999; Battin et

al., 2003). It has been shown that the presence of convective

flow paths is important for the transport of fluorescent latex

microbeads into biofilms (Okabe et al., 1998). These observa-

tions explained rapid particle transfer rates from the bulk

fluid into the biofilm which were not explained in previous

diffusion-only biofilm models (Characklis and Marshall, 1990).

An important factor determining different biofilm architec-

tures is the production and accumulation of EPS (Kuehn et al.,

2001). Extracellular polysaccharides, for example, due to their

structural characteristics and production rates on many

biofilms have been considered important on this respect

(Sutherland, 2001). The development and architecture of P.

aeruginosa SG81 sand-grown biofilms are very different when

compared to agar-grown biofilms. When grown on agar

plates, this organism formed highly mucoid microbial colo-

nies, which developed into confluent biofilms several milli-

metres thick. This phenotype was not observed in fully

packed sand-grown biofilms, the type used in this study.

sence of monovalent, divalent cations and biofilms on the

olid phase, e.g., sand grains, to which colloids can be

ARTICLE IN PRESS

WAT E R R E S E A R C H 41 (2007) 2059– 2068 2067

However, in separate experiments, when the biofilms were

grown in columns simulating a river sediment bed, i.e.,

horizontally placed, half filled with sediment and with liquid

flowing on top, these developed a highly mucoid state at the

water body–sediment interface. The relatively low amount of

polysaccharides produced and the high ratio of cells/EPS

found in our sand columns with time (Fig. 2), may be related

to the influence of the porous environment on biofilm

development and architecture. It was found that despite a

relatively low EPS production, biofilm development decreased

the hydraulic conductivity of the sand columns (Fig. 1). Even

though this rate of decrease was constant during the first 7

days of biofilm growth, a minimum permeability persisted

after this period of time. These findings suggest a biofilm

accumulation plateau, which allowed the maintenance of a

minimum nutrient transport within the system.

As the results indicate, EPS carbohydrate content is not

significantly related to the retention of colloids and colloid-

bound contaminants in porous media. The results show that

an incremental increase in EPS protein content, as well as cell

concentrations, with biofilm age coincided with colloid

retention. Normally, retention processes are attributed almost

exclusively to polysaccharides (Sutherland, 2001). Conditions

of low EPS production with relative high extracellular protein

contents are likely to be encountered in many natural and

engineered subsurface environments (Flemming and Win-

gender, 2003).

4.3. Significance for natural environments

Seawater infiltration, de-icing events and contamination, e.g.,

caused by hydrocarbon spilling accidents or collapsing of

liners in landfills can result in drastic changes on the ionic

composition of ground waters. A migrating plume of organic

contaminants without a constant source, i.e., in a spilling

accident, will stimulate confluent biofilm development in the

places through which it is moving. Biofilm presence therefore

can result in considerable longer times for Ca2+ and other

divalent cations retention and exchange. This has obvious

implications on the prediction of colloid and colloid-bound

contaminant transport and on the stability and detachment

rates of subsurface biofilms. EPS determinations (Fig. 6) and

preliminary laponite RD transport experiments in columns

packed with aged filter sand (Fig. 4) are consistent with

findings in model quartz sand systems. It is hypothesized,

that the effect of Ca2+ on the remobilization patterns of

natural colloids might be importantly influenced by the

presence of biofilms both in oligotrophic and contaminated

environments.

5. Conclusions

Sterile/non-sterile colloid transport experiments through

sand columns demonstrated that:

�

Biofilm formation increased the retention of colloids after

the ionic strength is decreased in Ca2+ presence. This also

results in:J increased stability of biofilms in the sand matrix,

J delayed colloid remobilization compared to sterile

systems.

�

Decreasing ionic strength in the presence of Na+ resulted

in:J decreased stability of biofilms,J alteration of colloid transport patterns rather than

retention processes.

�

Even at low EPS concentrations biofilms alter not only the

hydraulic conditions but also the transport of colloidal

particles in subsurface environments.

�

EPS analysis showed that the production of extracellular

proteins coincided with increased colloid retention in

biofilm columns.

�

The role of biofilms in the subsurface has to be taken into

account in the predictions of colloid transport.

Acknowledgements

This work forms part of the cooperative research project

KORESI—‘‘Kolloidaler Stofftransport bei der Regenwasserver-

sickerung—financed by the German Research Society (DFG).

We thank Professors U. Forstner, F.H. Frimmel and P.A.

Wilderer, and their staff, for constructive input and criticisms.

R E F E R E N C E S

Battin, T.J., Sengschmitt, D., 1999. Linking sediment biofilms,hydrodynamics, and river bed clogging: evidence from a largeriver. Microb. Ecol. 37, 185–196.

Battin, T.J., Wille, A., Sattler, B., Psenner, R., 2001. Phylogenetic andfunctional heterogeneity of sediment biofilms along environ-mental gradients in a glacial stream. Appl. Environ. Microbiol.67 (2), 799–807.

Battin, T.J., Kaplan, L.A., Newbold, D.J., Hansen, C.M.E., 2003.Contributions of microbial biofilms to ecosystem processes instream mesocosms. Nature 426, 439–441.

Bouwer, E., Rijnaarts, H.H.M., Cunningham, A.B., Gerlach, R., 2000.Biofilms in porous media. In: Bryers, J.D. (Ed.), Biofilms II:Process Analysis and Applications. Wiley-Liss, pp. 123–158.

Brummer, I.H.M., Fehr, W., Wagner-Dobler, I., 2000. Biofilmcommunity structure in polluted rivers: abundance of domi-nant phylogenetic groups over a complete annual cycle. Appl.Environ. Microbiol. 66 (7), 3078–3082.

Characklis, W.G., Marshall, K.C. (Eds.), 1990. Biofilms. Wiley Seriesin Ecological and Applied Microbiology. Wiley, New York.

Chen, B.M., Kojouharov, H.V., 1998. Accurate numerical simula-tion of biobarrier formation in porous media. In: Conferenceon Hazardous Waste Research. Snowbird, UT, USA.

Cherrey, K.D., Flury, M., Harsh, J.B., 2003. Nitrate and colloidtransport through coarse Hanford sediments under steadystate, variably saturated flow. Water Resour. Res. 39 (6), 1–10.

Corapcioglu, M.Y., Jiang, S., Kim, S.-H., 1999. Comparison ofkinetic and hybrid-equilibrium models simulating colloid-facilitated contaminant transport in porous media. Transp.Porous Media 36, 373–390.

Davis, C.J., Eschenazi, E., Papadopoulos, K.D., 2001. Combinedeffects of Ca2+ and humic acid on colloid transport throughporous media. Colloid Polym. Sci. 280 (1), 52–58.

Derjaguin, B.V., Landau, L., 1941. Theory of stability of stronglycharged lyophobic solutions and of the adhesion of stronglycharged particles in solutions of electrolytes. Acta Physico-chim. USSR 14, 346–354.

ARTICLE IN PRESS

WAT E R R E S E A R C H 4 1 ( 2 0 0 7 ) 2 0 5 9 – 2 0 6 82068

Deshpande, P.A., Shonnard, D.R., 1999. Modeling the effects ofsystematic variation in ionic strength on the attachmentkinetics of Pseudomonas fluorescens. Water Resour. Res. 35 (5),1619–1627.

Fetter, C.W., 1998. Contaminant Hydrogeology, second ed. Pre-ntice Hall, Upper Saddle River, New Jersey, USA, pp. 57–74.

Flemming, H.C., 1995. Sorption sites in biofilms. Water Sci.Technol. 32 (8), 27–33.

Flemming, H.C., Wingender, J., 2003. The crucial role of extra-cellullar polymeric substances in biofilms. In: Wuertz, S.,Bishop, P., Wilderer, P. (Eds.), Biofilms in Wastewater Treat-ment. An Interdisciplinary Approach. IWA Publishing, London.

Flemming, H.-C., Wingender, J., Mayer, C., Korstgens, V., Borchard,W., 2000. Cohesiveness in biofilm matrix polymers. In: Allison,D.G., Gilbert, P., Lappin-Scott, H.M., Wilson, M. (Eds.), Com-munity Structure and Cooperation in Biofilms. CambridgeUniversity Press, Cambridge, pp. 87–105.

Frolund, B., Palmgren, R., Keiding, K., Nielsen, P.H., 1996. Extrac-tion of extracellular polymers from activated sludge using acation exchange resin. Water Res. 30 (8), 1749–1758.

Grobe, S., Wingender, J., Truper, H.G., 1995. Characterization ofmucoid Pseudomonas aeruginosa strains isolated from technicalwater systems. J. Appl. Bacteriol. 79, 94–102.

Grolimund, D., Borkovec, M., Bartmettler, K., Sticher, H., 1996.Colloid-facilitated transport of strongly sorbing contaminantsin natural porous media: a laboratory column study. Environ.Sci. Technol. 30, 3118–3123.

Grolimund, D., Elimelech, M., Borkovec, M., Barmettler, K.,Kretzschmar, R., Sticher, H., 1998. Transport of in situmobilized colloidal particles in packed soil columns. Environ.Sci. Technol. 32, 3562–3569.

Grolimund, D., Elimelech, M., Borkovec, M., 2001. Aggregation anddeposition kinetics of mobile colloidal particles in natural porousmedia. Colloids Surf. A: Physicochem. Eng. Aspects 191, 179–188.

Kim, S., Corapcioglu, M.Y., 1997. The role of biofilm growth inbacteria-facilitated contaminant transport in porous media.Transp. Porous Media 26, 161–181.

Koerstgens, V., Flemming, H.C., Wingender, J., Borchard, W., 2001.Influence of calcium ions on the mechanical properties of amodel biofilm of mucoid Pseudomonas aeruginosa. Water Sci.Technol. 43 (6), 49–57.

Kretzschmar, R., Sticher, H., 1998. Colloid transport in naturalporous media: influence of surface chemistry and flowvelocity. Phys. Chem. Earth 23 (2), 133–139.

Kretzschmar, R., Borkovec, M., Grolimund, D., Elimelech, M., 1999.Mobile subsurface colloids and their role in contaminanttransport. Adv. Agronomy 66, 121–193.

Kuehn, M., Mehl, M., Hausner, M., Bungartz, H.-J., Wuerty, S., 2001.Time-resolved study of biofilm architecture and transportprocesses using experimental and simulation techniques: therole of EPS. Water Sci. Technol. 43 (8), 143–151.

Leon Morales, C.F., Leis, A.P., Strathmann, M., Flemming, H.C.,2004. Interactions between laponite and microbial biofilms inporous media: implications for colloid transport and biofilmstability. Water Res. 38 (16), 3614–3626.

Mayer, C., Moritz, R., Kirschner, C., Borchard, W., Maibaum, R.,Wingender, J., Flemming, H.C., 1999. The role of intermole-cular interactions: studies on model systems for bacterialbiofilms. Int. J. Biol. Macromol. 26 (1), 3–16.

Okabe, S., Kuroda, H., Watanabe, Y., 1998. Significance of biofilmstructure on transport of inert particulates into biofilms.Water Sci. Technol. 38 (8–9), 163–170.

Orvain, F., Galois, R., Barnard, C., Sylvestre, A., Blanchard, G.,Sauriau, P.-G., 2003. Carbohydrate production in relation tomicrophytobenthic biofilm development an integrated ap-proach in a tidal mesocosm. Microb. Ecol. 45, 237–251.

Stoodley, P., Boyle, J.D., DeBeer, D., Lappin-Scott, H.M., 1999.Evolving perspectives of biofilm structure. Biofouling 14 (1),75–90.

Sutherland, I., 2001. Biofilm exopolysaccharides: a strong andsticky framework. Microbiology 147 (Part 1), 3–9.

Tapia Estevez, M.J., Lopez Arbeloa, F., Lopez Arbeloa, T., LopezArbeloa, I., 1993. Absorption and fluorescence properties ofrhodamine 6G adsorbed on aqueous suspensions of wyomingmontmorillonite. Langmuir 9, 3629–3634.

Tatalovich, M.E., Lee, K.Y., Chrysikopoulos, C.V., 2000.Modeling the transport of contaminants originating fromthe dissolution of DNAPL pools in aquifers in the presenceof dissolved humic substances. Transp. Porous Media 38,93–115.

Vandevivere, P., Baveye, P., 1992. Effect of bacterial extracellularpolymers on the saturated hydraulic conductivity of sandcolumns. Appl. Environ. Microbiol. 58 (5), 1690–1698.

Wingender, J., Neu, T.R., Flemming, H.-C., 1999. What are bacterialextracellular polymeric substances? In: Wingender, J., Neu,T.R., Flemming, H.-C. (Eds.), Microbial Extracellular PolymericSubstances. Springer, Berlin, pp. 1–19.

Wingender, J., Strathmann, M., Rode, A., Leis, A., Flemming, H.-C.,2001. Isolation and biochemical characterization of extracel-lular polymeric substances from Pseudomonas aeruginosa.Methods Enzymol. 336 (25), 302–314.

Yao, K.M., Habibian, M.T., O’Melia, C.R., 1971. Water and wastewater filtration—concepts and applications. Environ. Sci.Technol. 5, 1105–1112.