Water Research 37 (2003) 4863–4872

ARTICLE IN PRESS

*Correspond

217-333-6968.

E-mail addr1 Current ad

Yale University

06510.

0043-1354/$ - se

doi:10.1016/j.w

Pore blockage effect of NOM on atrazine adsorption kineticsof PAC: the roles of PAC pore size distribution and NOM

molecular weight

Qilin Lia,1, Vernon L. Snoeyinka,*, Benito J. Marinasa, Carlos Camposb

a Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, 205 N Mathews Avenue, Urbana,

IL 61801, USAb Suez Environment-CIRSEE, 38 rue du pr!esident Wilson, 78230 Le Pecq, France

Received 28 August 2002; received in revised form 22 July 2003; accepted 14 August 2003

Abstract

Natural organic matter (NOM) in natural water has been found to have negative effects on the adsorption of various

trace organic compounds by activated carbon through two major mechanisms: direct competition for sites and pore

blockage. In this study, the pore blockage effect of NOM on atrazine adsorption kinetics was investigated. Two types of

powdered activated carbon (PAC) and three natural waters were tested to determine the roles of PAC pore size

distribution and NOM molecular weight distribution in the pore blockage mechanism. When PAC was preloaded with

natural water, the pore blockage effect of the NOM was found to cause a reduction of up to more than two orders of

magnitude in the surface diffusion rate of atrazine compared to simultaneous adsorption of atrazine and NOM with

fresh PAC. The surface diffusion coefficient of atrazine for preloaded PAC decreased with a decrease in PAC dose or an

increase in NOM surface concentration. Because of the pore blockage effect of NOM, a 30% drop in atrazine removal

was observed in a continuous flow PAC/microfiltration (MF) system after 7 days of contact compared to the removal

predicted from the batch isotherm test. Large micropores and mesopores were found to play an important role in

alleviating the effect of pore blockage. A PAC with a relatively large fraction of large micropore and mesopores was shown

to suffer much less from the pore blockage effect compared with a PAC that had a much smaller fraction of large pores.

Natural waters with different NOM molecular weight distribution caused different extent of pore blockage. The NOM

molecules with molecular weight between 200 and 700 Dalton appeared to be responsible for the pore blockage effect.

r 2003 Elsevier Ltd. All rights reserved.

Keywords: Powdered activated carbon; Natural organic matter; Atrazine; Pore blockage; Molecular weight distribution; Pore size

distribution

1. Introduction

Activated carbon has been widely used in drinking

water treatment to remove dissolved organic com-

ing author. Tel.: +1-217-333-4700; fax: +1-

ess: snoeyink@uiuc.edu (V.L. Snoeyink).

dress: Department of Chemical Engineering,

, 9 Hillhouse Avenue, New Haven, Connecticut

e front matter r 2003 Elsevier Ltd. All rights reserve

atres.2003.08.018

pounds, including background natural organic matter

(NOM) and a number of synthetic trace organic

compounds. However, due to a limited understanding

of the mechanisms of competitive adsorption, there is

yet the need for the development of an accurate model

that will predict trace organic compound adsorption

from natural water. Previous studies often use the ideal

adsorbed solution theory (IAST) [1] to model the

adsorption equilibrium of a multiple solute system,

which assumes equal access of all adsorbates to all

adsorption sites and interaction of adsorbates through a

d.

ARTICLE IN PRESSQ. Li et al. / Water Research 37 (2003) 4863–48724864

single mechanism: direct competition for adsorption

sites. It has been realized only recently that NOM affects

trace organic compound adsorption not only by directly

competing for adsorption sites but also by blocking

carbon pores [2–5].

A common example of the pore blockage effect of

NOM in continuous flow adsorption systems is the

‘‘preloading’’ phenomena found in fixed bed GAC

columns. Due to their slower adsorption kinetics,

NOM compounds move down the column faster than

trace organic compounds. As a result, the GAC at the

effluent end of the bed is preloaded with NOM before it

is contacted with the trace organic molecules. In fact, all

the adsorption systems in which activated carbon is held

in the system while water continuously flows through are

vulnerable to the pore blockage effect. In these systems,

including granular activated carbon (GAC) adsorbers,

floc blanket reactors with powdered activated carbon

(PAC) addition, and PAC/membrane systems, the

activated carbon is partially loaded with NOM and

trace organic compounds before more trace organic

compounds enter the system, resulting in pore blockage

when the surface concentration of NOM gets high

enough. The ‘preloading’ effect in GAC adsorbers is

usually attributed to occupation of adsorption sites by

NOM or to pore blockage by NOM that makes

adsorption sites in small pores practically unavailable

to trace organic compounds [6–15]. Therefore, most

competitive adsorption research on the preloading effect

has focused primarily on its impact on the adsorption

equilibrium of trace organic compounds [2–5,15,16].

However, a recent mechanistic study by Li et al. [17]

showed that, in addition to practically reducing adsorp-

tion capacity, pore blockage by large molecules had a

strong effect on the adsorption kinetics of atrazine. It

was also found by Lebeau et al. [18] that both

adsorption capacity and the surface diffusion coefficient

of atrazine decreased as the PAC age in an immersed

microfiltration system increased. The result is that

higher carbon doses are required to achieve given

treatment goals compared to those predicted by batch

isotherm and kinetic tests using fresh carbon.

Since adsorption equilibrium is rarely reached in

continuous flow adsorption systems such as PAC/

membrane reactors and GAC columns, it is critical to

fully understand the effect of NOM on adsorption

kinetics of a trace organic compound so that mean-

ingful, accurate design tools may be developed. How-

ever, NOM in different natural waters varies so much

that it can have dramatically different competitive

adsorption effects [19]. One of the most important

characteristics of NOM that affects adsorption is

molecular weight. Moreover, the heterogeneity of

activated carbon surface is also an important factor in

competitive adsorption. The pore size distribution of

activated carbon relative to the molecular sizes of

adsorbates (e.g. trace organic compounds and NOM)

has been found to determine the dominant mechanism

of competitive adsorption [15,17,20,21]. Therefore, an

accurate evaluation of the roles of NOM molecular

weight distribution (MWD) and PAC properties is

necessary for optimizing system design and operation

to achieve maximum pollutant removal.

The objectives of this study are to: (1) demonstrate the

pore blockage effect of NOM on adsorption kinetics of

trace organic compounds in batch as well as in

continuous flow systems; (2) determine the effect of

NOM surface loading on adsorption kinetics of trace

organic compounds; (3) evaluate the roles of NOM

molecular weight and carbon pore size distributions in

pore blockage. This information is needed to improve

our understanding of the competitive effect of NOM,

which is necessary for better modeling of competitive

adsorption of trace organic compounds in natural water.

The results from this study will also help water utilities

choose the best adsorbent based on NOM characteristics

and the physical/chemical properties of the target

compound.

2. Materials and methods

2.1. Water

Organic-free water was obtained by passing deionized

water through a NANOpure ultrapure water system

(Barnstead, Dubuque, Iowa). The dissolved organic

carbon (DOC) concentration of the organic-free water is

lower than 0.3 mg/L. A central Illinois ground water

(referred to as GW) and two surface waters were used to

study the effect of NOM. The GW was taken from the

source immediately before use. It was treated with a

greensand filter to remove dissolved iron and manganese

and filtered through a 0.45 mm-pore-size microfiltration

cartridge to remove particles. The surface waters were

both taken from Lake Decatur, Decatur, IL. One was

taken shortly before it was used (referred to as FLDW)

and the other was taken and stored for one year before

use (referred to as DLDW). Both waters were stored at

4�C and passed through a 0.45mm-pore-size nylon

membrane filter to remove particulate matter immedi-

ately before use. The DOC concentrations of all three

natural waters were determined using a Phoenix 8000

TOC Analyzer (Tekmar-Dohrmann, Cincinnati, OH).

The initial DOC concentrations were 2.770.2 mg/L for

GW, 4.470.2 mg/L for FLDW and 3.070.0 mg/L for

the DLDW.

2.2. Adsorbents

Two types of commercial PAC, PAC A (Calgon

Carbon Corp., Pittsburgh, PA) and PAC B (NORIT

ARTICLE IN PRESS

Table 1

Relevant characteristics of PACs A and B [17]

Property PAC A PAC B

BET surface area (m2/g) 903 1110

Primary micropore (o8 (A)

volume (cm3/g)

0.278 0.343

Secondary micropore (8–20 (A)

volume (cm3/g)

0.142 0.194

Mesopore (20–500 (A) volume

(cm3/g)

0.072 0.357

Micropore surface area (m2/g) 888 733

Mesopore surface area (m2/g) 15 379

Q. Li et al. / Water Research 37 (2003) 4863–4872 4865

France, S.a.r.l., Le Blanc Mesnil Cedex, France), were

used to demonstrate the effect of pore size distribution

on competitive adsorption mechanisms. Important pore

size distribution parameters of these carbons, obtained

using the N2 adsorption equilibrium technique [22], are

presented in Table 1.

2.3. Adsorbates

Atrazine was used as the target adsorbate. The stock

solution was prepared by mixing 14C-labeled atrazine

(Sigma, St. Louis, MO) and non-labeled atrazine (Chem

Service, Chester, PA) to yield a specific radioactivity of

33.68mCi/mg. The solution was refrigerated until use.

Atrazine samples were analyzed with a liquid scintilla-

tion analyzer (Tri-Carb Model 1600A, Packard Instru-

ment Co., Downers Grove, IL). The analytical

procedures are described elsewhere [22]. The detection

limit was determined to be 0.03mg/L with a confidence

level of 95%.

2.4. MWD analysis of NOM

The MWD of NOM was measured with a high

performance liquid chromatography system (HP 1090,

Series II, Hewlett-Packard Co., Wilmington, DE) using

a size exclusion chromatography (SEC) column (Pro-

tein-PakTM 125, Waters Corporation, Milford, MA) at

40�C and a mobile phase flow rate of 1 mL/min. The

mobile phase consisted of 0.02 M orthophosphate at pH

6.8. NaCl was added to yield an ionic strength of 0.1 M.

Before the analysis, the mobile phase was degassed for at

least two hours to prevent air bubbles from entering the

SEC column. The calibration curve was generated using

poly(styrene sulfonate) (PSS) with molecular weight

(MW) ranging from 1800 to 35,000 Da and acetone

(MW=58 Da) as the standards. The UV absorbance of

PSS and acetone was monitored at 224 nm. When NOM

was analyzed, a 6 mL sample was taken and filtered

through a 0.22 mm nylon membrane filter. The first 4 mL

was wasted to eliminate error due to adsorption of

NOM on the membrane, and the final 2 mL was

transferred to a vial for analysis. The UV absorbance

of the sample was monitored at 260 nm.

2.5. Batch atrazine adsorption kinetic tests with fresh and

preloaded PAC

Atrazine adsorption kinetic tests were conducted

using PAC preloaded with natural water to determine

the effect of NOM preloading on adsorption kinetics of

atrazine. Different doses of fresh PAC were first

preloaded with natural water in 2 L square jars for 4

days to obtain different surface coverage. DOC samples

were taken after preloading to determine the surface

concentrations of NOM. Mixing was maintained when

the samples were taken so that the carbon dose remained

the same after sampling. Atrazine was then spiked into

the solution to obtain the desired initial concentration

and the change of atrazine concentration was monitored

for at least 4 h by taking samples at predetermined times.

Atrazine samples were also taken after 11 and 25 days of

contact to determine the equilibrium concentration of

atrazine, considering the slow adsorption kinetics caused

by NOM preloading. During the 25-day contact, the

square jars were covered to avoid concentration change

by evaporation. In parallel, atrazine adsorption kinetic

tests were also conducted using fresh PAC in natural

water to determine atrazine adsorption kinetics during

simultaneous adsorption of atrazine and NOM. The

experimental procedure is similar to that with the

preloaded PAC and was described elsewhere [22].

Adsorption tests conducted in natural water with fresh

PAC are referred to as simultaneous adsorption hereon.

It should be noticed that for the purpose of comparing

the results for preloaded and simultaneous adsorption

tests, the same total amount of NOM and atrazine,

including the corresponding fractions in liquid and solid

phases, per unit mass of carbon was used for both tests.

2.6. Continuous flow experiment

The continuous flow experiment was conducted in a

bench-scale PAC/MF system. The schematic of the

experimental set-up is shown in Fig. 1. The membrane

unit is a 400 mL Amicon stirred cell (Model 8400,

Amicon Inc., Beverly, MA). A 1mm-pore-size Nucle-

pores MF membrane (Corning, Acton, MA) was

installed at the bottom of the cell that allowed water

to pass while retaining the PAC. Prior to the experiment,

PAC was added to the cell in a slurry form. The pressure

release valve was closed when the cell was filled with

the influent water so that the influent water could be

pumped through the cell at a constant flow rate. The

effluent was continuously circulated back to the influent

tank, and the experiment was run for 7 days. Samples

were taken from the effluent at the end of the 7th day

ARTICLE IN PRESS

PAC

MF Membrane

Stirred cell

Influent reservoir Stirrer

plate

Effluent

Pressure release valve

Compact tubing pump

Fig. 1. Schematic of the bench-scale PAC/MF system.

Q. Li et al. / Water Research 37 (2003) 4863–48724866

and analyzed for atrazine to determine its adsorption

capacity under a continuous flow condition. A pre-

liminary test was conducted by contacting atrazine and

the MF membrane in organic-free water to determine

the atrazine adsorption capacity of the membrane.

3. Results and discussion

3.1. Effect of NOM preloading on atrazine adsorption

kinetics: batch experiments

Batch atrazine adsorption kinetic tests were con-

ducted using PAC preloaded with NOM in natural

water to determine the effect of NOM preloading on

atrazine adsorption kinetics. Doses of 4, 8, 12 and

16 mg/L of PAC A and 2, 4, 8.3 and 12 mg/L of PAC B

were preloaded in 2 L of fresh Lake Decatur water

(FLDW) for 4 days to obtain a range of NOM surface

loading. Following the same procedure, 4, 8, 12 and

20 mg/L of PAC A was preloaded with the one year old

Lake Decatur water (DLDW), and 2 and 4 mg/L of

PAC B was preloaded with the ground water (GW). The

surface concentration of NOM after preloading was

determined from the change of DOC concentration

before and after preloading. Then, atrazine was spiked

into each test jar to obtain initial concentrations of

10mg/L in FLDW and DLDW, and 5mg/L in GW. The

kinetic tests were run for at least 4 h. In parallel,

simultaneous adsorption kinetic tests were conducted

using 4 mg/L of each fresh PAC in all three natural

waters. Atrazine samples were taken after 11 and 25

days to determine its equilibrium concentrations. The

atrazine adsorption capacity of both carbons in GW was

determined in a previous study [22].

The Freundlich model (Eq. (1)) was used to describe

adsorption equilibrium of atrazine and the homoge-

neous surface diffusion model (HSDM) [23] was used to

describe atrazine adsorption kinetics.

qeq ¼ KC1=neq ; ð1Þ

where, Ceq(mg/L) is the equilibrium liquid phase

concentration, qeq(mg/mg) is the equilibrium solid

phase concentration; and K ((mg/mg)(L/mg)1/n) and 1/n

(dimensionless) are Freundlich parameters.

3.1.1. The pore blockage effect of NOM

Figs. 2 and 3 present atrazine adsorption kinetic

curves in FLDW for fresh and preloaded PAC A and

PAC B, respectively. For both PACs, preloading with

FLDW showed a dramatic impact on atrazine adsorp-

tion. For the same PAC dose, 4 mg/L, fresh PAC A and

PAC B achieved 44 and 60% atrazine removal,

respectively, at the end of the 4-hour test, while the

preloaded PACs only obtained 6% and 10% removal of

atrazine, respectively. This discrepancy was attributed to

the pore blockage effect of NOM on both atrazine

adsorption capacity and kinetics.

The adsorption capacity for atrazine was determined

from the equilibrium concentrations of atrazine. A slight

concentration decrease was measured between the 11th

and the 25th day of contact. Therefore, equilibrium was

assumed after contact of 25 days, and the equilibrium

concentrations of atrazine for both preloaded and fresh

PAC are shown in Fig. 4. Also shown in Fig. 4 are

atrazine equilibrium concentrations obtained in DLDW

with fresh and preloaded PAC A. It has been found that

simultaneous adsorption isotherms of atrazine in natural

water are parallel to those in organic-free water at

low concentrations, i.e. the 1/n values are the same

[22,26,27]. Therefore, the 1/n values obtained in organic-

free water were used to evaluate atrazine adsorption

capacity, K, for different doses of preloaded and fresh

PACs. The 1/n values used were 0.41 for PAC A and

0.44 for PAC B [22]. The lines shown in Fig. 4 are

calculated isotherms for the fresh PACs using these 1/n

values based on the data obtained at the carbon dose of

4 mg/L. As shown in Fig. 4, the adsorption capacities of

the preloaded PACs agree with those of the fresh PACs

at high PAC doses, i.e. low Ceq range, showing that

pore blockage by NOM may not affect atrazine

adsorption equilibrium although it seriously impact

ARTICLE IN PRESS

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 60 120 180 240

Time, min

Atr

azin

e re

mai

nin

g, C

/C0

Fresh PAC, 4mg/L

Preloaded PAC, 4mg/L

Preloaded PAC, 8mg/L

Preloaded PAC, 12mg/L

Preloaded PAC, 16mg/L

HSDM fit

Fig. 2. Atrazine adsorption kinetic curves of PAC A in FLDW

0.0

0.2

0.4

0.6

0.8

1.0

1.2

0 60 120 180 240 300

Time, min

Atr

azin

e re

mai

nin

g,C

/C0

Fresh PAC, 4mg/L

Preloaded PAC,2mg/L

Preloaded PAC,4mg/L

Preloaded PAC,8mg/L

Preloaded PAC,12mg/L

HSDM fit

Fig. 3. Atrazine adsorption kinetic curves of PAC B in FLDW.

0.1

1

10

0.1 1 10Equilibrium atrazine concentration, µg/L

Atr

azin

e ad

sorb

ed, µ

g/m

g

DLDW Preloaded PAC AFLDW preloaded PAC AFresh PAC A in DLDWFresh PAC A in FLDWFLDW preloaded PAC BFresh PAC B in FLDWFresh PAC A, calculatedFresh PAC B, calculated

PAC dose = 4 mg/L

Fig. 4. Atrazine adsorption equilibrium concentrations obtained in natural water (25-day contact).

Q. Li et al. / Water Research 37 (2003) 4863–4872 4867

ARTICLE IN PRESS

1.0E-13

1.0E-12

1.0E-11

1.0E-10

0 50 100 150 200

NOM surface concentration, mg/g

Ds

of

atra

zin

e, c

m2 /m

in

PAC A

PAC B

Fig. 5. Change of atrazine Ds with surface concentration of

NOM in FLDW.

Q. Li et al. / Water Research 37 (2003) 4863–48724868

atrazine adsorption kinetics. This observation is con-

sistent with the occurrence of reversible adsorption.

Since the total amounts of NOM and atrazine used in

the adsorption tests, including the fractions in the liquid

and the solid phases, was the same for the fresh and the

preloaded PAC, the same equilibrium condition was

reached under the reversible adsorption conditions.

However, when the PAC dose decreased (high Ceq

range), the preloaded PAC isotherms deviate from those

of the fresh PACs, as demonstrated by the higher liquid

phase concentration and lower solid phase concentra-

tion obtained at the same carbon dose of 4 mg/L (see

Fig. 4), even though a long contact time of 25 days was

used. This may have been caused by extremely slow

adsorption kinetics of atrazine in some pores when these

pores were severely blocked by a certain fraction of

NOM apparently not able to reach critical solid phase

concentration at higher PAC doses.

The surface diffusion coefficient, Ds, of atrazine was

determined by fitting the kinetic data shown in Figs. 2

and 3 with the pseudo-single solute HSDM [24]. The

pseudo-single solute HSDM treats the trace organic

compound as if it was the only adsorbate in the system,

and takes into account the effect of NOM by using

adsorption equilibrium and kinetic parameters obtained

in the corresponding natural water. The K and 1/n

values used in the pseudo-single solute HSDM were

evaluated as described above. The Ds values obtained

with the fresh PACs (simultaneous adsorption of

atrazine and NOM), 4.0� 10�11 cm2/min for PAC A

and 8.0� 10�11 cm2/min for PAC B, were similar to

those obtained in organic-free water, indicating a small

effect of pore blockage in simultaneous adsorption. This

is consistent with the findings of Li et al. [17] using

atrazine and model pore-blocking compound PSS-1.8k.

It has been found that the Ds of atrazine in natural

water is independent of PAC dose when fresh PAC is

used [22,24,25]. However, when the preloaded PAC was

used, the Ds of atrazine was not only much lower than

that of the fresh PAC due to pore blockage, but it also

changed greatly with the PAC dose. This is because the

surface concentration of NOM was different for

different PAC doses. The lower the carbon dose, the

higher the NOM surface concentration, and therefore

the greater the effect of pore blockage. The Ds values

corresponding to different surface concentrations of

NOM in FLDW are shown in Fig. 5 for both PAC A

and PAC B. It was found that the Ds of atrazine

decreased with a decrease in PAC dose or an increase in

NOM surface concentration, from 4.0� 10�11 to

7.5� 10�13 cm2/min for PAC A and from 8.0� 10�11

to 3.9� 10�13 cm2/min for PAC B. This result is in

agreement with that of a previous study when PAC was

preloaded with PSS-1.8k [22]. Apparently the preloaded

NOM molecules in FLDW blocked the openings of

some pores where atrazine could adsorb. When the

preloaded PAC was contacted with atrazine, the surface

diffusion of atrazine into the pores that were blocked

was much slower compared to that with fresh carbon

because atrazine molecules had to move around or even

displace NOM from the pore openings before they could

enter these pores.

3.1.2. The effect of PAC pore size distribution

Fig. 5 shows that the pore blockage effect is more

pronounced for PAC A than for PAC B. For the same

surface concentration of NOM, the Ds of atrazine for

PAC A was much lower than that for PAC B. This can

be explained by the difference in pore size distributions

of the two PACs. As shown in Table 1, PAC B has

379 m2/g of surface area (34% of the total) in

mesopores, and a higher fraction of micropore surface

area in the larger secondary micropores (actual micro-

pore surface area breakdown not shown), where most

NOM molecules adsorb, while PAC A only has 15 m2/g

(1.7% of the total) of mesopores and a lower fraction of

secondary micropores. Therefore, for the same surface

concentration of NOM (amount of NOM adsorbed per

unit mass of PAC), the actual surface coverage of NOM

(amount of NOM per unit accessible surface area) is

much lower for PAC B than for PAC A. And the actual

surface coverage of NOM determines how much it can

impede atrazine surface diffusion. Consequently, the

pore blockage effect caused by NOM molecules is much

more severe on PAC A than PAC B. From this result, it

is clear that a carbon with appropriate surface area

distribution in small and large micropores and meso-

pores will have both high adsorption capacity and fast

adsorption kinetics for atrazine, even when the carbon is

preloaded.

3.1.3. Effect of NOM molecular weight distribution

The effect of NOM with different MWDs on atrazine

adsorption kinetics was also investigated. Fig. 6 shows

atrazine adsorption kinetic curves of fresh PAC A in

DLDW and FLDW. Although the DOC content of

ARTICLE IN PRESSQ. Li et al. / Water Research 37 (2003) 4863–4872 4869

DLDW was 32% lower than that of FLDW, the same

atrazine adsorption curve was achieved in the two

waters. It is shown in Fig. 4 that the same atrazine

equilibrium concentration was reached in the two

waters, indicating that the fraction of NOM in FLDW

that competes for adsorption sites with atrazine was the

same as that in DLDW. Similar atrazine Ds values,

4.0� 10�11 and 4.7� 10�11 cm2/min, were obtained in

FLDW and DLDW, respectively, close to the value

obtained in organic-free water using PAC A. As

explained above, this is because the impact of pore

blockage in simultaneous adsorption is very small.

However, the NOM in the two waters had different

effect on atrazine adsorption kinetics when the PAC was

preloaded. Fig. 7 compares atrazine adsorption kinetic

curves of PAC A preloaded in DLDW and FLDW. As

found for FLDW, preloading PAC with DLDW showed

great impact of pore blockage. The adsorption kinetics

of the DLDW preloaded PAC A was significantly lower

than that of the fresh PAC A, and the Ds of atrazine

decreased with an increase in NOM surface concentra-

0.0

0.2

0.4

0.6

0.8

1.0

0 60 120 180 240 300

Time, min

Atr

azin

e r

em

ain

ing

, C

/C0

DLDW, PAC dose=4mg/L

FLDW, PAC dose=3.9mg/L

HSDM fit

Fig. 6. Atrazine adsorption kinetics of fresh PAC A in FLDW

and DLDW.

0.0

0.2

0.4

0.6

0.8

1.0

0 60 120 180

Time, min

Atr

azin

e re

mai

nin

g, C

/C0

Fig. 7. Atrazine adsorption kinetic curves of p

tion (see Fig. 8). However, the two waters were found to

cause a different extent of pore blockage on PAC A. As

shown in Fig. 7, preloading PAC A in FLDW resulted in

slower atrazine removal than preloading PAC A in

DLDW for the three PAC doses that were common to

both waters, 4, 8 and 12 mg/L, indicating that FLDW

caused more serious pore blockage than DLDW did.

Fig. 8 demonstrates that for the same surface concen-

tration, the NOM in FLDW caused more reduction in

atrazine Ds than the NOM in DLDW. Although the

one-year storage at 4�C did not cause much change in

the site-competing fraction as discussed above, it did

remove some NOM in FDLW that causes pore

blockage. Fig. 9 compares the MWDs of FLDW and

DLDW. The major difference between the two waters

was a fraction of NOM mostly in the MW range from

200 to 700 Da, indicating that the pore blocking NOM

was likely in this MW range. This hypothesis was

supported by an MWD analysis of the DLDW NOM

that was adsorbed on PAC A during preloading,

presented in Fig. 10. The MWD of NOM that was

240

FLDW, PAC dose=4mg/L

FLDW, PAC dose=8mg/L

FLDW, PAC dose=12mg/L

DLDW, PAC dose=4mg/L

DLDW, PAC dose=8mg/L

DLDW, PAC dose=12mg/L

DLDW, PAC dose=20mg/L

HSDM fit

reloaded PAC A in FLDW and DLDW.

1.0E-13

1.0E-12

1.0E-11

1.0E-10

0 20 40 60 80 100 120

DOC surface concentration, mg/g

Ds

of

atra

zin

e, c

m2 /m

in

FLDW preloading

DLDW preloading

Fig. 8. Effect of NOM in FLDW and DLDW on Ds of atrazine

for PAC A.

ARTICLE IN PRESS

0.0

0.5

1.0

1.5

2.0

0 500 1000 1500 2000

Molecular weight, Dalton

UV

ab

sorb

ance

at

260n

m, m

AU DLDW

FLDW

Difference

Fig. 9. Molecular weight distributions of NOM in FLDW and

DLDW.

0.0

0.1

0.2

0.3

0.4

0.5

0 500 1000 1500 2000

Molecular weight, Dalton

UV

ab

so

rban

ce a

t 260n

m,

mA

U

PAC dose=4mg/L

PAC dose=8mg/L

PAC dose=12mg/L

PAC dose=20mg/L

Fig. 10. Molecular weight distribution of DLDW NOM

adsorbed on PAC A at different PAC doses.

0.0

0.2

0.4

0.6

0.8

1.0

0 60 120 180 240

Time, min

Atr

azin

e re

mai

nin

g,

C/C

0

Fresh PAC, 4mg/L

Preloaded PAC, 4mg/L

Preloaded PAC, 2mg/L

Fig. 11. Atrazine adsorption kinetic curves of PAC B in GW.

0.0

0.5

1.0

1.5

2.0

0 500 1000 1500 2000

Molecular weight, Dalton

UV

ab

sorb

ance

at

260n

m, m

AU

GW

FLDW

Fig. 13. MWDs of NOM in FLDW and GW.

1.0E-13

1.0E-12

1.0E-11

1.0E-10

0 50 100 150 200 250 300 350

NOM surface concentration, mg/g

Ds

of

atra

zin

e, c

m2 /m

in

FLDW preloading

GW preloading

Fig. 12. Effect of NOM in FLDW and GW on Ds of atrazine.

Q. Li et al. / Water Research 37 (2003) 4863–48724870

adsorbed on PAC was obtained by subtracting the

MWD of NOM remaining after adsorption from the

MWD of raw water NOM. It was found in Fig. 10 that

the adsorbed NOM, i.e. the NOM that caused pore

blockage, was mostly in the MW range from 200 to

700 Da.

When GW was used, the NOM in GW showed similar

pore blockage effect as that in FLDW and DLDW. The

atrazine adsorption kinetic curves are shown in Fig. 11.

The atrazine Ds was reduced by up to more than two

orders of magnitude with the PAC dose used. However,

the NOM in GW has a smaller effect on atrazine

adsorption kinetics compared to that in FLDW. Higher

atrazine Ds values were obtained in GW than in FLDW

for the same NOM surface concentration (see Fig. 12).

A comparison of the MWDs of the NOM in GW and

FLDW reveals that the concentration of the NOM

fraction with MW from 200 to 700 Da was much higher

in FLDW than in GW, as shown in Fig. 13. This further

support the earlier hypothesis that the pore blocking

NOM is mostly in the MW range of 200 to 700 Da.

3.2. Effect of NOM on atrazine adsorption in the

continuous flow PAC/MF system

The continuous flow experiment was conducted using

the bench-scale PAC/MF system (see Fig. 1) to

demonstrate the pore blockage effect of NOM on

atrazine adsorption under continuous flow condition.

A pulse dose of 30 mg of PAC B was added directly to

ARTICLE IN PRESSQ. Li et al. / Water Research 37 (2003) 4863–4872 4871

the stirred cell and was retained in the reactor by the MF

membrane. The influent was 2 L of the GW containing

289mg/L of atrazine. The influent water was pumped

through the stirred cell and the effluent was circulated

back to the influent tank. Atrazine concentration in the

system was measured after 7 days. A separate study

found negligible atrazine adsorption by the MF

membrane. A parallel experiment was conducted using

GW in a batch reactor with the same PAC dose, solution

volume, atrazine concentration, and contact time. After

7 days’ contact, 98.9% of atrazine was adsorbed in the

batch reactor. The removal measured in the batch

system was in good agreement with the prediction (99%

removal) by the equivalent background compound

(EBC)-IAST model using the EBC parameters obtained

in an earlier study [22]. However, in the continuous flow

PAC/MF system, only 68.8% atrazine removal was

achieved for the same 7-day contact.

The difference between the removals in the batch and

the continuous flow system is attributed to the pore

blockage effect of NOM. In a batch system, all

the atrazine and NOM molecules were contacted with

the PAC at the same time. Since atrazine is a small

molecule, it diffuses into the carbon pores faster than

large NOM molecules that can potentially block the

openings of the pores in which atrazine can adsorb.

Therefore, much of atrazine can adsorb before the large

NOM molecules cause significant pore blockage and the

maximum adsorption capacity may be reached. In the

continuous flow PAC/MF system, the PAC in the stirred

cell was only exposed to a small amount of atrazine and

NOM that was in the reactor at any given time. Some of

the NOM pore-blocking molecules thus have the

opportunity to adsorb before the PAC is exposed to

atrazine molecules that enter the reactor later, thus

causing an effect similar to the ‘preloading effect’ and

leading to much slower atrazine adsorption kinetics.

Consequently, the maximum adsorption capacity of

atrazine might not have been reached in 7 days as it was

in the batch system. Some pores might have even been

Table 2

Doses of fresh and preloaded PAC A required for 90% and 99% atr

Contact time (min) Dose of PAC A required for 90% removal

No preloading 109.6 mg/g NO

preloading

15 44 659

30 32 469

60 23 303

90 20 240

120 18 206

150 16 185

180 15 170

permanently (in a practical sense) blocked, which could

result in adsorption capacity loss.

3.3. Impact of the pore blockage effect on PAC dose

The reduction in adsorption kinetics caused by the

pore blockage effect of NOM means an increase in the

carbon dose required to achieve a desired removal of a

trace compound if a reactor has a limited carbon–water

contact time. Table 2 compares the carbon doses of PAC

A required to achieve 90% and 99% atrazine removal

from the FLDW when the PAC was preloaded with zero

or 109.6 mg/g NOM (the surface concentration obtained

when 4 mg/L of PAC A was preloaded with the FLDW).

The initial concentration of atrazine was 10mg/L. When

the PAC is preloaded with NOM, 10 to 16 times

more carbon is required to remove 90% or 99% of

atrazine compared to when fresh PAC is used, and the

difference in the required dose increases with decreasing

carbon–water contact time. Although it is not likely that

the PAC will be fully preloaded before it contacts the

trace compound, the potential increase in dosage and

the related cost is tremendous. In order to minimize the

treatment cost, water utilities should minimize the pore

blockage effect by selecting a suitable reactor type and a

PAC that is less vulnerable to pore blockage.

4. Conclusions

By studying the effect of preloading PAC with natural

water, it was found that atrazine adsorption kinetics of

PAC could be severely impeded by NOM molecules

through the pore blockage mechanism. When PAC was

preloaded in natural water, the surface diffusion

coefficient of atrazine decreased with increasing surface

loading of NOM on PAC, and was reduced by up to

more than two orders of magnitude depending on the

PAC dose or NOM surface concentration. The compar-

ison of the two PACs showed that large pores play the

azine removal

(mg/L) Dose of PAC A required for 99% removal (mg/L)

M No preloading 109.6 mg/g NOM

preloading

107 1780

79 1177

59 695

51 550

45 483

41 442

39 412

ARTICLE IN PRESSQ. Li et al. / Water Research 37 (2003) 4863–48724872

very important role of alleviating the pore blockage

effect because NOM mainly adsorbs in mesopores and

large micropores. NOM with MW between 200 and

700 Da was found to be most likely responsible for the

pore blockage mechanism. Natural water that contains a

higher concentration of NOM in this MW range showed

a greater potential to cause pore blockage. The pore

blockage effect was found to have serious impact on

atrazine adsorption in continuous flow adsorption

systems where PAC is kept in the system while water

passes through. Due to the continuous exposure of PAC

to natural water, the accumulation of NOM on the PAC

surface caused the same effect as when PAC was

preloaded in a batch system, resulting in a decreasing

surface diffusion coefficient of atrazine, possibly reduc-

tion of adsorption capacity.

References

[1] Radke CJ, Prausnitz JM. Thermodynamics of multisolute

adsorption from dilute liquid solutions. J Amer Inst Chem

Eng 1972;18(4):761–8.

[2] Carter MC, Weber Jr WJ. Modeling adsorption of TCE

preloaded by background organic matter. Environ Sci

Technol 1994;28(4):614–23.

[3] Newcombe G, Drikas M, Hayes R. Influence of character-

ized natural organic material on activated carbon adsorp-

tion: II. Effect on pore volume distribution and adsorption

of 2-methylisoborneol. Water Res 1997;31(3):1065–73.

[4] Kilduff JE, Karanfil T, Weber Jr WJ. Competitive effects

of nondisplaceable organic compounds on trichloroethy-

lene uptake by activated carbon. I. Thermodynamic

predictions and model sensitivity analyses. J Colloid

Interface Sci 1998;205(2):271–9.

[5] Kilduff JE, Karanfil T, Weber Jr WJ. Competitive effects

of nondisplaceable organic compounds on trichloroethy-

lene uptake by activated carbon. II. Model Verification

and applicability to natural organic matter. J Colloid

Interface Sci 1998;205(2):280–9.

[6] Zimmer G, Crittenden JC, Sontheimer H, Hand DW.

Design considerations for fixed-bed adsorbers that remove

synthetic organic chemicals in the presence of natural

organic matter. AWWA Annual Conference, Orlando, FL,

1988.

[7] Summers RS, Haist B, Koehler J, Ritz J, Zimmer G. The

influence of background organic matter on GAC adsorp-

tion. J AWWA 1989;81(5):66–84.

[8] Speth TF, Miltner RJ. Effect of preloading on the scale-up

of GAC microcolumns. J AWWA 1989;81(4):141–8.

[9] Wilmanski K, Van Breemen AN. Competitive adsorption

of trichloroethylene and humic substances from ground-

water on activated carbon. Water Res 1990;24(6):773–9.

[10] Speth TF. Evaluating capacities of GAC preloaded with

natural water. J Environ Eng 1991;117(1):66–79.

[11] Najm IN, Snoeyink VL, Suidan MT, Lee CH, Richard Y.

Effect of particle size and background natural organics

on the adsorption efficiency of PAC. J AWWA 1990;82(1):

65–73.

[12] Najm IN, Snoeyink VL, Richard Y. Effect of initial

concentration of a SOC in natural water on its adsorption

by activated carbon. J AWWA 1991;83(8):57–63.

[13] Carter MC, Weber Jr WJ, Olmstead KP. Effect of

background dissolved organic matter on TCE adsorption

by GAC. J AWWA 1992;73(8):81–91.

[14] Narbaitz RM, Benedek A. Adsorption of 1,1,2-trichlor-

oethane from river water. J Environ Eng 1994;120(6):

1400–15.

[15] Pelekani C, Snoeyink VL. Competitive adsorption in

natural water: role of activated carbon pore size. Water

Res 1999;33(5):1209–19.

[16] Kilduff JE, Karanfil T, Weber Jr WJ. TEC adsorption by

GAC preloaded with humic substances. J AWWA 1998;

90(5):76–89.

[17] Li Q, Snoeyink VL, Marinas BJ, Campos C. Elucidating

competitive adsorption mechanisms of atrazine and NOM

using model compounds. Water Res 2003;37(4):773–84.

[18] Lebeau T, Lelievre C, Wolbert D, Laplanche A, Prados M,

C #ot!e P. Effect of natural organic matter loading on the

Atrazine adsorption capacity of an aging activated carbon

slurry. Water Res 1999;33(7):1695–705.

[19] Sontheimer H, Crittenden JC, Summers RS. Activated

carbon for water treatment, 2nd ed. University of

Karlsruhe, West Germany: DVGW-Forschungstelle, 1988.

[20] Pelekani C, Snoeyink VL. Competitive adsorption between

Atrazine and methylene blue on activated carbon: the

importance of pore size distribution. Carbon 2000;

38(10):1423–36.

[21] Pelekani C, Snoeyink VL. A kinetic and equilibrium study

of competitive adsorption between atrazine and congo red

dye on activated carbon: the importance of pore size

distribution. Carbon 2001;39(1):25–37.

[22] Li Q, Snoeyink VL, Campos C, Marinas BJ. Displacement

effect of NOM on Atrazine adsorption by PACs with

different pore size distributions. Environ Sci Technol

2002;36(7):1510–5.

[23] Hand DW, Crittenden JC, Thacker WE. User-oriented

batch reactor solution to the homogeneous surface

diffusion model. J Environ Eng 1983;109(1):82–101.

[24] Qi S, Adham SS, Snoeyink VL. Prediction and verification

of atrazine adsorption by powdered activated carbon.

J Environ Eng 1994;120(1):202–18.

[25] Campos C, Marinas BJ, Snoeyink VL, Baudin I, Lain!e

JM. PAC-Membrane filtration process. II: model applica-

tion. J Environ Eng 2000;126(2):104–11.

[26] Campos C, Marinas BJ, Snoeyink VL, Baudin I, Lain!e

JM. Atrazine removal by powdered activated carbon in

floc blanket reactors. Water Res 2000;34(16):4070–80.

[27] Knappe DRU, Matsui Y, Snoeyink VL, Roche P, Prados

MJ, Bourbigot MM. Predicting the capacity of powdered

activated carbon for trace organic compounds in natural

waters. Environ Sci Technol 1998;32(11):1694–8.