Effect of chlorine on adsorption ultrafiltration treatment for removing natural organic matter in drink2

Effect of chlorine on adsorption ultrafiltration treatment for removing natural organic matter in drink2

Effect of chlorine on adsorption/ultrafiltration treatment for removing

natural organic matter in drinking water Tae-Wook Ha,a Kwang-Ho Choo,b,∗ and Sang-June Choib

aDepartment of Environmental Engineering, Taegu Science College, Buk-Gu, Daegu 702-724, South Korea

bDepartment of Environmental Engineering, Kyungpook National University, Buk-Gu, Daegu 702-701, South Korea

Received 29 July 2003; accepted 5 March 2004 Available online 16 April 2004

Abstract

In drinking water treatment, prechlorination is often applied in order to control microorganisms and taste-and-odor-causing materials, which may influence organics removal by adsorption and membrane filtration Thus, the addition of chlorine into an advanced water treat-ment process using a hybrid of adsorption and ultrafiltration (UF) was investigated in terms of natural organic matter (NOM) removal and membrane permeability A comparison between two adsorbents, iron oxide particles (IOP) and powdered activated carbon (PAC), was made

to understand the sorption behavior for NOM with and without chlorination Chlorine modified the properties of dissolved and colloidal NOM in raw water, which brought about lower TOC removal, during IOP/UF The location of IOPs, whether they were in suspension or in a cake layer, affected NOM removal, depending on the presence of colloidal particles in feedwater Chlorine also played a role in reducing the size of particulate matter in raw water, which could be in close association with a decline in permeate flux after chlorination

2004 Elsevier Inc All rights reserved

Keywords: Chlorination; Iron oxide; Adsorption; Ultrafiltration; Water treatment

1 Introduction

Ultrafiltration (UF) processes are increasingly popular in

drinking water treatment to produce better quality water and

meet more stringent treatment regulations, particularly

con-cerning removal of pathogens and turbidity [1–10]

How-ever, the widespread use of UF membranes in drinking water

treatment is still limited due to two major drawbacks,

mem-brane fouling and insufficient removal of disinfection

by-products (DBPs) precursors [11–16] Furthermore, several

studies on the membrane treatment of surface, lake, and river

waters have demonstrated that the DBPs precursors of

nat-ural organic matter (NOM) were one of the major foulants

of the membranes [17–22]

To overcome such problems caused by NOM in UF

ap-plications, conjunctive use of adsorbents and membranes is

thus becoming more attractive for water treatment because

the adsorbents can capture and retain NOM before it reaches

the membrane surface [13–15,23–28] The combination of

UF with powdered activated carbon (PAC) adsorption

re-* Corresponding author Fax: +82-53-950-6579.

E-mail address: chookh@knu.ac.kr (K.-H Choo).

vealed that PAC could play a role in removing NOM from water but is not always helpful in reducing membrane foul-ing [14,29,30] Additionally, a large amount of spent carbon sludge that should be disposed of was generated in the com-bined process Thus, an alternative adsorbent that can be readily regenerated as such iron oxide particles (IOP) has re-cently been developed and applied to water treatment in con-junction with UF membranes [26–28] The addition of IOP into UF systems contributed to both an increase in NOM re-moval efficiency and a decrease in membrane fouling How-ever, the interaction of IOPs with NOM and membranes was not fully understood, nor was the effect of water chemistries

on IOP adsorption well evaluated

Oxidation of raw water would have been done using chlo-rine or ozone to control metals and microbial growth in potable water treatment, which exerted an additional effect

on the coagulation efficiency [31,32] The oxidation step im-proved or deteriorated the coagulation process while chang-ing the nature of the water to be treated No information is available, however, on whether preoxidation could affect the treatment efficiency by adsorption and membrane filtration for water treatment

0021-9797/$ – see front matter  2004 Elsevier Inc All rights reserved.

doi:10.1016/j.jcis.2004.03.010

The comparison of sorption behavior between PAC and

IOP would help understand the characteristics of adsorbents

with different surface structures and functionalities

Analy-ses of the effects of chlorination of raw water during

adsorp-tion and UF could also establish some correlaadsorp-tion of NOM

removal and membrane permeability with alteration of the

physicochemical nature of NOM and colloids It may thus

provide insight into the phenomena occurring at the interface

among the membranes, the adsorbents, and the adsorbates

present in intact and chlorinated waters

In this work, therefore, the characteristics of adsorptive

NOM removal by IOPs were investigated in a combined

IOP/UF system with and without chlorination The

adsorp-tion behavior of IOP was also compared with that of PAC

at different adsorbent doses In particular, the influence of

prechlorination on IOP/UF was evaluated in terms of NOM

removal and membrane permeability, since chlorine was

of-ten dosed into raw water to control microbial growth and

taste-and-odor-causing materials Also, the effect of

parti-cles in the system, such as IOPs and particulate colloids, on

NOM removal efficiency was explored during UF

2 Materials and methods

2.1 Raw water

Water samples used for this study were obtained from the

Maegok Water Treatment Plant, which receives raw water

from the Nakdong River, the major water source of the city

of Daegu, Korea The water samples collected were moved

to the laboratory and then stored at 4◦C before use The key

water quality characteristics are given in Table 1 To remove

colloidal particles from raw water, it was pretreated, using a

0.45-µm filter, and so the filtrate was defined as prefiltered

water

2.2 Adsorbents

Two adsorbents, powdered activated carbons (PAC) and

iron oxide particles (IOP), were employed in this work

A commercially available PAC adsorbent was purchased

from Dae Jung Chemicals and Metals (Korea), while an

amorphous IOP slurry (10 g/L as Fe) was prepared in the

laboratory by neutralizing a ferric chloride solution using

5 N NaOH [33] The PAC and IOP adsorbents used had

av-erage surface areas of 1400 and 240 m2/g, respectively.

Table 1

Quality of raw water from the Nakdong River

Parameter pH Alkalinity

(mg/L

as

CaCO3)

Hardness

(mg/L

as CaCO3)

Turbidity (NTU)

DOC

(mg/L)

UV absorbance

at 254 nm (cm −1)

Value 7.4–7.8 59–71 95–97 31–39 2.4–2.9 0.071–0.079

2.3 Chlorination

A sodium hypochlorite dosing solution with 1000 mg/L

free chlorine was prepared from a 12% sodium hypochlo-rite solution Chlorine dosages were changed over the range

from 0 to 11 mg/L as Cl2at native pH Normally, a high

con-centration of chlorine (11 mg/L as Cl2) was applied in order

to expedite the chlorination reactions within a short period

of time The chlorination steps lasted 30 min and then resid-ual free chlorine was immediately quenched by the addition

of sodium sulfite

2.4 Adsorption tests

To evaluate efficiencies of NOM removal by different ad-sorbents, such as IOP and PAC, adsorption isotherm tests were performed at various adsorbent doses For IOP adsorp-tion tests, a certain amount of the stock IOP slurry

(cor-responding to 0–100 mg/L as Fe) was placed into several

300-ml Erlenmeyer flasks and the solution was then mixed

at 200 rpm and 20◦C for 30 min The solution was filtered through a 0.45-µm filter (Millipore, USA) and the filtrate samples collected were used for analyses PAC adsorption

tests with a PAC dosing range of 0–100 mg/L were

per-formed in the same manner as described above

2.5 UF membranes and stirred cell UF tests

The UF membranes used in this study was made of poly-ethersulfone and had a molecular weight cutoff of 100,000 and an effective surface area of 28.7 cm2 All the membranes used were initially rinsed, using pure water as mentioned in the instructions provided by the manufacturer As shown in Fig 1, batch UF experiments were performed using a 180-ml stirred cell plus an 800-ml reservoir (Amicon 8200, USA)

So the overall working volume of the batch unit was 980 ml The applied pressure was kept at 0.49 bar using a nitrogen cylinder, while the stirring speed was adjusted to 160 rpm using a magnetic stirrer During UF, the mass of permeate was measured using an electronic balance and recorded si-multaneously on an on-line personal computer to calculate permeation flux

Fig 1 Schematic of a stirred-cell UF system unit with a reservoir.

2.6 Analytical methods

Feedwater and permeate samples were analyzed for UV

absorbance at 254 nm and total organic carbon (TOC)

con-centrations Light absorbance at 254 nm, which is associated

with the aromatic groups in NOM, is used as a surrogate

parameter for monitoring the concentration of dissolved

or-ganic matter in a fast and easy manner during water

treat-ment [34] The UV absorbance was determined using a

UV/vis spectrophotometer (Hewlett Packard 8452A, USA)

and the TOC concentration was measured using a TOC

ana-lyzer (Sievers 8200, USA) In particular, the UV absorbance

and the dissolved organic carbon (DOC) concentration of

raw water were measured after removal of particulate matter

using a 0.45-µm membrane (Millipore, USA) Free

chlo-rine concentrations were determined by a DPD colorimetric

method using a Hach spectrophotometer (DR2500, USA)

and reagent pillows

When the prechlorination of raw water was applied in the

experiments, the initial UV absorbance was defined as the

UV absorbance of raw water after chlorination and used for

the evaluation of UV absorbance removal efficiency

3 Results and discussion

3.1 Comparison of sorption behavior of NOM onto IOP

and PAC

Fig 2 shows UV removal efficiencies for treatment of

prefiltered Nakdong River water at various dosages of IOP

and PAC The UV removal efficiency increased sharply, to

approximately 40%, with increasing IOP dosages and nearly

leveled off at such a small dosage as 10 mg Fe/L, whereas

a gradual increase in UV removal was achieved with higher

PAC dosages The distinctively different trends in

adsorp-tive NOM removal for IOP and PAC could be attributed

primarily to the discrepancy in structures and sorption

mech-Fig 2 Comparison of UV removal efficiency between IOP and PAC

adsor-bents: adsorption time, 30 min.

anisms of the adsorbents The majority of active sorption sites of nonporous IOP are located on the outer surface lay-ers, with an amorphous structure which is wide open to the bulk solution phase So NOM removal by IOP sorption can occur rapidly through surface coordinative reactions with-out any limitation on mass transfer of NOM molecules In case of PAC, however, most of the active surface sites are present inside the pores in which NOM sorption normally occurs through van der Waals attraction The rate of sur-face diffusion of NOM molecules inside PAC pores should

be very low because of high molecular weights, so the ad-sorption equilibrium would take more than a few days In fact, the initial external mass transfer coefficient of IOP

(1.17× 10−7 m/s), which was determined in a complete mixed reactor, was nearly one order of magnitude higher

than that of PAC (1.26× 10−8 m/s) In this regard, IOP could be a better adsorbent for removing NOM within lim-ited hydraulic residence times If the combination of adsorp-tion and membrane separaadsorp-tion (e.g., UF or MF) is applied

to drinking water treatment, residence times in the systems must be less than a few hours and thereby IOP would be more desirable due to faster adsorption

Fig 3 compares the effect of prechlorination on UV re-moval efficiency by IOP and PAC during the treatment of prefiltered water The UV removal efficiency for river wa-ter by IOP decreased with higher dosages of chlorine, while

no reduction in UV removal efficiency occurred for PAC ad-sorption A possible explanation of the difference between IOP and PAC adsorption abilities after chlorination is that chlorine reacts with NOM to form DBPs such as THMs, which are not adsorbable onto IOP but still adsorbable onto PAC As a result, it was found that the prechlorination of raw water could have an effect on NOM removal efficiency

by IOP However, further studies on the influence of chlorine during treatment of raw water by IOP adsorption and UF are needed (which will be discussed in the next sections)

Fig 3 UV removal efficiencies using IOP and PAC adsorbents at different

chlorine dosages: chlorination time, 30 min; IOP dose, 100 mg/L as Fe; PAC dose, 80 mg/L.

(b)

Fig 4 Effect of prechlorination on UV and TOC removal efficiencies in

IOP/UF and UF only, treating (a) prefiltered and (b) raw waters: IOP dose,

100 mg/L as Fe; chlorine dose, 11 mg/L as Cl2; chlorination time, 30 min.

3.2 Interaction of chlorine with dissolved and colloidal

NOM in IOP/UF

To examine the interaction of chlorine with dissolved

NOM and colloidal particles during IOP/UF and UF alone,

UV and TOC removal efficiencies using prefiltered and

raw water were compared and are shown in Fig 4 During

IOP/UF treatment of prefiltered water samples, there was a

negligible decrease in TOC removal efficiency with

chlori-nation, though the UV removal efficiency decreased by

ap-proximately 10% For raw water, however, the TOC removal

efficiency also decreased comparatively significantly with

prechlorination in IOP/UF On the other hand, in UF

treat-ment alone, the UV removal efficiency declined slightly with

prechlorination, but negligible TOC removal occurred for

both waters Thus, the lower UV and TOC removal

efficien-cies with prechlorination in IOP/UF might be attributed to a

decrease in the reactivity (adsorption ability) of NOM after

chlorination The specific UV absorbance (SUVA) related

Fig 5 Variation of
UV of different water samples with time during chlo-rination: chlorine dose, 11 mg/L as Cl2 Particles collected from raw water were obtained using a 0.45-µm filter and then resuspended in pure water.

to the dissolved portion of organic matter in river water

de-creased with higher chlorine doses (e.g., from 1.95 L/mg m with no chlorine addition to 1.57 L/mg m with a chlorine dose of 11 mg/L as Cl2), suggesting that the reactivity of TOC in river water was lowered by chlorination

However, it was not enough to explain why the TOC re-moval efficiency for raw water decreased more significantly than that for prefiltered water Thus, the variation of UV

ab-sorbance differentials (
UV) for three feedwater samples

of raw and prefiltered waters and a particle suspension was examined in the course of chlorination time (Fig 5) The

UV value for the prefiltered and raw water samples

de-creased from 0 to −2.1 m−1 and 0 to−1.2 m−1, whereas

UV increased from 0 to 0.65 m−1 with the suspension containing 30 NTU of colloidal particles from river water These results implied that during chlorination the structure

of dissolved NOM was modified, leading to a decrease of

UV absorbance, but some organic matter could be released from colloidal particles to supply more carbon into the bulk liquid phase When a feedwater sample containing 100 NTU

of particles was chlorinated,
UV increased up to 1.0 m−1 within 30 min of reaction time (data not shown) Conse-quently, during prechlorination the breakage of aromatic moieties in NOM molecules occurred without mineraliza-tion of NOM, which reduced the sorpmineraliza-tion capacity of NOM The conversion of particulate matter to dissolved matter by chlorination should contribute to an increase of carbon con-tent in the dissolved NOM pool, resulting in lower DOC removal during IOP/UF

3.3 Effect of IOP deposition and colloidal particles on NOM rejection

Since IOPs injected into the IOP/UF system exist either

in the bulk liquid phase or at the membrane surface, the ef-fect of IOP locations on NOM removal during IOP/UF was examined As shown in Fig 6, the UV and DOC removal

Fig 6 UV and DOC removal efficiencies using raw water during UF with

IOPs in suspension and in deposited layers: IOP dose, 100 mg/L as Fe.

efficiencies with deposited IOPs were always at higher

lev-els than those with suspended IOPs It could be hypothesized

that the formation of IOP cake layers at the membrane would

contribute to further NOM rejection due to a sieving effect

during IOP/UF However, it was not clear that IOPs in the

deposited layer had such similar sorption capacity for NOM

as those in suspension

Any changes of the sorption capacity of IOPs located

at different places were examined for raw water and are

compared in Fig 7 Different trends in NOM removal

us-ing IOP-deposited and suspended systems were observed

when prechlorination and/or prefiltration runs were

per-formed The DOC removal efficiency for the IOP-deposited

system declined substantially when the colloidal particles

present in raw water were removed, whereas that for the

IOP-suspended system was not so much dependent on the

absence of colloidal particles This indicated that a

signifi-cant loss of adsorption sites took place when IOPs formed a

cake layer at the membrane surface Thus it supported the

above reasoning that higher DOC removal with the

IOP-deposited system was mainly caused by the formation of

a denser cake layer associated with colloids in raw water

In addition, it could be thought that crossflow UF that can

keep IOPs in suspension would be better in regard to NOM

removal compared to dead-end membrane filtration if raw

water had a very low turbidity

3.4 Effect of chlorination on membrane permeability

Relative flux decline profiles for different treatment

sys-tems using raw and filtered waters are shown in Fig 8 The

IOP/UF system had a flux more than two times that of UF

alone It was thus clear that IOP addition helped to enhance

the membrane flux substantially, since IOP removed NOM

that can otherwise cause membrane fouling [26–28] On the

other hand, the fluxes with the prechlorination of raw water

were always kept at a relatively low level in both IOP/UF and

(a)

(b)

Fig 7 Effect of chlorine and colloidal particles on DOC removal during IOP/UF when (a) IOPs are in deposited cake layers and (b) in suspension:

chlorine dose, 11 mg/L as Cl2; chlorination time, 30 min.

UF-only systems, whereas the permeate flux was indepen-dent of chlorination when colloidal particles were removed prior to UF Although a change of the nature of dissolved organic matter was obvious during chlorination, it did not affect membrane flux so much (Fig 8) Considering the lower DOC removal efficiency with prechlorination shown

in Fig 4, a possible explanation for the above result was that the interaction of chlorine with colloidal particles in raw wa-ter may bring about a change of their characwa-teristics (e.g., particle size reduction) leading to flux decline As shown

in Fig 9, the size distribution of particles in raw water was shifted to lower ranges with chlorination, which can make the cake layer of particles at the membrane surface denser This supports the above explanation about flux decline with chlorination Accordingly, it was found that membrane flux was not sensitive to the change of physicochemical proper-ties of the DOC by chlorination, but it was affected by that

of particulate matter characteristics

(b)

Fig 8 Effect of prechlorination on flux in IOP/UF and UF alone treating

(a) raw and (b) filtered waters: chlorine dose, 11 mg/L as Cl2; chlorination

time, 30 min.

Fig 9 Particle size distributions of raw water before and after chlorination:

chlorine dose, 11 mg/L as Cl; chlorination time, 30 min.

4 Summary and conclusions

Prechlorination of river water during UF in combination with adsorption was conducted and evaluated with respect to NOM removal and membrane permeability IOP had a larger sorption capacity for NOM at lower dosages than PAC, though the sorption capacity of IOP declined slightly after chlorination of raw water The interaction of chlorine with dissolved NOM in raw water caused a decrease in UV ab-sorbance, while that with particulate matter did an increase

in it During the chlorination reactions, part of the particu-late matter was converted to dissolved matter, leading to an increase of dissolved carbon content in the NOM total These results contributed to the lower TOC removal efficiency by IOP/UF after chlorination When the IOPs injected into an

UF system formed a cake layer at the membrane surface, further NOM removal was achieved due to a sieving effect

by the physical barrier As colloidal particles were removed from raw water, however, a significant decrease in NOM re-moval efficiency occurred in the IOP-deposited system It was revealed that IOPs in the cake layer were not so effective for adsorption as those in suspension because of their ag-gregation During prechlorination, colloidal particles in raw water became smaller in size, so flux decline was relatively large due to the formation of a denser cake layer on top of the membrane

Acknowledgments

This work was supported by the Korea Research Foun-dation (Grant 2002-003-D00171) The authors thank public servants of the Maegok Water Utility for their help in ob-taining water samples The laboratory assistance provided

by Suck-Ki Kang and Jang-Hyun Kim at Daegu University

is appreciated

References

[1] M Lacey, J AWWA 95 (6) (2003) 2.

[2] A.A Karimi, S Adham, S.-C Tu, J AWWA 94 (12) (2002) 120–128 [3] K Glucina, A Alvarez, J.M Laîné, Desalination 132 (1–3) (2000) 73–82.

[4] K Hagen, Desalination 119 (1–3) (1998) 85–91.

[5] J.A.M.H Hofman, M.M Beumer, E.T Baars, J.P van der Hoek, H.M.M Koppers, Desalination 119 (1–3) (1998) 113–125.

[6] Y Magara, S Kunikane, M Itoh, Water Sci Technol 37 (10) (1998) 91–99.

[7] T Hirata, A Hashimoto, Water Sci Technol 38 (12) (1998) 103–107 [8] P Lipp, G Baldauf, R Schick, K Elsenhans, H.-H Stabel, Desalina-tion 119 (1–3) (1998) 133–142.

[9] S Nakatsuka, I Nakate, T Miyano, Desalination 106 (1–3) (1996) 55–61.

[10] S.W Lee, T.S Song, M.C Jo, S.J Park, J Ind Eng Chem 6 (1) (2000) 53–58.

[11] K Khatib, J Rose, O Barres, W Stone, J.-Y Bottero, C Anselme,

J Membr Sci 130 (1–2) (1997) 53–62.

[12] G.F Crozes, J.G Jacangelo, C Anselme, J.M Laîne, J Membr Sci 124 (1) (1997) 63–76.

[13] W Tsujimoto, H Kimura, T Izu, T Irie, Desalination 119 (1–3) (1998)

323–326.

[14] C.-F Lin, S.-H Liu, O.J Hao, Water Res 35 (10) (2001) 2395–2402.

[15] A Yuasa, Water Sci Technol 37 (10) (1998) 135–146.

[16] P.K Park, C.H Lee, S.J Choi, K.H Choo, S.H Kim, C.H Yoon,

De-salination 145 (1–3) (2002) 237–245.

[17] W Yuan, A.L Zydney, Environ Sci Technol 34 (2000) 5043–5050.

[18] A Maartens, P Swart, E.P Jacobs, Water Sci Technol 40 (9) (1999)

113–120.

[19] A Maartens, P Swart, E.P Jacobs, J Colloid Interface Sci 221 (2000)

137–142.

[20] M.R Teixeira, M.J Rosa, Desalination 151 (2) (2003) 165–175.

[21] A.I Schäfer, A.G Fane, T.D Waite, Desalination 131 (1–3) (2000)

215–224.

[22] H Lee, G Amy, J Cho, Y Yoon, S.H Moon, I.S Kim, Water

Res 35 (14) (2001) 3301–3308.

[23] J.S Kim, S.J Lee, S.H Yoon, C.H Lee, Water Sci Technol 34 (9)

(1996) 223–229.

[24] M Tomaszewska, S Mozia, Water Res 36 (16) (2002) 4137–4143 [25] S.S Adham, V.L Snoeyink, M.M Clark, C Anselme, J AWWA 85 (12) (1993) 58–68.

[26] Y.J Chang, M.M Benjamin, J AWWA 88 (12) (1996) 74–88 [27] Y.J Chang, K.H Choo, M.M Benjamin, S Reiber, J AWWA 90 (5) (1998) 90–102.

[28] K.W Lee, K.H Choo, S.J Choi, K Yamamoto, Water Sci Technol Water Supply 2 (5–6) (2002) 293–300.

[29] C.-F Lin, Y.-J Huang, O.J Hao, Water Res 33 (5) (1999) 1252– 1264.

[30] S.J Lee, K.H Choo, C.H Lee, J Ind Eng Chem 6 (6) (2000) 357– 364.

[31] C.R O’Meliaa, W.C Beckerb, K.-K Aua, Water Sci Technol 40 (9) (1999) 47–54.

[32] C Camel, A Bermond, Water Res 32 (11) (1998) 3208–3222 [33] K.H Choo, S.K Kang, Desalination 154 (2003) 139–146.

[34] G.V Korshin, C.-W Li, M.M Benjamin, Water Res 31 (7) (1997) 1787–1795.