Effect of Partial Flocculation and Adsorption as

Pretreatment to Ultrafiltration
H. K. Shon, S. Vigneswaran, and H. H. Ngo
Faculty of Engineering, University of Technology, Sydney, NSW 2007, Sydney, Australia

DOI 10.1002/aic.10597
Published online September 6, 2005 in Wiley InterScience ([Link]).

In this study, the sufficiency of flocculation and adsorption with reduced doses of ferric
chloride and powdered activated carbon (PAC) as pretreatment to ultrafiltration (UF)
was investigated. A 50 mg/L dose of FeCl3 and 0.5 g/L of PAC removed a majority of
organic matter (OM; 88%), thus reducing the organic loading on UF that was used as
posttreatment. Although flocculation with lower doses of FeCl3 (10 mg/L) followed by
PAC adsorption of 0.5 g/L and UF removed the same amount of organics (⬃88% OM),
the majority of the OM removal was by the posttreatment of UF rather than by pretreat-
ment, resulting in a significant decline of flux in UF. A detailed relative molecular mass
(RMM) distribution analysis was made with flocculation, adsorption, and UF permeates.
PAC adsorption decreased the majority of the relative intensity of smaller RMM of OM
from the preflocculated water. Flocculation with ⬍40 mg/L FeCl3 was not sufficient to
remove the relative intensity of the large RMM even after a postadsorption. The detailed
analysis on weight-averaged RMM (Mw) indicated that the Mw values of OM in the
wastewater and in the flocculated effluent were 29,800 daltons (initial), ⬎25,000 (after
flocculation with ⱕ40 mg/L FeCl3), and ⬍1000 (after flocculation with ⱖ50 mg/L FeCl3).
© 2005 American Institute of Chemical Engineers AIChE J, 52: 207–216, 2006
Keywords: semiadsorption, semiflocculation, relative molecular mass distribution, pre-
treatment, organic matter, ultrafiltration

Introduction ranging from 3 to 20 nm in diameter, appears to constitute the

important membrane foulants.1
Wastewater reuse will have a major impact in maintaining
Many researchers found that FeCl3 flocculation followed by
the environmental quality and the unrelenting pressure on
powdered activated carbon (PAC) adsorption is one of the most
conventional and natural freshwater sources. The most suitable
effective pretreatments to membrane filtration.2,3 Abdessemed
processes for reuse are the membrane processes such as ultra-
and Nezzal4 experimentally investigated the treatability of do-
filtration (UF), nanofiltration (NF), and reverse osmosis (RO).
mestic wastewater by a coagulation–adsorption process using
UF is an effective process for the removal of colloidal matter,
FeCl3 and PAC. It removed 86% of chemical oxygen demand
macromolecules, and pathogens. It removes only a part of
(COD) from domestic wastewater. Shon et al.3 also observed
dissolved organic matter (OM). Further, membrane fouling is that FeCl3 flocculation followed by PAC adsorption removed
the major problem in UF. Howe and Clark1 reported that 91% of dissolved organic carbon (DOC) from a biologically
particulate matter (⬎0.45 ␮m) was relatively unimportant in treated wastewater effluent. The FeCl3 flocculation process can
fouling compared to dissolved matter in UF. Dissolved OM, be used to aggregate colloids and suspended solids in the size
range of 0.1–10 ␮m. A slight portion of small relative molec-
ular mass (RMM) organic matter in the wastewater effluent can
Correspondence concerning this article should be addressed to S. Vigneswaran at also be removed by complexation with ferric hydroxides. PAC
[Link]@[Link].
adsorption can successfully remove the majority of the small
© 2005 American Institute of Chemical Engineers RMM organic matter such as refractory organic matter, hydro-

AIChE Journal January 2006 Vol. 52, No. 1 207

 Table 1. Constituents of Synthetic Wastewater the mixed synthetic wastewater ranged from 300 to about
Concentration Fraction 34,100 Da with the highest fraction of 900 –1200 Da. Although
Compound (mg/L) RMM (Da) by DOC sodium lignin sulfornate and tannic acid showed peaks at
Beef extract 1.8 300, 100, 70 0.065
12,100 and 6300 Da, respectively, the corresponding peaks
Peptone 2.7 34,300, 100, 80 0.138 were not found in the mixed synthetic wastewater, a phenom-
Humic acid 4.2 1500, 300 0.082 enon that may be explained by aggregations between OM and
Tannic acid 4.2 6300 0.237 inorganic and/or organic compounds in the synthetic wastewa-
Sodium lignin ter. Hong and Elimelech10 found that the chemical composition
sulfonate 2.4 12,100 0.067
Sodium lauryl sulfate 0.94 34,300 0.042 of feed water greatly influenced the structure of OM. In the
Arabic gum powder 4.7 900, 300 0.213 presence of high ionic strength and divalent cations, the OM
Arabic acid structure was changed from a stretched and linear configuration
(polysaccharide) 5.0 38,900 0.156 to one that was coiled and compact. Verwey and Overbeek11
(NH4)2SO4 7.1 0
K2HPO4 7.0 0
reported that particle attraction and repulsion can be explained
NH4HCO3 19.8 0 using DLVO (Derjaguin–Landau–Verwey–Overbeek) theory:
MgSO4 䡠 7H2O 0.71 0 (1) van der Waals (attraction) and (2) electrical double layer
(repulsion). The wastewater used in this study has humics,
tannic acids, polysaccharides, and divalent cations (such as
phobic organic matter in the range of 200 –3500 daltons (Da), magnesium). Thus, electrostatic interactions would change the
and a small portion of the large RMM organic matter. There- RMM distribution, which will have a major impact on the flux
fore, the pretreatment of flocculation followed by adsorption decline.
can remove the majority of dissolved organic matter in the
wastewater. Synthetic organic matter characterization
Abdessemed and Nezzal4 also stated from their experiments DOC was measured using a Dohrmann Phoenix 8000 ™
that flocculation is more beneficial than adsorption as pretreat- UV-persulfate TOC analyzer (Tekmar-Dohrmann, Mason, OH)
ment in terms of increase in permeate flux. There was an equipped with an autosampler. All samples were filtered
increase of 46.6% in permeate flux with flocculation and 22.2% through a 0.45-␮m membrane before DOC measurement. After
with PAC adsorption. Shon et al.3 showed that flocculation pretreatment, the synthetic wastewater was examined for the
with an optimum dose of FeCl3 of 68 mg/L improved the UF RMM distribution of organics by high-pressure size-exclusion
permeate flux by 50%. Al-Malack and Anderson5 found the chromatography (HPSEC, Shimadzu Corp., Kyoto, Japan) with
optimum FeCl3 dose as 200 mg/L FeCl3 at pH 9 to remove the an SEC column (Protein-pak 125®, Waters Chromatography
organic matter in wastewater. The COD removal was 99.3%. Division/Millipore, Milford, MA). Standards of RMM of var-
According to the result of Aguiar et al.,6 the optimum dose of ious polystyrene sulfonates (PSS: 210, 1800, 4600, 8000, and
coagulant was 2.1 ⫾ 0.2 mg Fe per mg of total organic carbon 18,000 Da) were used to calibrate the equipment.
(TOC). The RMM distribution represents a normalized fraction per-
PAC adsorption removed 60 –75% of DOC from biologi- centage, obtained by dividing each incremental height of the
cally treated wastewater effluent.3,7 Lin et al.8 studied the use of chromatogram with a sum of the heights when the chromato-
PAC adsorption at a dose up to 400 mg/L as pretreatment to gram is divided into incremental mass intervals.12,13 The mo-
remove humic substances of 20 mg/L in concentration. In their lecular weight distribution (MWD) can also be represented by
study, the PAC was ineffective in removing the RMM fractions a UV response (mV intensity) with time. The latter represen-
of ⬍300 or ⬎17,000 Da. The flux decline in UF for the tation is similar to that obtained with normalized fraction
PAC-treated streams was worse than that without preadsorp- percentages, making it considerably easier to visualize the
tion.
In all the above studies, large doses of flocculants (FeCl3)
and PAC were used, which is not economical. Further, it will
result in production of substantial amounts of sludge. In this
study, a pretreatment of flocculation with reduced doses of
FeCl3 followed by adsorption with reduced doses of PAC was
investigated in terms of flux decline in UF (used as posttreat-
ment). Different RMM ranges of organic matter removed by
different treatments were also studied. An attempt was made to
optimize the pretreatment requirement based on the results of
organic matter removal and effective RMM.

Experimental
Synthetic wastewater
This study was carried out with synthetic wastewater. The
composition of the synthetic wastewater is presented in Table
1. This synthetic wastewater represents the biologically treated
sewage effluent (BTSE).9 Synthetic wastewater has a number Figure 1. Partial flocculation followed by partial adsorp-
of known compounds at known concentrations. The RMM of tion and UF unit.

208 January 2006 Vol. 52, No. 1 AIChE Journal

 Table 2. Characteristics of Powdered Activated Carbon were 735 and 5.33 Pa, respectively. A Nitto Denko NTR-7410
(PAC) Used* membrane was used in this study (Table 3).
Specification PAC-WB
Moisture content (%) 5 max Results and Discussion
Surface area (m2 g⫺1) 882
Nominal size 80% min finer than 75 micron Removal of DOC by FeCl3 flocculation and PAC
Type Wood-based adsorption
Mean pore diameter (Å) 30.61 The efficiency of PAC adsorption and FeCl3 flocculation
Mean diameter (␮m) 19.71
was investigated in terms of dissolved organic carbon (DOC)
*MD3545WB powder, James Cumming & Sons Pty. Ltd., Australia. removal from the synthetic wastewater (Figures 2i and 2ii). As
can be seen in Figure 2i, DOC removal was the highest (78.2%)
with the FeCl3 flocculation at a dose of 68 mg/L. The optimum
reduction of a peak of organic matter by different pretreat-
FeCl3 dose was 68 mg/L from the laboratory jar test experi-
ments, such as flocculation and adsorption. Thus, in this study
ments. PAC also removed the DOC from 57.1 to 66.6% when
both representations were used in interpreting organic removal
PAC was added, at a dose of 1 to 2 g/L. For the synthetic
by different pretreatments.
wastewater used in this study, DOC removal was far better by
In general, UV absorbance at 254 nm detects limited com-
flocculation than by adsorption, suggesting that the majority of
ponents (mostly ␲-bonded molecules) of organic matter so this
organic matter (OM) in the synthetic wastewater is constituted
method is mostly applied to the MW estimations of humic and
of large molecular weight compounds.
fulvic acids as well as hydrophobic (aromatic) organic matter.
Experiments were also conducted with the addition of partial
The UV detector used in this study had a limitation in detecting
optimum concentration of FeCl3 and PAC to study the effect of
low UV-absorptivity components, such as proteins and poly-
reduced doses of chemicals in the DOC removal. The removal
saccharides. Thus, a relative intensity of UV response was used
efficiency of OM by flocculation and adsorption at different
to interpret the results.12,13 With an on-line DOC detector, all
doses and by posttreatment of UF is presented in Figure 3. For
organic compounds can be recognized. According to a previous
example, with the addition of an optimum dose of FeCl3 and
study, the RMM of 43,100 Da may be attributable to polysac-
reduced doses of PAC (0.05– 0.5 mg/L), removal of DOC from
charide, 580 and 870 Da to humic substances, 300 Da to
the synthetic wastewater was observed as follows: 82.6, 89.5,
building blocks, 250 Da to acids, and ⬍200 Da to amphiphi-
89.8, and 92.2% for PAC doses of 0.05, 0.1, 0.3, and 0.5 g/L
lics.14 Here, the building blocks refer to humic-substance hy-
PAC, respectively. The above results indicate that PAC adsorp-
drolysates (350 –500 Da). Details on the measurement meth-
tion after flocculation with an optimum dose of FeCl3 increased
odology are given elsewhere.12,15
the removal of DOC. The increase was 15% when a PAC dose
of 0.5 g/L was added. In the same manner, experiments were
Experimental setup of partial flocculation followed by conducted with reduced doses of FeCl3 and PAC. For instance,
partial adsorption and filtration
when the doses of FeCl3 and PAC were kept at 30 mg/L and 0.5
Figure 1 presents the experimental setup. Flocculation was g/L, respectively, removal of DOC was still high— up to 76%.
first carried out with ferric chloride (FeCl3) of different doses The posttreatment of UF had a significant effect in the
(10 – 68 mg/L). The samples were stirred rapidly for 1 min at removal of DOC for pretreated waters with lower FeCl3 and
100 rpm, followed by 20 min of slow mixing at 30 rpm, and 30 PAC doses. For example, the posttreatment of UF led to
min of settling. Adsorption with powdered activated carbon additional DOC removal of 40% for the pretreated water with
(PAC, 0.05– 0.5 g/L) was then conducted using the prefloccu- 10 mg/L FeCl3 and 0.5 g/L of PAC. The pretreatment led to
lated synthetic wastewater. The characteristics of PAC used are only 50% DOC removal, whereas the pretreatment followed by
given in Table 2. A predetermined quantity of PAC was stirred UF led to 90% removal. On the other hand, the pretreatment
with a mechanical stirrer at 100 rpm for 1 h. A prefilter (pore alone with 68 mg/L of FeCl3 and 0.5 g/L of PAC led to ⬎90%
size 0.45 ␮m) was used to separate PAC particles before the DOC removal, which allowed the posttreatment of UF to
application to UF unit. remove ⬍4% additional DOC.
In this study, the cross-flow UF unit (Nitto Denko Corp.,
Osaka, Japan) was used to study the organic removal from
synthetic wastewater. The synthetic wastewater was pumped to Flux decline of UF with pretreated wastewater
a flat-sheet membrane module (effective membrane area 0.006 The performance of UF was also studied in terms of nor-
m2). The operating pressure and cross-flow velocity were con- malized permeate flux (J/J0) with and without pretreatment
trolled at 300 kPa and 0.5 m/s by means of bypass and regu- (Figure 4). After the pretreatment of flocculation with the
lating valves. The Reynolds number and shear stress at the wall optimum dose of FeCl3, the UF did not undergo any flux

Table 3. Characteristics of UF Membrane Used

Contact Specific Flux R m (Membrane
MWCO* Angle Zeta Potential at 300 kPa Resistance,
Code Material (daltons) (°) at pH 7 (mV) (m/d) ⫻1012 m⫺1)
NTR Sulfonated
7410 polysulfones 17,500 69 ⫺98.63 1.84 14.1

*MWCO, molecular weight cutoff.

AIChE Journal January 2006 Vol. 52, No. 1 209

 Figure 2. DOC removal by (i) FeCl3 flocculation and (ii) PAC adsorption at different doses of FeCl3 and PAC.

decline (Figure 4i). The pretreatment of PAC adsorption also moved from the wastewater by flocculation, adsorption, and
helped in the reduction of flux decline. For example, the J/J0 in UF. This will help in the selection of a suitable pretreatment
the UF after a pretreatment of PAC adsorption (with 1 g/L) was method and a correct membrane for a given application.16
from 1 to 0.71 after 6 h of operation (that is, 29% decline). The Figures 6i and 6ii present the RMM distribution of OM [in
decline with no pretreatment was 0.66 (that is, 34% decline). terms of fraction percentage (peak heights divided by sum of
Figure 5 presents the permeate flux of UF with the waste- peak heights) and UV response with time] in wastewater clar-
water that has undergone flocculation with 68, 50, 30, and 10 ified with varying dosages of FeCl3. It should be noted that
mg/L of FeCl3 followed by PAC adsorption of known concen- settled flocs were removed and only supernatant was used in
tration of PAC. The flux decline was minimal, especially for RMM distribution measurements. In some cases, peaks in the
preflocculated waters with FeCl3 of ⱖ50 mg/L. RMM distribution chart are higher after treatment than they
were for raw wastewater. This is correct because peaks here do
Relative molecular mass (RMM) distribution of organic not represent absolute values but relative proportions. This is
matter the reason that UV response (mV) with time is also presented
To understand the phenomenon of the flux decline, it is along with these figures (Figure 6ii). The latter clearly shows
necessary to know the range of RMM of organic matter re- removal of OM of different RMM values.
Flocculation with FeCl3 in the 50 to 68 mg/L dosage range
(near optimum dose) efficiently removed large RMM OM,
tannic acid, sodium lignin sulfonate, sodium lauryl sulfate, and
arabic acid (Figures 6i and 6ii). Focculation using higher doses
of FeCl3 also removed some smaller RMM OM compounds
(600 –1000 Da; that is, peptone, beef extract, and humic acid).
The phenomenon of the small RMM OM removal (600 –1000
Da) by FeCl3 flocculation may be explained by the complex-
ation of Fe.17 However, the relative intensity of the smallest
RMM (at 250 Da) was not reduced, showing that flocculation
was not effective in removing small molecules. On the other
hand, lower FeCl3 dosages (ⱕ40 mg/L) did not decrease the
relative intensity of all the large RMM organic compounds and
the majority of small RMM compounds (Figures 6i and 6ii).
The peaks corresponding to the relative intensity of the large
RMM (36,300 Da) and small RMM (1000 and 900 Da) were
clearly present in wastewater supernatant treated with ⬍30
mg/L of FeCl3. This trend of RMM distribution is in agreement
with that of flux decline that occurred with 30 mg/L floccula-
tion (Figure 4i). The relative intensity of the large RMM OM
remaining in the synthetic wastewater may have been respon-
Figure 3. DOC removal of partial flocculation followed sible for the flux decline. The indicative relationship between
by partial adsorption and ultrafiltration. the particle size and the RMM is presented in Table 4. Permi-
UF membrane used: NTR-7410; MWCO: 17,500 Da; cross- nova et al.18 observed that large molecules (humic and fulvic
flow velocity: 0.5 m/s; transmembrane pressure: 300 kPa;
Reynolds number: 735.5; shear stress: 5.33 Pa; DOC removal acids) in the range of 4700 –30,000 Da were responsible for the
with UF alone: 75.3%. fouling of membranes by organic matter. Howe and Clark1

210 January 2006 Vol. 52, No. 1 AIChE Journal

Figure 4. Temporal variation of filtration flux of different pretreatments with UF NTR-7410: (i) after PAC adsorption; (ii)
after flocculation.
J0 ⫽ 1.84 m/d at 300 kPa; cross-flow velocity: 0.5 m/s; MWCO: 17,500 Da; Reynolds number: 735.5; shear stress: 5.33 Pa.

found that the dissolved organic matter, which was smaller ages (0.5 g/L) removed the majority of the relative intensity of
than about 3 nm, caused only minimal fouling. the small RMM OM (200 – 600 Da). However, the majority of
Figures 6iii and 6iv present the RMM distribution of waste- the relative intensity of larger RMM OM could not be removed
water pretreated with PAC. Adsorption using larger PAC dos- by adsorption alone. The PAC used had a pore radius from 1 to

Figure 5. Temporal variation of filtration flux and DOC ratio with partial flocculation followed by partial adsorption
with UF NTR-7410: (i) after 68 mg/L flocculation; (ii) after 50 mg/L; (iii) after 30 mg/L; (iv) after 10 mg/L.
J0 ⫽ 1.84 m/d at 300 kPa; cross-flow velocity: 0.5 m/s; MWCO: 17,500 Da; Reynolds number: 735.5; shear stress: 5.33 Pa.

AIChE Journal January 2006 Vol. 52, No. 1 211

Figure 6. RMM distribution after (i) flocculation with the large FeCl3 doses (fraction % vs. time); (ii) flocculation with
small FeCl3 doses (UV response vs. time); (iii) adsorption with large PAC doses (fraction % vs. time); and (iv)
adsorption with small PAC doses (UV response vs. time).

5 nm with mean radius of 1.8 nm. The observed removal of a with more than 50 mg/L FeCl3 (Figures 7i and 7ii). However,
portion of a large RMM organics by PAC may have been a the peak corresponding to 1000 Da remained at high intensity,
result of adsorption onto the larger pores, or the outer surface which indicates a difficulty in removing humic acid, tannic
of PAC.3 The RMM distribution results are consistent with the acid, and arabic gum powder (which has a peak at 1000 Da).
trend in flux decline (Figure 4ii). Flocculation using ⬍30 mg/L FeCl3 was not sufficient to
Figure 7 shows the RMM distribution results after partial remove the large RMM OM, even after a postadsorption (Fig-
flocculation with 30 and 50 mg/L FeCl3 doses and by partial ures 7iii and 7iv).
adsorption. PAC adsorption removed the majority of smaller Figure 8 presents the RMM distribution after FeCl3 floccu-
RMM of 900, 600, and 200 Da from wastewater preflocculated lation followed by PAC adsorption as pretreatment, and UF
NTR-7410 filtration as posttreatment. The absolute disappear-
Table 4. Relationship between the Size (in nm and ␮m) and ance of the relative intensity of OM at the peak corresponding
RMM (in Da) to 36,300 Da was found after UF. Flocculation with a low
Size (daltons) Size (␮m) Size (nm) dosage of 30 mg/L FeCl3 and consecutive adsorption did not
500* 0.00039 0.39 remove the majority of the large RMM OM. This resulted in
1000* 0.00050 0.50 relatively rapid flux decline in UF (Figure 5iii). Pretreatment
5000* 0.00085 0.85 with decreasing FeCl3 doses increased the relative intensity of
7000* 0.00095 0.95 the peak corresponding to 1000 Da.
10,000* 0.00107 1.07
20,000* 0.00134 1.34 Table 5 presents the weight-averaged RMM (Mw) values of
100,000** 0.01000 10.00 OM in the pretreated effluent and in the UF effluent. The Mw
500,000** 0.05000 50.00 values of OM in the wastewater and in the flocculated effluent
* The equation used to compute the size is: Size (␮m) ⫽ 0.0001 (MW) 0.3321 / 2. were 29,800 Da (initial), ⬍1000 (after flocculation with ⱖ50
**Adapted from Mulder.19 mg/L FeCl3), and ⬎25,000 (after flocculation with ⱕ40 mg/L

212 January 2006 Vol. 52, No. 1 AIChE Journal

 Figure 7. RMM distribution of organic matter after flocculation followed by PAC adsorption.

FeCl3). Thus, a flocculation with ⬎50 mg/L FeCl3 and PAC significant difference in the Mw of the pretreated effluent and
adsorption is essential in removing large and small RMM OM for UF effluent. For example, when a FeCl3 dose of 40 mg/L and
the synthetic wastewater in this study. The UF as posttreatment a PAC dose of 0.5 g/L were used, the effluent from pretreat-
could not remove smaller RMM OM in a noticeable manner. ment had Mw ⫽ 25,000 Da. When this effluent was filtered
When the FeCl3 concentration was decreased from 50 to 40 through UF, the Mw decreased to 900 Da. This clearly shows
mg/L, the increase of Mw was significant. As the FeCl3 con- that preflocculation with insufficient doses of flocculants del-
centration was further decreased from 40 to 20 mg/L, the egates (or passes) the removal of large RMM organic matter to
increase of Mw values of OM was not so significant. The same the posttreatment of UF. This, in turn, results in severe flux
trend was observed even with the PAC adsorption of the decline of membrane because the UF membrane fouling is
flocculated effluent (with FeCl3 dose of 10 – 40 mg/L). This mainly caused by large nanosized RMM organic matter.
phenomenon is explained by the following two reasons: (1) the A correlation between the amount of FeCl3 dose and the Mw
FeCl3 dose of ⱕ40 mg/L was not sufficient to decrease the value is presented in Figure 9. The deviant crease circles show
relative intensity of the large RMM OM; and (2) the PAC that the ranges of flocculant (FeCl3) and adsorbent (PAC) are
adsorbed only smaller RMM organic matter. When lower doses necessary to reduce the membrane fouling and to obtain supe-
of FeCl3 and PAC were used in the pretreatment, there was a rior DOC removal. Minimum concentrations of flocculant

AIChE Journal January 2006 Vol. 52, No. 1 213

Figure 8. RMM distribution after flocculation and adsorption as pretreatment and UF as posttreatment.
J0 ⫽ 1.84 m/d at 300 kPa; cross-flow velocity: 0.5 m/s; MWCO: 17,500 Da; Reynolds number: 735.5; shear stress: 5.33 Pa.

(FeCl3) and adsorbent (PAC) are necessary to obtain high DOC

removal and thus the minimum decreases in flux were 50 mg/L
Table 5. Weight-Averaged RMM Values of Organic Matter and 0.05 g/L, respectively, for the wastewater used in this
after Pretreatment of Flocculation and Adsorption after study.
Posttreatment of UF*
PAC Concentration of 0.5 g/L Conclusions
FeCl3 After Semiflocculation Followed by
(mg/L) PAC Adsorption UF Permeate
In this study, the sufficiency of flocculation and adsorption
with reduced doses of ferric chloride (FeCl3) and PAC as
68 1100 880 pretreatment to UF was investigated. The effectiveness of
60 1100 870
50 1000 850 pretreatment was evaluated in terms of reduction in the decline
40 25,000 910 of performance flux and the removal of organic matter of
30 31,100 910 different RMMs. The finding can be summarized as follows:
20 31,700 920 (1) A FeCl3 dose of ⬍40 mg/L was not sufficient to de-
10 32,000 930
crease the relative intensity of the large RMM organic matter.
*All units: daltons. FeCl3 and PAC concentrations of 50 mg/L and 0.5 g/L, respec-

214 January 2006 Vol. 52, No. 1 AIChE Journal

Figure 9. Correlation of partial flocculation concentration vs. DOC concentration and averaged-weight RMM: (i) FeCl3
concentration vs. DOC concentration of partial flocculation followed by partial adsorption; (ii) FeCl3 con-
centration vs. RMM of partial flocculation followed by partial adsorption.
J0 ⫽ 1.84 m/d at 300 kPa; cross-flow velocity: 0.5 m/s; MWCO: 17,500 Da; Reynolds number: 735.5; shear stress: 5.33 Pa.

tively, removed a majority of DOC (88%), thus reducing the Literature Cited
organic loading to UF used as posttreatment. Although floccu- 1. Howe K, Clark MM. Fouling of microfiltration and ultrafiltration
lation with lower doses of FeCl3 (10 mg/L) followed by PAC membranes by natural waters. Environ Sci Technol. 2002;36:3571-
adsorption of 0.5 g/L and UF removed the same amount of 3576.
organic matter, the majority of the DOC removal was achieved 2. Mameri N, Abdessemed D, Belhocine D, Lounici H. Treatment of
fishery washing water by ultrafiltration. AIChE J. 1996;67:169-175.
by the posttreatment of UF rather than by pretreatment. This 3. Shon HK, Vigneswaran S, Kim InS, Cho J, Ngo HH. The effect of
resulted in a significant decrease of flux in UF. For example, 10 pre-treatment to ultrafiltration of biologically treated sewage effluent:
mg/L of FeCl3 flocculation and 0.5 g/L of PAC adsorption A detailed effluent organic matter (EfOM) characterization. Water Res.
removed only 50% of DOC, thus delegating UF to remove 2004;38:1933-1939.
another 40% of DOC. 4. Abdessemed D, Nezzal G. Treatment of primary effluent by coagula-
tion–adsorption– ultrafiltration for reuse. Desalination. 2002;152:367-
(2) Flocculation with FeCl3 doses from 50 to 68 mg/L 373.
decreased the relative intensity of the majority of the large 5. Al-Malack MH, Anderson GK. Coagulation– crossflow microfiltration
RMM organic matter, such as tannic acid, sodium lignin sul- of domestic wastewater. J Membr Sci. 1996;121:59-65.
fornate, sodium lauryl sulfate, and arabic acid, and some of the 6. Aguiar A, Lefebvre E, Rahni M, Beguebe B. Relationship between raw
water TOC and the optimum coagulant dose (iron III chloride). Envi-
small RMM organic matter. However, the relative intensity of ron Technol. 1996;17:381-389.
the smallest RMM organic compounds (⬃300 Da) could not be 7. Arana J, Melian JAH, Rodriguez JMD, Diaz OG, Viera A, Pena JP,
removed by flocculation. The flocculation with lower FeCl3 Sosa PMM, Jimenez VE. TiO2-photocatalysis as a tertiary treatment of
doses (⬍40 mg/L) could not remove the relative intensity of naturally treated wastewater. Catal Today. 2002;76:279-289.
the large RMM organic matter (36,300 Da) even after a post- 8. Lin CF, Huang YJ, Hao OJ. Ultrafiltration processes for removing
humic substances: Effect of molecular weight fractions and PAC
adsorption. Thus, the preflocculated effluent (with ⬍40 mg/L treatment. Water Res. 1999;33:1252-1264.
FeCl3) led to a significant decline of flux in UF operation. This 9. Seo GT, Ohgaki S, Suzuki Y. Sorption characteristics of biological
also shows that the large RMM organic matter in the waste- powdered activated carbon in BPAC-MF (biological activated carbon–
water is the main cause for the flux decline in UF. microfiltration) system for refractory organic removal. Water Sci Tech-
nol. 1997;35:163-170.
(3) PAC adsorption removed the relative intensity of the 10. Hong S, Elimelech M. Chemical and physical aspects of natural
majority of smaller RMM of 900, 600, and 200 Da from the organic matter (NOM) fouling of nanofiltration membranes. J Membr
preflocculated water with FeCl3 dose of ⱖ50 mg/L. Sci. 1997;132:159-181.
(4) The Mw values of OM in the wastewater and in the 11. Verwey EJW, Overbeek JTG. Theory of the Stability of Lyophobic
flocculated effluent were 30,000 Da (initial), ⬎25,000 (after Colloids. Amsterdam, The Netherlands: Elsevier Scientific; 1948.
12. Cho J, Amy G, Pellegrino J. Membrane filtration of natural organic
flocculation with FeCl3 dose of ⱕ40 mg/L), and ⬍1000 (after matter: Factors and mechanisms affecting rejection and flux decline
flocculation with ⱖ50 mg/L FeCl3). with charged ultrafiltration (UF) membrane. J Membr Sci. 2000;164:
89-110.
13. Lee S, Park G, Amy G, Hong SK, Moon SH, Lee DH, Cho J.
Determination of membrane pore size distribution using the fractional
Acknowledgments
rejection of nonionic and charged macromolecules. J Membr Sci.
This research was funded by Australian Research Council (ARC) dis- 2002;201:191-201.
covery grant. The supports of the Kwanjeong Educational Foundation for 14. Huber SA. Evidence for membrane fouling by specific TOC constit-
the scholarship and ␮HM laboratory in GIST for the analysis of molecular uents. Desalination. 1998;119:229-234.
weight distribution are greatly appreciated. 15. Her NG. Identification and Characterization of Foulants and Scalants

AIChE Journal January 2006 Vol. 52, No. 1 215

 on NF Membrane. PhD Thesis. Boulder, CO: University of Colorado; 18. Perminova IV, Frimmel FH, Kudryavtsev AV, Abbt-Braun G, Hesse
2002:40-49. S, Petrosyan VS. Molecular weight characteristics of humic substances
16. Tandanier CJ, Berry DF, Knocke WR. Dissolved organic matter ap- from different environments as determined by size exclusion chroma-
parent molecular distribution and number-average apparent molecular tography and their statistical evaluation. Environ Sci Technol. 2003;
weight by batch ultrafiltration. Environ Sci Technol. 2000;34:2348- 37:2477-2485.
2353. 19. Mulder M. Basic Principles of Membrane Technology. 2nd Edition.
17. Vilge-Ritter A, Rose J, Masion A, Bottero JY, Laine JM. Chemistry Boston, MA: Kluwer Academic; 1996.
and structure of aggregates formed with Fe-salts and natural organic
matter. Colloid Surf A. 1999;147:297-308. Manuscript received Oct. 5, 2004, and revision received Apr. 29, 2005.

216 January 2006 Vol. 52, No. 1 AIChE Journal