w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6

Available online at www.sciencedirect.com

ScienceDirect

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

Dependence of floc properties on coagulant type,

dosing mode and nature of particles

Wenzheng Yu a,b, John Gregory a,*, Luiza C. Campos a, Nigel Graham b

a
Department of Civil, Environmental and Geomatic Engineering, University College London, Gower Street, London
WC1E 6BT, UK
b
Department of Civil and Environmental Engineering, Imperial College London, South Kensington Campus, London
SW7 2AZ, UK

article info abstract

Article history: Kaolin suspensions were coagulated with AlCl3 and a high-basicity PACl at pH 7, at dosages
Received 26 June 2014 that gave zeta potentials close to zero. The actions of the two coagulants were completely
Received in revised form different. With AlCl3, the formation of an amorphous hydroxide precipitate played a
23 September 2014 dominant role. When the coagulant was added to the suspension, flocs grew rapidly and
Accepted 25 September 2014 incorporated most of the kaolin particles within the hydroxide precipitate. When the
Available online 7 October 2014 suspension was added some time after the coagulant, the clay particles were found to be
mainly on the outer floc surfaces, although the floc size was about the same. The light
Keywords: scattering properties of the flocs were very dependent on the number and location of
Coagulation particles in the precipitate. With PACl, delaying the addition of kaolin had no influence on
Al(III) the final floc properties.
PACl In further tests, different suspensions over a range of concentrations were coagulated
Hydroxide precipitate with alum at pH 7. Monitoring by a turbidity fluctuation technique showed an apparent
Turbidity fluctuations increase in floc size with increasing particle concentration. However, floc sizes determined
from microscope images were very nearly constant, independent of particle nature and
concentration. With different particle types, the monitoring results were greatly dependent
on the light scattering properties of the particles.
Particles incorporated within hydroxide flocs appeared to have no influence on floc
properties, such as size and strength.
2014 Elsevier Ltd. All rights reserved.

been investigated in many previous studies (Dempsey et al.,

1. Introduction
Coagulation is a widely-used process for conventional water
treatment, and Al(III) is a very common coagulant in treating
drinking water. Pre-hydrolyzed coagulants such as poly-
aluminum chloride (PACl) have been known for many years
(e.g. Bottero et al., 1980) and their use in water treatment has
1984; Gray et al., 1995; Matsui et al., 1998; Xiao et al., 2008;
Hussain et al., 2013).
Al(III) and PACl have different coagulation mechanisms
and efficiencies. PACl has often been found to be more effi-
cient in the removal of natural organic matter (NOM) from
high alkalinity and high pH water than traditional coagulants
based on Al(III) and Fe(III) (Yan et al., 2008a) or in neutral pH

* Corresponding author. Tel.: 44 20 7689 7818.

E-mail addresses: j.gregory@ucl.ac.uk, johngregory15@hotmail.com (J. Gregory).
http://dx.doi.org/10.1016/j.watres.2014.09.045
0043-1354/ 2014 Elsevier Ltd. All rights reserved.
120 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6
water through charge neutralization and bridging (Yan et al.,
2008b), especially at low dosage (Zouboulis et al., 2008). PACl
proved to be a more efficient coagulant than alum and pro-
duced treated water with low turbidity and residual
aluminum content without requiring acid addition for pH
adjustment and subsequent base addition for re-stabilizing
the water (Zouboulis et al., 2008; Zarchi et al., 2013). PACls
with higher basicity (OH/Al) are more efficient for turbidity
and NOM removal (Yan et al., 2008b,c). PACls with a high
content of the Al tridecamer (Al13) can give larger flocs, higher
floc growth rate and shorter floc settling times than other Al-
based coagulants (Hussain et al., 2013). However, some re-
searchers have come to different conclusions. Trinh and Kang
(2011) found that PACl was more effective in removing parti-
cles from water, but less effective in removal of NOM
compared with alum. Shi et al. (2007) also found that co-
agulants with preformed Al species were less effective than
conventional Al coagulant in removing humic acid with high
molecular weight and hydrophobic character, giving smaller
flocs.
The coagulation mechanisms of alum and PACl are
different. Under most practical conditions, precipitate for-
mation and sweep flocculation occur with alum, whereas
charge neutralization by the Al13 polycation, electrostatic
patch and bridging interactions are more important for PACl
(Wu et al., 2009). After dosing into water, Al(III) undergoes a
series of hydrolysis, polymerization, precipitation and aggre-
gation processes. Amorphous Al(OH)3 has a low solubility
around neutral pH and can remove impurities by a combina-
tion of adsorption and precipitate enmeshment. PACl also
forms precipitates at sufficiently high pH, but the polymeric
structure of PACl is retained upon precipitation yielding a
solid phase with different light scattering characteristics,
electrophoretic mobility and solubility than amorphous
Al(OH)3 (Van Benschoten and Edzwald, 1990). Klute (1990)
found that the preformed species in PACls are more stable
than those formed in situ after alum addition; Matsui et al.
(1998) assumed that those preformed species could destabi-
lize particles faster. The preformed polymeric species con-
tained in PACl are fairly stable and highly positively charged
and these species can be adsorbed on surfaces of clay and
other particles and undergo surface aggregation and rear-
rangement to form electrostatic patches (Ye et al., 2007). The
Al13 polycation ([Al13O4(OH)24(H2O)12]7) has a tetrahedral
Al(O)4 center surrounded by 12 octahedrally coordinated Al
atoms (Johansson, 1960). Further hydrolysis or rearrangement
of Al13 subunits can occur (Bottero et al., 1987; Bertsch, 1987).
Both Al(OH)3-rich and Al13-aggregate flocs do not possess
long-range crystalline structure, but Al13-like crystals have
been found in the flocs formed by PACl and the chemical
compositions of the Al(OH)3 precipitates and Al13 aggregates
are markedly different (Lin et al., 2009).
In recent years, some researchers have focused on the floc
breakage and re-growth process to better understand floc
characteristics and coagulation mechanisms. It has been
found that only limited re-growth of broken flocs occurred for
alum and PACl, indicating a significant irreversibility of the
floc break-up process (Yukselen and Gregory, 2004). It is likely
that there are some changes on the surface of broken flocs.
According to our previous research (Yu et al., 2011b), if sweep
coagulation dominates the coagulation mechanism, a small
additional dosage of alum added at the time of floc breakage,
can greatly improve the re-growth ability of broken flocs.
However a similar small addition of high-basicity PACl has no
effect on floc re-growth, indicating that floc surfaces are quite
different in the two cases.
In this work, coagulation of clay suspensions with Al(III)
coagulants and a high-basicity PACl has been carried out with
different dosing conditions and with different suspensions.
The main aims of the work were:
 To compare the effects of Al(III) and PACl coagulants on the
structure of flocs, by optical microscopy
 To gain further insight into the mechanism of floc forma-
tion by adding particles at different times after coagulant
addition
 To investigate the effect of the nature and number of par-
ticles included within hydroxide flocs on the apparent floc
size determined by an optical technique
2. Materials and methods
2.1. Materials
Kaolin clay (Imerys, St Austell, Cornwall, UK) was used as a
model suspension. 200 g of kaolin was dispersed in 500 mL of
deionized water in a high-speed blender. To obtain full
dispersion it was necessary to raise the pH of the suspension
to about 7.5, which was achieved by adding 5 mL of 0.1 mM
NaOH. After blending at 4000 rpm for 10 min the clay sus-
pension was diluted to 1 L with deionized water and allowed
to stand overnight (12 h) in a measuring cylinder. The top
800 mL was decanted and its solids content was determined
gravimetrically and found to be 133 g/L.
Spherical silica particles (Geltech Inc) were dispersed in
dilute NaOH solution to give a concentration of 10 g/L. These
particles were porous, with a density of 2 g/cm3 and a mean
diameter of 1.5 mm, with a narrow size distribution.
Polystyrene latex was prepared by emulsifier-free poly-
merization of styrene, with potassium persulfate as initiator.
The particles were highly uniform with a mean diameter of
1.6 mm, determined by an Elzone particle counter. The sample
had been purified by repeated sedimentation, decantation and
redispersion in deionized water and the final concentration
was 40 g/L.
AlCl3 and PACl25 (OH/Al ratio 2.5) were used as co-
agulants. AlCl3 was purchased from Aldrich. The PACl25 was
prepared using a laboratory base titration method at room
temperature (Wang et al., 2004) as described in detail in our
previous paper (Yu et al., 2011a). The final concentration of
both coagulants is 0.1 M (as Al). The samples, after being left to
age one week, were analyzed using the ferron method as
described by Wang et al. (2004). AlCl3 was found to contain
97.1% Ala (mainly monomer Al) and PACl25 93.6% Alb (mainly
Al13).
Aluminum sulfate, Al2(SO4)3, NaHCO3 and HCl were of
analytical reagent grade and prepared as a 0.1 M solutions in
deionized water.
 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6
2.2. Jar test
For the most of the flocculation tests, the stock kaolin sus-
pension was diluted in water (800 mL) to give a clay concen-
tration of 50 mg/L. The flocculation tests were carried out in
5 mM NaHCO3 solutions, made up with deionized (DI) water.
During the jar test, water pH was maintained at 7.0 via prior
addition of a pre-determined amount of 0.1 M HCl. The dosage
of AlCl3 and PACl25 was 0.1 mM and 0.02 mM respectively (as
Al), since the zeta potential of the flocs was near zero at these
dosages (see below). The temperature throughout was
25 1  C. The jar tests were carried out with a Flocculator 2000
(Kemira Kemi, Helsingborg, Sweden), which enables stirring
speeds and times to be set in advance.
In order to investigate the effect of delaying the addition of
kaolin suspension, a fixed dosage of alum or PACl25 was added
into the test solution and at the same time the stirring of
flocculator started. The stirring speed was set at 200 rpm
(G z 184 s1) for 60 s, to give rapid mixing and then 50 rpm
(23 s1) to allow floc growth. Kaolin suspension (50 mg/L) was
added either initially (i.e. before rapid mixing) or at 0.5, 1, 2, 5
or 10 min after coagulant addition. After kaolin addition,
50 rpm stirring was maintained for a further 15 min.
The effects of different types of particle at different con-
centrations were investigated by a similar jar test procedure.
However, in this case, the flocs were broken at high shear
(200 rpm for 1 min) and then allowed to re-grow at 50 rpm. In
these tests alum at 0.1 mM Al was the coagulant and the pH
was 7.
2.3. PDA measurement
Experiments on the kinetics of formation, breakage and sub-
sequent re-growth of flocs were performed using the
Turbidity Fluctuation method (PDA3000, Rank Brothers, UK).
The procedure was similar to that of Yukselen and Gregory
(2004). The PDA 3000 measures the average transmitted light
intensity (dc value) and the root mean square (rms) value of
the fluctuating component of the intensity. The ratio (rms/dc)
provides a sensitive measure of particle aggregation. The ratio
value is strongly correlated with floc size and always increases
as flocs grow larger. It is referred to here as the Flocculation
Index (FI), which significantly increases as aggregation occurs,
and decreases when aggregates are broken. After the FI value
reached an initial steady value, coagulant was added to the
suspension and the dc and rms values recorded every second.
FI values depend essentially on the number concentration
of particles (flocs) and their extinction (or scattering) cross
sections (Gregory, 2009a). The FI value always increases as
particles aggregate and decreases when aggregates are
broken. It depends essentially on the number concentration,
N, and effective light scattering cross-section, C, of aggregates
(flocs). For quite large flocs, the scattering cross section should
be proportional to the geometric cross-sectional area of a floc.
If all the flocs are of equal size then the FI value is given by:
1 =
FI KCN 2
(1)
where K is a constant that depends only on the instrument
(essentially on the dimensions of the light beam).
121
As flocs grow, their number concentration decreases, but
the increasing scattering cross section more than compen-
sates, so that the FI value increases.
Of course, in a real flocculating suspension, flocs will have
a range of sizes, and hence a range of scattering cross sections.
In that case, C can be regarded as an average scattering cross
section, but larger flocs have a greater influence on the
average value.
When clay suspensions are dosed with hydrolyzing co-
agulants, the resulting flocs are composed of amorphous
hydroxide precipitate with included clay particles. In such
cases, the clay particles can give the dominant contribution
to the optical properties of the flocs and the scattering cross
section may be considerably increased, even though the
actual floc size may be little changed (Gregory, 2009b).
2.4. Floc images and electrophoretic mobility
During the slow stirring period, samples of flocs were taken
from below the surface of the suspension with a hollow glass
tube with an inner diameter of 5 mm. One end of tube was
inserted 3.0 cm below the surface and the other end was
covered by a finger, and then the sample was carefully with-
drawn. After transferring the sample onto a flat microscope
slide, an image of the flocs was captured by an optical mi-
croscope with a CCD camera (GE-5, Aigo, China). The images
were analyzed by Image J software and floc diameters were
determined from the floc area.
Electrophoretic mobility (EM) was measured on the kaolin
suspension before coagulant addition and for samples after
coagulant dosing and 1 min of rapid mixing, by a Particle
Electrophoresis Apparatus Mk 2 (Rank Brothers Ltd, Cam-
bridge, UK). In addition, the EM of flocs after 1 min of breakage
at high stirring speed was measured. The delay time of the
measurement was about 40 s. The average EM value for a
sample was determined from 20 measurements.
3. Result and discussion
3.1. Electrophoretic mobility of flocs
AlCl3 and PACl25, at a given dosage, cause different EM of flocs.
Prior to dosage, the EM of kaolin particles was negative
(around 3.9 mm s1 V1 cm1). The addition of low dosages of
AlCl3 caused a rapid increase of EM and the charge became
nearly stable at the dosage of 0.1 mM Al. Higher dosage caused
only a slight further increase of EM. For PACl25, very low
dosages caused a steep rise in EM and charge reversal
occurred at about 0.02 mM Al. Breakage of flocs caused no
significant change in EM. The dosages of PACl25, and AlCl3
required to give EM z 0 (the isoelectric point) were 0.02 mM
and 0.1 mM, respectively. In the following results, just these
two dosages were used (Fig. 1).
3.2. Effect of delayed addition of kaolin suspension
The effect of different delay times with AlCl3 and PACl25 on the
apparent coagulation efficiency was explored by adding kaolin
122 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6
Fig. 1 e Variation of electrophoretic mobility of kaolin flocs
with coagulant dosage at pH 7.
particles at different times after coagulant addition and the
results are shown in Fig. 2.
The results were significantly different for AlCl3 and
PACl25. From Fig. 2a, when the kaolin and AlCl3 were added at
the same time, the FI value was the highest, and the final
plateau FI value decreased markedly as the delay time
increased. However, we cannot assume that the lower FI value
implies smaller flocs, because of different light scattering
properties of the flocs (see 3.3).
With PACl25 as the coagulant, the results were very
different. As shown in Fig. 2b, the rate of increase of FI did not
change as the delay time increased. Also the plateau FI values
for different delay times were nearly the same, whereas for
AlCl3, the coagulation rate decreased with delay time. It can be
assumed that PACl25, in solution at pH 7, does not change
significantly during the delay time because of its high positive
charge. Thus, its ability to coagulate clay particles remains
Fig. 2 e The effect of delay time before adding kaolin: (a) 0.1 mM AlCl3, (b) 0.02 mM PACl25. AlCl3 or PACl25 was added into the
test solution (800 mL) and at the same time the stirring of flocculator started. The stirring speed was set at 200 rpm for 60 s
and then at 50 rpm. Kaolin suspension was added at 0, 0.5, 1, 2, 5 and 10 min, after coagulant addition. Note that the 0.5 and
1 min additions were during the rapid mix (200 rpm) phase.
unaffected by the delay. With AlCl3 at pH 7, precipitation to
give an amorphous precipitate occurs during the delay time
and this markedly reduces the coagulation efficiency.
3.3. Images of flocs
In order to see any changes in floc characteristics when the
kaolin was added later than coagulants, images of flocs were
captured after flocculation. Usually, after AlCl3 is added into
the kaolin suspension at pH 7, kaolin particles will adhere to
the colloidal precipitate particles, which then aggregate to
form larger flocs. These flocs essentially consist of amorphous
precipitate with embedded kaolin particles, distributed
randomly throughout the floc (sweep floc). This idea is
confirmed by the micrograph in Fig. 3a, where the clay parti-
cles can be seen within the amorphous hydroxide precipitate.
By contrast, when kaolin was added 10 min after addition
of AlCl3, the appearance of flocs was very different (Fig. 3b).
The size of flocs in Fig. 3b is nearly the same as those in Fig. 3a,
but the kaolin particles were distributed in a very different
way. In Fig. 3b, most of the particles are located on the outside
of the flocs, with very few in the interior. Also each floc in
Fig. 3b had fewer kaolin particles than those in Fig. 3a. During
the 10 min after coagulant addition hydroxide precipitates are
formed and these grow to large flocs. Kaolin particles added
subsequently have difficulty in entering these flocs and
adhere mainly to the outer floc surface. Comparing Fig. 2a
with Fig. 3a and b, it seems that the FI value was not deter-
mined by the size of flocs, but by their kaolin content. The very
small increase in FI before kaolin addition is because hy-
droxide flocs scatter little light. Only when kaolin is added
does the FI value rise significantly.
The images of flocs produced by PACl25 (Fig. 3c and d) are
completely different from those for AlCl3. Firstly, the PACl
flocs are considerably smaller and denser. Also it is clear that
there is no significant effect of delaying the addition of kaolin
suspension. These observations are consistent with a charge-
neutralization mechanism for PACl coagulation, either by Al13
polycations or small aggregates of these. It may be that
 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6
Fig. 3 e Images of flocs with different delay times before the addition of kaolin particles: (a) AlCl3 with no delay, (b) AlCl3 with
10 min delay, (c) PACl25 with no delay, (d) PACl25 with 10 min delay.
polycation aggregates are able to adsorb in a patchwise
manner and bind particles by electrostatic attraction or by a
bridging mechanism. It seems that the nature of the coagulant
species formed by PACl25 does not change significantly during
the 10 min delay time and there is no apparent difference in
their behavior as coagulants.
The flocculation results in Fig. 2 can be explained by the
images in Fig. 3. In the case of AlCl3, with no delay before
adding kaolin, the FI value reaches about 1.4 and the corre-
sponding value for PACl25 is only slightly less, despite the
great difference in floc size shown in Fig. 3. Eq (1) shows that
the FI value depends inversely on the number concentration,
N, and directly on the light scattering cross section, C, of
particles or flocs. In the case of amorphous hydroxide flocs,
without included particles, C is very low, so that, even for large
flocs (and hence low N) the FI value does not increase greatly
until kaolin is added. However, when kaolin particles are
included in the amorphous flocs, they can greatly increase the
effective value of the scattering cross section. So, the scat-
tering power of a hydroxide floc depends largely on the
included particles and different FI values reflect different
particle content and not necessarily floc size (Gregory, 2009b).
In Fig. 2a, the decrease of FI value with increasing delay time
can be explained in this way. It is likely that the flocs have
about the same size, independent of delay time, but the
decreasing FI values are due to the smaller number of kaolin
particles in each floc. Similar effects with other particles are
shown in the next section.
The flocs produced by PACl25 (Fig. 3c and d) are more dense
and have higher scattering cross-sections than the hydroxide
123
flocs, so that, despite their smaller size, they give about the
same FI value.
3.4. Effect of different particle concentrations
Suspensions of, kaolin, silica and polystyrene latex at
different concentrations (2e10 mg/L) were coagulated by alum
at a dosage of 0.1 mM Al and at pH 7. The results are shown in
Fig. 4. In all cases, the FI value increased during the floccula-
tion process, but to very different extents, depending on the
particle concentration. As the concentration of kaolin
increased from 2 mg/L to 10 mg/L, the first plateau of the FI
value increased as well as the corresponding value for the re-
grown flocs. With the other particles, similar trends were
observed. The right-hand graphs in Fig. 4 show plateau FI
values plotted against particle concentration. In all cases good
linear correlations were found for the flocs before and after
breakage.
There are differences between FI values for different
particles at equivalent concentrations. For the flocs before
breakage, the slopes of the regression lines in Fig. 4 show that
silica particles give about 25% higher FI values than kaolin,
and that latex particles show a 7-fold increase over kaolin.
Some of this effect can be explained by the different densities
of the particles. These are about 2.7, 2.0 and 1.05 g/cm3 for
kaolin, silica and latex respectively. The particle concentra-
tions in Fig. 4 are expressed on a mass basis, but the volume
concentrations differ and these are more relevant to light
scattering and turbidity. For the same mass concentration,
the volume fractions of kaolin, silica and latex would be in
124 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6

Fig. 4 e Formation breakage and re-growth of alum flocs at pH 7, with different particles at different concentrations. The left-
hand plots show the variation of FI value with time. The corresponding maximum FI values before floc breakage and after
re-growth are shown on the right-hand side. (a, b) kaolin, (c, d) silica, (e, f) latex.

the ratio 1:1.3:2.6. So, for silica, the increase in FI value over
kaolin is nearly the same as the volume fraction increase.
However, for latex, the higher volume fraction accounts for
less than half of the increase in FI value, indicating that the
polystyrene particles are more optically dense than kaolin
and silica.
Images of flocs formed for the different particle concen-
trations, after breakage and re-growth, were analyzed and floc
sizes derived as before. At least 100 flocs were sampled in each
case and the average sizes and standard deviations are shown
in Table 1. For low concentrations of kaolin and silica, the flocs
were quite difficult to resolve by light microscope and some
values are omitted from the Table. All of the floc sizes in Table
1 are close to 170 mm, despite the very different FI values
in Fig. 4. This confirms the idea that the FI value does not give
an absolute indication of size for hydroxide flocs, since it
 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6
Table 1 e Average floc size after re-growth with different
particles at different concentrations. Coagulation with
alum (0.1 mM Al) at pH 7. Standard deviations given in
brackets. (ND: not determined).
Particles Floc size (mm)
mg/L
2 4 6 8 10
Kaolin ND ND 174 (65) 170 (62) 171 (62)
Silica ND 174 (67) 172 (62) 168 (61) 165 (58)
Latex 168 (68) 167 (70) 164 (62) 161 (63) 165 (74)
depends greatly on the nature and number of particles
included within the flocs. Nevertheless, results such as those
in Figs. 2 and 4 can be very useful in deriving relative infor-
mation on flocculation rate, floc breakage and re-growth.
It is also noteworthy that the presence of different par-
ticles within hydroxide flocs has no significant effect on floc
properties, such as strength, breakage and re-growth. The
plateau floc size under given shear conditions is a good
indication of floc strength. This is not affected by included
particles, over the concentration range 2e10 mg/L. The floc
properties are determined by the nature of the hydroxide
precipitate.
4. Conclusions
The main conclusions of this work are:
1. With AlCl3 as coagulant at pH 7, hydroxide precipitation
plays the dominant role. If kaolin suspension is added
some time after the coagulant, when some precipitation
has occurred, then the resulting flocs are very different in
structure from those formed with addition of coagulant to
a kaolin suspension. In the latter case, clay particles are
distributed throughout the hydroxide floc, whereas with a
delay time particles are found mostly on the outer floc
surface and fewer particles are removed. For PACl25,
delaying the addition of kaolin had no significant effect on
floc growth, because the coagulation mechanism was quite
different, being mainly based on charge neutralization.
2. With alum at pH 7 and different suspension concentra-
tions, the response of a turbidity fluctuation monitor de-
pends directly on the included particles within hydroxide
flocs. Over a certain concentration range, there is a linear
relationship between Flocculation Index and the particle
concentration. However, floc size is independent of particle
concentration. The FI value is also influenced by the nature
of the particles. Those that scatter more light give higher FI
values.
3. The strength of flocs produced by sweep coagulation is
almost entirely determined by the nature of the hydroxide
precipitate and not by the included particles.
4. The light scattering properties of hydroxide flocs depend
mostly on the included particles and this has important
implications for monitoring of floc size by light scattering
methods.
125
Acknowledgments
This work was supported by National Natural Science Foun-
dation of China (Grants 51108444). The authors wish to
acknowledge the support of the European Commission in
providing a Marie Curie International Incoming Fellowship for
Dr Wenzheng Yu.
references
Bertsch, P.M., 1987. Conditions for Al-13 polymer formation in
partially neutralized aluminum solutions. Soil Sci. Soc. Am. J.
51, 825e828.
Bottero, J.Y., Cases, J.M., Fiessinger, F., Poirier, J.E., 1980. Studies of
hydrolyzed aluminum-chloride solutions .1. Nature of
aluminum species and composition of aqueous-solutions. J.
Phys. Chem. 84, 2933e2939.
Bottero, J.Y., Axelos, M., Tchoubar, D., Cases, J.M., Fripiat, J.J.,
Fiessinger, F., 1987. Mechanism of formation of aluminum
trihydroxide from keggin Al 13 polymers. J. Colloid Interface
Sci. 117, 47e57.
Dempsey, B.A., Ganho, R.M., O'Melia, C.R., 1984. The coagulation
of humic substances by means of aluminum salts. J. Am.
Water Works Assoc. 76, 141e150.
Gray, K.A., Yao, C.H., O'Melia, C.R., 1995. Inorganic metal
polymers - preparation and characterization. J. Am. Water
Works Assoc. 87, 136e146.
Gregory, J., 2009a. Monitoring particle aggregation processes. Adv.
Colloid Interface Sci. 147-48, 109e123.
Gregory, J., 2009b. Optical monitoring of particle aggregates. J.
Environ. Sci. 21, 2e7.
Hussain, S., van Leeuwen, J., Chow, C., Beecham, S.,
Kamruzzaman, M., Wang, D.S., Drikas, M., Aryal, R., 2013.
Removal of organic contaminants from river and reservoir
waters by three different aluminum-based metal salts:
coagulation adsorption and kinetics studies. Chem. Eng. J. 225,
394e405.
Johansson, G., 1960. On the crystal structures of some basic
aluminium salts. Acta Chem. Scand. 14, 771e773.
Klute, R., 1990. Destabilization and aggregation in turbulent pipe
flow. In: Hahn, H.H., Klute, R. (Eds.), Chemical Water and
Wastewater Treatment. Springer-Verlag, Berlin, pp. 33e54.
Lin, J.L., Huang, C.P., Chin, C.J.M., Pan, J.R., 2009. The origin of
Al(OH)(3)-rich and Al-13-aggregate flocs composition in PACl
coagulation. Water Res. 43, 4285e4295.
Matsui, Y., Yuasa, A., Furuya, Y., Kamei, T., 1998. Dynamic
analysis of coagulation with alum and PACl. J. Am. Water
Works Assoc. 90, 96e106.
Shi, B.Y., Wei, Q.S., Wang, D.S., Zhu, Z., Tang, H.X., 2007.
Coagulation of humic acid: the performance of preformed and
non-preformed Al species. Colloids Surf. A Physicochem. Eng.
Aspects 296, 141e148.
Trinh, T.K., Kang, L.S., 2011. Response surface
methodological approach to optimize the coagulation-
flocculation process in drinking water treatment. Chem. Eng.
Res. Des. 89, 1126e1135.
Van Benschoten, J.E., Edzwald, J.K., 1990. Chemical aspects of
coagulation using aluminum salts .1. Hydrolytic reactions of
alum and polyaluminum chloride. Water Res. 24, 1519e1526.
Wang, D.S., Sun, W., Xu, Y., Tang, H.X., Gregory, J., 2004.
Speciation stability of inorganic polymer flocculant-PACl.
Colloids Surf. A Physicochem. Eng. Aspects 243, 1e10.
Wu, X.H., Ge, X.P., Wang, D.S., Tang, H.X., 2009. Distinct
mechanisms of particle aggregation induced by alum and
126 w a t e r r e s e a r c h 6 8 ( 2 0 1 5 ) 1 1 9 e1 2 6
PACl: floc structure and DLVO evaluation. Colloids Surf. A
Physicochem. Eng. Aspects 347, 56e63.
Xiao, F., Ma, J., Yi, P., Huang, J.C.H., 2008. Effects of low
temperature on coagulation of kaolinite suspensions. Water
Res. 42, 2983e2992.
Yan, M.Q., Wang, D.S., Qu, J.H., Ni, J.R., Chow, C.W.K., 2008a.
Enhanced coagulation for high alkalinity and micro-polluted
water: the third way through coagulant optimization. Water
Res. 42, 2278e2286.
Yan, M.Q., Wang, D.S., Ni, J.R., Qu, J.H., Chow, C.W.K., Liu, H.L.,
2008b. Mechanism of natural organic matter removal by
polyaluminum chloride: effect of coagulant particle size and
hydrolysis kinetics. Water Res. 42, 3361e3370.
Yan, M.Q., Wang, D.S., Yu, J.F., Ni, J.R., Edwards, M., Qu, H.H.,
2008c. Enhanced coagulation with polyaluminum chlorides:
role of pH/Alkalinity and speciation. Chemosphere 71,
1665e1673.
Ye, C.Q., Wang, D.S., Shi, B.Y., Yu, J.F., Qu, J.H., Edwards, M.,
Tang, H.X., 2007. Alkalinity effect of coagulation with
polyaluminum chlorides: role of electrostatic patch. Colloids
Surf. A-Physicochem. Eng. Aspects 294, 163e173.
Yu, W.Z., Gregory, J., Campos, L., Li, G.B., 2011a. The role of mixing
conditions on floc growth, breakage and re-growth. Chem.
Eng. J. 171, 425e430.
Yu, W.Z., Gregory, J., Campos, L.C., 2011b. Breakage and re-growth
of flocs: effect of additional doses of coagulant species. Water
Res. 45, 6718e6724.
Yukselen, M.A., Gregory, J., 2004. The reversibility of floc
breakage. Int. J. Miner. Process. 73, 251e259.
Zarchi, I., Friedler, E., Rebhun, M., 2013. Polyaluminium chloride
as an alternative to alum for the direct filtration of drinking
water. Environ. Technol. 34, 1199e1209.
Zouboulis, A., Traskas, G., Samaras, P., 2008. Comparison of
efficiency between poly-aluminium chloride and aluminium
sulphate coagulants during full-scale experiments in a
drinking water treatment plant. Sep. Sci. Technol. 43,
1507e1519.