Environment
International, Vol. 24, No. 8, pp. 899-910, 1998
Copyright 01998 Elsevier Science Ltd
Printed in the USA. All rights resewed
0160-4120/98 $19.00+.00
PIISO160-4120(98)00073-7
SYNTHESIS AND SPECIATION OF
POLYALUMINUM
CHLORIDE FOR WATER
TREATMENT
Yun-Hwei Shen
Department of Environmental Protection Technology, National Pingtung University of Science
and Technology, Pingtung, Taiwan 91207, ROC
Brian A. Dempsey
Department of Civil and Environmental
University Park, PA 16802, USA
email: [email protected]
Engineering, The Pennsylvania State University,
EI 9803-135 M (Received29 March 1998; accepted 8 August 1998)
In this study, synthesis and speciation of polyaluminum chloride (PAC) for application in water
treatment was investigated using a calorimetric speciation method. It was possible to produce
stable preparations of PAC solutions in which a relatively stable cationic polymer predominated.
The mode of preparation has a dramatic effect on the composition of PAC preparation. Some
important parameters such as hydroxyl ligand number, mixing intensity, base injection rate and
hia
sbencc ~td
method, and aging were identified in this study. @l$@s
INTRODUCTION
Alum (aluminum sulfate) is one of the most widely
used coagulant for water treatment in the United States
and has been proven to be an effective coagulant for
the removal of certain contaminants, turbidity and
color. In recent years, the preformed polymeric aluminum salts have been used with some success. Polyaluminum chloride (PAC) may be produced by adding
base to aluminum chloride until an empirical formula
ofAl(OH
(with n from 1.O to 2.5) is achieved. A
variety of species can be formed when stock solutions
of PAC are added to a raw water. O’Melia and
Dempsey (1982), in a review of work done using a
wide variety of PAC coagulants, proposed that some
PAC formulations may contain aluminum precipitates.
The positively charged precipitates of AI(OH)
may
improve flocculation kinetics in turbidity removal and
899
adsorb humic substances. Based on the analysis of
redissolved precipitates, solubility test, turbidity data
and electrokinetic measurements, Van Benschoten and
Edzwald (1990) concluded that alum and PAC precipitate to form different solid phases; the polymeric
structure remains intact within the PAC precipitate and
particles are more positively charged and produce
lower turbidity than for alum floe. Dempsey et al.
(1985) have investigated the benefits of PAC relative
to alum. The results indicate that PAC is especially
effective when the concentration
of fulvic acid or
other species with high coagulant demand is low, or
when the pH falls outside the range of 5.5 to 7:O.
O’Melia et al. (1989) concluded that PAC coagulants
are effective at lower dosages than other aluminum
preparations for the coagulation of high turbidity
Y.-H. Shen and B.A. Dempsey
900
waters, particularly at low temperature or acidic pH,
also that PAC is an effective filter aid for low turbidity
waters, providing for destabilization and subsequent
filtration of particles at acidic and neutral pHs. Bottero
et al. (1980) reported that partially neutralized aluminum chloride solutions were effective coagulants for
clay suspensions. Some French investigators (Bottero
et al. 1980; Leprince et al. 1984) advocate on-site preparation of these coagulants by partial neutralization
or by heating, permitting a specific tailoring of the
inorganic “polymeric” coagulant to the water to be
treated. Several investigators have successfully prepared polyaluminum coagulants in the laboratory (Van
Benschoten and Edzwald 1990; Dempsey et al. 1985;
Parthasarthy and Buffle 1985; Yao 1987). There is still
confusion regarding the actual dominant aluminum
species in PAC preparations and regarding the relationship between some important synthesis parameters
and PAC speciation. In this study, effects of hydroxyl
ligand number, mixing intensity, base injection rate
and method, and aging on the speciation of synthetic
PAC were examined using a calorimetric speciation
method.
length. Only monomeric aluminum can form a complex with ferron, but other species may dissociate
sufficiently in the acidic ferron reagents so that these
species can also eventually produce color. Thus this
reagent permits the evaluation of species of aluminum
based on the time required for development of color.
Different categories of aluminum species, for example,
monomers (Al”), fast reacting polymer (Alb’), slow
reacting polymer (Alb”), and AI(OH)
precipitate
(Al”) can be identified by plotting the absorbance
against time (Fig. 1). The time dependence of the reaction between aluminum species and ferron has been
described as (Smith 197 1):
Al,,, = Al” + Al,b( 1 - eekt)
where
Al,,, = Al that has reacted with ferron at time t,
Al” = monomeric aluminum,
Al,b = Alb in solution at time 0,
k = first order rate constant for Alb (min-‘)
Equation 1 can be rearranged
as follows:
ln(A1” + Al,b - Al,) = -kt + In(Al,b)
EXPERIMENTAL APPROACH
Chemicals
All chemicals utilized in this study were analytical
reagent grade or better. Distilled water was further
treated with a membrane filtering system (Millipore).
Stock solutions containing 1.0 and 0.1 mol L’ Al
were prepared with reagent-grade AlCl,.6Hz0 in CO,free distilled and demineralized water. Sodium hydroxide solutions were freshly prepared by dissolving
reagent grade NaOH pellets in distilled and demineralized water and were standardized against standard
HCl solutions.
Calorimetric speciation
test
The nature of the monomeric, polymeric, and solid
aluminum hydroxide species can be measured using
timed calorimetric speciation techniques such as the
ferron test (Smith and Hem 1972). The ferron test is an
indirect calorimetric analysis, in which the absorbance
at a wavelength of 370 nm is related to the concentration of reactive aluminum. Ferron is 7-iodo, 8-hydroxy
quinoline sulfonic acid and is soluble in water. The
complex compound formed between A13+and three
ferron molecules is non-ionic, soluble in organic solvents, and has an absorption peak at 370 nm wave-
(1)
(2)
The pseudo first-order rate constant k for specie Alb
can be then estimated by the slope of the first order
regression line on a ln(A1” + Al,b - Al,,) vs time plot.
A higher rate constant indicates smaller size polymer
(i.e., fast reacting polymer). In this study the ferron
test technique after Smith and Hem (1972) was used.
The original sample containing Al was directly injected into the ferron reagent with no intermediate
dilution step.
Reacfion apparatus
A 2-L plastic jar was fitted with a lid in which 4
inlets were drilled was used in the synthesis of PAC.
A Ross sureflow combination electrode was inserted
through one of the ports in the lid. The pH readings
were obtained with a Fisher Accumet pH meter (model
910). Another inlet contained a tygon tube through
which high quality nitrogen gas was bubbled vigorously through the solution to prevent CO, from accumulating in the system and affecting the pH. A
plastic capillary was inserted through a third opening
and used to inject base (NaOH) into the solution. The
aperture at the tip of the capillary was less than or
equal to 0.15 mm in diameter to minimize diffusion
of aluminum solution into the capillary during the
PAC synthesis and speciation
901
AIC Precipitate
Alb Polymeric
Species
Ala Monomeric
Species
1
i
Time
Fig. 1Schematic of calorimetric speciation method used in this study.
titration. Base was stored in a 50-mL plastic syringe
mounted on a syringe pump (Sage model 35 1) and was
pumped into the reactor at the desired flow rate. A
stainless steel mixing shaft with a dual-bladed
propeller was inserted through the fourth opening in
the lid; this shaft was driven by a variable speed motor
(Stir-pak mechanical stirrer) operated at a speed of
1800 rpm to provide vigorous mixing. No effort was
attempted to maintain a constant temperature but the
room temperature remained between 25 OCto 27.5 OC
during the course of this study.
Synthesis of PAC
One liter of 0.1 mol L-i [Al,] AlCl, solution was
titrated with final total volume of 20, 30, 40, 50, and
60 mL of 5 mol L’ NaOH, which represent ligand
numbers (n) of 1.O, 1.5,2.0,2.5, and 3.0, respectively.
The rate of base addition in every case was 0.09 mL
min“. The initial pH value was 2.9 and the final pH
values were 3.5,3.6,3.8,4.2,
and 6.3, respectively. All
the PAC preparations were clear at the end of the
titration except for the sample with ligand number 3.0.
Irreversible AI(OH)
precipitates occur at the final
stage of titration for this sample. All the PAC samples
were characterized using the timed calorimetric method described previously.
RESULTS AND DISCUSSION
Specia tion of PA C
Results for synthesis and speciation of PAC tests are
shown in Fig. 2. The gradual increase in absorbance
over time indicates the presence of Al polymer. For
the n=O sample, the species are primarily monomers,
and for n=3.0 Al (OH),,, precipitates dominate. The
distributions of Al species for PAC with different n
values were derived and are plotted in Fig. 3. The
concentration of polymeric species increases with n
until n exceeds 2.5. There is.a corresponding linear
increase in polymer concentration accompanied by a
linear decrease in monomer concentration. These data
indicate that the monomeric species condense into
polymeric species according to a fixed pattern and that
the polymer species is relatively stable during synthesis until n > 2.5.
Rate constants for the polymeric species in samples
were estimated using Equation 2. Results are shown in
Fig. 4. All first order regression lines in Fig. 4 have 8
valuesof0.90orhigher.
Theconsistencyoftherate
constants for the different PAC samples indicates the existence of a single stable polymeric species. The average
rate constant of this polymer observed in this study
(0.065min~‘withastandarddeviationofO.O058)agrees
reasonably well with determinations made by several
902
Y.-H. Shen and B.A. Dempsey
I
V
n = 0.0
V
0
n=l.O
n=1.5
:
?? n = 2.0
a
n = 2.5
A
n = 3.0
L
0.0 yYAAA
A
0
/
.
20
Time
Fig. 2. Absorbance
.-----
/
.I
60
40
80
100
( Min )
vs. time for PAC (0.1 M [Al& with different ligand number.
Monomer
Polymer
Precipitate
100
90
80
70
60
40
30
30
10
0
v.v
0.3
1.0
1.5
2.0
2.5
3.0
PHI / [AlI
Fig. 3. Al species frequency distributions
of PAC (0.1 M [Al& as a function of OH ligand number.
PAC synthesis and speciation
investigators for Al,, polymer (Van Benschoten and
Edzwald 1990; Bottero et al. 1980; Parthasarthy and
Buffle 1985). In order to better understand the PAC
synthesis process, an experiment was set up to study
the stochiometric
relationship between monomeric
Al species and added base. In this experiment, three
0.1 mol L-’ [Al,] AlCl, solution samples with different
sample volumes(2.0 L, 1.5 L, and 1.O L) were titrated
with 5 mol L“ NaOH in a manner similar to PAC synthesis procedures discussed previously. The amount of
total monomeric Al was monitored during the course
of titration. The results are shown in Fig. 5. The
stochiometric curves for three different tests tend to
have a unique slope (mole Al/mole OH) -0.5. This
indicates that for every two mole of NaOH added, one
mole of monomeric Al species will be consumed. The
result of this experiment supports the argument that a
relatively stable polymer is formed in a fixed reaction
path during PAC synthesis.
Effect of mixing
Mixing produces shear force to disperse the injected
base. In addition,mixing
affects the rate of micromixing and may control which of several competitive
reactions may dominate at the interface between the
mixed solutions. The effect of mixing on PAC
synthesis was studied by timed calorimetric tests.
Results are shown in Fig. 6 where all tests have
[AlJ = 0.1 mol L-‘, final n = 2.5, and total time of
base (5 mol L-’ NaOH) addition 8 h. Polymer production efficiency was 96 % at 1800 rpm (approx.
velocity gradient of 2800 s-l). The efficiency dropped
to 8.5 % at 300 rpm (approx. velocity gradient of 80 s-‘)
to 600 rpm (approx. velocity gradient of 200 s-‘), and
at 150 rpm (approx. velocity gradient of 25 S’ ) a
small amount of polymer is formed and AI(OH)
precipitates become dominant.
Inadequate mixing causes the persistence of local
concentration heterogeneities.
These local regions of
high OH-concentration
seem to favor the nucleation of
Al(OH),,,,. Inadequate mixing also tends to produce
polymeric species with smaller reaction rate constants
with ferron (k = 0.042 and 0.050 min-‘) when compared to polymers formed using more rapid mixing
(k = 0.070 and 0.065 min“). This observation is supported by data shown in Fig. 7.
If the stable polymer that is formed from AlCl, and
NaOH under intensive mixing is Al,,, then it is reasonable to propose that inadequate mixing produces
heterogeneous
OR ion distribution and favors the
903
formation of larger polymers that are more inert with
respect to reaction with ferron. Carrying the hypothesis a bit further, it is possible that inadequate mixing favors the formation of hexameric rings or fragments of an AI(OH)
structure.
Effect of Al concentration
The effect of Al concentration on PAC synthesis
was determined using the timed calorimetric
test.
Figure 8 shows the absorbance as a function of time
for three initial Al concentrations.
Lowering the Al
concentration from 10-l to 1O-‘.’ mol L-’ results in
about 8 % reduction in formation of polymer and a
similar increase in precipitated Al. Further lowering
Al concentration to lo”.’ mol L-‘, which is typical
for water treatment practice, reduces the polymer
content to 60 % of total Al and this is accompanied
by a 23 % increase in concentration
of monomers.
Once again, there is an increase in the formation of
Al that is precipitated.
The polymers
that are
produced in lo”.’ mol L’ [Al,] PAC are more inert
with respect to reaction with ferron (Fig. 9). Lowering
the Al concentration to 1O-*.’mol L-’ seems to produce
the same effect as inadequate mixing (i.e., reduced
polymer production
and increased precipitation).
Decreasing the initial [Al,] to lo”.’ mol L-’ extends
these trends, results in more inert “polymeric” material
and increases the fraction of [AI,] that is monomeric.
Effect of base injection
The effect of base injection on the synthesis of PAC
was studied by varying the base injection rate from
0.09 to 1.O mL min- 1, The upper end of this range was
constrained by the build up of high pressure within the
capillary tube, resulting in rupture of the feed line.
Figure 10 shows the results for tests performed on
PAC samples with 0.1 mol L-’ [Al,] and with n=2.5.
There is no significant change in Al speciation (as determined by the ferron reaction) due to changes in base
injection rate.
The effect of extreme concentration heterogeneities
on PAC synthesis was tested by employing a droplet
method for addition of base. In this method, base was
added from a lOO-mL Pyrex burette at a flow rate of
about 5 mL min-‘. It is well known that large shear
forces can not be exerted at gas-liquid interfaces. Thus
drop-wise addition of base into agitated solution results in extremely heterogeneous
base distribution.
This concentration heterogeneity is expected to favor
the formation of relatively inert polymeric species and
904
Y.-H. Shen and B.A. Dempsey
-
0.0
E
c -0.5
: -1.5
c
RS
2 -2.0
0
II)
2
-2.5
7
< -3.0
.W
2
p= -3.5
Y
2
-4.0
Time
( Min
)
Fig. 4. Estimate of the first-order rate constant of the reaction of polymeric species with ferron for 0.1 M [Al ;I PAC.
0.1
0.2
Mole
0.3
NaOH Added
Fig. 5. The stochiometry relationship between monomeric Al species and added base during PAC synthesis.
PAC synthesis and speciation
905
0.6
0.5
2
c
g
CT
-G
cu
0.4
0.3
:
s
m
P
L
3
0.2
.z
d
!
10
v
W
/A
20
1800 RPM
600 RPM
300 RPM
150
RPM
60
Time
( Min
100
)
Fig. 6. Absorbance vs. time for PAC (0.1 M [Al,], n=2.5) at different mixing intensities.
-
0.0
E
c
E
m
4
-0.5
-1.0
cc
Q,
2
-1.5
2
:
-2.0
:
4
-2.5
‘;;;
al
.-
-3.0
iz
= -3.5
-
c
-4.0
u
10
20
Time
( Min
30
)
Fig. 7. Estimate of the first-order rate constants of reactions of polymeric species with ferron.
906
Y.-H. Shen and B.A. Dempsey
0.6
0.0
’
0
I
20
I
I
40
60
I
80
100
Time (min)
Fig. 8. Absorbance vs. time for PAC (n=2.5) at different total aluminum concentrations.
-0.5
-1.0
-1.5
!5
p”
z
-2.0
2
-2.5
T;;
21
-3.0
:
lx
-
-3.5
2
-4.0
I
:
0
10
20
30
40
Time ( Min )
Fig. 9. Estimate of the first-order rate constants of reactions of polymeric species formed in different [Al J PAC preparations with ferron.
PAC synthesis and speciation
907
0.6
20
40
Time
60
80
100
(min)
Fig. 10. Absorbance vs. time for PAC (0.1 M [AI,], n=2.5) at different base injection rates.
Al(OH),,,, precipitates, and in fact this was observed.
Figures 11 and 12 show the effect of drop-wise addition of base on the speciation of three PAC samples
with [Al,] equal 1O-l, 1O-*.‘1, O”.’mol L-l, respectively,
and n equal to 2.5. In all cases, the drop-wise base addition increased the amount of AI(OH)
precipitate
(Fig. 11) and also resulted in more inert polymer
(Fig. 12). For [AlJ equal to 10-l mol L-‘, and [OH]
equal 5 mol L-l, drop-wise addition produced almost
90 % AI(OH
precipitate in contrast to about 90 %
polymer production with capillary injection.
Effect of aging
It has been suggested that the polymeric species in
PAC solutions are metastable (Smith and Hem 1972;
Smith 1971), but the reactions that take place during
aging are not well understood. The three PAC samples
from the concentration effect study were also used in
this aging study. Samples were aged at room temperature for 60 days. During the entire 60 days, samples
were clear to the naked eye. The solution pH decreased slightly with time, but the magnitudes of
change were small and within the range of experimental uncertainty (0.1 pH unit). Nevertheless, the
compositions of these solutions were different after
60 days of aging, as shown by their reaction with
ferron (Figs. 13 and 14). For lo-*.’and 1O”.7 mol L-’
PAC solutions, about 40 % to 50 % of the polymeric
species converted to AI(OH)
precipitate after 60
days of aging. The polymeric species in 10-l mol L-’
PAC solution was stable against precipitation.
The
remaining polymeric species in three samples were
converted to more inert species with respect to reaction with ferron. These results suggest that the
initially formed relatively stable polymers are only
transitional species that slowly convert to more stable
species during aging.
CONCLUSIONS
PAC were prepared by titration of aluminum chlorite
salts by NaOH solutions where concentrations, rate of
titration, mixing intensity, and counter-ion composition were the major independent variables. The coagulant species were determined using color development
after addition ofthe ferron reagent. The following conclusions were made.
First, it was possible to produce stable, clear preparations of partially neutralized aluminum chloride
908
Y.-H. Shen and B.A. Dempsey
0.6
v [A~]=O.OOZM CapillaryInjection
??[Al]=O.O002M
f
V [Al]=O.O02M Dropwise Addition
0 [Al]=O.O002M
"
A
,c
20
A
40
60
Time
Fig. 11. Absorbance
VS. time
A
A
3
a0
100
(min)
for PAC (n=2.5) formed by different base injection methods.
-0.5
E
c
-1.0
R
4
-1.5
Ld
:
cl
-2.0
2
2
-2.5
2
‘;;;
-3.0
2
.2
P=
-3.5
-c
-4.0
v
10
20
Time
( Min
)
Fig. 12. Estimate of the first-order rate constants of reactions of polymeric species in PAC preparations formed by different base injection
methods with ferron.
PAC synthesis and speciation
909
[
0.3
:
c
z
0
d
0.2
p”
4
.
v
0.1
[Al]=O.l
(M)
[Al]=O.OOZ (M)
?? [Al]=O.OOOZ
0
(M)
60 day
60 day
aging
aging
0
60 day
aging
80
60
40
20
0
v
Time
100
(min)
Fig. 13. Effect of aging on absorbance vs. time plots for different [AlJ PAC preparations.
-
0.0
E
c
.
v
-0;5
[Al]=O.l
(M)
[A1]=0.002
(M)
0
60 day
aging
v
60 day
aging
1
?
c*5
-1.0
c,
(d
:
c
z
-1.5
-2.0
0
-2.5
-
c
-4.0
0
10
20
Time
30
40
( Min )
Fig. 14. Effect of aging on the estimate of the first-order rate constants of reactions of polymeric species in PAC preparations with ferron.
Y.-H. Shen and B.A. Dempsey
910
(PAC) in which “polymeric” materials predominated.
These polymeric species resemble an aluminum specie
frequently
described
as Al,,O,(OH),,7+ and written as
Al,, according to the reaction rate constant with
ferron.
Second, the mode of preparation has a dramatic
effect on the composition
of PAC. Some important
parameters such as concentrations
(both [Al,] and
[OK]), mixing intensity, base injection method and
aging were identified
in this study. In general, a continuously
linear
is observed
increase
concentration
increasing [OR] added until a
of 2.5 is reached. Lowering
the Al concentration
or inadequate mixing both
caused a reduction in the formation of polymeric
material that was similar to Al,,. In addition, inadequate mixing favored the formation of AI(OH)
precipitates and lowering the Al concentration promoted the formation of monomeric species. Inadequate micromixing resulting from drop-wise addition
of base favors the formation of polymeric species
that react more slowly with ferron and Al(OH),(,, precipitates.
Finally, the results of the aging study suggest that
the initially formed polymers in PAC solution are only
transitional species that slowly convert to more inert
species during aging.
molar
with
in polymeric
ratio ([OH-]/[Al,])
REFERENCES
Bottero, J.Y.; Poirier J.E.; Fiessinger F. Study of partially neutralized aqueous aluminum chloride solutions: identiticationof
aluminum species and relation between composition of the solutions and their efficiency as a coagulant. Prog. Water Technol.
12: 601-612; 1980.
Dempsey, B.A.; Sheu H.; Ahmed T.T.M.; Mentink J. Polyaluminum chloride and alum coagulation of clay-fulvic acid susnensions. J. Am. Water Work Assoc. 77: 74-80: 1985.
Leprince, A.; Fiessinger F.; Bottero J.Y. Polymerized iron
chloride: an improved inorganic coagulant J. Am. Water Work
Assoc. 76: 93-97; 1984.
O’Melia, CR.; Gray, K.; Yao, C. Polymeric inorganic coagulants.
AWWARF Final Report. Denver, CO: American Water Works
Association Research Foundation; 1989.
O’Melia, C.R.; Dempsey, B.A. Coagulation using polyaluminum
chloride. In: Rantke, S.J.; VanProyen, A., eds. Proc, 24thAnnual
public water supply engineering conference. 1982; 5-14. Available from: American Water Works Association, Denver, CO.
Parthasarthy, N.; Buffle, J. Study of polymeric aluminum (III)
hydroxide solutions for application in wastewater treatment:
Properties ofthe polymer and optimal conditions of preparation.
Water Res. 19: 25-36; 1985.
Smith, R.W.; Hem, J.D. Effect of aging on aluminum hydroxide
complexes in dilute aqueous solutions. U. S. Geological Survey,
Water supply paper. 1827-D. Washington, D.C.: U.S. Govemment Printing Office; 1972.
Smith, R.W. Relations among equilibrium and nonequilibrium
aqueous species of aluminum hydroxy complexes. Am. Chem.
Sot. Adv. Chem. Ser. 106: 250-279; 1971.
Van Benschoten, J.E.; Edzwald, J.K. Chemical aspect of coagulation using aluminum salts-I.: Hydrolytic reactions of alum and
polyaluminum chloride. Water Res. 24: 15 19-1526; 1990.
Yao, C. The preparation of polymeric aluminum chloride (PACl)
and its application in water treatment. Doctoral dissertation, The
Johns Hopkins University, Baltimore, MD; 1987.