J O U RN A L OF E N V I RO N ME N TA L S CI EN CE S 7 4 (2 0 1 8 ) 9 5– 1 0 6

Available online at www.sciencedirect.com

ScienceDirect
www.elsevier.com/locate/jes

Aluminum formate (AF): Synthesis, characterization and

application in dye wastewater treatment

Moxi Xue, Baoyu Gao⁎, Ruihua Li, Jianzhang Sun

Shandong Key Laboratory of Water Pollution Control and Resource Reuse, School of Environmental Science and Engineering, Shandong University,
No.27 Shanda South Road, Jinan 250100, Shandong, China

AR TIC LE I N FO ABS TR ACT

Article history: Aluminum formate (AF), a degradable and non-corrosive coagulant, was synthesized from
Received 17 August 2017 aluminum hydroxide and formic acid. Polyamidine (PA), as a coagulation aid, was combined
Revised 11 February 2018 with AF for dye wastewater treatment. AF was characterized by XPS, FT-IR, viscosity, zeta
Accepted 22 February 2018 potential, mass spectrum and XRD, and the flocculation properties of the dual-coagulation
Available online 3 March 2018 system were characterized by FT-IR and SEM. The results showed that COOH, Al2O3-Al and
O2-Al bonds were formed in the AF synthesis process, and AF had a higher molecular weight
Keywords: and higher charge neutralization ability than PAC. The hydrolysates of AF were determined
Aluminum formate to contain Al13 Al11 and Al2, and the components of AF were confirmed to comprise a
Polyamidine mixture including aluminum formate (C3H3AlO6) and its hydrate. When the color removal
Dual-coagulation efficiency reached 100% in jar tests, the optimized dosage of AF/PA was 18.91/0.71 mg/L,
Phase analysis while the optimized dosage of PAC/PA was 21.19/0.91 mg/L. According to the variance
Interaction analysis, the interaction between AF/PA and PAC/PA were insignificant in macroscopic
view. FT-IR spectrum indicated AF captured pollutant by means of C_C_O bond, PAC
captured pollutant by δ _CH, \C_C\ and δ _CH. Overall, although the coagulation
mechanism of AF was different from that of PAC, AF/PA showed better coagulation
efficiency than PAC/PA in dye wastewater treatment.
© 2018 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences.
Published by Elsevier B.V.

Introduction dyes from wastewater, such as adsorption, photolysis, chemical

Dye wastewater is a complex, high-yield and heterogeneous
pollutant that has enormous environmental consequences,
for example, increasing the chrominance in drainage, jeopar-
dizing aquatic life and generating health hazards for humans
(Goncaf. Talu, 2004; Malachova et al., 2013; Palacio et al., 2009;
Salima et al., 2013). Reactive blue (K-GL), which belongs to the
category of reactive dyes, is widely utilized in the textile
industry because of the tight combination between textile
fibers and its reactive group (Şahinkaya, 2013). A number of
physical, chemical and biological methods are used to remove
⁎ Corresponding author.
E-mail address: bygao@sdu.edu.cn (B. Gao).
precipitation, chemical oxidation and reduction, and electro-
chemical precipitation (Albuquerque et al., 2013). However,
those technologies are usually not effective at color removal, or
lack practical utilization due to high cost; thus, an efficient,
inexpensive and convenient wastewater treatment system is
needed for remediation of large amounts of dye wastewater
(Pathak et al., 2015; Seo et al., 2015). Coagulation could remove
90% of dye molecules from dye wastewater, and is potentially
cost-effective given the large volume of output containment
required (Albuquerque et al., 2013; Furlan et al., 2010; Liang
et al., 2014).
Various common coagulants such as aluminum sulfate,
ferric sulfate, and ferric chloride are currently the most suitable
coagulants for this application. However, chloride and sulfate

https://doi.org/10.1016/j.jes.2018.02.013
1001-0742/© 2018 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences. Published by Elsevier B.V.
96 J O U RN A L OF E N V I RO N ME N TA L S CIE N CE S 7 4 (2 0 1 8) 9 5– 1 0 6
have been shown to cause corrosion with the participation of
calcium, which might damage water supply equipment (Mikola
and Tanskanen, 2015). Compared to aluminum chloride and
aluminum sulfate, aluminum formate has a higher coagulation
effect and it is able to effectively decrease the corrosive
effect compared to the other counter anions (Hellstén et al.,
2005; Parrino et al., 2014). Furthermore, formate salts can be
decomposed under UV irradiation at 253.7 nm, which would
avoid the accumulation of anions in wastewater treatment
(Goncaf. Talu, 2004). The water containing formate can be used
for irrigation without causing land salinization. However, when it
is used in treatment of water supply, formate groups would
contribute to the dissolved organic carbon content. As a result, in
combination with UV light and ozone processes, aluminum
formate (AF) can be compatible with the treatment process
(Mikola et al., 2013; Väänänen et al., 2012). In practical application,
AF can be used in wastewater treatment as part of a complete
process including coagulation, biological treatment and UV
disinfection. Väänänen et al. (2012) reported a method for AF
synthesis and its application in water treatment. However,
characterization of this process has been lacking, such as
determining the changes of chemical species after synthesis.
Thus, in order to understand the chemical reaction mechanism,
more powerful proof is needed with additional characterization
methods.
In addition to the corrosion of equipment by low-molecular-
weight aluminum salts, excess aluminum poses a serious threat
to humans and other living organisms as well (Cheng and Chi,
2002; Mena et al., 2013; Roque-Ruiz et al., 2016). Therefore,
efficient coagulant aids have been adopted in water treatment
processes in order to increase the coagulation effect. Many
studies have provided strong evidence for the improvement of
coagulation efficiency using coagulation aids (Lee and
Westerhoff, 2006; Yang et al., 2011; Ying et al., 2011). Polyamidine
(PA), a cationic polymer containing the amidine unit, has a high
absorption bridging capacity for heavy metal ions and has been
applied in treatment of high-concentration industrial wastewater
(Toktonov et al., 2006). However, there are only a few studies on
the coagulation performance of systems utilizing PA. Therefore,
the application of PA in combination with other coagulants is
worthy of study. In this work, in order to understand the effect of
PA in the coagulation process, the AF/PA and aluminum
polychloride (PAC)/PA coagulation systems were studied to
investigate the efficiency of dual coagulation using PA and Al
coagulants. Guo et al. (2015) investigated the role of PA in the
coagulation performance and floc properties in combination with
aluminum salts for the removal of natural organic matter (NOM).
However, the increase in floc dimension induced by PA was not
the main mechanism of its function as a coagulation aid,
according to Guo et al. (2015). Therefore, more microscopic
characterization and data analyses are necessary to gain a better
understanding.
In this study, the synthesis of AF was investigated in detail
by rigorous characterization, such as X-ray photoelectron
spectroscopy (XPS), Fourier Transform Infrared Spectroscopy
(FT-IR), viscosity and zeta potential measurements. Further-
more, the interaction between AF and PA was exhibited by
scanning electron microscope (SEM) and statistical analysis.
PAC was adopted as a comparison in this research to evaluate
the efficiency of AF.
1. Experiment
1.1. Material
1.1.1. Synthesis of aluminum formate and PAC
AF solution was directly obtained from the reaction of aluminum
hydroxide and formic acid. The synthesis method was a
modification of that in a patent application by Bårström et al.
(2008), and this method was adopted by Mikola et al. (2013) as
well. The HCOOH to Al molar ratio was set to 3.0 during this
study.
Firstly, Aluminum hydroxide containing 5.1% (W/W) of
aluminum (aluminum hydroxide, reagent grade, Sigma Aldrich,
CAS: 21645-51-2), was suspended in UP water. Secondly, formic
acid ((98% (W/W), Tianjin Damao Chemical Reagents Co.,China)
was diluted to 80% in water (W/W), and then added to the
suspension to reach a Al-HCOOH molar ratio of 1/3. After steady
stirring for 2 hr at 60°C, the mixture turned into a translucent
solution (Mikola and Tanskanen, 2015). In this study, PAC was
synthesized by titrating sodium carbonate into aluminum
chloride solution, where the molar ratio of [OH−]/[Al3+] was
adjusted to 2.2, the aging temperature was 25°C and the aging
time was 30 days.
1.1.2. Preparation of polyamidine
The synthesis of polyamidine followed the method described
by Guo et al. (2017).
1.1.3. Preparation of simulated dye wastewater
The simulated dye wastewater was obtained by dissolving
Reactive Blue K-GL powder in tap water, and the relevant
characteristics are shown in Table 1.
1.2. Characterization
1.2.1. FT-IR and XPS measurements to characterize the
synthesis process
The AF powder was achieved by vacuum freeze-drying of the
above-described AF product for 24 hr (−60°C). FT-IR and XPS
experiments were conducted to observe the chemical variations
occurring during the synthesis process. The reaction mecha-
nism was revealed by disappearance or formation of chemical
species by comparing results for AF and Al(OH)3.
1.2.2. Intrinsic viscosity and zeta potential measurements for AF
Characterization methods, such as viscosity and zeta potential
measurements, helped elucidate the difference between AF and
PAC in terms of coagulation behavior. Viscosity measurements
show the relationship between molecular weight and the
intrinsic viscosity of the solution. Generally, organic coagulants
with higher viscosity have larger molecular weight, as well as
better coagulation ability. This is because coagulants with large
molecular weight can aggregate the pollutants and precipitate
more easily. The viscosity of samples with different concentra-
tions was measured by a Ubbelohde viscometer, and the
intrinsic viscosity was calculated by extrapolation to character-
ize the molecular weight of different samples.
Charge neutralization, an important property for coagu-
lants, is reflected by the zeta potential. To compare the charge
 J O U RN A L OF E N V I RO N ME N TA L S CI EN CE S 7 4 (2 0 1 8 ) 9 5– 1 0 6
Table 1 – Relevant parameters of Reactive Blue (K-GL).
Name Concentration Wave length
(mg/L) (nm)
Reactive Blue 100 608
neutralization ability and adaptability of AF and PAC under
different pH conditions, the pH of coagulant samples was
adjusted from 1.6–13.4 using NaOH or HCl, and measured by a
Zetasizer Nanoseries instrument (Malvern Co., Ltd., UK).
1.2.3. MS detection for AF and XRD substance identification for
AF
The mass spectrum (MS), which can detect samples at low
concentrations (mg/L), was able to characterize the hydrolyzed
coagulants in terms of m/z (mass charge ratio) distribution,
quantity of electric charge and aluminum species present (Bi et
al., 2013; Feng et al., 2011a; Zhang et al., 2017). To detect the
characteristics of hydrolyzed AF using MS measurements, AF was
dissolved in RO water to obtain a dilute solution of concentration
10 mg/L. Then, the AF solution was injected into the MS
spectrometer (LCQ Fleet™, Thermo Scientific™). In addition,
PAC was measured for comparison in MS analysis, and all the
MS spectrum data were set in positive ion mode. The XRD (X-ray
Diffraction) pattern of the freeze-dried AF powder for scattering
angles (2θ) 5°–90° is shown in Fig. 6. The chemical components of
AF were detected by an X-ray fluorescence spectrometer.
1.3. Jar tests
Jar tests were carried out using a program-controlled jar tester
(Zhongrun Water Industry Technology Development Co. Ltd.),
and each jar in the tests had a volume of 1.0 L. The coagulation
process contained three segments: rapid mixing (200 r/m), slow
mixing (40 r/m, 15 min) and sedimentation (0 r/m, 30 min). In
these experiments, the dosages of coagulants and the dosages
of PA were designed by Design-Experts (Version 8.0.6) according
to response surface methodology. The related parameters and
designed test units are shown in Tables 2 and 3. For the AF/PA
Table 2 – Results of viscosity measurements.
C (g/L) t (sec) η ηr ηSP lnηr ηSP/c lnηr/c
PAC
10 421.55 1.179 1.182 0.182 0.167 0.018 0.0167 0.009
5 378.82 1.060 1.062 0.062 0.060 0.012 0.012
1 360.21 1.008 1.010 0.010 0.009 0.010 0.009
AF 3.0
10 428.72 1.199 1.202 0.202 0.184 0.020 0.018
5 389.32 1.089 1.091 0.091 0.087 0.018 0.017
1 360.91 1.010 1.012 0.012 0.011 0.012 0.011
RO
356.63 0.9982 1
PAC: aluminum polychlorid; AF 3.0: aluminum formate for the
HCOOH/Al molar ratio at 3.0; RO: reverse osmosis.
97
Absorption Molecular structure
1.030 ± 0.03
system, AF was dosed before PA in the rapid mixing segment,
and the interval was 30 sec. For the PAC/PA system, PAC was
dosed 30 sec before PA addition as well. To weaken the effect of
pH change on coagulation behavior, the UV608 and zeta
potential data were measured after adjusting the pH to 8.16 ±
0.05 (initial pH of raw water). The interaction between coagu-
lants and coagulant aid was studied by variance analysis, and
the optimal dosages of coagulants and coagulant aid were
ascertained by a 3D surface model.
1.4. Floc properties
To demonstrate the interaction between the coagulants and
coagulant aid, FT-IR spectroscopy was used as an effective
method to observe the changes in chemical species during the
coagulation process, and the results could provide possible
combinations to explain how the coagulant aid functioned in
the coagulation process. The sedimentary flocs were freeze-
dried at − 60°C in a vacuum freeze-dryer for 24 hr, and then
the pulverized flocs were subjected to FT-IR and SEM analysis.
2. Result and discussion
2.1. Characterization of aluminum formate
In this study, FT-IR and XPS were chosen to measure the
chemical changes occurring during the synthesis process, and
the Al-HCOOH molar ratio in AF was selected as 1/3. The XPS
results are shown in Fig. 1. The main element (Al, C and O)
measurement was able to discriminate the difference between
AF and Al(OH)3 by the displacement and proportion of the
peaks. Because Al2O3·3H2O existed in the form of Al(OH)3, Al2O3
was the sole Al species observed in Al(OH)3. Compared with Al
(OH)3, AF showed new chemical species such as AlOx-Al, Al2O3-
Al, O2-Al, Al-HOOC, which suggested that AF was synthesized
[η]
successfully. In addition, the XPS result proved the hypothesis
that formic acid not only played a role in adjusting acidity, but
also participated in the polymerization of aluminum species by
the formation of Al2O3-Al and O2-Al. This conclusion could also
be demonstrated by the corresponding peaks in the FT-IR
spectra (Fig. 2). When Al2O3-Al and O2-Al bonds appeared,
0.012 another hypothesis was put forth that the speciation of
aluminum might be more complicated than reported in recent
research. According to Väänänen's work (Väänänen et al., 2012),
dimeric aluminum is the dominant species of AF at pH 4, and
some trimeric species exist as well. However, although Al2O3-Al
and O2-Al were found in AF, the speciation of AF still needed
further characterization to verify the species present. Al2O3-Al
and AlOx-Al might exist in complex polymeric forms, and
98 J O U RN A L OF E N V I RO N ME N TA L S CIE N CE S 7 4 (2 0 1 8) 9 5– 1 0 6

Table 3 – Comparison between PAC and AF 3.0 in m/z distribution, positive charge and aluminum species from MS spectral
analysis.
Distribution of m/z (%) 80–300 300–500 500–700 700–1000

PAC 8.27803 2.706951 67.63884 21.37617883

AF 3.0 58.835546 4.085802 20.73544 16.34320735
Distribution of positive charge (%) 1–3 3–9 >9 Unidentified substances
PAC 2.4100594 12.0503 10.02445 75.51519385
AF 3.0 19.509704 3.677222 9.499489 67.31358529
Distribution of aluminum species (%) Al1-Al4 Al5-Al12 Al13-Al16 Unidentified substances
PAC 2.8990569 9.308418 0.279427 87.51309815
AF 3.0 10.214505 21.34831 13.89173 54.54545455

the aluminum hydrolyzation at different pH should also be
considered. Fig. 1d and e illustrate that the \COO\ group was
formed in this synthesis process, which verified that the
\COO\ group was attached to aluminum and combined into
stable chemical species, not in a volatile acid (Davies et al.,
1991).
As shown in Fig. 2, the Fourier Transform Infrared Spectros-
copy (FT-IR) proved the hypothesis that the synthesis of AF not
only formed \COOH groups bound with aluminum, but also
formed polymeric aluminum species. These conclusions also
agreed with the results derived by XPS. The wide peak in the Al
(OH)3 spectrum at 3470 cm−1 represented the \OH group in Al
(OH)3 before the synthesis of AF. After the synthesis took place,
the peak for the \OH group weakened and the characteristic
peaks of carboxyl groups appeared in the AF FT-IR curve at 3490,
2939, 1616 and 1400 cm−1. The strong peaks at 1616 and
1400 cm−1 might suggest vibration coupling of the \COO\
group, due to delocalization of electrons on the two oxygen
atoms. In addition, the peak which appeared at 1090 cm−1
indicated the existence of R\O\R′ groups. Considering the
reaction products, it was possible that Al\O\Al was formed in
this reaction system. The Al\O\Al bond provided more
evidence for the existence of polymeric aluminum species in
the reaction products. This conclusion was in accordance with
the XPS results suggesting that the polymeric aluminum was
formed during the synthesis.
In order to characterize the coagulation ability of AF, the
viscosity and zeta potential of samples were measured.
According to recent research (Ko et al., 2017; Stratiev et al.,
2016), intrinsic viscosity ([η]) has a strong relationship with
molecular weight, and coagulants with high intrinsic viscosity
usually have high molecular weight. So, the molecular weight of
AF samples was characterized by intrinsic viscosity measure-
ments in this study, which could be calculated by extrapolation
of the plotted data (Hobbs, 1980). Based on previous research
provided by Väänänen et al. (2012), the dominant aluminum
species of AF dimeric aluminum. Therefore, AF should have a
smaller intrinsic viscosity than PAC, which is constituted by
species containing 13 aluminum atoms. According to the results
shown in Fig. 3 and Table 2 ([η]PAC < [η]AF3.0), so the molecular
weight of AF 3.0 (the Al-HCOOH molar ratio of AF was 1/3) was
clearly larger than that of PAC.
Charge neutralization is the most dominant mechanism in
the coagulation process of aluminum salts. When a coagulant
has a higher surface charge, the charge neutralization ability
is better consequently. As a result, zeta potential was adopted
as an effective and convenient measurement for optimizing
AF samples. Furthermore, zeta potential measurements could
help in studying the aluminum hydrolysis process over a wide
pH range. Therefore, the zeta potential of AF samples was
measured at pH ranging from 1.6 to 13.4, with PAC also
measured for comparison, to better understand the aluminum
hydrolysis process.
According to Fig. 4, the trend of AF curve was similar to
that of PAC in the pH range of 1.6–8.0, except for the higher
zeta potential at pH 8.0–13.4, showing an obvious difference

Fig. 1 – X-ray photoelectron spectroscopy (XPS) spectra for Al(OH)3 Al (a), O (b); XPS spectra for aluminum formate (AF) Al (c), O
(d), C (e).
 J O U RN A L OF E N V I RO N ME N TA L S CI EN CE S 7 4 (2 0 1 8 ) 9 5– 1 0 6 99

50
100
40 PAC
AF 3.0
80 R-CH
30
Transmittance(%)

Zeta potential (mV)

60
20

40 R-OH
10
R-COOH

20 0
AF
Al(OH)3
R-O-R' R-CH -10
0 R-OH
R-CO-R' R-OH
4000 3000 2000 1000 -20
0 2 4 6 8 10 12 14
-1
Wavenumbers cm pH

Fig. 2 – Comparison of FT-IR spectra for Al(OH)3 and AF. Fig. 4 – Zeta potential results for the pH range of 1.4–13.6.

between the coagulants in terms of surface electric charge.
The hydrolysis of aluminum salts showed a complex dynamic
balance as a function of pH: When the pH was less than 4,
the dominant hydrolysate was Al(H2O)3+, which has a weaker
Fig. 3 – Viscosity results of aluminum polychloride (PAC), AF
3.0 plotted with concentration as the X axis, and ηSP/C, lnηr/C
as Y axis; the intercept of the intersection gives the intrinsic
viscosity, which can be used to represent weight.
ability in bridge and sweep flocculation than polymeric
aluminum. Generally, highly complex polymeric aluminum,
such as PAC, has better ability to adapt to a wide pH range
than simple aluminum salts. As a result, the proportion of the
Al(H2O)3+ hydrolysate in PAC is less than that in solutions of
monomers (such as AlCl3) and simple polymeric aluminum
(Al2(SO4)3). Therefore, in Fig. 4, AF presented weak acidic
adaptability, which suggested that AF contained polymeric or
monomeric aluminum hydrolysates. In the pH range of 5 to 9,
Al(OH)(H2O)2+ 4+
5 , Al(OH)2(H2O) , and Al(OH)3 began to appear,
and the zeta potential curve reached a peak. In this condition,
the charge neutralization ability reaches its peak, and the
optimized pH condition is always within this pH region (Wang
et al., 2009). In this region, the trends for PAC and AF were
similar, which indicated that the hydrolysates of AF and PAC
were consistent with those of PAC. Furthermore, both AF and
PAC exhibited better coagulation behavior in the pH range 5 to
9 than in strong acidic (pH < 5) and basic (pH > 9) conditions.
However, in the pH range of 9–13.4, AF was different from PAC.
In alkaline conditions, the dominant hydrolysate of PAC was
Al(OH)−4, which contributed a negative charge to the zeta
potential. The higher zeta potential of AF than PAC indicated
the presence of a different hydrolysis mechanism under these
conditions.
On the basis of the results in Fig. 5, the distribution of
molecular weight, positive charge and aluminum species are
shown in Table 3. Because the test concentration used for MS was
10 mg/L, the substances detected should be the hydrolysates of
the aluminum coagulants. According to Table 3, the hydrolysates
of AF were concentrated in the m/z range of 80 to 300 and the
most intense peak was at m/z 213 (100%). In addition, Al13 species
at m/z 213, 337 and 328 were also detected. Therefore, the
dominant species of AF was found to be Al13. In addition, Al11 (m/z
226) and Al2 (m/z 138) were also detected as main products in the
hydrolysis of AF (Feng et al., 2011b). Based on Sarpola's research
(Sarpola et al., 2007), the Al13 core is very stable and requires a
high collision energy (75 eV) to break its structure. However,
Zhao's research (Zhao et al., 2009) proved that dimers, small
polymers and Al13 could transform into each other during the
hydrolysis and polymerization process of aluminum salts. Thus,
100 J O U RN A L OF E N V I RO N ME N TA L S CIE N CE S 7 4 (2 0 1 8) 9 5– 1 0 6

Fig. 5 – Mass spectrum of AF 3.0 (a) and PAC (b) at 10 mg/L.

the aluminum species of AF could be seen as a series of 2.2. Application of AF in coagulation

transitional species between small polymers and Al13. Accord-
ingly, the mechanism of PAC hydrolysates was more ambiguous
than that of AF. Because of the lack of peaks for Al13 species and
the appearance of a large amount of unidentified species, the
hydrolysates of PAC might aggregate into bigger polymeric
amorphous hydroxide species (Zhao et al., 2015). The phase
compositions of AF and PAC were determined by XRD, and
results are exhibited in Fig. 6. Based on Fig. 6, the composition
of AF was identified as a mixture including aluminum formate
(C3H3AlO6), aluminum formate hydrate (C3H3AlO6·3H2O) and
aluminum hydrogen formate (C3H3AlO6·2H2O2). By comparison,
PAC was deemed to be a mixture dominated by aluminum
chloride (AlCl3).
To ascertain the optimized proportions for PAC/PA or AF/PA, a
central composite design (CCD) (designed by Design-Expert
software (version 8.0.6)) was adopted for the response surface
experiment. The relevant parameters are shown in Table 4.
The dosage of coagulant and dosage of PA were designed as
arguments respectively. The dye removal efficiency and zeta
potential were set as dependent variables, and each numeric
factor was varied in 5 levels. According to the design requirement,
12 jar test groups were needed in this design, and the methods
and results of the design are listed in Tables 5 and 6, respectively.
For visualization of the coagulation type differentiation, the
dosages for both PAC/PA and AF/PA were executed as the design
 J O U RN A L OF E N V I RO N ME N TA L S CI EN CE S 7 4 (2 0 1 8 ) 9 5– 1 0 6 101

Table 5 – Results of jar test of PAC combined with PA.

No. Dosage of Dosage of Removal Zeta potential
PAC (mg/L) PA (mg/L) efficiency (%) (mV)

1 19 1.25 93.123 −20.8

2 19 1.25 94.556 −19.2
3 8 2 17.574 −19.85
4 19 1.25 94.174 −18.3
5 3.443651 1.25 0.621 −19.75
6 19 1.25 95.033 −21.25
7 19 2.31066 91.117 −21.3
8 19 0.18934 93.314 −19.55
9 30 2 97.947 −13.4
10 8 0.5 18.72 −19.5
11 34.55635 1.25 98.567 −10.305
12 30 0.5 98.185 −13.4

the optimized dosage was 21.19/0.91 mg/L, and the dye removal
efficiency reached 99.9%. Compared with the PPA^2 value
(0.2336), the smaller PPAC^2 value (0.0002) suggested that PAC
offered a stronger effect in the coagulation process. Corre-
spondingly, the optimized AF/PA dosage was 18.91/0.71 mg/L,
where the removal efficiency was 99.9%. Due to the lower
p-value of AF (PAF^2 = 0.0049, PPA^2 = 0.6942), the dominant
Fig. 6 – The X-ray diffraction analysis results of PAC and AF factor was AF in the AF/PA system. For the same removal
3.0. efficiency, both the dosage of AF and dosage of PA added were
less than the dosages of PAC/PA. The comparison between the
PAC/PA and AF/PA systems revealed that AF had a better
coagulation effect in reactive dye wastewater treatment, and
specified, using variance analysis to judge the interaction the combination of AF and PA was better than PAC/PA.
between the coagulant and coagulation aid and compare the According to Fig. 8, the 2D removal efficiency graph could
coagulation effects between AF and PAC (Tables 7 and 8). show the effects of the coagulant and coagulation aid
The correlation result of the designed model could be separately. The color removal efficiency on Line A or Line B
represented by the p-value: if the p-value is less than 0.05, it could provide the separate effect of the coagulant aid (PA) or
can be seen that the entire model is significant in accordance coagulant on the coagulation process. For example, the data
with the 5% confidence interval in the analysis of variance. parallel with line A represent the influence of PAC on the
The smaller the p-value, the higher the model significance, coagulation process; the data parallel with line B represent the
which means that the fit of the equation is better. Based on influence of PA on the coagulation process. According to the
the p-values in Tables 7 and 8, the model was in accordance 2D graph, we found that the effect of the coagulation aid was
with the significance testing. Therefore, the design was insignificant in the PAC/PA and AF/PA systems, and that the
reliable. Apart from judging the reliability of the model, coagulant aid in the PAC/PA system shows better coagulation
variance analysis could be used to analyze the interaction ability than PA in the AF/PA system. However, the better
between the coagulant and coagulant aid. If the p-value of A*B separate effect of PA in the PAC/PA system might not reflect
is less than 0.05, it can be seen that the coagulant and better color removal efficiency for the PAC/PA system. AF/PA
coagulant aid (factor A and factor B) have coefficient interac-
tion. However, both PAC/PA and AF/PA revealed no coefficient
interaction (PPAC⁎PA = 0.9567, PAF⁎PA = 0.8039). Although the
Table 6 – Results of jar test of AF combined with PA.
interaction was insignificant, PA still showed a positive effect
in the coagulation process, which will be discussed further on NO. Dosage of Dosage of Removal Zeta potential
AF (mg/L) PA (mg/L) efficiency (%) (mV)
in Section 2.3. According to Fig. 7, the response surface model
revealed the optimized of PAC/PA and AF/PA as well as the 1 34.55635 1.25 96.512 −5.545
dominant in the coagulation processes. In the PAC/PA system, 2 19 1.25 96.124 −13.9
3 19 1.25 96.512 −14.5
4 19 1.25 97.481 −9.85
5 19 0.18934 95.155 −12.95
Table 4 – Levels of removal efficiency test with aluminum
6 19 1.25 97.384 −14.65
coagulant combined with PA for CCD design.
7 30 0.5 97.287 −6.245
Argument Level 8 19 2.31066 96.221 −14.7
9 8 2 65.019 −16.5
−α −1 0 1 α
10 3.443651 1.25 1.066 −18.5
Dosage of coagulant (mg/L) 3.4436 8 19 30 34.5565 11 30 2 98.062 −5.58
Dosage of PA(mg/L) 0.1893 0.5 1.25 2 2.3106 12 8 0.5 70.736 −16.55
102 J O U RN A L OF E N V I RO N ME N TA L S CIE N CE S 7 4 (2 0 1 8) 9 5– 1 0 6

Table 7 – Results of variance analysis for PAC combined with PA.

Sum of Mean F p-Value

Source squares df Square value Prob > F

Model 15,301.01 5 3060.201 47.47364 < 0.0001 significant

A-PAC 11,126.93 1 11,126.93 172.6148 < 0.0001
B-PA 2.521166 1 2.521166 0.039111 0.8498
A*B 0.206116 1 0.206116 0.003198 0.9567
A2 4165.946 1 4165.946 64.62733 0.0002
B2 113.0338 1 113.0338 1.753521 0.2336
Residual 386.7664 6 64.46106
Lack of fit 384.787 3 128.2623 194.3976 0.0006 Significant
Pure error 1.979381 3 0.659794
Cor total 15,687.77 11

had better color removal efficiency, but PA showed an
inconspicuous separate effect in the AF/PA system.
The addition of PA did not improve coagulation efficiency,
which was different from previous research showing that the
coagulation efficiency was increased significantly when PA
was used with aluminum sulfate (Guo et al., 2015). This was
because different coagulants were used in the two studies.
The combined efficiency between simple polymeric alumi-
num (Al2(SO4)3) and PA was much better than the interaction
between complex polymeric aluminum (PAC) and PA. The
zeta potential results in Fig. 7 show that the AF/PA system was
closer to the isoelectric point than PAC/PA, which indicated
that the charge neutralization ability of AF was better than
that of PAC. Based on the experimental design, PA was added
in the same dosage range when it was combined with PAC or
AF, therefore, any charge influence due to PA dosage can be
ignored, and the zeta potential differences between PAC/PA
and AF/PA totally resulted from the charge neutralization
ability of the coagulants. Furthermore (Guo et al., 2015),
the flocculating effect of PA was dependent on the UV254
and DOC (dissolved organic carbon) removal efficiencies. The
improvement due to PA was shown clearly in the DOC
removal efficiency (from 42% to 59%), but in the UV254 removal
efficiency, the removal efficiency only increased by 4%,
which was a very slight advance. In this research, the dye
removal efficiency was expressed as UV608, which might
be another reason for the apparently insignificant effect of
PA addition. Besides, the DOC removal efficiency is an
unnecessary index in dye wastewater treatment, and UV608
is sufficient to represent the color removal efficiency. However,
in water supply treatment studies, such as that of Guo et al.
(2015), DOC removal efficiency is an indispensable index for
representing the heterogeneous organic matter in natural
water.
obviously, and the dimension of the flocs could exceed
800 μm. PA, which is highlighted in blue, showed a large
branched pattern composed of cubic units, and PA could
easily be distinguished from the small branch-like AF flocs
and the dye pollutant, with nubby appearance and vague
edges. Based on Fig. 9c and d, the floc structure pattern
changed a great deal when PA combined with small AF flocs to
form larger flocs when PA was added in the system. Although
the assisting role of PA was proven by the SEM images, the
optimum combination ratio for AF/PA formed flocs needed to be
confirmed. Therefore, variance analysis was an indispensable
tool for analyzing the interaction between the coagulant and
coagulant aid from the macroscopic viewpoint.
To investigate the changes in chemical species during the
coagulation process, FT-IR spectroscopy was used to measure
the flocs in raw wastewater, those for PAC added alone, those
AF added alone, those in PAC/PA, and flocs in AF/PA. As
shown in Fig. 10a, compared with the raw wastewater flocs, the
spectra for PAC or PAC/PA flocs showed the disappearance of
many peaks in the range 700–1600 cm−1, which are considered
to be related to the meta-substitution and para-substitution in
the benzenoid structure of the reactive blue dye molecule.
When PAC or PAC/PA was added into the system, all of the
peaks (910, 743 cm−1) in raw wastewater were replaced by a
strong peak at 805 cm−1, which was indicative of 1,2,3,4,5
substitution in the benzene ring, and forcefully proved that
the substitution reaction occurred in the benzene ring. Accord-
ing to Fig. 10b, when AF or AF/PA was added in dye wastewater
treatment, the same substitution took place as well. When PAC
Table 8 – Results of variance analysis for AF combined
with PA.
Sum of Mean F p-Value

Source squares df square value Prob > F

2.3. The effect of PA addition on floc properties
Model 7944.276 5 1588.855 10.15243 0.0068 significant
A-AF 4732.43 1 4732.43 30.23917 0.0015
The finding of insignificant removal efficiency improvement
B- PA 1.474429 1 1.474429 0.009421 0.9258
by PA suggested that UV608 might not express the coagulant A*B 10.53652 1 10.53652 0.067326 0.8039
aid assistance completely. Therefore, SEM was selected to A2 2933.341 1 2933.341 18.7434 0.0049
examine the interaction between coagulant and PA through B2 26.65383 1 26.65383 0.170312 0.6942
microscopic images. As shown in Fig. 9a, the dimension of Residual 939 6 156.5
flocs was usually no greater than 50 μm when AF was added Lack of fit 937.6779 3 312.5593 709.2408 < 0.0001 significant
Pure error 1.322087 3 0.440696
alone, and the flocs exhibited a branch-like structure. When
Cor total 8883.276 11
PA was added in the system (Fig. 9b), the floc size increased
 J O U RN A L OF E N V I RO N ME N TA L S CI EN CE S 7 4 (2 0 1 8 ) 9 5– 1 0 6
Fig. 7 – Response surface for removal efficiency of aluminum coagulant combined with PA: PAC/PA (a), AF/PA (b). Response
surface for zeta potential of aluminum coagulant combined with PA: PAC/PA (c), AF/PA (d).
was added into the system, the peak at 3400 cm−1 became
broader than that of raw water. This phenomenon suggested
that the quantity of hydrogen bonds was increased and
Fig. 8 – The separate effects of PAC and PA in the PAC/PA
system (a), and the separate effects of AF and PA in the AF/PA
system (b).
103
concluded that PAC bridged by means of hydrogen bonds.
Moreover, new peaks appeared at 2410, 1950, 805 cm−1, which
were the characteristic peaks for δ _CH, \C_C\ and δ _CH.
This revealed that alkene was formed in the coagulation
process. However, compared with the AF system, the flocs of
the AF/PA system did not show obvious differences in the FT-IR
spectra, even though the interaction of AF and PA was better
than that between PAC and PA. One explanation was that the
interaction between AF and PA was not dominated by chemical
bonds but by physical absorption. Besides the influence of
physical absorption on the coagulation ability of AF/PA,
some changes in the AF floc spectrum might indicate that the
pattern of connection of AF with the pollutant was different
that of PAC. In addition to the \COOH group of AF (peaks
at 3060, 3400 and 1640 cm−1), a peak for the newly formed
chemical matters attributable to C_C_O was found at
2400 cm−1, which indicated that AF captured the pollutant
by means of C_C_O bonds rather than δ _CH, \C_C\ and
δ _CH like PAC.
According to Fig. 10, the coagulation ability of PA was
questionable, in that none of the floc samples containing PA
showed PA's characteristic groups (such as \C`N, \NH3X−)
in the FT-IR spectrum. Therefore, the insignificant interaction
observed between the coagulant and PA might be attributed to
the poor purity of PA.
3. Conclusions
(1) AF was synthesized from aluminum hydroxide and
formic acid. FT-IR and XPS results found that COOH,
Al2O3-Al and O2-Al bonds were formed in the synthesis
process of AF, and the formation of Al2O3-Al and O2-Al
in AF indicated that formic acid participated in the
polymerization of aluminum.
104 J O U RN A L OF E N V I RO N ME N TA L S CIE N CE S 7 4 (2 0 1 8) 9 5– 1 0 6

b
a PA
added

c

d

Fig. 9 – Image of flocs without coagulant aid used for SEM analysis (a), PA coagulation aid added (b); image of PA combined with
coagulant at PA molecule tip (c), at PA molecule trunk (d).

(2) Viscosity measurements revealed that the molecular range 1.6–8.0; however, the higher zeta potential of AF
size of AF was larger than that of PAC. Based on the than PAC implied that the formate groups in AF could
results of zeta potential measurements, AF showed a raise the zeta potential to improve the charge neutral-
hydrolysis mechanism similar to that of PAC in the pH ization ability in the pH range 8.0–13.4.

Fig. 10 – FT-IR spectra of flocs, with PAC as coagulant (a), AF as coagulant (b).
 J O U RN A L OF E N V I RO N ME N TA L S CI EN CE S 7 4 (2 0 1 8 ) 9 5– 1 0 6
(3) The hydrolysates of AF were determined by mass spec-
troscopy measurements, which showed that the main
hydrolysis product was Al13, while the main hydrolysate
product of PAC was Al9. The compositions of aluminum
salts were identified by XRD phase measurements, which
proved that AF consisted of aluminum formate (C3H3AlO6),
aluminum formate hydrate (C3H3AlO6·3H2O) and alumi-
num hydrogen formate (C3H3AlO6·2H2O2).
(4) The jar test results, as analyzed by the response surface
model, showed that the optimized dosage of AF/PA
(when the color removal efficiency reached 100%) was
18.91/0.71 mg/L, and the optimized dosage of PAC/PA
was 21.19/0.91 mg/L.
(5) The interaction between coagulants and PA was proved by
SEM images, which showed that the flocs became larger
when PA was added. FT-IR spectra indicated that AF
captured the pollutant by means of C_C_O bonds,
while PAC captured the pollutant by δ _CH, \C_C\ and
δ _CH.
Acknowledgments
This work was supported by the Tai Shan Scholar Foundation
(No. ts201511003).
Appendix A. Supplementary data
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.jes.2018.02.013.
REFERENCES
Albuquerque, L.F., Salgueiro, A.A., Melo, J.L. d S., Chiavone-Filho, O.,
2013. Coagulation of indigo blue present in dyeing wastewater
using a residual bittern. Sep. Purif. Technol. 104, 246–249.
Bårström, R., Gillberg, L., Martikainen, M., 2008. Sätt att framställa
polyaluminiumföreningar. Pat. SE 530 908 C2, Kemira Kemi AB,
Helsingborg Sweden (In Swedish).
Bi, Z., Feng, C., Wang, D., Ge, X., Tang, H., 2013. Transformation of
planar Mögel Al13 coagulant during the dilution and aging
process. Colloids Surf. A Physicochem. Eng. Asp. 416, 73–79.
Cheng, W.P., Chi, F.H., 2002. A study of coagulation mechanisms of
polyferric sulfate reacting with humic acid using a
fluorescence-quenching method. Water Res. 36, 4583–4591.
Davies, P., Roberts, M.W., Shukla, N., 1991. The Reactive
Chemisorption of Formic Acid at A1(111) Surfaces and the
Influence of Surface Oxidation and Coadsorption with Water:
A Combined XPS and HREELS Investigation.
Feng, C., Ge, X., Wang, D., Tang, H., 2011a. Effect of aging condition
on species transformation in polymeric Al salt coagulants.
Colloids Surf. A Physicochem. Eng. Asp. 379, 62–69.
Feng, C., Zhao, S., Bi, Z., Wang, D., Tang, H., 2011b. Speciation of
prehydrolyzed Al salt coagulants with electrospray ionization
time-of-flight mass spectrometry and 27Al NMR spectroscopy.
Colloids Surf. A Physicochem. Eng. Asp. 392, 95–102.
Furlan, F.R., de Melo da Silva, L.G., Morgado, A.F., de Souza, A.A.U.,
Guelli Ulson de Souza, S.M.A., 2010. Removal of reactive dyes
from aqueous solutions using combined coagulation/
flocculation and adsorption on activated carbon. Resour.
Conserv. Recycl. 54, 283–290.
105
Goncaf. Talu, V., 2004. formate ion decomposition in water under
UV irradiation at 253.7 nm. Environ. Sci. Technol. 38,
3984–3993.
Guo, B., Yu, H., Gao, B., Rong, H., Dong, H., Ma, D., et al., 2015.
Coagulation performance and floc characteristics of aluminum
sulfate with cationic polyamidine as coagulant aid for kaolin-
humic acid treatment. Colloids Surf. A Physicochem. Eng. Asp.
481, 476–484.
Guo, B., Yu, H., Gao, B., Zhang, S., Yue, Q., Xu, X., 2017. Novel
cationic polyamidine: synthesis, characterization, and sludge
dewatering performance. J. Environ. Sci. 51, 305–314.
Hellstén, P., Salminen, J., Jørgensen, K., Nystén, T., 2005. Use of
formate in road winter deicing can reduce groundwater
deterioration. Environ. Sci. Technol. 39, 5095–5100.
Hobbs, D.S.D.W.F., 1980. Physical and enzymatic properties of
myosin from porcine brain. Biophys. Soc. 32, 705–718.
Ko, Y.G., Kim, C.J., Cho, Y.H., Chung, K.H., Kang, M.J., 2017.
Combustion/absorption process for the separation of 14C and
3H in radwastes released from nuclear power plants and their
analysis. J. Hazard. Mater. 331, 13–20.
Lee, W., Westerhoff, P., 2006. Dissolved organic nitrogen removal
during water treatment by aluminum sulfate and cationic
polymer coagulation. Water Res. 40, 3767–3774.
Liang, C.-Z., Sun, S.-P., Li, F.-Y., Ong, Y.-K., Chung, T.-S., 2014.
Treatment of highly concentrated wastewater containing
multiple synthetic dyes by a combined process of coagulation/
flocculation and nanofiltration. J. Membr. Sci. 469, 306–315.
Malachova, K., Rybkova, Z., Sezimova, H., Cerven, J., Novotny, C.,
2013. Biodegradation and detoxification potential of rotating
biological contactor (RBC) with Irpex lacteus for remediation of
dye-containing wastewater. Water Res. 47, 7143–7148.
Mena, J.B., Ubices de Moraes, A.A., Benito, Y.R., Ribatski, G., Parise,
J.A.R., 2013. Extrapolation of Al2O3-water nanofluid viscosity for
temperatures and volume concentrations beyond the range of
validity of existing correlations. Appl. Therm. Eng. 51, 1092–1097.
Mikola, M., Tanskanen, J., 2015. Preparation and coagulation
performance of solid aluminium formate. J. Water Process Eng.
5, 1–5.
Mikola, M., Rämö, J., Sarpola, A., Tanskanen, J., 2013. Removal of
different NOM fractions from surface water with aluminium
formate. Sep. Purif. Technol. 118, 842–846.
Palacio, S.M., Espinoza-Quinones, F.R., Modenes, A.N., Oliveira, C.
C., Borba, F.H., Silva Jr., F.G., 2009. Toxicity assessment from
electro-coagulation treated-textile dye wastewaters by
bioassays. J. Hazard. Mater. 172, 330–337.
Parrino, F., Camera-Roda, G., Loddo, V., Palmisano, G., Augugliaro,
V., 2014. Combination of ozonation and photocatalysis for
purification of aqueous effluents containing formic acid as
probe pollutant and bromide ion. Water Res. 50, 189–199.
Pathak, V.V., Kothari, R., Chopra, A.K., Singh, D.P., 2015.
Experimental and kinetic studies for phycoremediation and
dye removal by Chlorella pyrenoidosa from textile wastewater.
J. Environ. Manag. 163, 270–277.
Roque-Ruiz, J.H., Cabrera-Ontiveros, E.A., González-García, G.,
Reyes-López, S.Y., 2016. Thermal degradation of aluminum
formate sol-gel; synthesis of α-alumina and characterization
by 1 H, 13 C and 27 Al MAS NMR and XRD spectroscopy. Results
Phys. 6, 1096–1102.
Şahinkaya, S., 2013. COD and color removal from synthetic textile
wastewater by ultrasound assisted electro-Fenton oxidation
process. J. Ind. Eng. Chem. 19, 601–605.
Salima, A., Benaouda, B., Noureddine, B., Duclaux, L., 2013.
Application of Ulva lactuca and Systoceira stricta algae-based
activated carbons to hazardous cationic dyes removal from
industrial effluents. Water Res. 47, 3375–3388.
Sarpola, A., Hellman, H., Hietapelto, V., Jalonen, J., Jokela, J., Rämö,
J., et al., 2007. Hydrolysis products of water treatment chemical
aluminium sulfate octadecahydrate by electrospray ionization
mass spectrometry. Polyhedron 26, 2851–2858.
106 J O U RN A L OF E N V I RO N ME N TA L S CIE N CE S 7 4 (2 0 1 8) 9 5– 1 0 6
Seo, Y.H., Park, D., Oh, Y.K., Yoon, S., Han, J.I., 2015. Harvesting of
microalgae cell using oxidized dye wastewater. Bioresour.
Technol. 192, 802–806.
Stratiev, D., Shishkova, I., Tsaneva, T., Mitkova, M., Yordanov, D.,
2016. Investigation of relations between properties of vacuum
residual oils from different origin, and of their deasphalted
and asphaltene fractions. Fuel 170, 115–129.
Toktonov, A.V., Mognonov, D.M., Mazurevskaya, Z.P., Botoeva, S.
O., 2006. Synthesis of polyamidines based on aromatic bis
(imidoyl) chlorides in solution. Polym. Sci. A 48, 1–10.
Väänänen, M., Kupiainen, L., Rämö, J., Sarpola, A., Tanskanen, J.,
2012. Speciation and coagulation performance of novel
coagulant — aluminium formate. Sep. Purif. Technol. 86,
242–247.
Wang, Y., Gao, B.Y., Xu, X.M., Xu, W.Y., Xu, G.Y., 2009.
Characterization of floc size, strength and structure in various
aluminum coagulants treatment. J. Colloid Interface Sci. 332,
354–359.
Yang, Z., Gao, B., Wang, Y., Wang, Q., Yue, Q., 2011. Aluminum
fractions in surface water from reservoirs by coagulation
treatment with polyaluminum chloride (PAC): influence of
initial pH and OH−/Al3 + ratio. Chem. Eng. J. 170, 107–113.
Ying, F., Bao-yu, G., Yi-feng, Z., Xin-yu, Z., Nan, S., 2011. Organic
modifier of poly-silicic-ferric coagulant: characterization,
treatment of dyeing wastewater and floc change during
coagulation. Desalination 277, 67–73.
Zhang, M., Xiao, F., Wang, D., Xu, X., Zhou, Q., 2017. Comparison of
novel magnetic polyaluminum chlorides involved coagulation
with traditional magnetic seeding coagulation: coagulant
characteristics, treating effects, magnetic sedimentation
efficiency and floc properties. Sep. Purif. Technol. 182, 118–127.
Zhao, H., Liu, H., Hu, C., Qu, J., 2009. Effect of aluminum speciation
and structure characterization on preferential removal of
disinfection byproduct precursors by aluminum hydroxide
coagulation. Environ. Sci. Technol. 43, 5067–5072.
Zhao, P., Gao, B., Yue, Q., Kong, J., Shon, H.K., Liu, P., et al., 2015.
Explore the forward osmosis performance using hydrolyzed
polyacrylamide as draw solute for dye wastewater reclamation
in the long-term process. Chem. Eng. J. 273, 316–324.