Accepted Manuscript Investigation on the simultaneous removal of fluoride, ammonia nitrogen and phosphate from semiconductor wastewater using chemical precipitation Haiming Huang, Jiahui Liu, Peng Zhang, Dingding Zhang, Faming Gao PII: DOI: Reference:
S1385-8947(16)31226-8 http://dx.doi.org/10.1016/j.cej.2016.08.134 CEJ 15700
To appear in:
Chemical Engineering Journal
Received Date: Revised Date: Accepted Date:
20 July 2016 27 August 2016 29 August 2016
Please cite this article as: H. Huang, J. Liu, P. Zhang, D. Zhang, F. Gao, Investigation on the simultaneous removal of fluoride, ammonia nitrogen and phosphate from semiconductor wastewater using chemical precipitation, Chemical Engineering Journal (2016), doi: http://dx.doi.org/10.1016/j.cej.2016.08.134
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Investigation on the simultaneous removal of fluoride, ammonia nitrogen and
phosphate from semiconductor wastewater using chemical precipitation
Haiming Huang, Jiahui Liu, Peng Zhang, Dingding Zhang, Faming Gao*
Hebei Key Laboratory of Applied Chemistry, School of Environmental and Chemical Engineering,
Yanshan University, Qinhuangdao 066004, PR China
Abstract: This study investigates the simultaneous removal of the total ammonia nitrogen (TAN),
phosphate (PO4-P) and fluoride (F–) from semiconductor wastewater by chemical precipitation.
The lab-scale experiment results revealed that the fluoride removal by using magnesium salts
produced a good performance. The fluoride present could significantly inhibit the struvite
crystallization, in this process. The inhibition ratio of the fluoride on struvite crystallization
remarkably increased with an increase in the fluoride concentration and a drop in the pH value.
The optimal pH for struvite precipitation in the semiconductor wastewater was taken as 9.5, the
value at which the fluoride effect significantly decreased. Therefore, to further lower the fluoride
effect, an overdose of the magnesium source was required in the process of struvite precipitation.
The experimental results thus indicated that overdosing the bittern was the more effective method
to treat the semiconductor wastewater compared with a brucite overdose; this was because large
amounts of un-reacted brucite remained in the solution, causing increased costs and operation
difficulty when it was employed as magnesium source. The pilot-scale study demonstrated that
97% of the PO4-P, 58% of the TAN and 91% of the F– could be removed from semiconductor
wastewater by a two-stage precipitation process. An economic analysis showed that the treatment
cost of the process proposed was approximately 1.58 $/m3.
Keywords: Ammonia nitrogen, phosphate, fluoride, struvite, semiconductor wastewater.
23 24 25 26 27
*Corresponding Author: Phone: +86 335 8387 741; Fax: +86 335 8061 569;
E-mail: [email protected]
As the global demand for electronic products has rapidly increased over the past decade,
the global semiconductor industries have made considerable strides in development [1, 2].
However, this speedy development of the semiconductor industry also triggers some
environmental risks including the generation of large amounts of wastewaters and high water
demand . The wastewater generated from the semiconductor manufacturing process generally
contains high levels of total ammonia nitrogen (TAN), fluoride (F–) and phosphate (PO4-P) .
The TAN and PO4-P are the well-known significant nutrient substances that induce water
eutrophication. When they exist in substantial quantities in the water bodies, large amounts of
algae and microorganisms would breed, resulting in a higher dissolved oxygen depletion and fish
toxicity. Although fluoride is one of the essential elements of the human body, the excessive
fluoride intake can result in dental and skeletal fluorosis . The safe prescribed fluoride level in
drinking water, according to WHO, is less than 1.5 mg/L . Therefore, a good efficient treatment
of semiconductor wastewater plays a crucial role in the prevention of environment pollution and
human health risk.
Commonly, biological treatment is accepted as the economical and feasible process to
remove nutrients from the wastewaters. However, biological processes may not be very feasible in
the treatment of semiconductor wastewater because of its high content of toxic substances, which
can inhibit the microorganism activity in the biological treatment system [7, 8]. Although fluoride
may be efficiently removed from aqueous solution by the electrodialytic method [9–11], this
process is difficult to be applied to the treatment of the semiconductor wastewater due to the
complexity of the wastewater. As an alternative, precipitation using calcium salts is often used to
treat the semiconductor wastewater . Unfortunately, this process cannot simultaneously
remove the TAN and PO4-P, because it is quickly interrupted by the presence of PO43–, SO42– and
NH4+ in the wastewater, resulting in a decrease in the recovery factor of the CaF2 for several
industrial purposes [13, 14]. Additionally, because the chemical precipitation produces very fine
CaF2 precipitates, flocculants like polyferric sulfate and polyaluminum chloride need to be added
to accelerate the solid separation process . Compared with the precipitation using calcium salts,
struvite crystallization can help remove both the TAN and PO4-P, and has been largely considered
a promising pretreatment method to remove nutrients from various types of wastewaters [16–20]; 2
this process also has several advantages including the high reaction rate, simple operation and
excellent solid-liquid separation performance. Besides, the struvite thus recovered finds use as a
valuable slow-releasing fertilizer. In Japan, struvite has been commercially recovered by Unitika
Ltd., and sold to American fertilizer companies . Hence, struvite crystallization appears to be
an attractive process to pre-treat semiconductor wastewater.
In the earlier literature, some papers reported that the struvite crystallization process is
significantly influenced by certain inorganic ions like Ca2+, K+, Fe3+, CO32–, etc. [22–24], which
could interfere with the nucleation of the struvite crystal or compete with the NH4 + and Mg2+ for
the HPO42–, inhibiting the struvite formation. Besides, some researchers also reported that some
organic substances like citric acid  and humic substances  have an observable inhibitory
impact on the struvite crystallization. However, some papers are available in the literature which
studied the effect of the F– ions on the TAN and PO4-P removal by struvite crystallization. The F–
concentration in the semiconductor wastewater is usually as high as several hundred to several
thousand mg/L, based on the operational conditions . Although Ryu et al.  have confirmed
that the high concentration of F– may inhibit the removal of TAN and PO4-P by struvite
precipitation, their investigation did not specifically report the influence mechanism and impact
strength of the F– on the struvite crystallization under different conditions. Therefore, in this study,
it is necessary to further investigate the mechanism of influence and strength of the fluoride on the
struvite crystallization and identify a process to eliminate the effect.
The objective of this study was to simultaneously remove F–, TAN and PO4-P from the
semiconductor wastewater. To achieve this, first the fluoride was removed from synthetic
wastewater by chemical precipitation utilizing magnesium salts. Second, the experiments were
performed to investigate the influence of fluoride on struvite crystallization at different pHs and
fluoride concentrations. Third, to eliminate the fluoride effect, overdosing with bittern and brucite
mineral powder during chemical precipitation was done; besides, the economic feasibility of both
the magnesium sources was evaluated and compared. Finally, the pilot-scale treatment of
semiconductor wastewater was conducted.
2. Materials and methods
2.1. Experimental materials
The raw semiconductor wastewater used in the experiments was supplied by a 3
semiconductor manufacturer from Shenzhen, China. To ensure sample stability, the wastewater
was stored below 4 oC before use. The main characteristics of the wastewater sample are described
as follows: pH, 7.5 ± 0.05; F–, 1280 ± 56 mg/L; PO4-P, 201 ± 3.4 mg/L; TAN, 131 ± 5.1 mg/L.
Besides, in the experiments, natural brucite mineral powder (particle size, 50-80 µm) and bittern
were used as the magnesium sources of chemical precipitation. The brucite mineral powder was
purchased from a mineral processing plant in Dandong city and about 94% of it was composed of
magnesium hydroxide. The bittern was collected from a solar salt field in Shenzhen, and the chief
constituents were as follows: Mg2+, 45.3 g/L; K+, 10.1 g/L; Ca2+, 142 mg/L. All the other
chemicals used in the experiments were of analytical grade and purchased from the Tianjin
Fengchuan Chemical Reagent Plant, China.
2.2. Experimental methods
1) Removal of fluoride: As the semiconductor wastewater has high fluoride content, the
removal of fluoride ions by co-precipitating with Mg2+ was first investigated. In this work,
synthetic wastewater containing a fluoride concentration of 1280 mg/L was used, which was
prepared by dissolving sodium fluoride (NaF) in deionized water. The experimental procedures are
described as follows: 100 ml of the synthetic wastewater was first added to a 200 ml beaker placed
on a magnetic stirrer, followed by the addition of magnesium chloride (MgCl2· 6H2O) in a
desirable Mg:F molar ratio (0.5–1), and then the wastewater was stirred at a designed pH (8–10)
for 30 min. After the stirring was completed, the mixture was left for 30 min to be precipitated,
and 5 ml of the supernatant was removed and filtered through 0.22-µ m filter membranes for
2) Effect of fluoride on the struvite crystallization: To determine the degree of the influence
of the fluoride on the struvite crystallization, a series of experiments was performed. The
wastewater used in the experiments were prepared by dissolving Na2HPO4· 12H2O and NH4Cl in
deionized water. It was found to contain PO4-P 201 mg/L and TAN 90.7 mg/L. The experiments
were performed as follows: 100 ml of synthetic wastewater was first taken in a 200 ml beaker and
placed on a magnetic stirrer. Next, several milliliters of the NaF stock solution was added to the
wastewater to adjust the concentration of the fluoride ions to be in the range of 0–2100 mg/L.
Then, magnesium chloride was added to the wastewater in a stoichiometric ratio of Mg:N:P. After
that, the mixed solution was stirred for 30 min and the solution pH was controlled at a designed 4
value (8–10) by the addition of 0.1 M NaOH. Finally, the solution at the end of reaction was
allowed to settle for 30 min and 5 ml of the supernatant was drawn and filtered through a 0.22-µ m
filter membrane for component analysis.
3) Simultaneous removal of F–, PO4-P and TAN from real semiconductor wastewater: In
this work, bittern and natural brucite were used as the magnesium source for the chemical
precipitation. The experimental procedures using the bittern were similar to those of the
experiments performed for the effect of fluoride on struvite crystallization. In the experiments, the
molar ratio of Mg:P ranged from 1 to 3, and the solution pH was maintained at 9.5. Regarding the
use of natural brucite, the experiments were conducted according to the following procedures. 500
ml of the semiconductor wastewater was first taken in a 1000 ml beaker, followed by the addition
of natural brucite in different doses (2.4–4.8 g/L, i.e. Mg:P molar ratio, 6–12). Next, the mixture
was stirred for 6 h. During the reaction, the solution pH was measured at every 5 minutes and 1 ml
of each sample was drawn at every 20 minutes and filtered through 0.22-µ m filter membranes for
2.3. Inhibition model of fluoride on struvite crystallization
To determine the degree of the effect of F– on the removal of TAN and PO4-P by struvite
crystallization, an inhibition model was introduced into this study . The inhibition ratios (IR)
of fluoride on the TAN and PO4-P removal efficiency are described as follows.
IR N =
NRE 0 − NRE i × 100% NRE 0
IR P =
PRE0 − PREi × 100% PRE 0
where NRE0 and NREi are represented as the TAN removal efficiency (%) in the absence and
presence of fluoride, respectively. Similarly, the PRE0 and PREi are the PO4 -P removal efficiency
(%) in the absence and presence of fluoride, respectively.
2.4. Pilot-scale treatment of semiconductor wastewater
The pilot-scale treatment of semiconductor wastewater was performed at a semiconductor
plant located in a suburban area of Shenzhen. In this study, the pilot-scale test was operated in a
continuous-flow mode and lasted for 100 h without interruption. The specific schematic diagram
is shown in Fig. 1. As observed in Fig.1, the semiconductor wastewater treatment was divided into 5
two phases, namely, the first and second stages of precipitation. 1) First stage precipitation: the
semiconductor wastewater was first continuously pumped into the Reaction Tank (a), which is a
barrel having a 70 L reacting zone. In this stage, the influent flux of the semiconductor wastewater
was maintained at 70 L/h, which implied a hydraulic retention time (HRT) of 1 h. To
simultaneously remove the phosphate, fluoride and ammonia nitrogen, the bittern stored in the
liquid storage tank was continuously added to the wastewater in a 2:1 molar ratio of Mg:P, and the
solution pH was constantly maintained at 9.5. The mixture in the Reaction Tank (a) flowed
through the metering valve to the Sedimentation Tank (a) (HRT, 2 h). After precipitation, the
precipitates collected at the bottom of the sedimentation tank were discharged at 3 L/h. 2) Second
stage precipitation: the supernatant in the Sedimentation Tank (a) was pumped into the Reaction
Tank (b) (a barrel with a reacting zone of 67 L; HRT, 1 h) for the further removal of phosphate,
fluoride and ammonia nitrogen. In this process, the bittern which served as the magnesium source
was continuously added to the supernatant, and the pH and the Mg:F molar ratio of the solution
were controlled at 9.5 and 1:1, respectively. Through the second precipitation reaction, the mixing
solution was continuously discharged into the Sedimentation Tank (b) (HRT, 2 h). The precipitates
collected at the bottom of the sedimentation tank were drained into a subsequent sludge
concentration tank through the hydrostatic pressure, and the resulting supernatant was discharged
into a subsequent treatment process for advanced treatment. In the tests, all the pumps and pH
meter were controlled by a Programmable Logic Controller and a computer. The stirring rate in
the Reaction Tanks (a) and (b) was kept at 200 rpm. To maintain the stable removal of the
pollutants from the semiconductor wastewater, the main parameters in the pilot-scale treatment
such as the solution pH, the Mg:P molar ratio and Mg:F molar ratio in the Reaction Tanks (a) and
(b) need to be accurately controlled. The deviations of the Mg:P and Mg:F molar ratios should be
controlled within 10%, and the deviation of the solution pH was maintained at ± 0.2. During the
test, at every 2 h, 10 ml of the supernatant in both the sedimentation tanks were drawn and filtered
through 0.22 µ m filter membranes for composition analysis.
Fig. 1 here
2.5. Chemical analysis
Various parameters of semiconductor wastewater were analyzed according to the American
Public Health Association (APHA) standard methods . The concentrations of TAN and PO4-P 6
of wastewater samples were colorimetrically determined using a 752N spectrophotometer (China).
The F– concentration in the sample was determined using an ion selective electrode (Orion 720
A+ Ion analyzer). During experiments, the solution pH was recorded using a pH meter (pHS-3C;
China). The precipitates formed in the experiments were collected and washed thrice with
deionized water, and oven dried at 40°C for 24 h. The morphology of the dried precipitates was
observed using a scanning electron microscope-energy dispersive spectrometer (SEM-EDS; FEI
Nova NanoSEM 450; American). In this study, all the tests were performed in triplicate, and their
average values were reported.
3. Results and discussion
3.1. Removal of fluoride by the co-precipitation of Mg2+
To determine the performance of the Mg2+ on the fluoride removal, a series of experiments
was performed. The removal efficiency of F– by Mg2+ is shown in Fig. 2. As observed in Fig. 2,
the F– removal efficiency increased as the solution pH and the Mg:F molar ratio increased. When
the Mg:F molar ratio was 1 and the pH of solution was 10, the removal efficiency of F– by Mg2+
reached > 90%. When the Mg2+ was added to the wastewater rich in fluoride, if the ionic product
of Mg2+ and F– was greater than the solubility product of the product (Ksp), F– could be removed
as insoluble fluoride precipitates, according to the following reaction equation .
2F – + Mg 2 + → MgF2
Furthermore, during the process of the experiments, the settleability of the fluoride precipitates
formed by Mg2+ was observed. The fluoride precipitates formed by the Mg2+ were observed to
show good settleability, which formed an obvious thin sedimentation layer after 90 min of free
sedimentation. However, in this study, we have performed control experiments using calcium salts
to precipitate the fluoride, and found that although the use of calcium salts as the precipitator of
fluoride can achieve fluoride removal efficiency comparable to that by the Mg2+, the settleability
of the formed precipitates was very poor. It was observed that the resulting solution system was
still turbid after 90 min of free sedimentation, without any obvious solid-liquid separation
interface. The CaF2 precipitate was very fine with a particle size of around 1µ m, which could have
been responsible for the difficulty in settling [15, 29]. Therefore, in the CaF2 precipitating process,
polyaluminum chloride and polyelectrolyte are normally required to be added for attaining a good
solid separation from water . Hence, from the viewpoint of the settleability, it was considered
that the fluoride removal with the Mg2+ appeared to be more feasible than it did with the Ca2+.
Fig. 2 here
3.2. Effect of fluoride on the removal of TAN and PO4-P by struvite crystallization
The experiments done to investigate the effect of fluoride on the struvite crystallization were
performed at the Mg:N:P molar ratio of 1:1:1. Figs. 3a and 3b show the changes in the TAN and
PO4-P removal efficiency with solution pH and fluoride concentration, respectively. The results
demonstrated that the fluoride present in solution could significantly influence struvite formation.
As observed in Fig. 3a, at a given fluoride concentration, the TAN removal efficiency increased
with increasing the solution pH from 8 to 9.5, and reached a maximum at pH 9.5; however, it
progressively decreased with a further increase in the pH in the range of 9.5–10. This finding was
consistent with reports in published investigations . Solution pH is an important factor that
influences struvite crystallization . It could critically influence the existing form and activity
of the NH4+, Mg2+ and HPO4 2– species of the struvite crystal . Struvite formation proceeds by
the following reaction equations:
Mg 2+ + NH 4 + + HPO 4 2 – + 6H 2 O → MgNH 4 PO 4 ⋅ 6H 2 O ↓ + H +
Within the pH range of 8–9.5, the H+ concentration in the solution gradually decreased with the
increase in the pH value, which is beneficial to the crystallization reaction from left to the right,
thus resulting in an obvious increase in the efficiency of TAN removal. However, at pH > 9.5, as a
large quantity of the NH4+ in the solution was converted to NH3 which cannot be precipitated by
struvite crystallization, and the phosphate began to react with Mg2+ to form Mg3(PO4)2 [32, 33],
these would markedly obstruct the reaction process of Eq. (4), causing a drop in efficiency of the
TAN removal in the pH range of 9.5–10. In addition, the decrease of Mg2+ for the formation of
Mg(OH)2 at pH > 9.5 may be another reason resulting in a drop in efficiency of the TAN removal.
Further, it is evident from Fig. 3a that at a given pH, the TAN removal efficiency rapidly
decreased with the increase in the fluoride concentration. When the solution pH was 9.5, the TAN
removal efficiency decreased from 85.6% to 56.3% with the rise in the fluoride concentration
from 0 mg/L to 1200 mg/L.
Fig. 3b shows the effects of the pH of the solution and fluoride concentration on the PO4-P 8
removal, which reveals that at a given fluoride concentration, the PO4-P removal efficiency
increases with an increase in the pH, whereas at a given pH it decreased with the increase in the
fluoride concentration. In this study, quite different from the removal of TAN, which was only
through struvite formation (the NH3 volatilization can be ignored), the PO4-P removal could
proceed through the formations of struvite and magnesium phosphate. At pH >9.5, although the
struvite crystallization was inhibited, magnesium phosphate formation was accelerated. This may
be responsible for the increase in the PO4-P removal efficiency with a rise in the pH value.
Furthermore, combining Figs. 3a and Fig. 3b, it became evident that at a given pH, the TAN and
PO4-P removal efficiency simultaneously decreased with the increase in the fluoride concentration.
Based on the results in Section 3.1, it can be confirmed that the F– can compete with NH4+ and
HPO42– for Mg2+ to form MgF2, which resulted in a decrease in the quantity of Mg2+ used for the
struvite crystallization. Besides this, F– as the foreign ions in the crystallization reaction can be
easily adsorbed onto the struvite crystal surface, which can induce the slowing down of the crystal
growth . In the earlier literature, Ryu et al.  reported that as the fluoride concentration was
above 600 mg/L, the removal of TAN and PO4-P by forming struvite were obviously inhibited,
which concurs with our findings.
To further determine the effect mechanism of the fluoride on the struvite crystallization, the
struvite precipitates obtained at pH 9.5 and different fluoride concentrations were characterized by
SEM-EDS (see Fig. 4). The results indicated that the shape of the struvite crystal produced was
markedly affected by the fluoride ions. Fig. 4 reveals that the morphology of the struvite crystals
obtained in the absence of the fluoride ions was columnar with smooth surfaces, whereas the
shape of the crystals formed at fluoride concentrations of 600 mg/L and 1500 mg/L gradually
became irregular blocks with fine amorphous particles on their surfaces. This suggested that the
presence of fluoride may influence the nucleation and growth of the struvite crystals, producing
different shaped products. Besides, the EDS patterns clearly demonstrated that the fluoride
element was present in the precipitates formed, and its peaks were intensified with the increase in
the fluoride concentration.
The IR values of fluoride on the TAN and PO4-P removal efficiencies are shown in Figs. 3c
and 3d, respectively. As evident, the IR values were closely related to the solution pH and fluoride
concentration. At a given fluoride concentration, the IR values of the fluoride on the TAN and 9
PO4-P removal efficiencies decreased rapidly with the increasing pH in the range of 8–9.5, and
plateaued in the pH range of 9.5–10. Otherwise stated, the lower the pH, the greater the inhibition
level of the fluoride ions on the struvite crystallization. For example, when the fluoride
concentration was 300 mg/L, the IR value of the fluoride on the TAN removal efficiency increased
from 5% at pH 9.5 to 31% at pH 8. Any rise in the fluoride concentration could rapidly increase
the IR values. When the fluoride concentration increased from 300 mg/L to 2100 mg/L at pH 9,
the IR values of the fluoride for the removal efficiencies of TAN and PO4-P increased from around
5% to around 70%. Therefore, it can be confirmed that maintaining a high pH level and lowering
the fluoride ion concentration could contribute to the struvite formation in the semiconductor
Fig. 3 here
Fig. 4 here
3.3. Simultaneous removal of F–, PO4-P and TAN from real semiconductor wastewater
Based on the results mentioned above, it can be confirmed that an excessive magnesium dose
is required to achieve the efficient removal of PO4-P and TAN from real semiconductor
wastewater by struvite crystallization. Therefore, in the subsequent experiments, bittern and
brucite mineral powder were used as the magnesium sources of chemical precipitation to
simultaneously remove F–, TAN and PO4-P from the semiconductor wastewater.
3.3.1. The use of bittern
In the investigation mentioned above, the inhibition of fluoride on struvite crystallization was
understood to be closely related to the solution pH and fluoride concentration. Therefore, it was
assumed that if the fluoride concentration in the semiconductor wastewater was decreased at a
given pH, the inhibition of fluoride on the struvite crystallization would correspondingly decrease,
resulting in the improvement of the removal efficiencies of the TAN and PO4-P. Hence, in this
work, the bittern was overdosed to eliminate the fluoride influence on the struvite crystallization.
The results of the removal efficiencies of F–, PO4 -P and TAN from real semiconductor wastewater
by chemical precipitation are shown in Fig. 5. As observed from Fig. 5a, the removal efficiency of
the F– rapidly escalated with the increase in the magnesium dose, implying that the fluoride
inhibition on the removal of PO4-P and TAN would be decreased. In fact, from Fig. 5b, it is noted
that with the increase in the molar ratio of Mg:P in the range of 1–1.75, the TAN removal 10
efficiency rose from 39.9% to 53.2%. This may be attributed to the lowered inhibition of the
fluoride and the increase in the quantity of magnesium utilized in the struvite crystallization.
Certain publications have reported that the correct increase of the magnesium dose could promote
the removal of TAN by struvite crystallization [34, 35]. However, a further increase in the molar
ratio of Mg:P in the range of 1.75–3 produced a drop in the removal of TAN. This result concurred
with the finding in an earlier publication in the literature . This may have occurred due to the
fact that an excessive magnesium dose may have induced the formation of magnesium phosphate,
which in turn could have competed with the NH4+ for HPO42– and Mg2+. Although the magnesium
phosphate formed could result in a decrease in the removal of TAN, it was advantageous in the
removal of phosphate. This could be corroborated by the results shown in Fig. 5c. Fig. 5c reveals
that the PO4-P removal efficiency rose as the Mg:P molar ratio increased. The PO4-P removal
efficiency achieved 90.1% when the molar ratio of Mg:P was 2:1, and it further increased to
95.5% when the ratio was increased to 3:1. This finding was consistent with the result reported by
Warmadewanthi and Liu , who used magnesium chloride to treat semiconductor wastewater
and achieved a PO4-P removal efficiency of 92.1% at a molar ratio of Mg:P of 3:1. Fig. 5 here
3.3.2. The use of brucite mineral powder
In order to compare the performance of bittern, the utilization of brucite mineral powder for
treating semiconductor wastewater was further investigated. In this investigation, the brucite
dosage ranged from 2.4 g/L to 4.8 g/L, implying that the Mg:P molar ratio is in the range of 6–12
or the Mg:F molar ratio is between 0.58 and 1.16. The results explaining the effects of the brucite
dose and reaction time on the pH value and the removal efficiencies of PO4-P, TAN and F– are
shown in Fig. 6(a–d). In Fig. 6a, the solution pH values for all the brucite doses tested it can be
observed to rapidly increase with the escalations of the brucite dose and reaction time in the first
240 min, and then gradually reach a plateau between 240 and 360 min. This plateau may be
attributed to the achievement of the dissolution equilibrium of Mg(OH)2. Fig. 6b shows that when
the brucite dose was 2.4 g/L, the TAN removal efficiency progressively increased within the
reaction time tested. However, when the brucite dose was greater than 2.4 g/L, the TAN removal
efficiency rapidly increased during the initial 160 min, followed by a slow drop in the reaction
time range of 160–360 min. Besides, from Fig. 6b it is evident that the TAN removal efficiency 11
decreased with an increase in the brucite dose between 240 min and 360 min. This event may have
resulted from the high pH, which was higher than 9.5 for the brucite dose, greater than 2.4 g/L in
the reaction time range of 160–360 min.
Figs. 6c and 6d show the changes in the PO4-P removal efficiency and the F– removal
efficiency with the brucite dose and reaction time, respectively. As observed in Fig. 6c, although
the PO4-P removal efficiency progressively increased with the increases in the brucite dose and
reaction time, it never exceeded 80% in the batch experiments. This finding does not concur with
the publications reported in the literature, in which the phosphate in wastewater can be almost
completely removed by overdosing brucite or Mg(OH)2 as the magnesium source [16, 36]. In this
study, the competition between F– and PO43– for Mg2+ may be the main reason for the occurrence,
which may induce a lack of Mg2+ for the phosphate removal. As seen in Fig. 6d, when the brucite
dose increased from 2.4 g/L to 4 g/L, the F– removal efficiency also increased; however, further
increases in the brucite dose caused only a very small rise in the fluoride removal. Furthermore,
for all the brucite doses tested, the F– removal efficiency gradually plateaued when the reaction
time crossed 200 min. This suggested that the reaction between F– and Mg2+ was almost
completed. However, in this study, the brucite dosage far exceeded that required for the removal of
phosphate and fluoride. This suggested that large quantities of brucite were not involved in the
From the results mentioned above and the published literatures [37, 38], we proposed a
reaction mechanism for the simultaneous removal of phosphate, ammonia nitrogen and fluoride by
using brucite as a magnesium source, the sketch map of which is described in Fig. 7. As observed
in Fig. 7, the reaction process was distinguishable into three stages: 1) in the first stage, the Mg2+
and OH– were rapidly released from the brucite and diffused into the bulk solution, while the F–,
PO43– and NH4+ diffused into the brucite surface at the same time, followed by the formation of
struvite and magnesium fluoride; 2) in the second stage, the struvite and magnesium fluoride grew
on the brucite surface and the pH of the bulk solution rapidly increased, which blocked any further
brucite dissolution; 3) in the third stage, the brucite dissolution achieved an equilibrium state and
the removal reactions of phosphate and fluoride culminated. In this reaction mechanism proposed,
only the formation of struvite and magnesium fluoride was considered. However, some other
compounds such as magnesium phosphate may also be formed. 12
Fig. 6 here
Fig. 7 here
3.3.3. Comparison analysis of both magnesium sources
In this investigation, a comparative analysis for the simultaneous removal of phosphate,
ammonia nitrogen and fluoride employing bittern and brucite as the magnesium sources was
performed. The analysis mainly related to the economic evaluation and the difficulty level of the
operation. In this comparative analysis, the reaction conditions were set as follows: bittern, at a
dose of 7g/L (i.e. Mg:P molar ratio, 2:1) and a pH of 9.5 for 60 min; brucite, at a dose of 4 g/L for
240 min. In the economic evaluation, to simplify the calculation, only the costs of the chemicals
utilized and the energy consumed were considered; the manpower and equipment maintenance
costs were disregarded. The results of the economic evaluation are shown in Table 1. The cost for
treating semiconductor wastewater using bittern was found to be 0.44 $/m3 , which has a better
economic advantage compared to that by using brucite (0.76 $/m3). From the operation aspect,
although the removal efficiencies of phosphate, ammonia nitrogen and fluoride employing brucite
were comparable to those by using bittern, a big difference was observed in the level of difficulty
of their operation. On the one hand, because the use of brucite as the magnesium source required a
longer stirring reaction time compared with the use of bittern, this implied that a bigger reactor
was needed, especially constructed for the semiconductor wastewater treatment. This would
greatly increase the investment cost for the equipment. On the other hand, when brucite was
employed as the magnesium source, more precipitates were produced when compared with the
utilization of bittern because a large amount of un-reacted brucite remained in the solution. This
increased the difficulty in the subsequent disposal of solid waste. Therefore, from the viewpoints
of economy and operability, it can be confirmed that the treatment of semiconductor wastewater
employing bittern as the magnesium source was optimal. Hence, in the subsequent pilot-scale
treatment, bittern alone was used as the magnesium source for the removal of phosphate, ammonia
nitrogen and fluoride from semiconductor wastewater.
Table 1 here
3.4. Pilot-scale treatment of semiconductor wastewater
To further determine the feasibility of the simultaneous removal of phosphate, ammonia
nitrogen and fluoride from semiconductor wastewater by chemical precipitation, a pilot-scale test 13
was performed based on the lab-scale experiment results. In this test, to maintain the stability of
water quality, all the wastewater required during test was stored in advance, in a 10 m3 storage
pool. The characteristics of the semiconductor wastewater used in the pilot-scale study were as
follows: pH, 7.55; F–, 1162 mg/L; PO4-P, 205 mg/L; TAN, 135 mg/L. The results of the pilot-scale
test are shown in Fig. 8, and the concentration changes of the pollutants in the remained solution
at different precipitation stages are given in Table 2. As observed in Fig. 8 and Table 2, the
removal performance of the phosphate, ammonia nitrogen and fluoride by the pilot-scale process
were found to be consistent with those in the lab-scale experiments. In the first precipitation stage,
the average removal efficiencies of the phosphate, ammonia nitrogen and fluoride reached 90 ±
3%, 50 ± 4% and 20 ± 2%, respectively, and the average remaining concentrations of the
phosphate, ammonia nitrogen and fluoride decreased to 21, 68 and 930 mg/L, respectively;
subsequently, in the second precipitation, the average removal efficiencies of the three pollutants
further rose to 97 ± 1.5%, 58 ± 2% and 91± 2.4%, respectively, and their average remaining
concentrations were further decreased to 6.2, 57 and 105 mg/L, respectively. In the published
literature, some papers are available on the treatment of semiconductor wastewater by struvite
crystallization. Ryu et al.  considered that struvite precipitation was an effective process for
the treatment of semiconductor wastewater, and can achieve 89% ammonia nitrogen removal
efficiency. Kim et al.  in their investigation on the effects of the mixing intensity and mixing
duration on the struvite precipitation in semiconductor wastewater, reported that under certain
conditions, the removal efficiencies of phosphate, ammonia nitrogen and fluoride can reach
approximately 70%, 80% and 20%, respectively. Warmadewanthi and Liu  reported that
84.2% of the PO4-P and 33.5% of the TAN were removed from semiconductor wastewater when
the ratio of Mg:P was 2.5:1 at pH 9. On comparison with these literatures, we can state that the
process proposed in this study can simultaneously achieve higher removal efficiencies of
phosphate and fluoride. Besides, based on the economic evaluation method in Section 3.3.3, the
cost incurred for treating 1 m3 semiconductor wastewater by the pilot-scale process involving two
stages of precipitations was calculated to be approximately $1.58. Overall, this cost is acceptable
by most semiconductor manufacturing enterprises in China. As for the final disposal of the
precipitates recovered by the proposed process, since the precipitates contained large amounts of
fluoride, it cannot be used as the agricultural slow release fertilizer; however, it can be served as a 14
phosphate rock to be reused in industrial production. Furthermore, through the pretreatment of the
proposed process, the amount of the harmful substances in the semiconductor wastewater will be
greatly reduced, which would be beneficial to the stable operation in the following biological
Fig. 8 here
Table 2 here
In the present study, chemical precipitation was practiced to simultaneously remove the
phosphate, ammonia nitrogen and fluoride from semiconductor wastewater. The lab-scale
conditional experiments were initially performed to investigate the effects of different parameters
on chemical precipitation. Subsequently, a pilot-scale study was performed to determine the
feasibility of treating semiconductor wastewater with a two-stage precipitation process. The main
conclusions drawn from the study are given below:
428 429 430 431
1) Magnesium salts as the precipitator exhibited a superior settleability with respect to fluoride removal when compared with the calcium salts. 2) Fluoride significantly inhibited struvite crystallization and its inhibition ratio increased with the rise in fluoride concentration and the lowering pH value.
3) Despite the usage of bittern and brucite as the sources of magnesium to attain comparable
phosphate, ammonia nitrogen and fluoride removal from semiconductor wastewater, the
utilization of bittern was more advantageous in terms of treatment cost and the operability
compared with the use of brucite.
4) The pilot-scale process proposed, involving a two-stage precipitation, was economical and
feasible for the treatment of semiconductor wastewater, as it could achieve a high degree of
success in the removal of PO4-P and F– (97% and 91%, respectively).
This work was financially supported by the National Natural Science Foundation of China
(Grant No. 51408529), the Natural Science Foundation of Hebei Province (Grant No.
E2014203080), the Outstanding Young Scholars Project of Colleges and Universities of Hebei
province (Grant No. BJ2014059), and China Postdoctoral Science Foundation Funded Project
(Grant Nos. 2015M580215, 2015M581319 and 2016T90215). 15
 S. Bang, J. Choi, K. Cho, C. Chung, H. Kang, S. W. Hong, Simultaneous reduction of copper
and toxicity in semiconductor wastewater using protonated alginate beads, Chem. Eng. J.
288 (2016) 525–531.
 Y. C. Liang, A. Daverey, Y. T. Huang, S. Sung, J. G. Lin, Treatment of semiconductor
wastewater using single-stage partial nitrification and anammox in a pilot-scale reactor, J. Taiwan
Inst. Chem. Eng.63 (2016) 236–242.
 C.J. Huang, B.M. Yang, K.S. Chen, C.C. Chang, C.M. Kao, Application of membrane
technology on semiconductor wastewater reclamation: A pilot-scale study, Desalination 278 (2011)
 Y. Xiao, T. Chen, Y. Hu, D. Wang, Y. Han, Y. Lin, X. Wang, Advanced treatment of
semiconductor wastewater by combined MBR–RO technology, Desalination 336 (2014) 168–178.
 L. Deng, Y. Liu, T. Huang, T. Sun, Fluoride removal by induced crystallization using
fluorapatite/calcite seed crystals, Chem. Eng. J. 287 (2016) 83–91.
 WHO, Guidelines for Drinking-water Quality, fourth ed., Geneva, 2011, pp. 370–373.
 D. Kim, J. Kim, H.D. Ryu, S. Lee, Effect of mixing on spontaneous struvite
precipitation from semiconductor wastewater, Bioresour. Technol. 100 (2009) 74–78.
 S.H. Lin, C.D. Kiang, Combined physical, chemical and biological treatments of wastewater
containing organics from a semiconductor plant, J. Hazard. Mater. 97 (2003) 159–171.
 R.S. Keri, K.M. Hosamani, H.R.S. Reddy, S.K. Nataraj, T.M. Aminabhavi, Application of the
electrodialytic pilot plant for fluoride removal, J. Water Chem. Technol. 33 (2011) 293–300.
 S. K. Nataraj, K. M. Hosamani, T. M. Aminabhavi, Electrodialytic removal of nitrates and
hardness from simulated mixtures using ion-exchange membranes, J. Appl. Polym. Sci. 99 (2006)
S.K. Nataraj, S. Sridhar, I.N. Shaikha, D.S. Reddy, T.M. Aminabhavi, Membrane-based
microfiltration/electrodialysis hybrid process for the treatment of paper industry wastewater, Sep.
Purif. Technol. 57 (2007) 185–192
 C.Y. Hu, S.L. Lo, W.H. Kuan, Y.D. Lee, Treatment of high fluoride-containing wastwater by 16
continuous electrocoagulation – flotation systemwith bipolar aluminum electrodes, Sep. Purif.
Technol. 39 (2008) 1–5.
 R. Aldaco, A. Garea, A. Irabien, Calcium fluoride recovery from fluoride wastewater in a
fluidized bed reactor, Water Res. 41 (2007) 810–818.
 Warmadewanthi, J.C. Liu, Recovery of phosphate and ammonium as struvite from
semiconductor wastewater, Sep. Purif. Technol. 64 (2009) 368–373.
 C. Liu, J.C. Liu, Coupled precipitation-ultrafiltration for treatment of high fluoride-content
wastewater, J. Taiwan Inst. Chem. Eng. 58 (2016) 259–263.
 S. G. Barbosa, L. Peixoto, B. Meulman, M. M. Alves, M. A. Pereira, A design of experiments
to assess phosphorous removal and crystal properties in struvite precipitation of source separated
urine using different Mg sources, Chem. Eng. J. 298 (2016) 146–153.
 O. Lahav, M. Telzhensky, A. Zewuhn, Y. Gendel, J. Gerth, W. Calmano, L. Birnhack,
Struvite recovery from municipal-wastewater sludge centrifuge supernatant using seawater NF
concentrate as a cheap Mg(II) source, Sep. Purif. Technol. 108 (2013) 103–110.
 O. Ichihashi, K. Hirooka, Removal and recovery of phosphorus as struvite from
swine wastewater using microbial fuel cell, Bioresour. Technol. 114 (2012) 303–307.
S. He, Y. Zhang, M. Yang, W. Du, H. Harada, Repeated use of MAP decomposition residues
for the removal of high ammonium concentration from landfill leachate, Chemosphere 66 (2007)
 H.M. Huang, D. Xiao, J.H. Liu, L. Hou, L. Ding, Recovery and removal of nutrients from
swine wastewater by using a novel integrated reactor for struvite decomposition and recycling, Sci.
Rep. 5 (2015) 10183, doi:10.1038/srep10183.
 J.E. Lee, M.M. Rahman, C.S. Ra, Dose effects of Mg and PO4 sources on the composting of
swine manure, J. Hazard. Mater. 169 (2009) 801–807.
 C. D. Iaconi, M. Pagano, R. Ramadori, A. Lopez, Nitrogen recovery from a stabilized
municipal landfill leachate, Bioresour. Technol. 101 (2010) 1732–1736.
 Y. Song, P. Yuan, B. Zheng, J. Peng, F. Yuan, Y. Gao, Nutrients removal and recovery by
crystallization of magnesium ammonium phosphate from synthetic swine wastewater, 17
Chemosphere 69 (2007) 319–324.
 H. Huang, Y. Chen, Y. Jiang, L. Ding, Treatment of swine wastewater combined with
MgO-saponification wastewater by struvite precipitation technology, Chem. Eng. J. 254 (2014)
 Y. Song, Y. Dai, Q. Hu, X. Yu, F. Qian, Effects of three kinds of organic acids on phosphorus
recovery by magnesium ammonium phosphate (MAP) crystallization from synthetic swine
wastewater, Chemosphere 101 (2014) 41–48.
 Z. Zhou, D. Hu, W. Ren, Y. Zhao, L. Jiang, L. Wang, Effect of humic substances on
phosphorus removal by struvite Precipitation, Chemosphere 141 (2015) 94–99.
 H.D. Ryu, D. Kim, S. Lee, Application of struvite precipitation in treating
ammonium nitrogen from semiconductor wastewater, J. Hazard. Mater. 156 (2008)
 APHA, 1998. Standard methods for the examination of water and wastewater, 19th ed.
American Public Health Association/American Water Works Association/Water Environment
Federation, Washington, DC, USA.
 S.V. Jadhav, C.R. Gadipelly, K.V. Marathe, V.K. Rathod, Treatment of fluoride concentrates
from membrane unit using salt solutions, J. Water Process Eng. 2 (2014) 31–36.
 L. Zhai, Y. Sun, C. He, Research on coagulation/sedimentation process for simulation of
fluorine-containing wastewater treatment, Appl. Mech. Mater. 361–363 (2013) 755–759.
 T. Zhang, L. Ding, H. Ren, X. Xiong, Ammonium nitrogen removal from coking
wastewater by chemical precipitation recycle technology, Water Res. 43(2009)
 H. Huang, J. Liu, J. Xiao, P. Zhang, F. Gao, Highly efficient recovery of ammonium nitrogen
from coking wastewater by coupling struvite precipitation and microwave radiation technology,
ACS Sustainable Chem. Eng. 4 (2016) 3688−3696.
 I. Stratful, M.D. Scrimshaw, J.N. Lester, Conditions influencing the precipitation of
magnesium ammonium phosphate, Water Res. 35 (2001) 4191–4199.
 T. Zhang, L. Ding, H. Ren, Pretreatment of ammonium removal from landfill leachate
by chemical precipitation, J. Hazard. Mater. 166 (2009) 911–915.
 S. Uludag-Demirer, M. Othman, Removal of ammonium and phosphate from the supernatant 18
of anaerobically digested waste activated sludge by chemical precipitation, Bioresour. Technol.
100 (2009) 3236–3244.
 H. M. Huang, X. M. Xiao, L. P. Yang, Removal of ammonium from rare-earth wastewater
using natural brucite as a magnesium source ofstruvite precipitation, Water Sci. Technol. 63 (2011)
 J.M. Chimenos, A.I. Fernández, G. Villalba, M. Segarra, A. Urruticoechea, B. Artaza, F.
Espiell, Removal of ammonium and phosphates from wastewater resulting from the process of
cochineal extraction using MgO-containing byproduct, Water Res. 37 (2003) 1601–1607.
 T. Chen, X. Huang, M. Pan, S. Jin, S. Peng, P. H. Fallgren, Treatment of coking wastewater
by using manganese and magnesium ores, J. Hazard. Mater. 168 (2009) 843–847.
541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 19
561 562 563
Table 1. The market prices of the chemicals used and energy consumed and economic evaluation
of using bittern and brucite as magnesium sources.
Table 2. The changes of the remaining concentrations of the phosphate, ammonia nitrogen and
fluoride in the remained solution at different precipitation stages.
Fig. 1. The flow diagram of the pilot-scale process for treating the semiconductor wastewater by
Fig. 2. The removal efficiency of the fluoride by Mg2+ at different solution pHs and Mg:F molar
Fig. 3. Struvite crystallization at different pH and fluoride concentrations (a) effect of fluoride on
TAN removal, (b) effect of fluoride on PO4-P removal, (c) IR value of fluoride on TAN removal
efficiency and (d) IR value of the fluoride on the PO4-P removal efficiency.
Fig. 4. SEM-EDS characterization results of the struvite crystals obtained at pH 9.5 and different
fluoride concentrations: (a) and (b) are the SEM micrograph and the EDS pattern at a fluoride
concentration of 0 mg/L, respectively; (c) and (d) are the SEM micrograph and the EDS pattern at
a fluoride concentration of 600 mg/L, respectively; (e) and (f) are the SEM micrograph and the
EDS pattern at a fluoride concentration of 1500 mg/L, respectively.
Fig. 5. The changes in the removal efficiency of fluoride, TAN and PO4 -P from semiconductor
wastewater by chemical precipitation with the Mg:P molar ratio (a) fluoride removal efficiency, (b)
TAN removal efficiency and (c) PO4-P removal efficiency.
Fig. 6. The changes in the solution pH (a), TAN removal efficiency (b), PO4-P removal efficiency
(c) and fluoride removal efficiency (d) with the brucite dose and reaction time.
Fig. 7. Diagrammatic sketch of simultaneously removing the phosphate, ammonia nitrogen and
fluoride with brucite.
Fig. 8. Pilot-scale treatment of semiconductor wastewater by chemical precipitation involving two
stages of precipitations: (a) the removal efficiency of PO4 -P, TAN and F– in the first stage of 20
precipitation, (b) the removal efficiency of PO4-P, TAN and F– in the second stage of precipitation.
Table 1. The market prices of the chemicals used and energy consumed and economic evaluation of using bittern and brucite as magnesium sources. Market price ($/kg)
Cost of using bittern ($/m3)
Cost of using brucite ($/m3)
Energy consumed Total
596 597 598 599 600 601 602
Table 2. The changes of the remaining concentrations of the phosphate, ammonia nitrogen and
fluoride in the remained solution at different precipitation stages. Initial concentration
Average remaining concentration
Average remaining concentration in
of the influent
in the first stage precipitation
the second stage precipitation
605 606 607 608 609 610 611 21
612 613 614
615 616 617
618 619 620
Fig. 1. The flow diagram of the pilot-scale process for treating the semiconductor wastewater by chemical precipitation.
621 622 623 624 625 626 627 628 629 630 22
631 632 633
634 635 636
637 638 639
Fig. 2. The removal efficiency of the fluoride by Mg2+ at different solution pHs and Mg:F molar ratios.
640 641 642 643 644 645 646 647 648 649 650 23
651 652 653
654 655 656 657
658 659 660 661
Fig. 3. Struvite crystallization at different pH and fluoride concentrations (a) effect of fluoride on TAN removal, (b) effect of fluoride on PO4-P removal, (c) IR value of fluoride on TAN removal efficiency and (d) IR value of the fluoride on the PO4-P removal efficiency.
662 663 664 665 24
666 667 668
669 670 671
672 673 674 675 676
Fig. 4. SEM-EDS characterization results of the struvite crystals obtained at pH 9.5 and different fluoride concentrations: (a) and (b) are the SEM micrograph and the EDS pattern at a fluoride concentration of 0 mg/L, respectively; (c) and (d) are the SEM micrograph and the EDS pattern at a fluoride concentration of 600 mg/L, respectively; (e) and (f) are the SEM micrograph and the 25
EDS pattern at a fluoride concentration of 1500 mg/L, respectively.
678 679 680
681 682 683 684 685
686 687 688 689
Fig. 5. The changes in the removal efficiency of fluoride, TAN and PO4 -P from semiconductor wastewater by chemical precipitation with the Mg:P molar ratio (a) fluoride removal efficiency, (b) TAN removal efficiency and (c) PO4-P removal efficiency.
690 691 692 693 26
694 695 696 697
698 699 700 701 702
703 704 705
Fig. 6. The changes in the solution pH (a), TAN removal efficiency (b), PO4-P removal efficiency (c) and fluoride removal efficiency (d) with the brucite dose and reaction time.
706 707 708 709 710 27
711 712 713 714
715 716 717 718
719 720 721
Fig. 7. Diagrammatic sketch of simultaneously removing the phosphate, ammonia nitrogen and fluoride with brucite.
722 723 724 725 726 727 728 729 730 731 28
732 733 734 735
736 737 738
739 740 741 742
Fig. 8. Pilot-scale treatment of semiconductor wastewater by chemical precipitation involving two stages of precipitations: (a) the removal efficiency of PO4 -P, TAN and F– in the first stage of precipitation, (b) the removal efficiency of PO4-P, TAN and F– in the second stage of precipitation.
743 744 745 746 29
747 748 749
754 755 756 757 758 759 760
Treatment of semiconductor wastewater by chemical precipitation was investigated. Removal of fluoride by Mg2+ showed good performance. Fluoride ions exert a significant inhibitory effect on struvite crystallization. Bittern is suitable to treat semiconductor wastewater. PO4-P and F– can be efficiently removed using the two-stage precipitation process.