You are on page 1of 14


Dominique Claveau-Mallet1*, Scott Wallace2 and Yves Comeau1

Department of Civil, Geological and Mining Engineering Ecole Polytechnique Montreal,

Montreal, QC, Canada, H3C 3A7
Naturally Wallace Consulting, 4276 Greenhaven Court, Vadnais Heights, Minnesota, 55127
* Email:
The capacity of a slag filter for extensive combined treatment of phosphorus, fluoride and metals
in a mining wastewater was assessed. Ten slags and one apatite were tested in short-term
individual batch tests to compare their treatment efficiencies. Two slags were chosen for their
high pH rise capacity (EAF slag from Forth Smith) and high fluoride treatment capacity (EAF
slag from Blytheville). Three treatment sequences containing one or two different slags were
tested in short-term sequential batch tests to compare their treatment efficiencies. The same three
sequences were tested in long-term column tests (43 to 95 days) fed with two reconstituted
mining wastewaters operated at 32 hours total hydraulic retention time. All the sequences
achieved high removal efficiency for all the contaminants. EAF slag from Blytheville was more
efficient for fluoride removal than EAF slag from Forth Smith, but its longevity was shorter.
Fluoride removal was possible only if phosphorus was present in the inlet water.
KEYWORDS: steel slag, phosphorus, fluoride, metals, mining wastewater.
The mining industry faces major environmental concerns, if it discharges effluents that have the
potential to damage ecosystems and human health. The various composition of mining
wastewaters has lead to the development of several treatment systems. Systems for leachate from
abandoned mines have to be efficient for a long period of time and cost effective, as they are
often implemented as a condition of post-closure plans. In this study, the potential of steel slag
filters as an extensive treatment for a closed gypsum mine leachate was assessed. The leachate
contained phosphorus, fluoride and metals.
Steel slag is a by-product material generated by the steel industry. It has previously been shown
capable of removing phosphorus efficiently from various types of wastewaters (Chazarenc et al.,
2008). As steel slag is a low-cost and abundant material, it can be combined to small secondary
treatment systems (such as wetlands) to improve the phosphorus removal of the system avoiding
the continuous addition of chemical products (Higgins et al., 2000; Drizo et al., 2002; Drizo et
al., 2006; Chazarenc et al., 2007). Steel slag filters can achieve phosphorus removal for periods
of more than 5 years (Pratt et al., 2010). However, little attention has been given to the capacity
of slag filter to treat multiple components of wastewater.

Fluoride removal from drinking water was studied by adsorption on various media such as coal
fly ash or natural mud (Xue et al., 2009; Chen et al., 2010; Lu et al., 2010). Adsorption on slag
was also used for wastewaters with higher fluoride concentration (Huang et al., 2011; Xu et al.,
2011). Fluoride removal by precipitation (mainly as CaF2) was tested with chemical additives
such as polymers or inorganic coagulants (Yang et al., 2001) or in fluidized bed of silica sand
(Aldaco et al., 2005). Several specific industrial wastewaters (semiconductor manufacturers and
electronic industries) contain both high fluoride and phosphorus the presence of which reduces
the removal efficiency of fluoride by CaF2 formation (Warmadewanthi et al., 2009). Some
authors reported selective precipitation of phosphorus or fluoride from wastewater (Yang, et al.,
2001; Grzmil et al., 2006; Warmadewanthi, et al., 2009). All these processes for fluoride removal
were efficient, but involved the continuous addition of chemicals and the production of chemical
sludge that was expensive to manage and required significant maintenance.
The use of slag for the treatment of metals was well documented in a recent critical review (Zhou
et al., 2010). Slag was mainly studied as an adsorbent in single-metal contaminant batch tests and
characterized with Langmuir or Freundlich equations.
The aim of this study was to assess the combined removal performance of a slag filter for various
contaminants (phosphorus, fluoride and metals) with minimal maintenance. Short-term batch
tests and long-term column tests were conducted in laboratory with a reconstituted mine
The experimental program included three parts: individual batch tests, sequential batch tests and
column tests. Individual batch tests were used to compare the treatment potential of different
media. Sequential batch tests are a slightly modified version of the individual batch tests, where
different treatment sequences are compared. Finally, column tests were conducted with the best
media and sequences selected from the first two steps. A detailed description of the three
experimental phases is given below.
Reconstituted Effluent and Media
Tests were conducted with two reconstituted mine effluents. These effluents were prepared using
laboratory-grade salts (e.g. KH2PO4, CaCl2, MnSO4, etc.) dissolved in tap water. Two
reconstituted effluents were prepared to represent the minimum and maximum concentrations
observed in the field. Salts were individually dissolved then mixed together. pH was adjusted to
the field value with 25 to 45 mL of H2SO4 6N. The solution was let to rest for at least 24 h to
reach equilibrium (a small amount of precipitate formed during the mixing of individually
dissolved salts) and settled. The supernatant was used in all tests. A sample of the initial solution
was filtered and used for analytical characterization. The average composition of the
reconstituted effluents is shown in Table 1.

Table 1. Average composition of the reconstituted effluents.


Units Min - max concentration

6.92 5.67
154 311
32 177
17 132
16 17
1.7 8.2
0.24 0.83
0.20 3.3
mg S/L
201 317
25 265
mg P/L
11 107
9 - 37

Nine types of slag from the United States and one from Quebec, Canada, and one type of apatite
was tested (Table 2). All slag samples were sieved to 5-10 mm, washed and air dried before use
in batch tests. Some slag samples had to be crushed to achieve 5-10 mm because of the original
size of the material (slag #2, 3, 4, 8 and 9). The apatite sample was tested in batch tests without
initial sieving (5-20 mm).
Table 2. Tested slags and apatite.
Slag/Steel compagny
Harsco Metals
Phoenix Services
Phoenix Services
Granite City Slag LLC
Tube City
Charleston Mill Service
Golden Triangle Mill Service
10 Stein
Ap1 Cargill



Blytheville, Arkansas
Indiana Harbor East, Indiana
Indiana Harbor East, Indiana
Granite City, Illinois
Decatur (Trinity), Alabama
Contrecoeur, Quebec
Forth Smith, Arkansas
Charleston, South Carolina
Columbus, Mississippi
Cleveland, Ohio (Arcelor Mittal)
Florida (Cargill)

Note: ACBF: air cooled blast furnace slag; BF; blast furnace slag; BOF: basic oxygen furnace slag; EAF: electric
arc furnace slag.

Experimental Program
Individual batch tests were conducted in duplicate using a 35 g media sample that was placed in
a 1000 mL Erlenmeyer flask containing 700 mL of solution. The Erlenmeyer flask was shaken
for 48 hours. Samples were taken after 0, 24 and 48 hours and filtered for the determination of P,
F, Al, Mn, Zn and pH. Two solutions (Table 1) were tested for each media and the removal
capacity of media were compared. A blank test containing the solution without media was
conducted with each test group (7 tests including one blank sample conducted at the same time).

For sequential batch tests, different treatment sequences were tested. A 24 h individual batch test
was performed with a first media and then the solution was filtered and used for a second
individual batch test of 24 h with a second media. The treatment performance of a two-media
sequence (24 h with first media and 24 h with second media) was compared with the treatment
performance of a single-media system (48 h of the same media, as tested in individual batch
tests). A blank test without media was conducted with each test group.
Two media were selected for the column tests: slag #1 (EAF slag from Blytheville) and slag #7
(EAF slag from Fort Smyth). Four series of two transparent plastic columns (17 cm long and 15
cm in diameter) were fed during 43 to 95 days with a peristaltic pump. Three treatment
sequences and two reconstituted effluents were tested, as presented in Figure 1. The mean initial
void hydraulic retention time (HRT) was 16 h for each column. The effluent of each column was
collected twice or once a week for pH, P, F, Ca, Mn, Zn and Al analyses. pH was measured in
the next half-hour following sampling to minimize interference with the atmosphere. Pictures of
column tests are shown in Figure 2.

Figure 1. Experimental setup for column tests.

Individual Batch Tests
Results of individual batch tests are shown in Figures 3 and 4. Eleven media are identified in the
legend as described in Table 2. Low and high concentration tests were compared. The removal
mechanism appeared to be mainly due to precipitation, as white precipitates were formed during

the tests. In a high pH environment, phosphorus precipitates in a variety of metastable calcium

phosphates that normally transforms into hydroxyapatite (Ca5(PO4)3OH), a stable mineral phase
(Lundager Madsen, 2008).
Results confirmed that phosphorus could be removed efficiently by slag, as ortho-phosphates
concentrations were below 0.1 mg P/L after 48 hours for several media (Figure 3B and 3E). The
phosphorus removal is strongly linked with a rise in pH, as reported in previous studies
(Chazarenc, et al., 2008; Vohla et al., 2011). A similar behavior was observed for manganese,
zinc and metal removal. The correlation between fluoride removal and a high pH was present but
not as clear.

Column 1B, t = 47 d

Column 1A, t = 47 d

Column 2B, t = 42 d

Column 2A, t = 73 d

Column 3B, t = 26 d

Column 4B, t = 42 d

Column 3A, t = 26 d

Column 4A, t = 42 d

Figure 2. Column tests. Columns are fed from the bottom. Slag media (black) and
phosphorus/metals precipitates (white) are visible.
The type of tested solution (low or high concentration) influenced the pH rise and treatment
performance. In the concentrated solution, the pH rise was slower and only three slags (#3, 7,
and 10) reached a pH of 11 or above after 48 hours (Figure 3D), yielding the best phosphorus
removal efficiency (Figure 3E). This observation indicated that the influent solution had a
buffering effect on the release of hydroxide from slag. Thus, slag dissolution kinetics are affected
by the wastewater composition.
Slag #1 was the most efficient for fluoride removal in low concentrations tests (Figure 3C).
However, its fluoride treatment capacity was comparable to other slags capacity in high
concentrations tests (Figure 3F). Slag #1 reached a pH of 8.5 after 48 hours in the high
concentrations test instead of 11.5 in the low concentrations test. The high pH value must have

favoured the precipitation of fluoroapatite (Ca5(PO4)3F), possibly explaining the poorer fluoride
removal in the high concentration test. It appeared that slag #1 has a special affinity that
catalyses the formation of fluoroapatite instead of hydroxyapatite.
Apatite was the least efficient media for phosphorus removal. As there was no pH rise with
apatite, the phosphorus removal was mainly due to adsorption at near neutral pH conditions
(Harouiya et al., 2011). However, the fluoride removal capacity of apatite was high, probably
due to ionic exchange between apatite hydroxides and aqueous fluorides.
Low concentration tests

High concentration tests

Figure 3. Monitoring of pH, ortho-phosphate concentration and fluoride concentration in
individual batch tests for low (A, B, C) and high concentration tests (D, E, F).

Low concentration tests

High concentration tests


Figure 4. Monitoring of manganese, zinc and aluminium concentrations in individual batch

tests for low (A, B) and high concentration tests (D, E, F).
Sequential Batch Tests
The sequences to test were based on the results of individual batch tests. Three slags had a strong
capacity to increase the pH and to remove phosphorus and metals (slags #3, 7 and 10), and one
slag had a strong capacity to remove fluoride (slag #1). Considering that the fluoride treatment
capacity of slag #1 was reduced in the concentrated solution, the main objective of sequential
batch tests was to improve the fluoride removal capacity of slag #1 by pretreating it with a slag
showing a high pH rise capacity. The tested sequences were #3-#1, #7-#1 and #10-#1. Sequential
batch tests were conducted with the high concentration solution only. Results of sequential batch
tests are presented and compared with individual batch tests in Figure 5.

Figure 5. Monitoring of pH, phosphorus, fluoride and metals concentration in sequential
batch tests (high concentration tests only).
The three tested sequences resulted in a higher removal of phosphorus, fluoride and manganese
then the single-slag test using only slag #1 (Figure 5C). This result indicated that it was possible
to improve the fluoride removal of slag #1 by increasing the pH with an upstream slag. However,
slag #1 leached some aluminum (Figure 5F).
Slag #10 was the most effective slag to increase the pH, but it was discarded because of its
mechanical instability. Slag #10 was friable and unsuitable for use as a filter. The retained
sequence for the best phosphorus and fluoride removal was #7-#1.

Column Tests
The experimental setup shown in Figure 1 was chosen according to the results of sequential
batch tests. Columns 1A to 3B were used to assess the performance of the chosen sequence (#7#1) compared with single-slag series. Columns 4A and 4B were used to assess the influence on
longevity and global removal performance of the filter if the inflow displayed a low
concentration in contaminants. Results are presented in Figures 6 and 7 and the global
performance is presented in Table 3.
Table 3. Global treatment performance of column tests for P, F, Mn, Zn and Al.
Column outlet

of operation (d)

Global removal
F Mn Zn


Global specific removal

(mg contaminant/g media)



99.9 86.5 97.0 98.8



99.9 90.3 97.6 99.4 35.6* 1.49 0.50 0.009 0.042 0.038*

2.87 0.91 0.018 0.081




94.3 81.1 88.7 92.8 82.9 4.83 1.55 0.031 0.144 0.327



99.9 87.8 98.7 99.8 86.1 2.62 0.92 0.017 0.079 0.174



89.1 93.2 89.3 97.4 57.4* 2.85 1.10 0.019 0.094 0.141*



99.9 98.6 99.3 99.0


1.38 0.50 0.010 0.043




98.1 28.5 90.5 94.6


0.26 0.06 0.004 0.005


99.9 44.3 98.0 95.8
* aluminum leaching observed in this column.


0.12 0.04 0.002 0.002


Phosphorus removal. Phosphorus removal was directly related to pH, as shown in Figures 6A
and 6B. A pH higher than 9 reduced the phosphorus concentration to less than 1 mg P/L. When
the effluent pH was higher than 11, the effluent phosphorus concentration was between 0.01 and
1.0 mg P/L while an effluent pH between 6 and 9 resulted in a partial phosphorus removal. All
the tested sequences resulted in phosphorus removal efficiencies of 99.9 % (columns 1B, 2B, 3B
and 4B). However, the removal efficiency of columns 2A (slag #7) and 3A (slag #1) decreased
after 70 and 30 days of operation, respectively. Slag #7 could maintain a pH higher than 10 for a
longer period than slag #1, whose efficiency was rapidly diminished by the buffering capacity of
the wastewater. This behavior had also been observed in sequential batch tests.
Metals removal. For all filters, the removal capacity of manganese and zinc was related with a
high pH, as was the removal capacity of phosphorus. All the tested sequences had a removal
efficiency of manganese and zinc higher than 95%. Again, there was a reduction in removal
efficiency after 70 and 30 days for columns 2A (slag #7) and 3A (slag #1). Aluminum removal
was not possible at the beginning of operation, as filters even leached some aluminum (Figure
7C). However, leaching of aluminum (attributed to the dissolution of aluminum oxides)
contributed to the rise in pH and to the global removal capacity.

Figure 6. Monitoring of pH, phosphorus and fluoride concentration in column tests.

Figure 7. Monitoring of manganese, zinc and aluminium concentrations in column test.

Fluoride removal. Partial fluoride removal was observed for the sequence slag #7 slag #7
(column 2B, outlet concentrations between 3 and 6 mg/L) during its first 70 days of operation.

The sequence slag #1 slag #1 had a higher fluoride removal efficiency (outlet concentrations
less than 1 mg/L) during its first 30 days of operation. This result was consistent with results
from individual and sequential batch tests, in which high fluoride removal was observed with
slag #1. However, slag #1 was efficient for fluoride removal only if it was used in the upstream
column (column 3A). If used in the downstream column (columns 1B and 4B), the outlet
fluoride concentration was higher than with the sequence slag #1 slag #1. There appeared to be
a relationship between fluoride removal and effluent pH, but some treatment was still achieved at
pH values under 8.
The proposed mechanism for fluoride removal was the formation of fluoride-phosphate
precipitates (such as fluoroapatite) either by direct precipitation or by ionic exchanges between
OH of hydroxyapatite and F in solution. Direct precipitation of fluoroapatite is in competition
with the formation of hydroxyapatite. It is believed that slag #1 catalyzed the formation of
fluoroapatite instead of hydroxyapatite. Ionic exchange of OH and F could explain the treatment
capacity of the filter at pH values below 8, when the precipitation of apatite is limited. Fluoridephosphate precipitates cannot be formed without the presence of phosphorus, explaining why
slag #1 is efficient for fluoride removal only when used in an upstream treatment. The sequence
slag #7 and slag #1 resulted in a high phosphorus removal efficiency (due to precipitation as
hydroxyapatite), but phosphorus became limiting in slag #1 filter preventing the formation of
fluoride-phosphates precipitates. Sequential batch tests indicated that fluoride removal could be
improved by using the sequence slag #7 slag #1 because there was still some phosphorus left in
solution to react with slag #1.
The influence of phosphorus on fluoride removal was illustrated by comparing the efficiency of
the sequence slag #7 slag #1 fed with either a low or a high concentration solution (columns
1A and 1B vs columns 4A and 4B). After 40 days of operation, the global fluoride removal
efficiency were 92 % and 44 % for the high and low concentration effluents, respectively. The
fluoride concentration at the outlet of column 1B was lower than at the outlet of column 4B. The
F/P molar ratios of the high and low concentration solutions were 1 F/1.7 P and 1 F/0.7 P. The
ratio F/P has a direct effect on fluoride removal efficiency considering that if phosphorus is in
deficit, the fluoride removal is poorer. Columns fed with the high concentration solution
achieved a high fluoride removal even if their molar F/P ratio was lower than the theoretical
fluoroapatite ratio (1F/3P) suggesting that fluoride may be adsorbed or precipitated in other
forms than fluoroapatite.
Recommendations for combined treatment. The best global removal efficiency was obtained by
the sequence slag #1 slag #1. This sequence achieved global removal of 99.9 %, 98.6 %,
99.3 % and 99.0 % for P, F, Mn and Zn, respectively, after 58 days of operation. However, a
decrease in pH and removal efficiency was observed in the first column after 30 days, limiting its
use as a long-term treatment system. The sequence slag #7 slag #7 resulted in a lower fluoride
removal efficiency, but in a greater longevity as the pH decrease in the first column was
observed after 70 days of operation. The global removal of sequence slag #7 slag #7 after 95
days of operation was 99.9 %, 87.8 %, 98.7 % and 99.8 % for P, F, Mn and Zn, respectively. The
longevity of the system can be increased by increasing the hydraulic retention time (16 h for each
filter in this study). The sequence slag #7 slag #1 did not improve the fluoride removal capacity
of sequence slag #7 slag #7 because of a phosphorus deficit in the feed to the downstream slag

#1 filter. A conceivable way to improve the fluoride removal capacity of slag #7 would be to test
mixes of slag #1 and slag #7. A low content of slag #1 would possibly catalyze the precipitation
of fluoroapatite without affecting the slag capacity for pH rise.
The capacity of a slag filter for combined treatment of phosphorus, fluoride and metals was
assessed. Ten slags and one apatite were tested in comparative individual batch tests: EAF slag
from Forth Smith was the most efficient for pH rise and global treatment efficiency, and EAF
slag from Blytheville was the most efficient for fluoride removal. Efficient combined removal of
phosphorus, manganese and zinc (>95 %) was observed in column tests for durations of 58, 62
and 95 days for the three tested sequences (Forth Smith Blytheville, Forth Smith Forth Smith,
Blytheville Blytheville, respectively). Removal of all pollutants was related to pH rise in the
filters. Fluoride removal was best for the sequence Blytheville Blytheville (98.6 % in 58 days)
than for the sequences Forth Smith Forth Smith 87.8 % in 95 days) and Forth Smith
Blytheville (90.3 % in 62 days). Fluoride removal in slag filters was attributed to two reactions:
direct precipitation of fluoride-phosphates precipitates (such as fluoroapatite) and ionic
exchanges between OH of hydroxyapatite and fluoride from solution. The F/P ratio in the
wastewater had an influence on the amount of precipitated fluoride and on fluoride removal
Technical support of Denis Bouchard, Manon Leduc, Mathieu Blisle and Marie Ferland from
cole Polytechnique is gratefully acknowledged. Scientific assistance was provided by Shaw
Environmental. This project was supported by the Missouri Remediation Trust, the Natural
Sciences and Engineering Research Council of Canada and the Fonds qubcois de la recherche
sur la nature et les technologies.
Aldaco, R.; Irabien, A.; Luis, P. (2005) Fluidized bed reactor for fluoride removal. Proceedings
of the Proceedings of the XVI International Conference on Chemical Reactors:
Chazarenc, F.; Brisson, J.; Comeau, Y. (2007) Slag columns for upgrading phosphorus removal
from constructed wetland effluents. Proceedings of the Wetland Systems for Water
Pollution Control X; 12 Caxton Street, London, SW1H 0QS, United Kingdom.
Chazarenc, F.; Kacem, M.; Gerente, C.; Andres, Y. (2008) 'Active' filters: a mini-review on the
use of industrial by-products for upgrading phosphorus removal from treatment wetlands.
Proceedings of the 11th Int. Conf. on Wetland Systems for Water Pollution Control;
Indore, India.

Chen, N.; Zhang, Z.; Feng, C.; Li, M.; Chen, R.; Sugiura, N. (2010) Removal of fluoride from
aqueous solution by adsorption onto Kanuma mud. Water Science and Technology, 62
(8), 1888-1897.
Drizo, A.; Comeau, Y.; Forget, C.; Chapuis, R. P. (2002) Phosphorus saturation potential: A
parameter for estimating the longevity of constructed wetland systems. Environmental
Science and Technology, 36 (21), 4642-4648.
Drizo, A.; Forget, C.; Chapuis, R. P.; Comeau, Y. (2006) Phosphorus removal by electric arc
furnace steel slag and serpentinite. Water Research, 40 (8), 1547-1554.
Grzmil, B.; Wronkowski, J. (2006) Removal of phosphates and fluorides from industrial
wastewater. Desalination, 189 (1-3 SPEC. ISS.), 261-268.
Harouiya, N.; Rue, S. M.; Prost-Boucle, S.; Lienar, A.; Esser, D.; Molle, P. (2011) Phosphorus
removal by apatite in horizontal flow constructed wetlands for small communities: pilot
and full-scale evidence. [Article]. Water Science and Technology, 63 (8), 1629-1637.
Higgins, J. P.; Hurd, S.; Weil, C. (2000) The use of engineered wetlands to treat recalcitrant
wastewaters. Journal of Environmental Science and Health - Part A Toxic/Hazardous
Substances and Environmental Engineering, 35 (8), 1309-1334.
Huang, Y.-H.; Shih, Y.-J.; Chang, C.-C. (2011) Adsorption of fluoride by waste iron oxide: The
effects of solution pH, major coexisting anions, and adsorbent calcination temperature.
Journal of Hazardous Materials, 186 (2-3), 1355-1359.
Lu, H.; Wang, B.; Ban, Q. (2010) Defluoridation of drinking water by zeolite NaP1 synthesized
from coal fly ash. Energy Sources, Part A: Recovery, Utilization and Environmental
Effects, 32 (16), 1509-1516.
Lundager Madsen, H. E. (2008) Influence of foreign metal ions on crystal growth and
morphology of brushite (CaHPO4, 2H2O) and its transformation to octacalcium
phosphate and apatite. Journal of Crystal Growth, 310 (10), 2602-2612.
Pratt, C.; Shilton, A. (2010) Active slag filters-simple and sustainable phosphorus removal from
wastewater using steel industry byproduct. Water Science and Technology, 62 (8), 17131718.
Vohla, C.; Koiv, M.; Bavor, H. J.; Chazarenc, F.; Mander, U. (2011) Filter materials for
phosphorus removal from wastewater in treatment wetlands-A review. Ecological
Engineering, 37 (1), 70-89.
Warmadewanthi; Liu, J. C. (2009) Selective precipitation of phosphate from semiconductor
wastewater. Journal of Environmental Engineering, 135 (10), 1063-1070.
Xu, X.; Li, Q.; Cui, H.; Pang, J.; Sun, L.; An, H., et al. (2011) Adsorption of fluoride from
aqueous solution on magnesia-loaded fly ash cenospheres. Desalination, 272 (1-3), 233239.
Xue, J.; Ma, Y. (2009) Removal of high fluoride from drinking water treated by new type fly ash
filter material. Proceedings of the 3rd International Conference on Bioinformatics and
Biomedical Engineering, iCBBE 2009, June 11, 2009 - June 13, 2009; Beijing, China.
Yang, M.; Zhang, Y.; Shao, B.; Qi, R.; Myoga, H. (2001) Precipitative removal of fluoride from
electronics wastewater. Journal of Environmental Engineering, 127 (10), 902-907.
Zhou, Y.-F.; Haynes, R. J. (2010) Sorption of heavy metals by inorganic and organic
components of solid wastes: Significance to use of wastes as low-cost adsorbents and
immobilizing agents. Critical Reviews in Environmental Science and Technology, 40
(11), 909-977.