You are on page 1of 15


discussions, stats, and author profiles for this publication at:

Effects of organic matter on crystallization of

struvite in biologically treated swine wastewater
Impact Factor: 1.56 DOI: 10.1080/09593330.2015.1088580



Aurlie Capdevielle

Fabrice Bline

Laboratory of Chemical Engineering - LGC

National Research Institute of Science and Tec





marie-line Daumer
National Research Institute of Science and Tec

All in-text references underlined in blue are linked to publications on ResearchGate,

letting you access and read them immediately.

Available from: Aurlie Capdevielle

Retrieved on: 02 March 2016

Environmental Technology

ISSN: 0959-3330 (Print) 1479-487X (Online) Journal homepage:

Effects of organic matter on crystallization of

struvite in biologically treated swine wastewater
Aurlie Capdevielle, Eva Skorov, Fabrice Bline & Marie-Line Daumer
To cite this article: Aurlie Capdevielle, Eva Skorov, Fabrice Bline & Marie-Line Daumer
(2015): Effects of organic matter on crystallization of struvite in biologically treated swine
wastewater, Environmental Technology, DOI: 10.1080/09593330.2015.1088580
To link to this article:

Published online: 23 Oct 2015.

Submit your article to this journal

Article views: 16

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at
Download by: [Aurlie Capdevielle]

Date: 17 December 2015, At: 17:58


Effects of organic matter on crystallization of struvite in biologically treated swine

Aurlie Capdeviellea,c, Eva Skorovb, Fabrice Blinea,c and Marie-Line Daumera,c
IRSTEA/Cemagref, 17 avenue de Cucill, CS 64427, 35044 Rennes Cedex, France; bICT Prague, Department Technology of water, Technick 5,
Prague 6, 166 28, Czech Republic; cUniversit europenne de Bretagne (UEB), 5 Boulevard Lannec, 35000 Rennes, France

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015



A sustainable way to recover phosphorus (P) in swine wastewater involves a preliminary step of P
dissolution followed by the separation of particulate organic matter (OM). The next two steps are
rstly the precipitation of struvite crystals done by adding a crystallization reagent (magnesia)
and secondly the ltration of the crystals. To develop the process successfully at an industrial
scale, the control of the mechanisms of precipitation is the key point in order to obtain high
value-added products, that is, big struvite crystals easy to harvest and handle. Experiments with
process parameters optimized previously in a synthetic swine wastewater were performed on
real swine wastewater to assess the role of the OM on struvite crystallization. After 24 h, with a
pH increase to 6.8 only, 90% of the initial P was precipitated and 60% was precipitated as
struvite. 80% of the solid recovered was in the fraction > 100 m. The other forms recovered
were brushite, amorphous calcium phosphate, NaCl, KCl and OM. The inuence of OM on
struvite precipitation in acidied swine wastewater was negative on the reaction kinetics but
positive on the size of the struvite crystals. The presence of colloidal particles increased the size
of the struvite crystals but slowed down the kinetics due to the viscosity induced by the
repulsive force of the colloids. The maximum size of single struvite crystals (200 m) was
observed with the presence of particulate OM.

Received 9 June 2015

Revised 17 August 2015
Accepted 24 August 2015

The issues in the recycling of phosphorus (P) are double:
(i) the protection of a limited resource and (ii) the ght
against eutrophication by adapting the application of P
to the crop requirements which would minimize the
leaks to the aquatic environment. The main sources for
P recycling are animal wastes and wastewater sludge.
Struvite (MgNH4PO46H2O), a crystalline form obtained
from dissolved P of liquid efuents, is marketed as a fertilizer in Japan or in Canada. It is usually obtained from
the liquid phase of dewatering anaerobic digested
wastewater sludge by crystallization but it can be
obtained from livestock waste.[1] However, most of the
systems developed at an industrial scale recycle only
the dissolved P from urban or agro-industrial efuents
and are seldom applied to concentrated efuents such
as swine wastewaters. One of the main limitations for
the development of such a process to recycle P from
swine wastewater is that P is mainly in particulate form
(80%) in the wastewater and mixed with a high level of
biodegradable organic matter (OM).[2]
So, as described by Daumer et al.,[3] a sustainable way
to recover P from swine wastewaters involves a
CONTACT Marie-Line Daumer

2015 IRSTEA. Published by Taylor & Francis.


Struvite; phosphorus
recovery; pig manure; swine
wastewater; organic matter

preliminary step of P dissolution by acidication followed

by a separation of residual particulate OM. The next two
steps of the recovery process are the precipitation of
struvite crystals (MgNH4PO46H2O) by adding a crystallization reagent (MgO) and the ltration of struvite crystals
in lter bags with a 100 m cut-off.
To successfully develop the process at an industrial
scale, the control of the mechanisms of precipitation is
the key point in order to obtain high value-added products, that is, big struvite crystals easy to harvest and
handle and to be able to predict mineral composition.
In this framework, our previous works were performed
in a synthetic solution comparable to swine wastewaters
as regards ionic strength, to dene the optimal parameters
to precipitate P as struvite [4] and to understand the mechanisms and kinetics of the precipitations.[5] The results
showed that the addition of MgO was directly followed
by the precipitation of amorphous calcium phosphate
(ACP) and struvite. Then, the struvite crystals grew was
limited by the dissolution of ACP. The addition of MgO
and the N:P ratio were the leading parameters that inuencedmost the precipitation of P as large crystals of struvite. A
high temperature increased the total volatilization of

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015


ammonia and so increased the dissolution of the struvite

crystals formed to restore the balance in ammonium.
However volatilization can be neglected at a temperature
usually found in an intensive livestock production area in
western and northern Europe.
Moreover, in real swine wastewaters, the precipitation
of struvite is not only inuenced by the process parameters but also by the residual OM. In urology, the
OM inhibits or promotes the struvite crystallization function of the length and the charge of the carbon chain.[6
9] Although the inuence of OM is noticed on struvite
precipitation in swine wastewaters,[1013] it has never
been, to our knowledge, quantied in processes
applied on farms.
The purpose of this work was to evaluate the inuence of OM of the biologically treated swine wastewater
(BTSW) on the precipitation kinetics and on the quality
and quantity of the solid fractions produced. For that,
the OM was divided into three fractions depending on
size and density: (1) the dissolved fraction, (2) the colloidal fraction and (3) the particulate fraction and the
inuence of each of these fractions was measured on
struvite precipitation during this study.

Materials and methods

Figure 1. OM fractionation.

Swine wastewater pre-treatment

Swine wastewater (BTSW) was collected under mixing
after the biological treatment (N removal by nitrication/denitrication) in a commercial farm in Brittany.
The BTSW was acidied with 8 g kg1 of formic acid
(80% from Carlos Erba) to reach a pH value of 4.5 to dissolve the mineral particulate P which is the main form of
P in swine wastewater. After 24 h, a cationic polymer occulant with a high molecular weight (Z8140 from CIBA)
was added (0.1% w/w) into the acidied BTSW as a solution at 2 g L1 to occulate OM. Afterwards, the ocks
were removed using a 250 m-mesh sieve. The liquid
fraction collected after the sieving is called total acidied
swine wastewater (TASW) and contained residual OM
(ne particulate, colloidal and dissolved). The particulate
fraction of the TASW was removed by centrifugation at
10,000 rpm at 4C for 20 min. The liquid from this separation step is called centrifuged ASW (CASW) and is supposed to contain colloidal and dissolved organic matter
(DOM). The largest colloidal fraction of the CASW was
removed by ltering at 0.45 m on Sartorius lters
(regenerated cellulose). The liquid obtained from this
step is called ltered ASW (FASW) and was supposed
to contain only DOM (Figure 1).
The precipitation study was carried out on three fractions: TASW, CASW and FASW.

In the text, ASW is used as a generic term to describe

the efuents obtained from BTSW (TASW, CASW and

Synthetic swine wastewater preparation

The synthetic swine wastewater (SSW) was prepared by
mixing different salts in distilled water to obtain the
ionic concentrations measured in the acidied swine
wastewater (ASW). The salts used for synthetic efuent
preparation were: KOH, K2SO4, NaNO2, KNO3, CaCl22H2O
(Carlo Erba), MgCl26H2O, and Na3PO412H2O (Merck).
Formic acid (80%, Carlos Erba) (7 g kg1) was added in
the solution to reach a pH value of 4.5. Five minutes
before the beginning of each run, NH4Cl powder (99%,
Carlo Erba) was added under vigorous stirring in order
to reach an N:P molar ratio of 3 which was the optimal
value determined by the previous work.[4]

Magnesia addition
The MgO suspension was prepared 1 h before the beginning of the run by mixing 10 g of MgO powder (97%,
VWR Prolabo) in 100 mL of distilled water. Magnesia
was kept in suspension under continuous stirring and


was thermostated at the same temperature as the runs.

The runs began when adding 11 ml of the suspension
of MgO to reach a total Mg:Ca molar ratio close to 2
according to Capdevielle et al.[4] But the dissolution of
magnesia is slow. The Mg:Ca ratio at the beginning of
the run was calculated by the sum of the dissolved magnesium in the suspension and in the samples. It was close
to 0.9 for all the trials.

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015

Precipitation runs were performed in an FC6s occulator
from Velp Scientica. Each beaker was lled with 800 mL
of efuent. The addition of MgO was carried out under
continuous stirring at 90 rpm for 30 s. Then, the stirring
rate was maintained at 60 rpm for 24 h. Previous works
have shown that the optimal temperature for struvite
crystal growth was 5C.[4] However a low temperature
is difcult to apply in an on-farm processing plant. Therefore, reactions were carried out in beakers thermostated
at 15C, which is the annual average temperature in Brittany, by water circulating through a constant temperature bath. pH and temperature were recorded by an 8channel data logger pH/mV/Ion/O2 from Fisher Scientic.
The experiments were performed in duplicate

Samples (4 ml) were collected from beakers at 15, 30, 45
min then at 1, 2, 3, 4, 6, 8, 22, and 24 h for dissolved com2+

pound analysis (PO3
4 , NH4 , Ca , Mg , K , Cl Na ).
After 24 h, samples were sieved on 25 and 100 m
lters. The three solid fractions ( < 25 m, 25100 m,
and > 100 m) were dried at room temperature. Each
fraction was then weighed and analysed as described

Organic charge
The total chemical oxygen demand (TCOD) was analysed
using the French standard NFT 90-101. TCOD analyses
were conducted on the ASW (total, centrifuged and ltered) before and after the precipitation step. The total
solids (TSs), the total suspended solids (TSSs) and the
volatile suspended solids (VSSs) were measured using
the american standard methods.[14]
Solid and liquid characterization
The compositions of the collected solid fractions were
analysed by ionic chromatography after dissolution by
formic acid and the structure by X-ray powder diffraction
(XRD). The rates of the precipitation/dissolution of the

solid fractions were evaluated by monitoring the dis2+

solved concentrations of Ca2+, PO3
4 , Mg , and NH4 .
The concentrations of dissolved PO4-P, NH4-N were
analysed by automate colorimetric methods on QuikChem FIA+ from Lachat Instruments with the QuikChem
method 10-115-01-1-P and 10-107-06-1-J, respectively.
The dissolved Mg2+ and Ca2+ were determined by cationic chromatography.
The crystal structures were investigated by powder
XRD using a Bruker AXS D8 Advance (2 BraggBrentano geometry) diffractometer working with monochromatized Cu-K1 radiation ( = 1.5406 ) and equipped
with a LynxEye detector. The experiments were performed within the 580 2 data range with 2 steps of
0.02 and counting time of 200 ms per step, each step
being measured 178 times. The diffraction data were
analysed using the FULLPROF program.[15]
The molar balance was checked by analysing the solid
and liquid compositions at the beginning and the end of
the precipitation.

NH3 volatilization
Two beakers were reserved for the analysis of the NH3
volatilization in each experiment: one for the SSW and
one for the ASW. The volatilization of ammonia was evaluated from the difference between the concentration of
ammonium at 0 h and its concentration at 24 h after dissolution of the solid fractions with formic acid (80%) until
pH reached 4.5. The NH+4 /NH3 content was analysed by
distillation (method modied from NFT 90-015-1, Bchi
distillation unit B-324).
Particle analysis
The particle analysis was carried out with a morphogranulometer (Occhio, Flowcell FC200M) using an 800 m
thick measuring cell. The liquid is circulated continuously
in the measuring glass cell. For each measurement, 20
pictures were taken through the glass cell and analysed
in order to identify the number and the morphometric
criteria of each particle. The equivalent volume of each
picture is 4.04 L.
When the samples were clear enough, the particle
analysis was carried out directly in the beaker. This
method is called direct.
If the samples were heavily loaded with small particles, the method dilution was applied. The samples
were diluted using the liquid phase from centrifuged
samples (47,900 G, 20 min at the same temperature of
runs). The solid/liquid equilibrium was then maintained
avoiding the solid dissolution or crystallization due to
For the morphometric analysis, Callisto 5.38 software
from Occhio was used. The struvite crystals were isolated


by applying a lter based on three shape factors: luminance, ISO circularity and ISO bre thickness. The luminance was in the range [00.45], the circularity in the
range [00.91] and the ISO bre thickness in the range


Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015

Initial conditions
The aim of the rst acidication/separation step was to
separate the P and the OM by dissolving mineral P
which is the main form in swine wastewater. In this
trial, dissolved P was increased from 312 to 928 mgP
L1. Ammonia was increased from 0 to 1233 mg L1. A
part of ammonia can be adsorbed on the biomass or precipitated as struvite.[16] Dissolution or desorption by the
acidication was responsible for this increase. About 85%
of the VS were removed and the VS/TS ratio in TASW was
reduced to 23% compared to 58% in treated slurry. The
sum of the ionic concentrations measured in this study
explained more than 85% of the dissolved mineral
matter composition (Table 1). At the end of the step,
OM estimated from chemical oxygen demand (COD)
was mainly dissolved (76%) and in particles (20%). Only
4% of the OM was in the colloid form as dened in this
paper. The COD due to formic acid addition was estimated by the concentration measured in synthetic efuent. However, a part of this oxidizing potential could
have been consumed by OM oxidation during acidication of the real efuent.

1 h, pH was between 6.3 and 6.4 for all the samples.

Afterwards, when the pH of the ASW was increasing
slowly, the pH of the SSW reached a plate at pH 6.6 in
less than 2 h and then increased again to reach the
nal plate close to 7 twelve hours later. Inexions were
also observed in ASW; however they were not as well
dened as in SSW. The nal plate was reached after 12
h in CASW and FASW at pH 6.65 and 6.42, respectively.
In TASW the pH increased slowly to pH 6.8 during the
entire test and the nal plate was not clearly reached
at the end of the experiments (24 h) (Figure 2). The pH
inexions can be explained by the equilibrium between
the protons released from the precipitation of struvite,
brushite and ACP [4,5,17,18] and the dissolution of
MgO into Mg2+ and OH. The kinetics of ionic species
presented below provide more information about the
mechanisms involved.

Kinetics of ionic species evolution

The evolution of ionic concentrations with time is
described in Figure 2 with a focus on the rst hours
on the right. Usually, the precipitation of phosphates,
ammonium, calcium and magnesium with MgO followed a rst-order kinetic.[5,19] In those experiments,
there were two steps in the precipitation of phosphate, before and after an inexion point as shown
in Figure 3. They both followed a rst-order kinetic
reaction which can be described by the following
d[PO4 P]t
= KP ([PO4 P]t [PO4 P]e ), (1)


Evolution of pH
In all the efuents, the pH increased from 4.5 to 6.1 in less
than 5 min. Then, the increase slowed down abruptly for
45 min only and speeded up again in SSW when it was
slowing down more progressively in all the ASW. After

[PO4 P]t = ([PO4 P]e [PO4 P]0 ) (1 eKP t )

+ [PO4 P]0 ,


yPO4 P = ([PO4 P]0 [PO4 P]e ) KP .


Table 1. Composition of the efuents.

TS (g L1)
VS (g L1)
COD(gO2 L1)
Cl(mg L1)

2 -N (mg L )

3 -N (mg L )

SO4 -S (mg L )
Na+(mg L1)
NH+4 -N(mg L1)
K+(mg L1)
Mg2+(mg L1)
Ca2+(mg L1)






40 3
23 2
27 3
2297 29
1.6 0.6
23.7 0.3
312 6
81 5
1112 18
1899 26
161 12
80 20

15 2
3.5 0.5
5.0 0.5
2108 28
14.1 0.5
0.9 0.8
928 19
69 1
1200 100
1233 20
2200 100
705 4
1289 25

4.0 0.3
5300 200
11.0 0.1
3.0 0.1
907 18
68 1
1100 100
1360 36
2100 100
718 10
1318 5

3.8 0.1
5700 100
12.0 0.1
4.0 0.1
1032 16
70 1
1300 200
1365 28
2200 100
726 5
1377 3

2.42 0.1
3864 59
26.8 4.2
6.7 0.8
1155 15
72 3
2301 26
1368 41
1764 27
563 7
1176 47


Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015

Figure 2. pH variation during the entire test with a focus on the rst hours.

The concentration of phosphates precipitated as struvite during the reaction (PO4-PStruvite mmol L1) was calculated according to Equation (4). The ammonia
volatilization was below the quantication limits of the
methods ( < 5% of the NH4-N measured by distiller
Bchi). In those experiments struvite precipitation followed a rst-order kinetics after the latency phase
where [NH4-N] did not vary signicantly.

of magnesium followed a rst-order reaction kinetic

(11) as it was related to the struvite precipitation.

PO4 PStruvite = [NH4 N]0 [NH4 N]t

[PO4-P]t, [NH4-N]t, [Ca]t and [Mg]t are the concentrations of the removed phosphates, ammonium,
calcium and magnesium during the reaction (mmol
L1). [PO4-P]0/[NH4-N]0/[Ca]0 and [PO4-P]e/[NH4-N]e/[Ca]e
are the removed concentrations at the beginning of
the reaction and at equilibrium, respectively, (mmol
L1). [Ca]m and [Mg]m (mmol L1) are the maximum concentration of calcium removed and magnesium dissolved, respectively, reached during the run. KP, KN, KCa
and KMg (h1) are the kinetic constant of precipitation;
1/K is the time when the rst reaction is slowing down.
The precipitation rates of phosphates (yPO4 P mmol
L1 h1, (3)) of struvite (yPO4 PStruvite mmol L1 h1, (6)), of
calcium (yCa mmol L1 h1, (8)) and of magnesium (yMg
mmol L1 h1, (12)) are the asymptotes of the curves at
the beginning of the precipitations.
For all the regressions, the r was 0.99. Table 2 shows
the coefcient calculated by regression of the different
steps of the reaction.
The rst step was the precipitation of calcium phosphate and, in a less part, of struvite. For SSW and
FASW, the PO4-P precipitation rate in the rst minutes
was the sum of the calcium and struvite precipitation
rate. For TASW and CASW, the PO4-P precipitation rate
was higher than this sum. Phosphate might have precipitated with magnesium.
The second step was mainly due to the precipitation
of ammonium (or struvite) which started less than 5
min after the MgO was added for SSW and 1550 min
after the addition of MgO for the ASW. The precipitation
rate of struvite was low in TASW compared to in the
other efuents. We have previously observed that a

[NH3 N]volatilized ,
[NH4 N]t =([NH4 N]e [NH4 N]0 )
(1 eKN t ) + [NH4 N]0 ,
yPO4 PStruvite = ([NH4 N]0 [NH4 N]e ) KN .


The evolution of calcium can also be divided in two

steps; the rst one is an apparent precipitation which followed a rst-order reaction kinetic (7).
[Ca2+ ]t = ([Ca2+ ]m [Ca2+ ]0 ) (1 eKCa t )
+ [Ca2+ ]0 ,
yCa = ([Ca2+ ]m [Ca2+ ]0 ) KCa .


The second one is an apparent dissolution of calcium

which also followed a rst-order reaction kinetic (9) as
Posner and Betts said.[20]
[Ca2+ ]t = ([Ca2+ ]e [Ca2+ ]m ) (1 eKdCa t )
+ [Ca2+ ]m ,
ydCa = ([Ca2+ ]e [Ca2+ ]m ) KdCa .


KdCa (h1) is the kinetic constant of calcium precipitate

dissolution. The dissolution rate of calcium (ydCa mmol
L1 h1) was described by Equation (10), the asymptote
at the beginning of the dissolution.
Magnesium evolution can also be described in two
steps with an apparent dissolution step followed by an
apparent precipitation step. The apparent precipitation

[Mg2+ ]t = ([Mg2+ ]e [Mg2+ ]m ) (1 eKMg t )

+ [Mg2+ ]m ,
yMg = ([Mg2+ ]e [Mg2+ ]m ) KMg .



Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015

Figure 3. Evolution of the ionic concentration during the entire test on the left with a focus on the rst ve fours on the right.

signicant amount of NH+4 can be trapped by the OM in

BTSW.[16] This effect could explain the difference.
The third step started with the apparent precipitation
of magnesium which was correlated to the inexion in
phosphate precipitation. The apparent magnesium precipitation started 5 min after the beginning of the

reaction in SSW and between 40 and 60 min in ASW.

The apparent precipitation rate of magnesium was
higher in FASW and CASW than in TASW or SSW.
The last step was the apparent dissolution of the
calcium precipitates which started after 6070 min in
SSW and FASW and after 120130 min in TASW and


Table 2. Kinetics parameters of the runs.

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015

PO4-P prec.* rate (mmol L1 h1)
1/KP (h)
PO4-P removed when struvite prec. started (mmol L1)
Time of the inexion (min)
Total PO4-P removed (mmol L1)
Time of struvite prec. (min)
Struvite prec. rate (mmol L1 h1)
1/KN (h)
Total NH4-N removed (mmol L1)
Ca prec. rate (mmol L1 h1)
1/KCa (h)
Max.* Ca removed (mmol L1)
Time of Ca diss.* (min)
Ca diss. rate (mmol L1 h1)
1/KdCa (h)
Total Ca removed (mmol L1)
Time of Mg prec. (min)
Max. Mg dissolved (mmol L1)
Mg prec. rate (mmol L1 h1)
1/KMg (h)
Final Mg (mmol L1)





62 1.2
4.8 1.2

168.1 6.2a 9.4 0.1b

0.1a 2b
12.6 0.1
28.7 0.1
5.8 0
21.9 2.1
82.3 1.5
13.5 0.5
2.6 0.1
5.9 0.1
19.8 2.6
3.4 0.0
10.6 3.2

148.6 4.1a 16.3 0.1b

0.1a 1.3b
15 0.1
26.8 0.6
9.4 0.1
30.7 2.4
66.8 0.5
11.6 0.1
6.3 0
7.2 0.1
18.9 2.0
5.5 0.1
10.1 1.3

114.3 1.5a 58 2.2b

0.2a 0.5b
12.8 2.2
29.5 0.1
14.2 0.4
29.7 1.1
102.4 0.6
11.0 1.3
9 0.1
7.8 0.1
19.6 1.5
7.2 0.2
6.8 1.2

34.1 0.1
10.4 0.3
36.1 2.6
50.7 0.8
18.4 1.3
13.5 0.2
8.2 0.5
15.4 2.8
2.6 0.1
4.6 1.4

before the inexion.

after the inexion.
*Prec., precipitation, diss., dissolution, max., maximum.

CASW. The apparent dissolution rate of calcium

increased in the function of the concentration of the

Particle size distribution

The mean size of the particles was rst estimated by the
equivalent disc diameter (D0) which is calculated from
the following equation:

D0 = 2. A/,
where A is the area of the particle.
Before MgO addition, particles were supposed to be
OM. After centrifugation the number of particles was
decreased by 1520. The ltration decreased the
number of the particles of CASW by 510 (Figure 4(a)).
The mean diameter of the particles of TASW was
95 m (volume).The mean diameters of particles of
CASW and FASW were 13 and 8.3 m (volume), respectively (Figure 4(b)).
The evolution of the particles size was analysed
during the precipitation. Due to OM particles the
direct morphometric analysis was not possible in
TASW. The dilution method described previously was
applied. It required a large amount of samples. It was
done only at 10 and 24 h.
The evolution of the size of the entire population
showed a difference between SSW and ASW. In SSW,
the maximum was 135 m and was reached after 2.5
3 h. Afterwards, the particle size decreased from 135 to
100 m. In CASW and FASW, the maximum mean size
was 82 and 89 m, respectively, and was reached also

after 22.5 h. Afterwards, the particle size decreased to

70 m (Figure 5).
After 24 h, the precipitation produced two populations of particles well distinct in CASW and TASW
and, in a lesser extent, in FASW as shown in Figure 6(a)
and (b)
The microscopic observations at 24 h showed that the
particles were more regular and well-structured in presence of OM (Figure 7). In SSW, particles were consisting
in agglomeration of small struvite crystals. This was conrmed by the roughness calculation. The regularity of the
surface of particles can be assessed by the global roughness index calculated from the following equation:
Rgy = A A(Oy )/A,


where A is the area of the particle and A(O y) the area that can be
covered with discs larger than y. In this calculation y was 0.75.

The roughness of the three ASWs was lower and more

stable in time than in SSW (Figure 8). The agglomerated
particles of SSW seemed to be easily breakable.
The lter described in the section Particle analysis
was applied on the pictures to counting and measuring
morphometric parameters of particles which were considered as struvite crystals. Due to struvite crystals
agglomeration the lter was not selective enough to
be applied to SSW. The mean diameter of struvite crystals
was increasing with organic load (Figure 9). In all the
cases the peak was corresponding to the second peak
of the total particle distribution.
In SSW, the microscopic observations showed small
translucent crystals appearing after 1 h and increasing
in number during the reaction. No translucent crystals


Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015

Figure 4. Distribution of the particles size in number (A) and in volume (B) in the different ASWs.

reach pH 7 in previous work performed on SSW only.

[4] This value was reached in SSW but not in ASW. The
nal production of struvite (mmol L1) is calculated
from Equation (4). The percentage of PO4-P precipitated
as struvite (PO4-PStruvite (%)) is the ratio of the production
of struvite and the initial concentration in PO4-P. The production of struvite was lower in TASW than in the other
efuents. There was no signicant difference in the production of struvite between SSW, CASW and FASW
(Table 3).
Figure 5. Evolution of the particle size.

were produced in ASW. The single translucent

crystals were supposed to be brushite as described by

Final products
pH and concentrations
The amount of magnesia added for precipitation calculated to have a Mg:Ca ratio of about 2 was sufcient to

Figure 6. Distribution of the particle size after 24 h.

Composition of solid fraction

After 24 h, samples were sieved on 25 and 100 m lters.
Table 4 shows that the mass of the solid fraction was two
times higher in TASW than in the other efuents. Most of
the solid product after 24 h was in the fraction 25100
m for CASW and FASW and in the fraction > 100 m
for TASW and SSW.
The three solid fractions ( < 25 m, 25100 m and
> 100 m) were analysed by ionic chromatography
after acid dissolution and by XRD on solid samples. The
mass balance was calculated in dried solids according
to our previous work,[4] phosphates are considered as
struvite and brushite, calcium was thought to precipitate

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015


Figure 7. Picture of the particles after 24 h.

Table 3. Struvite production.


Final pH

PO4-PStruvite (% of P total)

Struvite (mmol L1)

7.00 0.66
6.81 0.20
6.65 0.15
6.42 0.08

67 6
60 0
76 7
64 5

23.8 1.9
18.0 0.4
22.2 1.7
21.4 1.4

MgO and struvite (15).

mdried solid = mMgO + mStr + mBr/ACP + mCaCO3
Figure 8. Roughness evolution during the experiments.

Figure 9. Distribution of struvite crystals size after 24 h.

as brushite in the fraction > 25 m, as ACP in the fraction

< 25 m and as CaCO3. Magnesium was considered as

+ mNa+ + mK+ + mCl + mNO2

+ mNO3 + mSO2


The difference between the calculated and the

weighed mass of dry solid can be due either to experimental error up to 20% or to the OM. Thirty-ve
percent of the TCOD was removed by the precipitation
of struvite in TASW and 25% in CASW and FASW. The
brown colour of recovered products indicated that
one part of the TCOD was retained in the solid fractions.
The difference between calculated and weighed mass
was above 20% in the fractions < 25 m in FASW and
CASW and in the 25100 m fraction in TASW. That is
in compliance with the distribution of OM which was
mainly dissolved or as small particles in FASW and
CASW and in larger ones in TASW. Struvite crystals
were mainly in the fractions 25100 and > 100 m
(Figure 10).
The XRD patterns conrmed the presence of struvite
in all the fractions except in the fractions < 25 m of
TASW and SSW. Brushite was identied in all the fractions



Table 4. Mass distribution in solid fractions.

Fraction < 25 m

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015


g L1

0.11 0.04
0.88 0.00
0.76 0.45 17 11
0.39 0.19

Fraction 25100 m
g L1

1.74 0.26
1.13 0.34
2.68 0.80
4.02 0.57

35 3
12 2
61 17
85 1

Fraction > 100 m

g L1

shown that struvite can be formed at pH 6.6 when

MgO is used as a precursor.

3.21 0.24 64 2
7.46 0.56 79 2
0.92 0.23 21 6
0.34 0.04 7 2

except in the fractions > 25 m of CASW where brushite

crystals remained in the fraction < 25 m. The XRD patterns of the fractions < 25 m showed a noisy background which indicates that those fractions contained
more amorphous phases than the other fractions. The
main crystalline phases of the fractions < 25 m were
NaCl and KCl (Figure 11). This was coherent with the dissolution analyses which show that 6.75 mmol g1 of NaCl
and KCl were crystallized into SSW and 5.34 mmol g1
into TASW. In FASW and CASW, 2 mmol g1 of NaCl
and KCl were crystallized. The crystallization of NaCl in
dried struvite samples was also observed by Li and
The characteristic peaks of struvite obtained in SSW
had a different intensity from those obtained in ASW.
In SSW, the main peak was at 33 2, in ASW, the main
peak was at 16 2. This conrms the microscopic observations; the structure of the struvite crystals in ASW was
different from in SSW (Figure 11). The optimal pH for
struvite precipitation is usually about 9 and more often
struvite crystallization is performed between 8 and
9. Due to the high buffer effect of pig slurry this pH
was under 7 in our experiments. However struvite was
clearly identied by X-ray and the presence of struvite
is consistent with chemical analysis of the solid. This is
in compliance with Stolzenburg et al. [23] who have

In these investigations, we found that OM had a strong
inuence on the kinetics of the reaction and on the
shape of the crystals formed. The dissolved fraction is
mainly composed of humic substances such as steroids
[24] and ions (all the volatile fatty acids were removed
by the biological treatment [25]). The colloidal fraction
mainly contains proteins. The particular fraction is
mainly composed of carbohydrates and lignocelluloses:
cellulose, hemicelluloses and lignin.[26]

Inuence of the DOM

The inuence of the DOM was deduced from the results
of the SSW and FASW. The TCOD of the DOM corresponded to 1.4 gO2 L1. The DOM took part in slowing
down the pH increase, speeded up the apparent precipitation rates of calcium and magnesium and affected the
shape of the solid products. Unlike in SSW, no translucent
crystals of brushite were observed in the pictures taken
from FASW. However, the XRD pattern of the solid fractions of FASW showed the characteristic peaks of brushite which indicated that brushite crystals had a different
habit. There are two possible explanations for the difference in the brushite shape. First, the functional groups of
the DOM might x on the positively charged site of the
brushite nuclei and block the faces which have a fast
growth.[27] Therefore, the brushite crystals are smaller
but more structured than in synthetic medium. Second,
the presence of DOM might increase the surface
energy of the brushite crystals which slowed down the
crystal growth.[28] The two hypotheses also explain
why the brushite crystals did not crystallize onto the struvite crystals as seen in SSW. The DOM seemed to prevent
the dissolution of MgO. Indeed, the pH increased less
with DOM. It has been already observed that the acidic
functional groups of the OM formed complexes with
the surface hydroxyl of metal-oxides such as Mg(OH)2
and prevent its dissociation into Mg2+ and OH.[29,30]

Inuence of the colloidal particles

Figure 10. Distribution of mineral components in fraction from

mass dried solid calculation.

The inuence of the colloidal particles was deduced from

the results of the three runs in ASW. The pH increased
more with the colloidal particles than with the DOM.
We also showed that all the precipitation kinetics were
slower in CASW than in FASW. The colloidal particles
increased the size of the struvite crystals in the solid fractions. The viscosity induced by the colloidal particles

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015



Figure 11. XRD patterns of the solids from the different fractions. a,f,i,l: > 25 m; b,j,g,m: 25100 m; c,h,k,n: > 100 m; d: struvite; e:
brushite; o:NaCl.

explains why all the reaction kinetics were slowed down

and why few or no agglomeration of crystals was
observed. The agglomeration of crystals is divided into
three steps: the collision between the crystals (aggregation), the association of the two particles and the formation of a crystalline bridge between them
(agglomeration). The aggregation kernel (Ki,j) is given
by Equation (16). ri,j (m) is the radius of the two particles,
(Pa s) is the dynamic viscosity of the uid, k is the Boltzmann constant (1.381.1023 j K1) and T the temperature (K).[31]

which lowered the aggregation kernel and few or no

agglomerated crystal was observed. Moreover, the colloidal particles of swine wastewaters are mainly negatively charged [26,32,33] and might have trapped the
positively charged ions, NH+4 , Mg2+ and Ca2+, by
forming complexes.[34] The supersaturation of struvite
dened by Equation (17), was then decreased which
favoured the formation of large crystals.[16,30,35]

1 1
(ri + rj ) + .
Ki,j =
ri rj

ai are the ion activities and KStruvite is the solubility

product of the struvite.


The aggregation kernel is very strong for small particles

in a low viscosity uid. This is why we observed a lot of
agglomerated crystals in SSW and FASW (Figure 5). In
CASW, the colloidal particles increased the viscosity


aMg2+ aNH4 + aPO4 3



Inuence of the particulate OM

The inuence of the particulate OM was deduced from
the results from the TASW and CASW. The particulate
OM slowed down the pH increase but the maximum

Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015



pH value reached was higher than in CASW and FASW.

All the precipitation kinetics were slowed down. The particulate OM decreased also the quality of the solid fraction (presence of OM) and the quantity of the struvite
crystals in the product. Our results conrm that a concentration of TSS higher than 1 g L1 and a high concentration in total or dissolved organic carbon disturbs the
struvite precipitation and lower the purity of the
product.[13,3640] However, the particle analysis also
shows that the struvite crystals were larger in TASW
than in the other efuents and none of them were
agglomerated. The low supersaturation induced by the
negatively charged particles of the particulate OM and
the steric hindrance explain the slow growth of the struvite crystals. Therefore, low amount of struvite crystals
were produced which explain the high pH values
recorded. Indeed, the protons released are proportional
to the quantity of struvite or brushite precipitated. It
was also possible that some particles of the polymer
added in the ASW remained in TASW. These long chain
positively charged could poison the crystals by adsorbing on their surface and prevent them from growing.
Hence, the crystals remained in nuclei form. As the
dynamic viscosity increases in the function of the concentration of the dry matter,[41] no agglomerated
crystal was observed. The effects of OM on struvite crystallization are summarized in Table 5.

The inuence of OM on struvite precipitation in acidied
swine wastewater was negative on the reaction kinetics
but positive on the size of the struvite crystals.
One explanation could be that the DOM prevented
the dissociation of Mg(OH)2 into Mg2+ and OH. The
DOM also modied the shape of the brushite crystals.
The presence of colloidal particles increased the size
of the struvite crystals but slowed down the kinetics
probably because of the viscosity induced by the repulsive force of the colloids.
Table 5. Inuence of OM on struvite crystallization compared to
synthetic media.
Final pH
MgO dissolution
Brushite crystal growth
Struvite crystal size
Fraction of Struvite in
nal product




DOM + collodal



Notes: 0, no effect; , negative effect; + , positive effect.

The particulate OM slowed down all the reaction kinetics. In addition to the effect of DOM on MgO dissolution
steric hindrance linked to the particles could explain this
result. However, the supersaturation was also lower,
which favoured the growth of large crystals of struvite.
The maximum size of single struvite crystals (200 m)
was observed with the presence of particulate OM.
After 24 h, in the acidied swine wastewater, with a
pH increase to 6.8, 90% of the initial P was precipitated
and 60% was precipitated as struvite. Eighty percent of
the solid recovered was in the fraction > 100 m. The
other forms recovered were brushite, ACP, NaCl, KCl
and OM.

Disclosure statement
No potential conict of interest was reported by the authors.

This work was supported by the French Research National
Agency [grant number ANR2009-ECOT-001401].

[1] Smit A, Bindraban P, Schrder J, et al. Phosphorus in agriculture: global resources, trends and developments.
Report to the Steering Committee Technology
Assessment of the Ministry of Agriculture. The
Netherlands, Wageningen; 2009.
[2] Daumer M-L, Bline F, Sprandio M, et al. Relevance of a
perchloric acid extraction scheme to determine mineral
and organic phosphorus in swine slurry. Bioresour
Technol. 2008;99:13191324.
[3] Daumer M.-L, Picard S, Saint-Cast P, et al. Technical and
economical assessment of formic acid to recycle phosphorus from pig slurry by a combined acidication-precipitation process. J Hazard Mater. 2010;180:361365.
[4] Capdevielle A, Skorov E, Biscans B, et al. Optimization of
struvite precipitation in synthetic biologically treated
swine wastewater - Determination of the optimal
process parameters. J Hazard Mater. 2013;244245:357
[5] Capdevielle A, Skorov E, Bline F, et al. Kinetics of
struvite precipitation in synthetic biologically treated
swine wastewater. Environ Technol. 2014. doi:10.1080/
[6] Grases F, Shnel O, Vilacampa AI, et al. Phosphates precipitating from articial urine and ne structure of phosphate renal calculi. Clin Chim Acta. 1996;244:4567.
[7] Grases F, Costa-Bauz A, Garca-Ferragut L. Biopathological
crystallization: a general view about the mechanisms of
renal stone formation. Adv Colloid Interface Sci.
[8] Grases F, Vilacampa AI, Costa-Bauz A. Ammonium and
sodium urates precipitating from synthetic urine and







Downloaded by [Aurlie Capdevielle] at 17:58 17 December 2015











ne structure of urate renal calculi. Urol Res. 1999;27:141

Costa-Bauz A, Ramis M, Montesinos V, et al. Type of renal
calculi: variation with age and sex. World J Urol.
Wilsenach JA, Schuurbiers CAH, van Loosdrecht MCM.
Phosphate and potassium recovery from source separated urine through struvite precipitation. Water Res.
Ryu H-D, Lee S-I. Application of struvite precipitation as a
pretreatment in treating swine wastewater. Process
Biochem. 2010;45:563572.
Liu Y, Kwag J-H, Kim J-H, et al. Recovery of nitrogen and
phosphorus by struvite crystallization from swine wastewater. Desalination. 2011;277:364369.
Schuiling RD, Andrade A. Recovery of struvite from Calf
Manure. Environ Technol. 1999;20:765768.
APHA. Standard methods for the examination of water
and wastewater. 18th ed. Washington (DC): American
Public Health Association; 1998.
Rodriguez-Carvajal J. FULLPROF: a program for Rietveld
renement and pattern matching analysis. in: Satellite
Meeting on Powder diffraction of the XV Congress of
the IUCr; 1990.
Boursier H. Etude et modlisation des processus biologiques au cours du traitement arobie du lisier de porcs
en vue dune optimisation du procd [Modelling of biological processes during aerobic treatment of pig slurry
aiming at process optimisation] [dissertation]. Toulouse
(FR): Institut National des Sciences Appliques de
Toulouse; 2003.
Le Corre KS. Understanding struvite crystallisation and
recovery [dissertation]. Craneld: Cranel University,
School of applied science Department of sustainable
system center for water science; 2006.
Abbona F, Lundager Madsen HE, Boistelle R. The nal
phases of calcium and magnesium phosphates precipitated from solutions of high to medium concentration. J
Cryst Growth. 1988;89:592602.
Quintana M, Sanchez E, Colmenarejo MF, et al. Kinetics of
phosphorus removal and struvite formation by the utilization of by-product of magnesium oxide production.
Chem Eng J. 2005;111:4552.
Posner AS, Betts F. Synthetic amorphous calcium phosphate and its relation to bone mineral structure. Acc.
Chem. Res. 1975;8:273281.
Daumer ML. Analyse des transformations du phosphore
lors du stockage et du traitement biologique et physicochimique des lisiers de porc en vue dun recyclage
[Study on phosphorus transformations during biological
and chemicophysical treatment of pig slury applied to
phosphrorus recycling] [dissertation]. Toulouse (FR):
Institut National des Sciences Appliques de Toulouse,
Li XZ, Zhao QL. Efciency of biological treatment affected
by high strength of ammonium-nitrogen in leachate and
chemical precipitation of ammonium-nitrogen as pretreatment. Chemosphere. 2001;44:3743.
Stolzenburg P, Capdevielle A, Teychen S, et al. Struvite
precipitation with MgO as a precursor: Application to
wastewater treatment. Chem Eng Sci. 2015;133:915.


[24] Derrien M, Jard E, Gruau G, et al. Extreme Variability of

Steroid Proles in Cow Feces and Pig Slurries at the
Regional Scale: Implications for the Use of Steroids to
Specify Fecal Pollution Sources in Waters. J Agric Food
Chem. 2011;59:72947302.
[25] Boursier H, Bline F, Paul E. Piggery wastewater characterisation for biological nitrogen removal process design.
Bioresour Technol. 2005;96:351358.
[26] Christensen ML, Hjorth M, Keiding K. Characterization of
pig slurry with reference to occulation and separation.
Water Res. 2009;43:773783.
[27] Koutsopoulos S, Dalas E. The Crystallization of Hydroxyapatite
in the Presence of Lysine. J Colloid Interface Sci.
[28] Kona AN, Demadis KD, Koutsoukos PG. The effect of
citrate and phosphocitrate on struvite spontaneous precipitation. Cryst Growth Des. 2007;7:27052712.
[29] Cicerone DS, Regazzoni AE, Blesa MA. Electrokinetic properties of the calcite/water interface in the presence of
magnesium and organic matter. J Colloid Interface Sci.
[30] Matabola KP, van der Merwe EM, Strydom CA, et al. The
inuence of hydrating agents on the hydration of industrial magnesium oxide. J Chem Technol Biotechnol.
[31] Mersmann A. Crystallization technology handbook. 2nd
ed. New York: Marcel Dekker; 2001.
[32] Hjorth M, Christensen ML, Christensen PV. Flocculation,
coagulation, and precipitation of manure affecting three
separation techniques. Bioresour Technol. 2008;99:8598
[33] Garcia MC, Vanotti MB, Szogi AA. Simultaneous separation
of phosphorus sludge and manure solids with polymers.
Biol Eng Trans. 2007;50:22052215.
[34] Schulze-Rettmer R, Metzen P, Alfter P, et al. MAP
Precipitation for recovering nutrients from manure; s. d.
[35] Mullin J. Crystallization. 4th ed. Oxford: Butterworth
Heinemann; 2001.
[36] Pang Z. Phosphorus enrichment in the treatment of pig
manure in China using anaerobic digestion technology
[dissertation]. Weimar: Bauhaus Universitat, Fakultt
Bauingenieurwesen; 2008.
[37] Zhang D, Chen Y, Jilani G, et al. Optimization of struvite crystallization protocol for pretreating the swine wastewater
and its impact on subsequent anaerobic biodegradation
of pollutants. Bioresour Technol. 2012;116:386395.
[38] Donnert D, Berg U, Weidler PG, et al. Phosphorus removal
and recovery from waste water by crystallisation. Wasserund Geotechnologie. 2002;3:16103645.
[39] Bhuiyan MIH, Mavinic DS, Koch FA. Phosphorus recovery
from wastewater through struvite formation in uidized
bed reactors: A sustainable approach. Water Sci Technol.
[40] Shen Y, Ogejo JA, Bowers KE. Reducing the effects of
calcium on struvite precipitation in liquid dairy manure.
in: American Society of Agricultural and Biological
Engineers Annual International Meeting; 2010.
[41] Thirion F. Liquid manure viscosity and ow properties, in:
2e dition des Journes de lpandage, Avances scientiques et technologiques de lpandage des efuents
liquides. Montoldre, France; 2005.