Manuscript May 2010

For Review Purposes Only/Aux fins d'examen seulement
1
Application of Polymath Chemical Equilibrium Model for Struvite Precipitation in Soils 1
Michael K. Miyittah
1
, Sachin Gadekar
2
, Pratap Pullammanappallil
2
, Craig D. Stanley
1
, Jean- 2
Claude Bonzongo
3
, Jack E. Rechcigl
1
3
4
1
Soil and Water Science Department, University of Florida, IFAS, Gulf Coast Research and 5
Education Center, Wimauma, FL. 33598. USA.
2
Department of Agricultural and Biological 6
Engineering, University of Florida, Gainesville, FL. 32611. USA.
3
Department of Environmental 7
Engineering Sciences, University of Florida, Black Hall, P.O. Box, 116450, Gainesville, FL. 8
32611. USA. 9
10
11
Environmental Context. Excess loss of phosphorus is a major concern for many areas laden 12
with manure. It is confirmed that combination of Al-based with Ca-Mg-based materials may help 13
reduce offsite release of phosphorus. The formation of phosphorus solid phases in soil may 14
prevent phosphorus runoff. Notably, solid phases difficult to predict in soil is struvite. Struvite is 15
considered as a slow release fertilizer. Polymath model predicted struvite f ormation well, but 16
prediction was relatively poor in Visual-Minteq. Struvite prediction in soil suggests that the 17
immobilized phosphorus can be re-used. The slow kinetics to release phosphorus by struvite may 18
be suitable for plants gradual uptake. The processes of fixing and slowly releasing phosphorus 19
for re-use, due to struvite gradual dissolution may help in promoting the sustainability of 20
environmental nutrient resource. 21
22
Abstract 23
24
A new speciation model is developed and implemented in Polymath was found to be successful 25
in predicting struvite precipitation in soils. Struvite (NH
4
MgPO
4
) has been identified as a mineral 26
for the recovery of nitrogen and phosphorus. Predicting struvite precipitation potential in soil is 27
important for optimal quantification of nutrient species. Polymath and visual-minteq model were 28
used for prediction of several solid-phases in the soil. One approach to immobilize P for solid 29
phase formation is by co-blending. Immobilization was achieved through the blending of an Al- 30
based water treatment residual (Al-WTR) and with Ca-Mg-based materials (slag and magnesium 31
oxide). The results suggest that Polymath model revealed solid phases of Dicalcium phosphate 32
pentahydrate (DCPP), Magnesium hydroxide (MHO), Magnesium orthophosphate (v) 33
docosahydrate (MP22), Magnesium orthophosphate (v) octahydrate MP8, and struvite, which 34
were lacking in the modeling from visual-minteq. Residual leachate from the co-blended 35
amendments; Soil+WTR+Slag, Soil+WTR+MgO, Soil+MgO, Soil+Slag, Soil+WTR, and the 36
control (without amendment) had struvite of 353, 199, 119, 90, 37, and 12 mg L
-1
respectively. 37
This implies that struvite a phosphate mineral can be precipitated in the soil and could be 38
released as nutrients for plant uptake. Struvite precipitation in soil and for re-use may reduce cost 39
and may be a safe practice for sustainable environmental nutrient management. 40
41
Keywords. Struvite, manure-impacted soil, phosphorus, modeling, 42
43
2
1
Corresponding address: 2
Gulf Coast Research & Education Center 3
14625 CR 672, University of Florida 4
Wimauma, Fl. 33598. 5
Phone: (1) - 813-633-4111 6
Fax: (1) -813-634-0001 7
*Corresponding email: 8
miyittah@ufl.edu 9
10
11
Introduction 12
13
Phosphorus (P) pollution from agriculture sources are primary contributor to more than 14
40% of impaired lakes in the United States.
[1]
Phosphorus from dairy manure-laden soils has 15
also been identified as non-poi nt sources leading to eutrophication and water quality 16
degradation.
[2]
Significant amount of research has been carried out in recent times to understand 17
P dynamics in animal manure and the interactions of the manure P components with soils.
[3]
18
Soil P has been found to be strongly associated with iron (Fe)-aluminum (Al) and 19
calcium (Ca)-magnesium (Mg) metal hydr(oxides) depending on the prevailing pH. In acid 20
soils, Fe-Al hydr(oxides) and other metals react with P to form solid phases. On the other hand, 21
under alkaline condition, Ca and Mg activities tended to control the solubility of P in soils.
[3,4]
22
In Florida, Spodosols are usually characterized by low contents of Fe-Al and Ca-Mg due 23
to coarse texture sandy soils. In addition, the prevailing high rainfall also leads to leaching of 24
basic cations and significant loss of applied P through runoff.
[5]
Various management practices 25
have been suggested to reduce potential P losses and subsequent impact on water bodies. One 26
approach is the use of amendments to increase the soil sorption capacity for P retention. 27
Successful soil amendments include industrial by-products and municipal wastes that contain 28
3
metals (e.g. Fe, Ca, Al) for P retention. Metals salts (e.g. Al
2
(SO
4
)
3
, FeCl
3
) have also been used 1
to mitigate P solubility and mobility in poultry manure and poultry manure amended soils.
[6,7,8]
2
The reaction of these metal cations of Fe-Al and Ca-Mg in soils containing high P has 3
been a subject of recent investigations. The importance of the metal cations is to form a solid 4
phase of P in order to prevent future release. However, phosphate solid phases vary greatly in 5
soils depending on the metal cations. It is therefore of great advantage to understand solubility of 6
P if metal cations are co-blended in various ratios. By co-blending we refer to the mixing or 7
blending of two or more by-products containing aluminum water treatment residual (Al-WTR) 8
with Ca-Mg based materials (slag and low activity magnesium oxide).
[9]
9
Geochemical model (MINTEQA2) has been used as a predictive tool for solid phase 10
speciation of P in manure-impacted soils using Al-WTR as sorbent.
[5,10]
However, the model 11
used failed to identify struvite (MgNH
4
PO
4
.6H
2
O) as one of the phosphate minerals in the soils. 12
The availability of struvite, a sparingly soluble mineral if present, suggests slow release of P for 13
plants uptake. Consequently, the slow release of nutrients from struvite may prevent the 14
likelihood of adverse risk of excess P into water bodies. In addition, the available struvite in the 15
soil if quantified could provide insight on how much P, N and Mg could be released. 16
In animal manure study (pig, cattle, sheep and poultry liter), struvite has been identified 17
using qualitative scanning electron microscopy/energy dispersive X-ray spectroscopy 18
(SEM/EDX), X-ray powder diffraction (XRD) and solid-state
31
P NMR techniques.
[11,12]
In 19
addition, sophisticated equipment like XANES has been used to evaluate the Mg phosphates 20
(struvite) in animal manure.
[3]
However, the use of these methods in investigating the presence 21
of struvite is labor intensive, expensive and quantitative assessment becomes difficult. To our 22
4
knowledge, to predict struvite in soil is a daunting task. However, polymath model has been used 1
to predict struvite in wastewater.
[13]
2
The Polymath model was developed for predicting magnesium and calcium related 3
minerals. It is a mathematical model for precipitation, which uses physicochemical equilibrium 4
expressions, mass balance equation for N, P, Mg, and charge balance. The equations derived 5
from the expressions are solved using polymath education version 6.1.
[13]
6
The objective of this study is to compare prediction of struvite in soil with polymath and 7
that of Visual-Minteq model, resulting from co-blending of aluminum residual and Ca-Mg-based 8
materials. 9
10
Experimental methods 11
Soil and Amendments Characterizations 12
A manure-impacted sample of Immokalee fine sand (sandy, siliceous, hyperthermic 13
Arenic Haplaquod) was obtained from a field site on a dairy cattle ranch located in the Lake 14
Okeechobee County watershed. This soil was selected for its high total P due to long-term 15
manure application and/or fecal deposition, as well as the soils geographic extent in South 16
Florida. Random samples were collected from A horizons (0-15 cm), and were thoroughly 17
mixed to yield a composite sample. The soil was air-dried and sieved through a 2 mm stainless 18
steel screen. 19
The Al-WTR was collected from a drinking water treatment plant in Bradenton Florida. 20
In characterizing the amendments, pH was measured in a 1:2 (w: v) aqueous suspension as 21
reported elsewhere.
[5]
Redox potential (Eh as mV) was also measured using two electrodes 22
5
connected to the pH meter (Accument XL60, Fisher Scientific, USA). The reference electrode 1
and temperature sensor were connected to an interface multi-channel (pH-meter/redox meter), 2
which enabled simultaneous readings of pH and Eh. After every 10 samples, Eh monitoring was 3
constantly evaluated to avoid drifts in voltage through quality control/assurance maintenance.
[14]
4
The Al-WTR samples were air-dried and sieved through < 850 m screen to minimize 5
slaking and increase reactivity.
[15]
Two Ca and/or Mg products, low activity magnesium oxide 6
(MgO) and slag were characterized in a similar manner as the Al-WTR. Total recoverable Fe, P, 7
Al, Ca and Mg were determined using inductively coupled plasma optimal-emission 8
spectrometry (ICP-OES, Perkin-Elmer Plasma 2100DV, Perkin-Elmer Inc, Waltham, MA., 9
USA.), following digestion according to EPA Method 3050B.
[16]
10
11
Soil Extractions, Co-blending and Leaching 12
The soil was sequentially extracted for P according to Chang et al.
[17]
, using 1:20 13
soil/solution ratio, with modification as in Silveira et al.
[5]
The sequential extraction procedure is 14
an operational defined scheme of fractionating t he soil P. The extraction procedure utilized: 1) 15
1M KCl to determine the soluble or exchangeable P. This P fraction is easily leachable and most 16
responsible for algal growth. 2) 0.1M NaOH to determine the chemisorbed P onto oxides and 17
hydroxides of Fe, Al and Mn. 3) 0.5M HCl to determine P associated with Ca-and Mg 18
complexes.
[18, 19, 20]
However, at the end of the analysis, the recalcitrant P was not analyzed. 19
Amendments containing Ca-, Mg- and Al-based were co-blended as described elsewhere.
[9]
20
In brief, amendments were applied at 0%, 1%, and 2% by mass to the weight (d.w.) of the soil. 21
6
Each of the soil mixes amended with Al-WTR, Al-WTR+MgO, and Al-WTR+Slag had rate of 1
(20 gkg
-1
or 2%, (10+10 g kg
-1
, or 1%+1%), or (10+10 g kg
-1
, or 1% +1%), respectively. In all, 2
the soil mixes had the same rates of 20 g kg
-1
amendment for comparison, with the exception of 3
the control (0%). All treatments were replicated three times in a completely randomized design. 4
Soil column set up was similar as previously reported in Silveira et al.
[5]
One hundred 5
milliliters of DDI water (adjusted to pH 5, t o mimic the pH of rainfall in South Florida) were 6
added to each column weekly. Each leaching event corresponded to approximately 1 pore 7
volume. Leachate of the column studies were collected and major anions (PO
4
, SO
4
, NO
3
, and 8
Cl) and cations (Al, Fe, Ca, Mg, K, Na, and NH
4
), pH plus EC were analyzed. Cations with 9
exception of ammonium were analyzed using Inductively Coupled Plasma Optical Emission 10
Spectrometers (ICP-OES). Ammonium was measured with ion selective electrode, Fisher 11
scientific, USA. Phosphate, SO
4
, Cl and NO
3
were analyzed using ion chromatograph, Metrohm 12
USA, Inc. 13
Quality assurance and quality control (QA/QC) protocols were followed, including the use 14
of 5% repeats and 5% spikes and blanks for each procedure.
[22]
Standard calibration curves, as 15
well as quality check standards, were prepared for each procedure. Repeats were within 10% 16
relative standard deviation. In situation where samples standard deviation falls outside the 10% 17
range, repeats were conducted to cross check any anomaly. Spike recovery varied with 18
procedure, but all fell within acceptable levels (95-110%). External quality control checks 19
achieved recoveries within (98-110%). Statistical tests were performed to validate the model 20
fitness. A root mean square error (RMSE), coefficient of determination (R
2
) and d-statistics or 21
index of agreement was used in the model evaluation. 22
23
7
Chemical Equilibrium Speciation Modeling 1
Phosphorus chemical speciation was calculated for the leachates from the control (without 2
amendment), and for 2% application rate of Al-WTR, Slag, and MgO, respectively to the 3
manure-impacted soil. In addition, speciations were calculated for manure-impacted soils co- 4
blended with (1%+1%) each of Al-WTR+MgO, and Al-WTR+Slag, respectively. Previous work 5
from Silveira et al.
[5]
using several weeks of leachates suggested that, number of weeks of 6
selected leachate used did not have any significance difference in chemical speciation. Based on 7
that, the 1st week of data was selected for the chemical speciation equilibrium modeling. In 8
addition, after 2nd to 4th week, due to P immobilizations by the amendments application, PO
4
9
measured was almost zero. The low observed PO
4
limits the prediction of possible phosphate 10
minerals in the leachates. Leachates anions analyzed were NO
3
, Cl, SO
4
, ammonium, and PO
4
. 11
Cations analyzed were Al, Mg, Ca, K, and Na. The pH, EC (electrical conductivity) and 12
alkalinity data were also analyzed as inputs for the chemical speciation. Ionic strength (I), as mol 13
L
-1
of the extracts was calculated from the EC as 0.013*EC. 14
The use of Visual-Minteq has a wide application in water quality assessments. The model 15
has borrowed and incorporated wide thermodynamic databases from MINTEQA2. Visual- 16
Minteq utilizes the activities of dissolved chemical species in calculating saturation index (SI) of 17
different mineral phases. The SI of each solid is defined as: 18
sp
K
IAP
SI log = (1) 19
where, IAP is the ion activity product of the respective chemical species, and K
sp
is the solubility 20
product of the estimated solid phase. For any particular mineral solid, if the calculated SI > 0, it 21
implies that solution/leachate is saturated with the respect to that mineral. Similarly, if calculated 22
8
SI < 0, it means that the leachate is under saturated with the mineral phase. On the other hand, if 1
SI = 0, it suggests that the solid has reached equilibrium.
[22]
2
3
Chemical Equilibrium Modeling with Polymath 4
A chemical equilibrium model was developed recently with the aid of polymath and 5
reported in details for predicting magnesium related minerals such as struvite.
[13]
It is a 6
mathematical model which uses physicochemical equilibrium expressions, charge balance, 7
solubility equilibrium expressions, and mass balance equation for N, P, and Mg. Values of 8
equilibrium constants and solubility products used in the model were reported in Gadekar and 9
Pratap
[13]
. The model is comprehensive in terms of number of solid species considered, wide 10
range of flexibility and applications. The model can be applied to reactors as in soil systems, to 11
explicitly determine the concentrations of all species (dissolved, ionic and solid) for 12
quantification of purity and yield of struvite under various conditions of pH, and initial 13
concentrations of NH4
+
, Mg
2+
, and PO4. Initial conditions which include pH, total concentrations 14
of nitrogen, magnesium, phosphorus, inorganic carbon and calcium are inputs along with 15
equilibrium constants. Initial guesses for Mg
2+
, NH
4
+
, PO
4
3-
, CO
3
2+
, and Ca
2+
were provided. The 16
Polymath Educational Version 6.1 program solved the expressions and gave concentrations of 17
dissolved, ionic species and concentrations of solid components. These expressions included as 18
mass balance equations for total magnesium (Mg
t
), calcium (Ca
t
), ammonia-N (N
t
), inorganic 19
carbon (TIC), and phosphorus (P
t
). The expressions were written as functions of corresponding 20
ionic species concentrations in molar quantities. In all, the model predicted the formation of 15 21
different solid phases when solutions containing ammonium, calcium, phosphate, and carbonates 22
species are mixed. Struvite is one of the solids predicted. Using the charge balance equation, the 23
9
appropriate [Exion] concentration was determined. This gives the acid or base requirement to 1
maintain the pH. The following charge balance equation applies to the system: 2
[NH
4
+
]+[K
+
]+2[Mg
2+
]+2[Ca
+
]
2
+[Na
+
]+[H
+
]+[MgH
2
PO
4
+
]+[MgOH
+
]=3[PO
4
-3
]+2[HPO
4
-2
] 3
+[H
2
PO
4
-
]+2[CO
3
2-
]+[HCO
3
-
]+[H
2
PO
4
-
]+[MgPO
4
-
]+[Cl
-
]+[OH
-
]+[Exions] (2) 4
[Exions] = [other cations]-[other anions] 5
[Exion] was calculated from the charge balance, a positive value indicated addition of alkali, and 6
a negative value indicated addition of an acid to maintain pH. The resulting outcome from the 7
Polymath computation suggests all possible Mg, Ca, K, N, P, and carbonates based on minerals 8
availability. 9
10
Results and discussions 11
Amendments and Soil 12
The physicochemical properties of Al-WTR taken from Bradenton Florida, showed 86 g 13
kg
-1
Al content, and 2.3 g kg
1
Fe content (Table 1). The relative high Al content in Al-WTR 14
suggests that the material might reasonably achieve high P sorption.
[23, 24]
The slag contained Ca, 15
Mg and Al (598, 106, 156 g kg
-1
), respectively. In comparison, MgO had a greater Mg content 16
(897 g kg
-1
) than slag. The pH of Al-WTR was 6.3, which was consistent with values reported 17
elsewhere (6.0-8.4).
[26]
The Al-WTR had P concentration of 0.02 g P kg
-1
, which was slightly 18
below the range reported for Al-WTRs (0.3-4.0 g P kg
-1
).
[26, 27]
The MgO and slag amendments 19
had alkaline pH values of 10.9 and 11.5, respectively, consistent with values reported by 20
Cucarella and Renman
[28]
. The soil pH was 7.1. Oxalate-extractable Fe and Al are usually 21
associated with amorphous phases of oxides, suggest that P will be attached to the sesquioxides 22
10
and other forms of hydr(oxides) present in the soil (Table 1). Total P in the Okeechobee test soil 1
was approximately 2800 mg kg
-1
indicating a high P load.
[2]
In the soil, most of the reactive P 2
was associated with Ca-Mg representing, 62% of total reactive P, whereas P associated with Al- 3
Fe was 23% and easily leachable P ~15% (Table 1). The above fractionation data are consistent 4
with results from Nair et al.
[2]
and Sharpley et al.
[25]
, showing that Ca-Mg bound P dominated the 5
forms in a manure-impacted soil. The total P concentration of MgO was below the instruments 6
detection limits, and slag P content was approximately 0.03 g P kg
-1
. 7
8
Simulation vs. Experimental data using Polymath model and Visual-Minteq 9
(1) Solid Phase Equilibria from the literature 10
Experimental, synthetic, and actual waste water data taken from the literature were 11
simulated using Visual-Minteq and Polymath model.
[13]
The inputs data for simulations utilized 12
the concentrations of total NH
4
-N, Mg, and PO
4
-P, Ca, total inorganic carbon and pH. The 13
current version of Visual-Minteq used did not include struvite in its database. Thermodynamic 14
parameters for struvite were added before running the computer model in Visual-Minteq for the 15
prediction of solid phases. The results are presented in Table 2 and 3. A graphical representation 16
of the amount of struvite predicted versus that of the experimental values is plotted (data not 17
shown). The results showed clearly, the superior strength of Polymath model (R
2
= 0.57) over 18
that of the Visual-Minteq (R
2
= 0.33) in predicting struvite (Table 3). However, to validate the 19
accuracy of the R
2
values, the use of Root Mean Square Error (RMSE) and index of agreement 20
(d-statistics) were obtained. The RMSE was used to differentiate the model having the most 21
significant influence, since R
2
may indicate inaccuracy when used as a measure of accuracy.
[29].
22
A lower value of 293.2 RMSE for Polymath suggests that the model fits the data well in terms of 23
11
struvite predictions (Table 3). Further, index of agreement or d-statistics was also evaluated. The 1
d of 0.78 was greater for Polymath compared to 0.69 for Visual-Minteq, suggesting t hat the 2
model is better in agreement for struvite prediction. The inability of Visual-Minteq to predict 3
struvite well may be due to greater stability of calcium ions such as calcium phosphate, 4
hydroxyapatite or due to Ca ions which compete with Mg for phosphate ions in solution.
[30, 31]
In 5
addition, struvite formation is affected by precipitation kinetics and wide degree of variabilities 6
such as chemical species, pH, EC and ions activity products.
[36]
These factors may have 7
accounted for the large amount of RMSE observed. A polymath chemical equilibrium model, 8
however, accounted for saturation indexes of phosphate minerals, and other amorphous forms 9
when Ca ions are greater than Mg into its formulations. Polymath model accounting for 10
phosphate saturation indexes might have led to the lower reduction of errors observed in the 11
model predictions. In further agreement, the polymath model was specifically designed for Mg-P 12
solid phases, thus the quality of the observed predictions. 13
14
(2) Solid phase Equilibria of Co-blended soil with Visual-Minteq 15
Simulation results of soil (control) and co-blended samples were calculated using Visual- 16
Minteq and Polymath model and are presented in Table 4 and 5, respectively. Although co- 17
blended samples were assumed to be thoroughly mixed, we recognized the complexities and the 18
uncertainties associated with non-equilibrium nature of soil samples, as well as the inherent 19
kinetic effects in chemical speciation. Bearing the uncertainties in mind, we restrict our 20
interpretation to major significant trends in the modeling results. 21
Phosphorus chemical speciation was calculated for the leachates collected from the samples 22
soil [Soil+MgO (2%), Soil+Slag (2%), Soil+Al-WTR (2%), Soil+Al-WTR+MgO (1+1%), 23
12
Soil+Al-WTR+Slag (1+1%)]. Visual-minteq was used to predict the saturation indices for 1
possible solid minerals formation. Table 2 indicates the soil solution saturation indices for 2
phosphate minerals whose dissolution-precipitation reactions may control P activity of the soil. 3
For the control, major fraction of P was associated with Ca-Mg ~60%, followed by HPO
4
2-
4
~30%, and H
2
PO
4
-
~4% of negatively charged complexes. The control soil solution was 5
undersaturated with respect to Mg-P minerals of farringtonite [Mg
3
(PO
4
)
2
] and newberryite 6
[MgHPO
4
.3H
2
O], but were saturated with Ca-P minerals of hydroxyl apatite [Ca
5
(PO
4
)
3
OH], - 7
tricalcium phosphate [- Ca
3
(PO
4
)
2
], and octacalcium phosphate [Ca
4
H(PO
4
)
3
.3H
2
O], except that 8
of dicalcium phosphate dihydrate [CaHPO
4
.2H
2
O] and dicalcium phosphate [CaHPO
4
] whose 9
mineral phases were almost at equilibrium with the soil. The results suggest that phosphate 10
solubility and activity in soil solutions from manure-impacted soils, without any amendment 11
were dominated mainly by the dissolution-precipitation reactions of under-saturated Ca- and Mg- 12
P solid phases (SI< 0). The above observation with the control soil without any amendment is 13
consistent with results from Immokalee soil, a spodosol.
[5, 10]
14
Co-blending the soil with Slag, Al-WTR and MgO at 2% rate, had similar mineral 15
predictions under visual-minteq modeling. All treatments were under saturated with newberryite 16
[MgHPO
4
.3H
2
O], farringtonite [Mg
3
(PO
4
)
2
] except that of Soil+MgO, indicating super- 17
saturation of newberryite. The super saturation with respect to newberryite supports the fact that 18
Mg-P bearing minerals are responsible for controlling P solubility in manure-impacted soils. On 19
the other hand, addition of Slag and Al-WTR co-blended soil had similar solid phases with Mg-P 20
minerals but with undersaturation indices (Table 4). In addition, undersaturation was also 21
observed with some calcium mineral phases of [CaHPO
4
.2H
2
O] and [CaHPO
4
], thus indicating 22
the role of Ca and Mg in controlling P solubility, which is consistent with the conclusions of 23
13
other studies involving manure-impacted soils
.[5, 10, 32]
Co-blended samples with Soil+WTR+Slag 1
and Soil+WTR+MgO, at 1% +1% amendments had solid phase similar to either Soil+MgO or to 2
Soil+Slag at 2%, respectively. Although, the solid solution of the co-blended samples were 3
thoroughly mixed, it was suggested that the chemical activity of solid phases in the soil system 4
dissolve nonstoichiometrically, thereby creating in-homogeneity and disequilibrium between the 5
solids surface and that of the interior.
[33]
Consequently, the true equilibrium in soil suspensions 6
maybe extremely difficult, and applying strict thermodynamics may be less effective for co- 7
blended samples. In all treatments, the most thermodynamically stable solid phase observed 8
appeared to be that of hydroxyapatite with SI > 0. 9
10
(3) Solid phase Equilibria of Co-blended soil with Polymath model 11
The types of solid phases identified by polymath model were identical to those predicted by 12
visual-minteq (Table 5). However, major differences revealed from Polymath were solid phases 13
of Dicalcium phosphate pentahydrate (DCPP), Magnesium hydroxide (MHO), Magnesium 14
orthophosphate (v) docosahydrate (MP22), Magnesium orthophosphate (v) octahydrate MP8, 15
and struvite. Polymath model shows clearly, the prediction of magnesium-based solids. Struvite 16
precipitation was observed to follow the following order: Soil+WTR+Slag > Soil+Slag > 17
Soil+WTR+MgO > Soil+MgO > Soil+WTR > control. The order of struvite precipitation might 18
suggest a combined effect of magnesium availability and pH. This is because in struvite 19
formation, magnesium ion and pH play a major role. The importance of struvite cannot be 20
overemphasized. This is because struvite as a sparingly soluble mineral can slowly release its 21
nutrients in the long run. The slow release of struvite may offer great advantage to forage 22
growers who are faced with excess laden of manure. The convectional school of thought in 23
14
addressing soil P pollution is to permanently fix the P with a sorbent in the soil. However, with 1
the identification of struvite in the amendment application which is aimed to immobilize the P 2
suggests that the unconventional thinking should be fixing the P and releases it later for future 3
use. This unconventional school of thought is beneficial in cost terms, as well as for sustainable 4
environmental management. Struvite is produced at industrial scale and used as a slow release 5
fertilizer. It can release the P and N at a slower rate which can be controlled by adjusting the pH 6
if needed. The released N and P can be beneficial to plant growth. However, struvite is not used 7
alone, but rather used by mixing with other forms of inorganic and organic fertilizers.
[34]
The 8
presence of struvite as revealed by the Polymath model suggests that cost will be saved 9
compared to synthetic/industrial struvite applied to the soil. This is due to the fact that the 10
kinetics of P release by struvite is so slow and a function of prevailing pH.
[35]
The slow release 11
of struvite may lead to saving cost to ranchers whose soils are heavily laden with manure. The 12
dwindling phosphate reserve in Florida is a concern thus, suggesting that cycling and re-use of 13
excess P in soil through co-blending techniques of amendments may be a safe practice. 14
15
Conclusions 16
Loss of phosphorus (P) is a major concern for many areas laden with manure. It is 17
confirmed that co-blending of Al-based materials with Ca-Mg-based materials may help to 18
reduce offsite release of P. The use of chemical equilibrium models of Visual-Minteq and 19
Polymath model revealed various solid phases formed in the soil. Polymath model however was 20
observed to add new information to that revealed by Visual-Minteq. The results suggest that 21
Polymath model revealed solid phases of Dicalcium phosphate pentahydrate (DCPP), 22
Magnesium hydroxide (MHO), Magnesium orthophosphate (v) docosahydrate (MP22), 23
15
Magnesium orthophosphate (v) octahydrate MP8, and struvite, which were lacking in the 1
modeling from visual-minteq. Notably, the solid phase that is difficult to predict in soils is 2
struvite. Struvite precipitation in soil suggests that the immobilized P can be re-used. Residual 3
leachate from the co-blended amendments; Soil+WTR+Slag, Soil+WTR+MgO, Soil+MgO, 4
Soil+Slag, Soil+WTR, and the control (without amendment) had struvite of 353, 199, 119, 90, 5
37, and 12 mg L
-1
respectively. The slow kinetics of struvite to release P may be harnessed for 6
future plants uptake. Such a process of P immobilization and release may help in promoting the 7
sustainability of environmental nutrient. 8
9
References 10
[1] USEPA, National Water Quality Inventory 2000 Report. USEPA, Washington, D.C. 11
Available at http://www.epa.gov/305b/2000report/. 2000, (Accessed July 22
nd
, 2009). 12
[2] V.D. Nair, D.A. Graetz, K.M. Portier, Forms of phosphorus in soil profiles from dairies of 13
South Florida. Soil Sci. Soc. Am. J. 1995, 59, 1244. 14
[3] K. Gngr, A. Jrgensen, K.G. Karthikeyan, Determination of phosphorus speciation of 15
dairy manure using XRD and XANES spectroscopy. J. Env. Qual. 2007, 36, 1856. 16
[4] W.L. Lindsay, Chemical equilibria in soils. The Blackburn Press. Caldwell, New Jersey. 17
USA 2001. 18
[5] M.L. Silveira, M.K. Miyittah, G.A. OConnor, Phosphorus release from a manure-impacted 19
spodosol: effects of a water treatment residual. J. Env. Qual. 2006, 35, 529. 20
16
[6] P.A. Moore, D.M. Miller, Decreased phosphorus solubility in poultry litter with aluminum, 1
calcium and iron amendments. J. Env. Qual. 1994, 23, 325. 2
[7] P.A. Moore, T.C. Daniel, D.R. Edwards, Reducing phosphorous runoff and improving 3
poultry production with alum. Poultry Sci. 1999, 78, 692. 4
[8] L.M. Malecki-Brown, J.R. White, Effect of aluminum-containing amendments on 5
phosphorus sequestration of wastewater treatment wetland soil. Soil Sci. Soc. Am. J. 2009, 6
73, 852. 7
[9] M.K. Miyittah, C.D. Stanley, C. Mackowiak, J. E. Rechcigl, Developing a remediation 8
strategy for phosphorus immobilization: Effect of co-blending Al-residual and Ca-Mg 9
amendments in a manure-impacted spodosol. Soil Sediment Contam. 2010 in Press. 10
[10] S. Agyin-Birikorang, G.A. OConnor, J-C. Bonzongo, Modeling solid phase control of 11
solubility of drinking water treatment residuals-immobilized phosphorus in soils. 12
Commun. Soil Sci. Plant Analysis, 2009, 40, 1747. 13
[11] K. Gngr, K.G. Karthikeyan, Probable phosphorus solid phases and their stability in 14
anaerobically digested dairy manure. Trans ASABE, 2005, 48, 1509. 15
[12] G.S. Toor, S. Hunger, J.D. Peak, J.T. Sims, D.L. Sparks, in Advances in Agronomy (Ed. 16
D.L. Sparks) 2006, p.1-72, Vol. 89. (Academic Press, Boston). 17
[13] S. Gadekar, P. Pullammanappallil, Validation and applications of chemical equilibrium 18
model for struvite precipitation. Environ. Modeling Assess. 2009. doi:10.1007/s10666- 19
009-9193-7 20
17
[14] M.C. Rabenhorst, Making soil oxidation-reduction potential measurements using 1
multimeters. Soil Sci. Soc. Am. J. 2009, 73, 2198. 2
[15] E. A. Dayton, N.T. Basta, A method for determining the phosphorus sorption capacity and 3
amorphous aluminum of aluminum-based drinking water treatment residuals. J. Env. Qual. 4
2005, 34, 1112. 5
[16] USEPA, Acid digestion of sediments, sludges, and soils [Online], Available at 6
http://www.epa.gov/epaoswer/hazwaste/test/pdfs/3050b.pdf (verified December 6th, 7
2009). USEPA, Cincinnati, OH. 1996. 8
[17] A.C. Chang, A. L. Page, F.H. Sutherland, E. Grgurevic, Fractionation of phosphorus in 9
sludge affected soils. J. Env. Qual. 1983, 12, 286. 10
[18] A.H. M. Hieltjes, L. Lijklema, Fractionation of inorganic phosphates in calcareous 11
sediments. J. Env. Qual. 1980, 9, 405. 12
[19] K.C. Ruttenberg, Development of a sequential extraction method for different forms of 13
phosphorus in marine sediments. Limnol. Oceanogr. 1992, 37, 1460. 14
[20] W. Stumm, J.J. Morgan, Aquatic chemistry: chemical Equilibria and rates in natural waters 15
3
rd
edition, 1996. (John Wiley & Sons, Inc. New York). 16
[21] V.H. Kennedy, A.P. Rowland, J. Parrington, Quality assurance for soil nutrient analysis: a 17
case study. Commun. Soil Sci. Plant Anal. 1994, 25, 1605. 18
[22] C. Zhu, G. Anderson, Environmental applications of geochemical modeling. 2002 19
(Cambridge University Press. Cambridge). 20
18
[23] D.G. Grubb, M.S. Guimaraes, R. Valencia, Phosphate immobilization using an acidic type 1
F fly ash. J. Hazard. Mater. 2000, 76, 217. 2
[24] M. Arias, J. Da Silva-Carballal, L. Garcia-Rio, J. Mejuto, A. Nunez, Retention of 3
phosphorus by iron and aluminum oxide-coated quartz particles. J. Coll. Interf. Sci. 2006, 4
295, 65. 5
[25] A. N. Sharpley, R.W. McDowell, P.J. A. Kleinman, Amounts, forms and solubility of 6
phosphorus in soils receiving manure. Soil Sci. Soc. Am. J. 2004, 68, 2048. 7
[26] K.C. Makris, G.A. OConnor, in Currents perspectives in environmental geochemistry, 8
(Eds D. Sakar, R. Datta, R. Hannigan), 2007, (Geological society of America Press, 9
Denver Col.) 10
[27] Dayton, E.A., N.T. Basta. C.A. Jakober, and J.A. Hattey, Using treatment residuals to 11
reduce phosphorus in agricultural runoff. J. Am. Water Works Assoc. 2003, 95,151. 12
[28] V. Cucarella, G. Renman, Phosphorus sorption capacity of filter materials used for on-site 13
wastewater treatment determined in batch experiments-a comparative study. J. Env. Qual. 14
2009, 38, 381. 15
[29] C.J. Willmott, Some comments on the evaluation of model performance. Bull. Amer. 16
Meteor. Soc. 1982, 63, 1309. 17
[30] S. Kristell, C. Le, E. Valsami-Jones, P. Hobbs, S.A. Parsons, Impact of calcium on struvite 18
crystal size, shape and purity. J. Crys. Growth. 2005, 283, 514. 19
19
[31] N. Ch. Bouropoulos, P.G. Koutsoukos, Spontaneous precipitation of struvite from aqueous 1
solutions. J. Crys. Growth. 2000, 213, 381. 2
[32] M.S. Josan, V.D. Nair, W.G. Harris, D. Herrera, Associated release of magnesium and 3
phosphorus from active and abandoned dairy soils. J. Env. Qual. 2005, 34, 181. 4
[33] H.L. Bohn, R.K. Bohn, Solid activities of trace elements in soils. Soil Sci. 1987, 143, 398. 5
[34] Y. Ueno, M. Fujii. Three years operating experience selling recovered struvite from full 6
scale plant. Environ. Sci. Technol. 2001, 22, 1373. 7
[35] M.I.H. Bhuiyan, D.S. Mavinic, R.D. Beckie, Dissolution kinetics of struvite pellets grown 8
in a pilot-scale crystallizer. Can. J. Civ. Eng. 2009, 36, 550. 9
[36] I. Celen, J.R. Buchanan, R.T. Burns, R.B Robison, D.R. Raman, Using chemical 10
equilibrium model to predict amendments required to precipitate phosphorus as struvite in 11
liquid swine manure. Water Res. 2007, 41, 1689. 12
13
14
15
16
17
18
20
Table 1 1
General properties of amendments and soil 2
Properties P Fe Ca Mg Al pH
Amendments .g kg
-1
..
Al-WTR 0.02 2.3 4.6 0.5 86 6.3
Slag 0.03 14 598 106 156 11.5
MgO
b
BDL 2.0 9.0 879 2.0 10.9
Sequentially extracted P Oxalate extractable
.... mg kg
-1
..
#
Soil pH
c
KCl
d
NaOH
f
HCl Fe Al
A-horizon 7.1 400 640 1720 348 210
3
a
Values are means of triplicate.
b
BDL (below detection limits of 1*10
-3
mg L
-1
).
c,d,f
Soil 4
fractionations (KCl-P (soluble or exchangeable P), NaOH-P (extractable P bound to Al-Fe) and 5
HCl-P (extractable P bound to Ca-Mg)) determined using the method of Chang et al., 1983. 6
#
Okeechobee soil. 7
8
9
21
1
Table 2 2
Data
a
taken from literature to simulate model predictions of Polymath and Visual-Minteq. 3
Initial concentrations
(mM)
Model Predictions
Struvite (mg L
-1
)
Type of wastewater pH
Mg
T
P
T
N
T
Exptl
Struvite
(mg L
-1
) Visual-
Minteq
Polymath
Solutions prepared by adding NH
4
Cl,
KHPO
4
, MgCl, carbonate and acetate
6.8 8.3 12.9 21.43 601 0 223.6
Synthetic urine containing PO
4
, NH
4
,
Na, Mg, K, Ca, Cl, citrate, carbonate
8.0 20.0 13.45 20.18 1685 0 1253.3
4
, NH
4
,
9.4 7.42 14.83 18.70 1045 1006 987.07
4
, NH
4
,
9.4 14.8 14.83 18.70 2011 1087 1845.0
Liquid manure 8.5 2.39 5.51 80.00 338 319 322.2
Supernatant from anaerobically digested
sludge dewatering centrifuge
8.5 1.51 1.97 43.88 195 171.05 200.9
Anaerobic digester effluent supernatant 8.5 7.03 6.38 24.5 805 825 818.2
Synthetic samples prepared by using
MgCl
2
, NaH
2
PO
4
, NH
4
Cl
9.0 14.3 14.26 14.26 1714 1581 452.4
Domestic wastewater + 2 % landfill
leachate
9.2 7.79 7.79 7.79 1036.9 681.7 207.4
Supernatant from sludge centrifuges in
a biological nutrient removal plant
8.1 1.54 2.00 44.5 210.98 143.6 198.2
Swine waste 9.0 9.74 6.09 12 758.6 713.5 705.5
a
Data taken from Gadekar and Pullammanappallil
[13]
. 4
22
1
Table 3 2
Statistical parameters for model prediction with coefficient of determination (R
2
), root mean 3
square error (RMSE), and index of agreement (d). 4
5
Where,
n
O P
RMSE
n
i
i i
=

=
1
2
) (
, n = number of observations, P
i
= predicted value for the i
th
6
measurement, and O
i
= observed value for the i
th
measurement. 7
8
1 0 ,
) (
) (
1
1
2
1
2
s s
(
(
(
(
' + '
=
=
d
O P
O P
d
n
i
i i
n
i
i i
, n = number of observations, P
i
= predicted value for the i
th
9
measurement, O
i
th
measurement, O
i
th
10
measurement, O = the overall mean of observed values, O P P
i i
=
'
and O O O
i i
= ' .
[29]
11
12
13
Statistics Polymath Visual-Minteq
R
2
0.57 0.33
RMSE 293.2 355.7
d 0.78 0.69
23
1
Table 4 2
Saturation indices calculated using Visual-Minteq for treatments with and without co-blending. 3
Predicted minerals
Treatments (-TCP) OCP DCP DCPD HA FNT NBT
Soil (Control) 4.17 3.99 0.40 0.10 13.57 -0.48 -0.39
Soil+MgO (2%) 3.13 1.03 -1.52 -1.81 13.42 1.89 -1.18
Soil+Slag (2%) 2.05 0.19 -1.28 -1.57 11.01 -2.33 -1.98
Soil+Al-WTR (2%) 3.22 2.59 -0.05 -0.35 12.13 -1.81 -0.97
Soil+WTR+MgO (1+1%) 2.37 0.14 -1.64 -1.94 12.01 0.16 -1.62
Soil+WTR+Slag (1+1%) 1.15 -0.62 -1.52 -1.80 9.58 -3.46 -2.28
Saturation index = log ion activity product (IAP)- log solubility product (Ks). 4
()-Ca
3
(PO
4
)
2
= -Tricalcium phosphate (-TCP) 5
Ca
4
H(PO
4
)
3
.3H
2
O = Octacalcium phosphate (OCP) 6
CaHPO
4
= Dicalcium phosphate (DCP) 7
CaHPO4.2H2O = Dicalcium phosphate dihydrate (DCPD) 8
Ca
5
(PO
4
)
3
OH = Hydroxyapatite (HA) 9
24
Mg
3
(PO)
2
= Farringtonite (FNT) 1
MgHPO
4
. 3H
2
O = Newberryite (NBT) 2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
25
1
Table 5 2
Calculated values of solid (mg) found per litter of residual leachate as determined by Polymath model 3
4
()-Ca
3
(PO
4
)
2
= -Tricalcium phosphate (-TCP) 5
Ca
5
(PO
4
)
3
OH = Hydroxyapatite (HA) 6
CaHPO
4
.5H
2
O = Dicalcium phosphate pentahydrate (DCPP) 7
CaHPO
4
= Dicalcium phosphate (DCP) 8
Mg(OH)
2
= Magnesium hydroxide (MHO) 9
Predicted mineral
Treatments -
TCP
HA DCPP DCP MHO NBT MP22 MP8 Struvite
(%Purity)
Total
(Solid)
Control 26.9 1.35E-9 622.3 280.9 22.5 23.2 0.15 1.135 12.24(1.2) 989.3
Soil+MgO (2%) 377.6 9.60E-7 173.4 78.3 22295.9 3.64 0.37 2.86 119.0(3.9) 3051.2
Soil+Slag (2%) 14.3 4.95E-10 484.4 218.6 96.7 87.9 9.2 70.2 90.29(8.4) 1071.7
Soil+Al-WTR (2%) 48.1 4.96E-9 544.4 245.7 96.0 36.9 1.61 12.28 37.18(3.6) 1022.2
Soil+WTR+MgO (1+1%) 9.2 6.58E-08 150.3 67.8 2080.7 8.79 1.98 15.1 199(7.4) 2610.9
Soil+WTR+Slag (1+1%) 8.8 8.23E-11 1092.6 493.0 19.0 324.1 24.6 187.4 353.1(14.1) 2502.6
26
MgHPO
4
. 3H
2
O = Newberryite (NBT) 1
Mg
3
(PO
4
)
2
. 22H
2
O = Magnesium orthophosphate (v) docosahydrate (MP22) 2
Mg3(PO4)2. 8H2O = Magnesium orthophosphate (v) octahydrate (MP8) 3
MgNH
4
PO
4
.6H
2
O = Struvite 4
5
6
27
1

Manuscript May 2010

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Manuscript May 2010

Uploaded by

Copyright:

Available Formats

For Review Purposes Only/Aux fins d'examen seulement

You might also like