- Email: [email protected]
Understanding colloidal FeSx formation from iron phosphate precipitation sludge for optimal phosphorus recovery
Journal of Colloid and Interface Science 403 (2013) 16–21
Contents lists available at SciVerse ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Understanding colloidal FeSx formation from iron phosphate precipitation sludge for optimal phosphorus recovery E. Mejia Likosova a, J. Keller a,⇑, R.A. Rozendal a, Y. Poussade b,c, S. Freguia a a
The University of Queensland, Advanced Water Management Centre (AWMC), St. Lucia, QLD 4072, Australia Veolia Water Australia, Level 15, 127 Creek Street, Brisbane, QLD, Australia c Seqwater, Level 2, 240 Margaret Street, Brisbane, QLD, Australia b
a r t i c l e
i n f o
Article history: Received 17 January 2013 Accepted 3 April 2013 Available online 18 April 2013 Keywords: Colloid Phosphorus recovery Sulfide Zeta potential
a b s t r a c t The use of sulfide to form iron sulfide precipitates is an attractive option for separation and recovery of phosphorus and ferric iron from ferric phosphate sludge generated in wastewater treatment. The key factors affecting the simultaneous generation and separation of iron sulfide precipitates and phosphate solution from ferric phosphate sludge have so far not been thoroughly investigated. This study therefore focuses on the recovery of phosphorus from synthetic sludge by controlled sulfide addition under different operating conditions. The factors that affect the phosphorus recovery, as well as the optimal process conditions to achieve an effective solid–liquid separation, were investigated. The separation of the FeSx particles is a significant challenge due to the colloidal nature of the particles formed. Faster separation and higher phosphorus recovery was achieved when operating at pH 4 with dosing times of at least 1 h. At this pH, phosphorus recovery of 70 ± 6% was reached at the stoichiometric S/Fe molar ratio of 1.5, increasing to over 90% recovery at a S/Fe molar ratio of 2.5. Zeta potential results confirmed the colloidal nature of the iron sulfide precipitate, with the isoelectric point around pH 4, explaining the fast separation of the FeSx particles at this pH. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Phosphorus is one of the most important nutrients required for plant growth. Nearly all the phosphorus used in agriculture comes from phosphate rock mines. However, it is estimated that the phosphate left from mining will last at most about 50–100 years [1]. This has been a driving force toward finding renewable sources. The utilization of wastewater as a nutrient source has increased in the last years [2], with digested sewage sludge being a major target as it contains 2–3% of elemental phosphorus. Methods for phosphorus recovery from wastewater sources such as struvite (magnesium–ammonium–phosphate, MAP) formation [3–6] and the recovery of phosphorus and aluminium from sewage sludge ash by sequential elution with acidic and alkaline solutions (SESAL-Phos-recovery process) [7] have been developed. Similarly, methods for coagulant recovery from ferric (Fe3+) and aluminium (Al3+) precipitation sludge have been proposed such as selective removal using ion exchange [8,9] and acidic extraction of coagulant [10]. However, limited work has been reported on simultaneous recovery of phosphate and ferric iron.
⇑ Corresponding author. Fax: +61 733654726. E-mail addresses: [email protected] (E. Mejia Likosova), j.keller@ awmc.uq.edu.au (J. Keller). 0021-9797/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2013.04.001
The advanced treatment of secondary effluent, e.g. for the production of purified recycled water, often includes a chemical precipitation process, in which coagulants such as FeCl3 are added in order to remove suspended solids, colloids, and phosphate. The sludge formed in this process contains mainly ferric oxy-hydroxides and phosphate, commonly termed ferric phosphate sludge. The simultaneous recovery of phosphate and ferric from ferric phosphate sludge using H2S was first proposed two decades ago [11]. However, in this initial study, no information was provided as to the factors affecting the phosphorus recovery process, i.e. pH, separation method, and performance data such as phosphate and iron recovery efficiencies. In this process, sulfide (H2S, HS, and S2) reacts chemically with iron phosphate, releasing phosphate into solution while precipitating iron sulfide and elemental sulfur [12]:
2FePO4ðsÞ þ 3H2 S ! 2FeSðsÞ þ S0ðsÞ þ 2H2 PO4 þ 2Hþ
ð1Þ
Wei and Osseo-Asare [13] described the formation of iron sulfides and elemental sulfur from the reaction of Fe3+ and HS in aqueous solution via the following reactions:
2Fe3þ þ HS ! 2Fe2þ þ S0 þ Hþ
ð2Þ
Fe2þ þ HS ! FeSðsÞ þ Hþ
ð3Þ
FeSðsÞ þ S0ðsÞ ! FeS2ðsÞ
ð4Þ
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
The formation of the different iron sulfide species strongly depends on the pH, as shown by Wei and Osseo-Asare, who found that in the pH range 3.6–5.7, mostly pyrite (FeS2, Eq. (4)) was produced. They attributed this to the high concentration of the precursors mackinawite (iron monosulfide, Eq. (3)) and elemental sulfur. Additionally, using Energy-Dispersive X-Ray Spectroscopy [EDS] analysis, a strong peak was seen at pH 4 for iron monosulfide and elemental sulfur. This agrees with the Fe/S Pourbaix diagram [14], where FeS exists as a stable compound at pH > 4 and redox potential lower than 0.001 V. Likewise, iron monosulfide formation was seen at pH up to 7.2. However, at pH P 7.2, FeS formation could not compete successfully with c-FeOOH formation, explaining the absence of pyrite as a reaction product [13,15–17]. The pH dependence on the stability of FeSx and soluble sulfur species was also observed by others [18–20]. In a more recent study, Kato et al. [12] evaluated phosphate recovery methods using sulfide based on the process proposed by Ripl et al. [11]. The focus was on maximizing phosphate recovery efficiencies and to validate the method itself. Experiments with pre-coagulated sewage sludge and FePO4 synthetic sludge were therein performed in the pH range of 5.3–6.9 and at S/Fe molar ratios between 1.0 and 2.0, leading to a phosphorus recovery up to 44% from the pre-coagulated sludge and 92.8% from the synthetic FePO4 after centrifugation. Even though the proposed phosphorus recovery process via iron sulfide generation has been further studied by others [12,21,22], the selected solid–liquid separation processes, e.g. precipitation and centrifugation [11,12] present significant challenges for the recovery efficiency and operating costs due to the poor settling characteristics of FeSx particles. When in contact with sulfide, phosphate is released into the solution and a black material is formed. The composition of this black material has been found to be a mixture of FeS and FeS2 [13]. Its exact composition depends on the pH. Fe(HS)+ was also found in solution in the pH range of 3.6–7.8 [13,19,20,23]. Despite the proven fast release of phosphate into solution, the separation of the formed suspension has been a challenge. At neutral pH or higher, the precipitated black material remains in suspension for more than a week and passes through a 0.22 lm filter, as preliminary tests in our lab have shown (data not shown). Among the parameters that may affect the solid–liquid separation in this system, mixing time in the reaction vessel and dosing
17
rate of sulfide have been previously identified as key factors for iron sulfide precipitation in different systems. Gutierrez et al. [24] demonstrated that the hydraulic retention time (HRT) of iron sulfide in sewers crucially impacts the formation of iron sulfide precipitates during primary settling, i.e. at shorter HRT a higher concentration of iron sulfide in the primary effluent was found. These findings suggest that mixing time in the reaction vessel is a crucial factor to enable effective separation of the formed iron sulfides. Metal sulfide salts have low solubility. Thus, supersaturated solutions by interaction of aqueous sulfide and metal are expected [16]. Under these supersaturation conditions, fast nucleation is expected to limit crystal growth, thus generating poor settling characteristics of the formed metal sulfides [18,23,25]. At a microscopic level, the dosing rate and mixing speed are key factors in the nucleation process, i.e., slow dosing rates and mixing speed are expected to hinder nucleation and enhance instead particle growth and thus precipitation [26]. Even though considerable effort has been placed on improving the efficiency of this process, little systematic research has been carried out to determine the factors that affect phosphate recovery and separation of the formed FeS colloidal particles. The aim of this research was to identify and characterize key factors that influence phosphate recovery and to determine the optimal process conditions to achieve an effective solid–liquid separation. In particular, the effects of pH, mixing time, sulfide dosing rate and settling time have been investigated in detail in this study. In order to elucidate the reaction chemistry and to understand the mechanisms of precipitation of the inorganic iron sulfide particles, most experiments in this work were carried out using synthetic FePO4 sludge, as a way to eliminate potential interference of organic solids, which are invariably present in ferric phosphate sludges obtained from wastewater treatment. 2. Materials and methods 2.1. Preparation of synthetic sludge and solutions In order to elucidate the stoichiometry of the iron phosphate reaction with sulfide and to determine the corresponding S/Fe ratio needed to achieve completion of the precipitation reaction, studies using synthetic iron phosphate suspensions were performed.
Fig. 1. Synthetic FePO4 sludge experimental setup: 1 – magnetic stirrer; 2 – magnetic bar; 3 – pH controller; 4 – syringe pump for sulfide addition; 5 – gasbag; 6 – acid addition pump; 7 – sampling port; and 8 – data logger.
18
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
A 0.1 M ferric phosphate suspension was prepared from FePO44H2O (97% pure from Sigma–Aldrich) to simulate the real sludge. Based on a typical Fe3+ concentration in real sludge, a 0.1 M suspension was prepared by mixing the salt in water deionized by reverse osmosis (RO). Sodium sulfide nonahydrate (Na2S9H2O, reagent grade) was purchased from Sigma–Aldrich. The Na2S9H2O crystals were previously washed with RO water to remove oxidized sulfur species at the surface of the crystals as described elsewhere [27]. A 0.8 M sodium sulfide solution was prepared with RO water purged with nitrogen to achieve anaerobic conditions. A 3 M HCl solution was used to control the pH in the precipitation vessel. 2.2. Reactor design Iron sulfide was precipitated in a 250 mL three-neck modified Schott bottle (Fig. 1). During the experiments, 100 mL of the FePO4 suspension was stirred using a magnetic stirrer at 2000 RPM, while a syringe pump added the sodium sulfide solution through a needle inserted at the bottom of the reactor near the magnetic bar, at a rate ranging from 18 to 113 mL/h. This method of addition ensures low local sulfide concentrations throughout the vessel, thus putatively minimizing nucleation and enhancing the iron sulfide particle growth. During the reaction, the pH was controlled using a pH meter/controller (Endress + Hauser Liquisys M CPM223/253) at a set point between 4.0 and 7.0 ± 0.2. HCl (3 M) was dosed with a peristaltic pump to control the pH. A gas collection bag was attached to the reactor in order to capture any gaseous sulfide that may evolve upon sulfide addition, especially during low pH experiments. The pH in the reactor was recorded using a data acquisition unit. 2.3. Sulfide addition experiments According to the iron sulfide precipitation reaction, 2 moles of iron react with 3 moles of sulfide as described by Eq. (1) (in pH range 3–6). This implies a stoichiometric S/Fe ratio of 1.5. Taking this into account, two different types of experiments were designed at different conditions and performed in triplicates. Table 1 shows the experimental conditions. To determine the required S/Fe ratio to achieve complete reaction, as well as to understand the fate of any excess sulfide added, a first set of experiments was proposed. In these experiments (all run at pH 4.0 ± 0.2), sodium sulfide was added in excess (up to S/Fe molar ratio of 2.5) in order to guarantee the completion of the reaction. The sulfide addition was done at a rate of 18.8 mL/h, reaching the S/Fe stoichiometric molar ratio of 1.5 after 1 h of addition. The subsequent ratios (1.75, 2, 2.25, and 2.5) were reached after further 10 min addition each (total addition time 100 min). 4-mL samples were taken at S/Fe molar ratios
of 1.5, 1.75, 2, and 2.5, split into two 2 mL Eppendorf tubes and left in quiescent conditions. After 1 h, the samples were centrifuged at 2147g for 10 min in a bench-top centrifuge in order to guarantee complete separation of the two phases. Analyses of the supernatant and solids were performed as described in Section 2.6. In the second set of experiments, sodium sulfide and iron phosphate solution reacted in stoichiometric S/Fe molar ratio at different pH values (6, 5 and 4 ± 0.2), dosing times (10, 30 and 60 min), and further mixing time after sulfide addition (1, 2 and 3 h). Following collection of 6 mL samples, these were placed in 10 mL plastic vials and left in quiescent conditions to allow for sedimentation of the solid phase (up to 24 h settling time), and analysis of the liquid and solid phases was performed as described in Sections 2.4 and 2.6. The error bars shown in the figures correspond to the standard deviation (SD) after performing the experiments in triplicates (n = 3). 2.4. Turbidity measurements In order to evaluate the rate of sedimentation, a turbidity test of the supernatant in the second set of experiments (Section 2.3) was performed every 5 min on the quiescent 6 mL samples using a LaMotte 2020 we/wi turbidimeter. 2.5. Particle size and zeta potential Two methods were used to measure the particle size of the iron sulfide suspensions: laser diffraction (Mastersizer 2000, Malvern Instruments) and dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments). Laser diffraction (also known as low angle laser light scattering) relies on the fact that the diffraction angle is inversely proportional to particle size [28]. This method is applicable for a particle size range between 0.1 and 3000 lm. This measurement assumes the volume of the particle as the important particle property, giving a volume/mass mean particle size distribution. Dynamic light scattering measures the speed at which the particles are diffusing due to the Brownian motion [29]. The rate of the Brownian motion will depend on the size of the particles, i.e., small particles fluctuate more rapidly than the larger ones. The Zetasizer Nano ZS measures the light intensity fluctuation of the particles and correlates them with a size measurement of a similar known particle, giving an intensity size distribution. The Zetasizer Nano ZS has accuracy in the range of 1 nm–10 mm. Along with the particle size, the Zetasizer Nano ZS also measures the zeta potential of the particles. Given the high density of the FeSx sludge suspension, a 500 times dilution was performed with RO water when using the Zetasizer Nano ZS. For the zeta potential measurement of the remaining particles in suspension after 24 h settling, a 20 times
Table 1 Experimental conditions and reagents concentrations. Initial Fe, P and S amounts and concentrations
pH
S/Fe molar ratio
Time to reach S/Fe molar ratio (min)
Na2S dosing rate (mL/h)
Experiment 1: iron sulfide precipitation reaction at pH 4 and sulfide overdose up to S/Fe molar ratio of 2.5 Fe (mol) 0.01 4 1.5 60 18.8 P (mol) 0.01 1.75 70 Vol. FePO4 (L) 0.1 2 80 Na2S (mol/L) 0.8 2.25 90 2.5 100 Experiment 2: iron sulfide precipitation reaction at different pH and dosing times, further mixing and sulfide addition up to S/Fe stoichiometric molar ratio of 1.5 Further mixing times (h) Sulfide dosing time (min) Fe (mol) 0.01 4 0, 1, 2 and 3 60 18.8 P (mol) 0.01 0 30 37.6 Vol. FePO4 (L) 0.1 0 10 112.8 Na2S (mol/L)
0.8
5 6
1, 2 and 3 1, 2 and 3
60 60
18.8 18.8
19
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
2.6. Chemical analysis The characterization of the liquid and the solid phases after the addition of sulfide was performed for each sample after centrifuging at 2147g for 10 min and decanting the liquid phase with a Pasteur pipette. Total phosphorus, sulfur and iron were determined using Inductively Coupled Plasma–Optical Emission Spectroscopy (ICP–OES, Perkin Elmer Optima 3300DV) analysis for both liquid and solid phases. The determination of total iron, sulfur, and phosphorus in the solid phase was achieved by acidic digestion of the dry solids obtained after drying the samples overnight at 105 °C. In all experiments, the fraction of the reacted sulfide was calculated as the ratio of the total moles of the element in the system (liquid plus solid phase) minus the moles of the element in the liquid phase over the total moles in the system %Sreacted ¼
Total moles of S in the system Moles of S in the liquid phase Total moles of S in the system 100%
2.5 2.0
S/Fe molar ratio
dilution was performed. The pH adjustment after dilution of the sludge was performed immediately before the measurement in the Zetasizer in order to avoid the settling of the particles before measurement. Either 3 M HCl or 2 M NaOH were used as required.
1.5 1.0 0.5 0.0 60
70
80
90
100
Time [min] S/Fe added
S/Fe solid phase
S/Fe system
S/Fe liquid phase
Fig. 2. Sulfur balance measured by ICP in the system (sum of solid and liquid phases) after sulfide overdose up to S/Fe molar ratio of 2.5 at pH 4. Sulfur concentrations are normalized to the total Fe in the system (constant) (n = 3). Solid line: S/Fe molar ratio added; squares and dotted line: S/Fe measured in the system (sum of S/Fe in the solid and liquid phases); triangles and solid line: S/Fe molar ratio in the solid phase; and dots and dashed line: S/Fe molar ratio in the liquid phase.
ð5Þ
Moles of P in liquid 100% Total moles of P in system
ð6Þ
The change of volume in the system due to sampling and sulfide addition was taken into account in each of the calculations.
100
100
90
90
80
80 %S [mass]
70
70
%P [mass]
60
%P recovered
%Precovered ¼
%S reacted
Similarly, the recovered phosphorus in the supernatant was calculated as the ratio of the P moles in the liquid phase over the total moles in the system
60
3. Results and discussion 3.1. Reaction stoichiometry and phosphorus recovery
50 1.50
50 1.75
2.00
2.25
2.50
S/Fe molar ratio Based on the stoichiometry (Eq. (1)), it is expected that 1.5 moles of sulfide react with 1.0 mole of FePO4. This was confirmed after overdosing sulfide up to a S/Fe ratio of 2.5, where the S/Fe molar ratio reached in the solid phase was approximately 1.5 regardless of sulfide overdosing, thus indicating full reaction of Fe, according to reactions (2)–(4) (Fig. 2). The measurements reveal some missing sulfur from the liquid phase at higher S/Fe ratios, possibly due to some loss as H2S during the acidic digestion procedure as required for ICP measurement. At pH 4 and continuous mixing, phosphorus recovery of 70 ± 6% was obtained with 1-h sulfide addition time up to S/Fe stoichiometric molar ratio. However, the phosphorus recovery increased up to 92 ± 6% as the S/Fe molar ratio increased to 2.5 (Fig. 3). This result agrees with the findings of Kato et al. [12]. However, a much faster separation of the solid and liquid phases was reached without the use of the centrifuge at pH 4, enhancing the efficiency of the process (Section 3.2). 3.2. Effect of pH on solids precipitation In the second set of experiments (Section 2.3), i.e., at the stoichiometric S/Fe molar ratio of 1.5 with differing pH levels of 6, 5, and 4, respectively, no separation of the liquid and solid phases was reached at pH 6 after 1 h settling (post sulfide addition). However, a more defined solid/liquid interface appeared as the pH became more acidic, i.e., pH 5 and 4. At pH 4, a clear separation of the phases was seen after only 9 min of sedimentation.
Fig. 3. Percentage of S reacted and P recovered in solution at pH 4 and different S/Fe molar ratios (up to 2.5). Solid line: %S reacted (present in solid phase); and dotted line: %P recovered (in liquid phase).
Turbidity analysis of the iron sulfide suspension as a function of the settling time revealed a much faster sedimentation rate of the iron sulfide particles at pH 4 compared to pH 6 (Fig. 4). After only 1 h, the resulting supernatant at pH 4 was significantly clearer than the one obtained after 1 day of settling at pH 6. In order to determine the reasons for the improved settleability of the iron sulfide particles (FeSx) at pH 4, an assessment of their particle size distribution (PSD) at different pH was performed using laser diffraction (Mastersizer 2000, Malvern Instruments) and dynamic light scattering (Zetasizer Nano, Malvern Instruments), as described in Section 2.5. Contrary to the expectations, as the pH became more acidic, the FeSx particle size decreased, i.e., the volume-based median particle size using laser diffraction was 49 ± 2 lm, 46 ± 2 lm, and 28 ± 2 lm for pH 6, 5, and 4, respectively. These particles are far larger than typical colloids, which may reflect the fact that they are the result of aggregation, a phenomenon that is well described in the literature for FeS particles and can lead to the increase in the observed particle size from 2– 10 nm (nanoclusters) to 1 lm (colloids) [18], and even up to 50 lm [30]. Results using dynamic light diffraction technique indicated a similar decreasing trend as the pH became more acidic. However, at pH 4, an increase in the particle size (higher than at
20
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
cles (presumed to be elemental sulfur according to Eq. (1)) remained in suspension in the supernatant. Colloidal particles typically hold a negative surface charge, which prevents aggregation and thus sedimentation. Therefore, the stability of such colloid systems depends on the balance of various interaction forces such as van der Waals attraction, double-layer repulsion and steric interaction [31]. Consequently, if the particles have a strong repulsion, the suspension will be stable. In Fig. 5, the zeta potential of the white particles in the supernatant at different pH values is shown. The point of zero charge of these particles was found to be between pH 2.5 and 4.2 (likely somewhat below pH 4), explaining their limited settleability and thus continued presence in suspension at this pH. These results largely agree with the findings of Wei and Osseo-Asare [13], who reported the isoelectric point of FeS2 at pH 3.
250
Turbidity [NTU]
200
150
100
pH 4 50
pH 6
0 0
20
40
60
80
100
120
1440
Settling Time [min] Fig. 4. Iron sulfide settling behavior at different pH levels after 1 h reaction time: turbidity analysis. Dots: pH 4 and squares: pH 6. A faster precipitation of the FeSx particles and much clearer supernatant was reached at pH 4 after 1 h sedimentation time compared to pH 6.
pH 6) was seen. This could be a result of the different stirring conditions and iron sulfide complexes present in the samples, i.e., samples are vigorously stirred in the laser diffraction technique compared to dynamic light diffraction (not stirred), hence allowing coagulation of the nanoclusters to larger particles with the latter technique. Fig. 5 shows the zeta potential of the FeSx particles at different pH. It can be seen that as the pH became more acidic, the zeta potential of the FeSx particles decreases to zero at around pH 4. This point of zero charge is known as the isoelectric point and the pH at which the suspension is least stable; thus, sedimentation of the particles occurs. Hence, the colloidal nature of the FeSx particles was confirmed, with an isoelectric point around pH 4. It can be deduced that the better settling characteristics of FeSx particles at lower pH were not a consequence of larger particle size. Fast precipitation at pH 4 is thus a consequence of the particles becoming neutral. The increase in the particle size at pH 4 as observed with the dynamic light diffraction technique could be explained by the occurrence of two phenomena: the aggregation of the FeSx nanoclusters [18] (which would scatter more light and therefore a larger particle size would be determined) and/or the formation of pyrite (FeS2) particles, which has been reported to happen between pH 3.6 and 5.7 [13]. Despite the visual differentiation of the two phases as the pH became more acidic, some black FeSx [13] and white-colored parti-
3.3. Effect of dosing time and further mixing time on reaction rate and solids precipitation The effect of different reaction times and further mixing after completion of the iron sulfide formation (S/Fe stoichiometric molar ratio 1.5) was studied after performing the second set of experiments described in Section 2.3 (Table 1) at pH 4. No significant change in the reacted iron sulfide was seen at different dosing rates, indicating the fast rate of particle formation and confirming the observations that nucleation and particle size are not key factors in achieving efficient phase separation here. However, a better phosphorus recovery (68 ± 9% against 49 ± 9%) was obtained with 1 h compared to 30 min dosing time at pH 4, suggesting a possible mass transfer or reaction rate limitation in the phosphate solubilization process at higher sulfide dosing rates. The settling times of the iron sulfide particles at the end of the sulfide addition for different dosing rates were also measured, as shown in Fig. 6. A slightly faster settling of the iron sulfide particles was seen after 1 h reaction time. Consequently, this indicates that dosing times of at least 1 h appear to achieve faster separation rates and higher phosphate recovery. No enhancement of the solid/liquid phase separation was achieved after further mixing at pH 6, 5, and 4. Contrary to the findings reported by Gutierrez et al. [24], further mixing times did not improve the settling of the FeSx particles, i.e., no clearer separation of the solid/liquid phases was reached with further mixing after completion of the iron sulfide reaction. On the contrary, a better separation of the two phases was reached when the sludge was allowed to settle right after completing the addition of sulfide in all cases. Only at pH 6 and at all mixing times, poor phase
250
30 min dosing time
White particles in supernatant
20
200
0 2
4
6
8
10
-20
Turbidity [NTU]
40
Zeta Potential [mV]
60 min dosing time
FeSx Particles
10 min dosing time
150 100 50
-40
0 0
-60
pH
Fig. 5. Dots: FeSx particles zeta potential at different pH using the Zetasizer and squares: S0 particles zeta potential at different pH.
30
60
90
120
1440
Settling Time [min] Fig. 6. Iron sulfide settling time at different dosing times at pH 4 based on turbidity analysis.
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
nature of the FeSx particles formed, charge neutralization of the FeSx particles at pH 4 (as shown by zeta potential based determination of the isoelectric point) appears to be the dominant factor in achieving efficient coagulation and settling of the particles.
100
P recovered in solution [%]
21
80 60 40 20 0 1
1.2
1.4
1.5
1.75
S/Fe molar ratio Fig. 7. Percentage of phosphorus recovered in solution from real ferric phosphate sludge at pH 4. S/Fe stoichiometric molar ratio of 1.5 was reached after 1 h reaction time.
Achieving an efficient precipitation and solid/liquid separation of iron sulfide sludge is important to enable processing of concentrated iron phosphate sludge, hence achieving a maximal concentration of the phosphate in the supernatant. This stream could most beneficially be re-used in a struvite precipitation process, which is typically phosphate limited, or even directly introduced into chemical fertilizer manufacturing. In both cases, achieving a high phosphate concentration and effective separation from the FeSx precipitants will be important. Additionally, the iron sulfide sludge can offer attractive avenues for iron recovery via sulfide oxidation to sulfur (or sulfate), hence creating further valuable opportunities for effective chemical recovery from this process. Acknowledgments
separation was achieved, even after 24 h sedimentation. Effectively, no change in the phosphate recovery, as well as a possibly slightly negative effect in the iron recovery, was seen with extended mixing times (data not shown). 3.4. Phosphorus recovery with real precipitation sludge In order to evaluate the robustness of the optimized process herein proposed, preliminary sulfide addition experiments (see Section 2.3) with real ferric phosphate sludge (with iron and phosphorus concentration of 10.6 and 0.7 g/L, respectively) from a local full-scale advanced water treatment plant (AWTP) were performed up to S/Fe molar ratio of 1.7 at pH 4. At the S/Fe stoichiometric molar ratio of 1.5, up to 63% phosphorus was recovered in solution (Fig. 7). As expected, better phosphorus recovery was reached at higher S/Fe molar ratio, i.e., up to 75% P was recovered at S/Fe molar ratio of 1.7. These results are quite similar those with synthetic sludge (reported above) and therefore confirm the efficiency of the optimized iron sulfide precipitation process for phosphorus recovery from ferric phosphate sludge. The point of zero charge of the FeSx particles formed from real ferric phosphate sludge was found to be slightly below pH 4 (data not shown), which supports the general applicability of the separation method. However, specific optimization of the process may be required for any given ferric sludge based on the particular composition in each situation. 4. Conclusions The aim of this research was to determine the optimal process conditions to achieve an effective solid–liquid separation and phosphate recovery from iron phosphate sludge via sulfide addition. After assessing the influence that pH, mixing time, and sulfide dosing rate, it was concluded that: Effective phosphorus recovery from iron phosphate sludge can be achieved by sulfide addition, reaching 70 ± 6% recovery at a S/Fe stoichiometric molar ratio of 1.5 and increasing up to 92 ± 6% as the S/Fe molar ratio increased to 2.5. pH does influence the separation of the solid/liquid phases, reaching fast settling of the iron sulfide particles at pH 4. Better settling characteristics of the FeSx particles at pH 4 were not a consequence of larger particle size. Given the colloidal
We thank Prof. Korneel Rabaey for his valuable input during the early stages of the Project, Mr. David Page for assisting with the particle size and zeta potential analysis and the AWMC Analytical Services Laboratory (ASL) for the ICP analyses. Elena Mejia Likosova thanks The University of Queensland for scholarship support. The Australian Research Council (LP100200122) together with the industry partners Seqwater and Veolia Water funded this work. References [1] D. Cordell, J.-O. Drangert, S. White, Global Environ. Change 19 (2009) 292. [2] F. Fischer, C. Bastian, M. Happe, E. Mabillard, N. Schmidt, Bioresour. Technol. 102 (2011) 5824. [3] A. Weidelener, W. Maier, J. Krampe, Proc. Water Environ. Fed. 2007 (2007) 417. [4] J.D. Doyle, S.A. Parsons, Water Res. 36 (2002) 3925. [5] N.A. Booker, A.J. Priestley, I.H. Fraser, Environ. Technol. 20 (1999) 777. [6] K. Gueney, A. Weidelener, Water Res. 42 (2008) 4692. [7] S. Petzet, B. Peplinski, S.Y. Bodkhe, P. Cornel, Water Sci. Technol. 64 (2011) 7. [8] J. Keeley, P. Jarvis, S.J. Judd, Desalination 287 (2012) 132. [9] D. Petruzzelli, A. Volpe, N. Limoni, R. Passino, Water Res. 34 (2000) 2177. [10] S.A. Parsons, S.J. Daniels, Environ. Technol. 20 (1999) 979. [11] W. Ripl, B. Stammert, U. Wiesmann, M. Karimnia, Vom Wasser 70 (1988) 179. [12] F. Kato, H. Kitakoji, K. Oshita, M. Takaoka, N. Takeda, T. Matsumoto, Water Sci. Technol. 54 (2006) 119. [13] D. Wei, K. Osseo-Asare, Colloids Surf., A 118 (1996) 51. [14] J. Horvath, M. Novak, Corros. Sci. 4 (1964) 159. [15] A.M. Jones, A.N. Pham, R.N. Collins, T.D. Waite, Geochim. Cosmochim. Acta 73 (2009) 2875. [16] A.E. Lewis, R. van Hille, Hydrometallurgy 81 (2006) 197. [17] D. Wei, K. Osseo-Asare, Colloids Surf., A 121 (1997) 27. [18] A.E. Lewis, Hydrometallurgy (2010) 222. [19] R. David, Geochim. Cosmochim. Acta 70 (2006) 5779. [20] D. Rickard, Chem. Geol. 78 (1989) 315. [21] A.H. Nielsen, P. Lens, J. Vollertsen, T. Hvitved-Jacobsen, Water Res. 39 (2005) 2747. [22] D. Firer, E. Friedler, O. Lahav, Sci. Total Environ. 392 (2008) 145. [23] N.G. Harmandas, P.G. Koutsoukos, J. Cryst. Growth (1996) 719. [24] O. Gutierrez, D. Park, K.R. Sharma, Z. Yuan, Water Res. (2010). [25] M.F.M. Bijmans, P.-J. van Helvoort, S.A. Dar, M. Dopson, P.N.L. Lens, C.J.N. Buisman, Water Res. 43 (2009) 853. [26] H.-H. Tung, E.L. Paul, M. Midler, J.A. McCauley, Crystallization of Organic Compounds: An Industrial Perspective, John Wiley & Sons, NJ, 2009. [27] P.K. Dutta, R.A. Rozendal, Z. Yuan, K. Rabaey, J. Keller, Electrochem. Commun. 11 (2009) 1437. [28] A. Rawle, in: Technical Paper Malvern Instruments, Malvern-Instruments, Malvern, 2010, p. 8. [29] Malvern-Instruments, in: DLS Technical Note, Malvern-Instruments, Malvern, 2011. [30] F. doRosario, A.M.T. Louvisse, S.M. Saraiva, G. Gonzalez, J.F. de Oliveira, J. Colloid Interface Sci. 181 (1996) 1. [31] T.F. Tadros, in: T.F. Tadros (Ed.), Colloid Stability: The Role of Surface Forces – Part I, Wiley-VCH, Weinheim, 2007, p. 1.
Contents lists available at SciVerse ScienceDirect
Journal of Colloid and Interface Science www.elsevier.com/locate/jcis
Understanding colloidal FeSx formation from iron phosphate precipitation sludge for optimal phosphorus recovery E. Mejia Likosova a, J. Keller a,⇑, R.A. Rozendal a, Y. Poussade b,c, S. Freguia a a
The University of Queensland, Advanced Water Management Centre (AWMC), St. Lucia, QLD 4072, Australia Veolia Water Australia, Level 15, 127 Creek Street, Brisbane, QLD, Australia c Seqwater, Level 2, 240 Margaret Street, Brisbane, QLD, Australia b
a r t i c l e
i n f o
Article history: Received 17 January 2013 Accepted 3 April 2013 Available online 18 April 2013 Keywords: Colloid Phosphorus recovery Sulfide Zeta potential
a b s t r a c t The use of sulfide to form iron sulfide precipitates is an attractive option for separation and recovery of phosphorus and ferric iron from ferric phosphate sludge generated in wastewater treatment. The key factors affecting the simultaneous generation and separation of iron sulfide precipitates and phosphate solution from ferric phosphate sludge have so far not been thoroughly investigated. This study therefore focuses on the recovery of phosphorus from synthetic sludge by controlled sulfide addition under different operating conditions. The factors that affect the phosphorus recovery, as well as the optimal process conditions to achieve an effective solid–liquid separation, were investigated. The separation of the FeSx particles is a significant challenge due to the colloidal nature of the particles formed. Faster separation and higher phosphorus recovery was achieved when operating at pH 4 with dosing times of at least 1 h. At this pH, phosphorus recovery of 70 ± 6% was reached at the stoichiometric S/Fe molar ratio of 1.5, increasing to over 90% recovery at a S/Fe molar ratio of 2.5. Zeta potential results confirmed the colloidal nature of the iron sulfide precipitate, with the isoelectric point around pH 4, explaining the fast separation of the FeSx particles at this pH. Ó 2013 Elsevier Inc. All rights reserved.
1. Introduction Phosphorus is one of the most important nutrients required for plant growth. Nearly all the phosphorus used in agriculture comes from phosphate rock mines. However, it is estimated that the phosphate left from mining will last at most about 50–100 years [1]. This has been a driving force toward finding renewable sources. The utilization of wastewater as a nutrient source has increased in the last years [2], with digested sewage sludge being a major target as it contains 2–3% of elemental phosphorus. Methods for phosphorus recovery from wastewater sources such as struvite (magnesium–ammonium–phosphate, MAP) formation [3–6] and the recovery of phosphorus and aluminium from sewage sludge ash by sequential elution with acidic and alkaline solutions (SESAL-Phos-recovery process) [7] have been developed. Similarly, methods for coagulant recovery from ferric (Fe3+) and aluminium (Al3+) precipitation sludge have been proposed such as selective removal using ion exchange [8,9] and acidic extraction of coagulant [10]. However, limited work has been reported on simultaneous recovery of phosphate and ferric iron.
⇑ Corresponding author. Fax: +61 733654726. E-mail addresses: [email protected] (E. Mejia Likosova), j.keller@ awmc.uq.edu.au (J. Keller). 0021-9797/$ - see front matter Ó 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.jcis.2013.04.001
The advanced treatment of secondary effluent, e.g. for the production of purified recycled water, often includes a chemical precipitation process, in which coagulants such as FeCl3 are added in order to remove suspended solids, colloids, and phosphate. The sludge formed in this process contains mainly ferric oxy-hydroxides and phosphate, commonly termed ferric phosphate sludge. The simultaneous recovery of phosphate and ferric from ferric phosphate sludge using H2S was first proposed two decades ago [11]. However, in this initial study, no information was provided as to the factors affecting the phosphorus recovery process, i.e. pH, separation method, and performance data such as phosphate and iron recovery efficiencies. In this process, sulfide (H2S, HS, and S2) reacts chemically with iron phosphate, releasing phosphate into solution while precipitating iron sulfide and elemental sulfur [12]:
2FePO4ðsÞ þ 3H2 S ! 2FeSðsÞ þ S0ðsÞ þ 2H2 PO4 þ 2Hþ
ð1Þ
Wei and Osseo-Asare [13] described the formation of iron sulfides and elemental sulfur from the reaction of Fe3+ and HS in aqueous solution via the following reactions:
2Fe3þ þ HS ! 2Fe2þ þ S0 þ Hþ
ð2Þ
Fe2þ þ HS ! FeSðsÞ þ Hþ
ð3Þ
FeSðsÞ þ S0ðsÞ ! FeS2ðsÞ
ð4Þ
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
The formation of the different iron sulfide species strongly depends on the pH, as shown by Wei and Osseo-Asare, who found that in the pH range 3.6–5.7, mostly pyrite (FeS2, Eq. (4)) was produced. They attributed this to the high concentration of the precursors mackinawite (iron monosulfide, Eq. (3)) and elemental sulfur. Additionally, using Energy-Dispersive X-Ray Spectroscopy [EDS] analysis, a strong peak was seen at pH 4 for iron monosulfide and elemental sulfur. This agrees with the Fe/S Pourbaix diagram [14], where FeS exists as a stable compound at pH > 4 and redox potential lower than 0.001 V. Likewise, iron monosulfide formation was seen at pH up to 7.2. However, at pH P 7.2, FeS formation could not compete successfully with c-FeOOH formation, explaining the absence of pyrite as a reaction product [13,15–17]. The pH dependence on the stability of FeSx and soluble sulfur species was also observed by others [18–20]. In a more recent study, Kato et al. [12] evaluated phosphate recovery methods using sulfide based on the process proposed by Ripl et al. [11]. The focus was on maximizing phosphate recovery efficiencies and to validate the method itself. Experiments with pre-coagulated sewage sludge and FePO4 synthetic sludge were therein performed in the pH range of 5.3–6.9 and at S/Fe molar ratios between 1.0 and 2.0, leading to a phosphorus recovery up to 44% from the pre-coagulated sludge and 92.8% from the synthetic FePO4 after centrifugation. Even though the proposed phosphorus recovery process via iron sulfide generation has been further studied by others [12,21,22], the selected solid–liquid separation processes, e.g. precipitation and centrifugation [11,12] present significant challenges for the recovery efficiency and operating costs due to the poor settling characteristics of FeSx particles. When in contact with sulfide, phosphate is released into the solution and a black material is formed. The composition of this black material has been found to be a mixture of FeS and FeS2 [13]. Its exact composition depends on the pH. Fe(HS)+ was also found in solution in the pH range of 3.6–7.8 [13,19,20,23]. Despite the proven fast release of phosphate into solution, the separation of the formed suspension has been a challenge. At neutral pH or higher, the precipitated black material remains in suspension for more than a week and passes through a 0.22 lm filter, as preliminary tests in our lab have shown (data not shown). Among the parameters that may affect the solid–liquid separation in this system, mixing time in the reaction vessel and dosing
17
rate of sulfide have been previously identified as key factors for iron sulfide precipitation in different systems. Gutierrez et al. [24] demonstrated that the hydraulic retention time (HRT) of iron sulfide in sewers crucially impacts the formation of iron sulfide precipitates during primary settling, i.e. at shorter HRT a higher concentration of iron sulfide in the primary effluent was found. These findings suggest that mixing time in the reaction vessel is a crucial factor to enable effective separation of the formed iron sulfides. Metal sulfide salts have low solubility. Thus, supersaturated solutions by interaction of aqueous sulfide and metal are expected [16]. Under these supersaturation conditions, fast nucleation is expected to limit crystal growth, thus generating poor settling characteristics of the formed metal sulfides [18,23,25]. At a microscopic level, the dosing rate and mixing speed are key factors in the nucleation process, i.e., slow dosing rates and mixing speed are expected to hinder nucleation and enhance instead particle growth and thus precipitation [26]. Even though considerable effort has been placed on improving the efficiency of this process, little systematic research has been carried out to determine the factors that affect phosphate recovery and separation of the formed FeS colloidal particles. The aim of this research was to identify and characterize key factors that influence phosphate recovery and to determine the optimal process conditions to achieve an effective solid–liquid separation. In particular, the effects of pH, mixing time, sulfide dosing rate and settling time have been investigated in detail in this study. In order to elucidate the reaction chemistry and to understand the mechanisms of precipitation of the inorganic iron sulfide particles, most experiments in this work were carried out using synthetic FePO4 sludge, as a way to eliminate potential interference of organic solids, which are invariably present in ferric phosphate sludges obtained from wastewater treatment. 2. Materials and methods 2.1. Preparation of synthetic sludge and solutions In order to elucidate the stoichiometry of the iron phosphate reaction with sulfide and to determine the corresponding S/Fe ratio needed to achieve completion of the precipitation reaction, studies using synthetic iron phosphate suspensions were performed.
Fig. 1. Synthetic FePO4 sludge experimental setup: 1 – magnetic stirrer; 2 – magnetic bar; 3 – pH controller; 4 – syringe pump for sulfide addition; 5 – gasbag; 6 – acid addition pump; 7 – sampling port; and 8 – data logger.
18
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
A 0.1 M ferric phosphate suspension was prepared from FePO44H2O (97% pure from Sigma–Aldrich) to simulate the real sludge. Based on a typical Fe3+ concentration in real sludge, a 0.1 M suspension was prepared by mixing the salt in water deionized by reverse osmosis (RO). Sodium sulfide nonahydrate (Na2S9H2O, reagent grade) was purchased from Sigma–Aldrich. The Na2S9H2O crystals were previously washed with RO water to remove oxidized sulfur species at the surface of the crystals as described elsewhere [27]. A 0.8 M sodium sulfide solution was prepared with RO water purged with nitrogen to achieve anaerobic conditions. A 3 M HCl solution was used to control the pH in the precipitation vessel. 2.2. Reactor design Iron sulfide was precipitated in a 250 mL three-neck modified Schott bottle (Fig. 1). During the experiments, 100 mL of the FePO4 suspension was stirred using a magnetic stirrer at 2000 RPM, while a syringe pump added the sodium sulfide solution through a needle inserted at the bottom of the reactor near the magnetic bar, at a rate ranging from 18 to 113 mL/h. This method of addition ensures low local sulfide concentrations throughout the vessel, thus putatively minimizing nucleation and enhancing the iron sulfide particle growth. During the reaction, the pH was controlled using a pH meter/controller (Endress + Hauser Liquisys M CPM223/253) at a set point between 4.0 and 7.0 ± 0.2. HCl (3 M) was dosed with a peristaltic pump to control the pH. A gas collection bag was attached to the reactor in order to capture any gaseous sulfide that may evolve upon sulfide addition, especially during low pH experiments. The pH in the reactor was recorded using a data acquisition unit. 2.3. Sulfide addition experiments According to the iron sulfide precipitation reaction, 2 moles of iron react with 3 moles of sulfide as described by Eq. (1) (in pH range 3–6). This implies a stoichiometric S/Fe ratio of 1.5. Taking this into account, two different types of experiments were designed at different conditions and performed in triplicates. Table 1 shows the experimental conditions. To determine the required S/Fe ratio to achieve complete reaction, as well as to understand the fate of any excess sulfide added, a first set of experiments was proposed. In these experiments (all run at pH 4.0 ± 0.2), sodium sulfide was added in excess (up to S/Fe molar ratio of 2.5) in order to guarantee the completion of the reaction. The sulfide addition was done at a rate of 18.8 mL/h, reaching the S/Fe stoichiometric molar ratio of 1.5 after 1 h of addition. The subsequent ratios (1.75, 2, 2.25, and 2.5) were reached after further 10 min addition each (total addition time 100 min). 4-mL samples were taken at S/Fe molar ratios
of 1.5, 1.75, 2, and 2.5, split into two 2 mL Eppendorf tubes and left in quiescent conditions. After 1 h, the samples were centrifuged at 2147g for 10 min in a bench-top centrifuge in order to guarantee complete separation of the two phases. Analyses of the supernatant and solids were performed as described in Section 2.6. In the second set of experiments, sodium sulfide and iron phosphate solution reacted in stoichiometric S/Fe molar ratio at different pH values (6, 5 and 4 ± 0.2), dosing times (10, 30 and 60 min), and further mixing time after sulfide addition (1, 2 and 3 h). Following collection of 6 mL samples, these were placed in 10 mL plastic vials and left in quiescent conditions to allow for sedimentation of the solid phase (up to 24 h settling time), and analysis of the liquid and solid phases was performed as described in Sections 2.4 and 2.6. The error bars shown in the figures correspond to the standard deviation (SD) after performing the experiments in triplicates (n = 3). 2.4. Turbidity measurements In order to evaluate the rate of sedimentation, a turbidity test of the supernatant in the second set of experiments (Section 2.3) was performed every 5 min on the quiescent 6 mL samples using a LaMotte 2020 we/wi turbidimeter. 2.5. Particle size and zeta potential Two methods were used to measure the particle size of the iron sulfide suspensions: laser diffraction (Mastersizer 2000, Malvern Instruments) and dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments). Laser diffraction (also known as low angle laser light scattering) relies on the fact that the diffraction angle is inversely proportional to particle size [28]. This method is applicable for a particle size range between 0.1 and 3000 lm. This measurement assumes the volume of the particle as the important particle property, giving a volume/mass mean particle size distribution. Dynamic light scattering measures the speed at which the particles are diffusing due to the Brownian motion [29]. The rate of the Brownian motion will depend on the size of the particles, i.e., small particles fluctuate more rapidly than the larger ones. The Zetasizer Nano ZS measures the light intensity fluctuation of the particles and correlates them with a size measurement of a similar known particle, giving an intensity size distribution. The Zetasizer Nano ZS has accuracy in the range of 1 nm–10 mm. Along with the particle size, the Zetasizer Nano ZS also measures the zeta potential of the particles. Given the high density of the FeSx sludge suspension, a 500 times dilution was performed with RO water when using the Zetasizer Nano ZS. For the zeta potential measurement of the remaining particles in suspension after 24 h settling, a 20 times
Table 1 Experimental conditions and reagents concentrations. Initial Fe, P and S amounts and concentrations
pH
S/Fe molar ratio
Time to reach S/Fe molar ratio (min)
Na2S dosing rate (mL/h)
Experiment 1: iron sulfide precipitation reaction at pH 4 and sulfide overdose up to S/Fe molar ratio of 2.5 Fe (mol) 0.01 4 1.5 60 18.8 P (mol) 0.01 1.75 70 Vol. FePO4 (L) 0.1 2 80 Na2S (mol/L) 0.8 2.25 90 2.5 100 Experiment 2: iron sulfide precipitation reaction at different pH and dosing times, further mixing and sulfide addition up to S/Fe stoichiometric molar ratio of 1.5 Further mixing times (h) Sulfide dosing time (min) Fe (mol) 0.01 4 0, 1, 2 and 3 60 18.8 P (mol) 0.01 0 30 37.6 Vol. FePO4 (L) 0.1 0 10 112.8 Na2S (mol/L)
0.8
5 6
1, 2 and 3 1, 2 and 3
60 60
18.8 18.8
19
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
2.6. Chemical analysis The characterization of the liquid and the solid phases after the addition of sulfide was performed for each sample after centrifuging at 2147g for 10 min and decanting the liquid phase with a Pasteur pipette. Total phosphorus, sulfur and iron were determined using Inductively Coupled Plasma–Optical Emission Spectroscopy (ICP–OES, Perkin Elmer Optima 3300DV) analysis for both liquid and solid phases. The determination of total iron, sulfur, and phosphorus in the solid phase was achieved by acidic digestion of the dry solids obtained after drying the samples overnight at 105 °C. In all experiments, the fraction of the reacted sulfide was calculated as the ratio of the total moles of the element in the system (liquid plus solid phase) minus the moles of the element in the liquid phase over the total moles in the system %Sreacted ¼
Total moles of S in the system Moles of S in the liquid phase Total moles of S in the system 100%
2.5 2.0
S/Fe molar ratio
dilution was performed. The pH adjustment after dilution of the sludge was performed immediately before the measurement in the Zetasizer in order to avoid the settling of the particles before measurement. Either 3 M HCl or 2 M NaOH were used as required.
1.5 1.0 0.5 0.0 60
70
80
90
100
Time [min] S/Fe added
S/Fe solid phase
S/Fe system
S/Fe liquid phase
Fig. 2. Sulfur balance measured by ICP in the system (sum of solid and liquid phases) after sulfide overdose up to S/Fe molar ratio of 2.5 at pH 4. Sulfur concentrations are normalized to the total Fe in the system (constant) (n = 3). Solid line: S/Fe molar ratio added; squares and dotted line: S/Fe measured in the system (sum of S/Fe in the solid and liquid phases); triangles and solid line: S/Fe molar ratio in the solid phase; and dots and dashed line: S/Fe molar ratio in the liquid phase.
ð5Þ
Moles of P in liquid 100% Total moles of P in system
ð6Þ
The change of volume in the system due to sampling and sulfide addition was taken into account in each of the calculations.
100
100
90
90
80
80 %S [mass]
70
70
%P [mass]
60
%P recovered
%Precovered ¼
%S reacted
Similarly, the recovered phosphorus in the supernatant was calculated as the ratio of the P moles in the liquid phase over the total moles in the system
60
3. Results and discussion 3.1. Reaction stoichiometry and phosphorus recovery
50 1.50
50 1.75
2.00
2.25
2.50
S/Fe molar ratio Based on the stoichiometry (Eq. (1)), it is expected that 1.5 moles of sulfide react with 1.0 mole of FePO4. This was confirmed after overdosing sulfide up to a S/Fe ratio of 2.5, where the S/Fe molar ratio reached in the solid phase was approximately 1.5 regardless of sulfide overdosing, thus indicating full reaction of Fe, according to reactions (2)–(4) (Fig. 2). The measurements reveal some missing sulfur from the liquid phase at higher S/Fe ratios, possibly due to some loss as H2S during the acidic digestion procedure as required for ICP measurement. At pH 4 and continuous mixing, phosphorus recovery of 70 ± 6% was obtained with 1-h sulfide addition time up to S/Fe stoichiometric molar ratio. However, the phosphorus recovery increased up to 92 ± 6% as the S/Fe molar ratio increased to 2.5 (Fig. 3). This result agrees with the findings of Kato et al. [12]. However, a much faster separation of the solid and liquid phases was reached without the use of the centrifuge at pH 4, enhancing the efficiency of the process (Section 3.2). 3.2. Effect of pH on solids precipitation In the second set of experiments (Section 2.3), i.e., at the stoichiometric S/Fe molar ratio of 1.5 with differing pH levels of 6, 5, and 4, respectively, no separation of the liquid and solid phases was reached at pH 6 after 1 h settling (post sulfide addition). However, a more defined solid/liquid interface appeared as the pH became more acidic, i.e., pH 5 and 4. At pH 4, a clear separation of the phases was seen after only 9 min of sedimentation.
Fig. 3. Percentage of S reacted and P recovered in solution at pH 4 and different S/Fe molar ratios (up to 2.5). Solid line: %S reacted (present in solid phase); and dotted line: %P recovered (in liquid phase).
Turbidity analysis of the iron sulfide suspension as a function of the settling time revealed a much faster sedimentation rate of the iron sulfide particles at pH 4 compared to pH 6 (Fig. 4). After only 1 h, the resulting supernatant at pH 4 was significantly clearer than the one obtained after 1 day of settling at pH 6. In order to determine the reasons for the improved settleability of the iron sulfide particles (FeSx) at pH 4, an assessment of their particle size distribution (PSD) at different pH was performed using laser diffraction (Mastersizer 2000, Malvern Instruments) and dynamic light scattering (Zetasizer Nano, Malvern Instruments), as described in Section 2.5. Contrary to the expectations, as the pH became more acidic, the FeSx particle size decreased, i.e., the volume-based median particle size using laser diffraction was 49 ± 2 lm, 46 ± 2 lm, and 28 ± 2 lm for pH 6, 5, and 4, respectively. These particles are far larger than typical colloids, which may reflect the fact that they are the result of aggregation, a phenomenon that is well described in the literature for FeS particles and can lead to the increase in the observed particle size from 2– 10 nm (nanoclusters) to 1 lm (colloids) [18], and even up to 50 lm [30]. Results using dynamic light diffraction technique indicated a similar decreasing trend as the pH became more acidic. However, at pH 4, an increase in the particle size (higher than at
20
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
cles (presumed to be elemental sulfur according to Eq. (1)) remained in suspension in the supernatant. Colloidal particles typically hold a negative surface charge, which prevents aggregation and thus sedimentation. Therefore, the stability of such colloid systems depends on the balance of various interaction forces such as van der Waals attraction, double-layer repulsion and steric interaction [31]. Consequently, if the particles have a strong repulsion, the suspension will be stable. In Fig. 5, the zeta potential of the white particles in the supernatant at different pH values is shown. The point of zero charge of these particles was found to be between pH 2.5 and 4.2 (likely somewhat below pH 4), explaining their limited settleability and thus continued presence in suspension at this pH. These results largely agree with the findings of Wei and Osseo-Asare [13], who reported the isoelectric point of FeS2 at pH 3.
250
Turbidity [NTU]
200
150
100
pH 4 50
pH 6
0 0
20
40
60
80
100
120
1440
Settling Time [min] Fig. 4. Iron sulfide settling behavior at different pH levels after 1 h reaction time: turbidity analysis. Dots: pH 4 and squares: pH 6. A faster precipitation of the FeSx particles and much clearer supernatant was reached at pH 4 after 1 h sedimentation time compared to pH 6.
pH 6) was seen. This could be a result of the different stirring conditions and iron sulfide complexes present in the samples, i.e., samples are vigorously stirred in the laser diffraction technique compared to dynamic light diffraction (not stirred), hence allowing coagulation of the nanoclusters to larger particles with the latter technique. Fig. 5 shows the zeta potential of the FeSx particles at different pH. It can be seen that as the pH became more acidic, the zeta potential of the FeSx particles decreases to zero at around pH 4. This point of zero charge is known as the isoelectric point and the pH at which the suspension is least stable; thus, sedimentation of the particles occurs. Hence, the colloidal nature of the FeSx particles was confirmed, with an isoelectric point around pH 4. It can be deduced that the better settling characteristics of FeSx particles at lower pH were not a consequence of larger particle size. Fast precipitation at pH 4 is thus a consequence of the particles becoming neutral. The increase in the particle size at pH 4 as observed with the dynamic light diffraction technique could be explained by the occurrence of two phenomena: the aggregation of the FeSx nanoclusters [18] (which would scatter more light and therefore a larger particle size would be determined) and/or the formation of pyrite (FeS2) particles, which has been reported to happen between pH 3.6 and 5.7 [13]. Despite the visual differentiation of the two phases as the pH became more acidic, some black FeSx [13] and white-colored parti-
3.3. Effect of dosing time and further mixing time on reaction rate and solids precipitation The effect of different reaction times and further mixing after completion of the iron sulfide formation (S/Fe stoichiometric molar ratio 1.5) was studied after performing the second set of experiments described in Section 2.3 (Table 1) at pH 4. No significant change in the reacted iron sulfide was seen at different dosing rates, indicating the fast rate of particle formation and confirming the observations that nucleation and particle size are not key factors in achieving efficient phase separation here. However, a better phosphorus recovery (68 ± 9% against 49 ± 9%) was obtained with 1 h compared to 30 min dosing time at pH 4, suggesting a possible mass transfer or reaction rate limitation in the phosphate solubilization process at higher sulfide dosing rates. The settling times of the iron sulfide particles at the end of the sulfide addition for different dosing rates were also measured, as shown in Fig. 6. A slightly faster settling of the iron sulfide particles was seen after 1 h reaction time. Consequently, this indicates that dosing times of at least 1 h appear to achieve faster separation rates and higher phosphate recovery. No enhancement of the solid/liquid phase separation was achieved after further mixing at pH 6, 5, and 4. Contrary to the findings reported by Gutierrez et al. [24], further mixing times did not improve the settling of the FeSx particles, i.e., no clearer separation of the solid/liquid phases was reached with further mixing after completion of the iron sulfide reaction. On the contrary, a better separation of the two phases was reached when the sludge was allowed to settle right after completing the addition of sulfide in all cases. Only at pH 6 and at all mixing times, poor phase
250
30 min dosing time
White particles in supernatant
20
200
0 2
4
6
8
10
-20
Turbidity [NTU]
40
Zeta Potential [mV]
60 min dosing time
FeSx Particles
10 min dosing time
150 100 50
-40
0 0
-60
pH
Fig. 5. Dots: FeSx particles zeta potential at different pH using the Zetasizer and squares: S0 particles zeta potential at different pH.
30
60
90
120
1440
Settling Time [min] Fig. 6. Iron sulfide settling time at different dosing times at pH 4 based on turbidity analysis.
E. Mejia Likosova et al. / Journal of Colloid and Interface Science 403 (2013) 16–21
nature of the FeSx particles formed, charge neutralization of the FeSx particles at pH 4 (as shown by zeta potential based determination of the isoelectric point) appears to be the dominant factor in achieving efficient coagulation and settling of the particles.
100
P recovered in solution [%]
21
80 60 40 20 0 1
1.2
1.4
1.5
1.75
S/Fe molar ratio Fig. 7. Percentage of phosphorus recovered in solution from real ferric phosphate sludge at pH 4. S/Fe stoichiometric molar ratio of 1.5 was reached after 1 h reaction time.
Achieving an efficient precipitation and solid/liquid separation of iron sulfide sludge is important to enable processing of concentrated iron phosphate sludge, hence achieving a maximal concentration of the phosphate in the supernatant. This stream could most beneficially be re-used in a struvite precipitation process, which is typically phosphate limited, or even directly introduced into chemical fertilizer manufacturing. In both cases, achieving a high phosphate concentration and effective separation from the FeSx precipitants will be important. Additionally, the iron sulfide sludge can offer attractive avenues for iron recovery via sulfide oxidation to sulfur (or sulfate), hence creating further valuable opportunities for effective chemical recovery from this process. Acknowledgments
separation was achieved, even after 24 h sedimentation. Effectively, no change in the phosphate recovery, as well as a possibly slightly negative effect in the iron recovery, was seen with extended mixing times (data not shown). 3.4. Phosphorus recovery with real precipitation sludge In order to evaluate the robustness of the optimized process herein proposed, preliminary sulfide addition experiments (see Section 2.3) with real ferric phosphate sludge (with iron and phosphorus concentration of 10.6 and 0.7 g/L, respectively) from a local full-scale advanced water treatment plant (AWTP) were performed up to S/Fe molar ratio of 1.7 at pH 4. At the S/Fe stoichiometric molar ratio of 1.5, up to 63% phosphorus was recovered in solution (Fig. 7). As expected, better phosphorus recovery was reached at higher S/Fe molar ratio, i.e., up to 75% P was recovered at S/Fe molar ratio of 1.7. These results are quite similar those with synthetic sludge (reported above) and therefore confirm the efficiency of the optimized iron sulfide precipitation process for phosphorus recovery from ferric phosphate sludge. The point of zero charge of the FeSx particles formed from real ferric phosphate sludge was found to be slightly below pH 4 (data not shown), which supports the general applicability of the separation method. However, specific optimization of the process may be required for any given ferric sludge based on the particular composition in each situation. 4. Conclusions The aim of this research was to determine the optimal process conditions to achieve an effective solid–liquid separation and phosphate recovery from iron phosphate sludge via sulfide addition. After assessing the influence that pH, mixing time, and sulfide dosing rate, it was concluded that: Effective phosphorus recovery from iron phosphate sludge can be achieved by sulfide addition, reaching 70 ± 6% recovery at a S/Fe stoichiometric molar ratio of 1.5 and increasing up to 92 ± 6% as the S/Fe molar ratio increased to 2.5. pH does influence the separation of the solid/liquid phases, reaching fast settling of the iron sulfide particles at pH 4. Better settling characteristics of the FeSx particles at pH 4 were not a consequence of larger particle size. Given the colloidal
We thank Prof. Korneel Rabaey for his valuable input during the early stages of the Project, Mr. David Page for assisting with the particle size and zeta potential analysis and the AWMC Analytical Services Laboratory (ASL) for the ICP analyses. Elena Mejia Likosova thanks The University of Queensland for scholarship support. The Australian Research Council (LP100200122) together with the industry partners Seqwater and Veolia Water funded this work. References [1] D. Cordell, J.-O. Drangert, S. White, Global Environ. Change 19 (2009) 292. [2] F. Fischer, C. Bastian, M. Happe, E. Mabillard, N. Schmidt, Bioresour. Technol. 102 (2011) 5824. [3] A. Weidelener, W. Maier, J. Krampe, Proc. Water Environ. Fed. 2007 (2007) 417. [4] J.D. Doyle, S.A. Parsons, Water Res. 36 (2002) 3925. [5] N.A. Booker, A.J. Priestley, I.H. Fraser, Environ. Technol. 20 (1999) 777. [6] K. Gueney, A. Weidelener, Water Res. 42 (2008) 4692. [7] S. Petzet, B. Peplinski, S.Y. Bodkhe, P. Cornel, Water Sci. Technol. 64 (2011) 7. [8] J. Keeley, P. Jarvis, S.J. Judd, Desalination 287 (2012) 132. [9] D. Petruzzelli, A. Volpe, N. Limoni, R. Passino, Water Res. 34 (2000) 2177. [10] S.A. Parsons, S.J. Daniels, Environ. Technol. 20 (1999) 979. [11] W. Ripl, B. Stammert, U. Wiesmann, M. Karimnia, Vom Wasser 70 (1988) 179. [12] F. Kato, H. Kitakoji, K. Oshita, M. Takaoka, N. Takeda, T. Matsumoto, Water Sci. Technol. 54 (2006) 119. [13] D. Wei, K. Osseo-Asare, Colloids Surf., A 118 (1996) 51. [14] J. Horvath, M. Novak, Corros. Sci. 4 (1964) 159. [15] A.M. Jones, A.N. Pham, R.N. Collins, T.D. Waite, Geochim. Cosmochim. Acta 73 (2009) 2875. [16] A.E. Lewis, R. van Hille, Hydrometallurgy 81 (2006) 197. [17] D. Wei, K. Osseo-Asare, Colloids Surf., A 121 (1997) 27. [18] A.E. Lewis, Hydrometallurgy (2010) 222. [19] R. David, Geochim. Cosmochim. Acta 70 (2006) 5779. [20] D. Rickard, Chem. Geol. 78 (1989) 315. [21] A.H. Nielsen, P. Lens, J. Vollertsen, T. Hvitved-Jacobsen, Water Res. 39 (2005) 2747. [22] D. Firer, E. Friedler, O. Lahav, Sci. Total Environ. 392 (2008) 145. [23] N.G. Harmandas, P.G. Koutsoukos, J. Cryst. Growth (1996) 719. [24] O. Gutierrez, D. Park, K.R. Sharma, Z. Yuan, Water Res. (2010). [25] M.F.M. Bijmans, P.-J. van Helvoort, S.A. Dar, M. Dopson, P.N.L. Lens, C.J.N. Buisman, Water Res. 43 (2009) 853. [26] H.-H. Tung, E.L. Paul, M. Midler, J.A. McCauley, Crystallization of Organic Compounds: An Industrial Perspective, John Wiley & Sons, NJ, 2009. [27] P.K. Dutta, R.A. Rozendal, Z. Yuan, K. Rabaey, J. Keller, Electrochem. Commun. 11 (2009) 1437. [28] A. Rawle, in: Technical Paper Malvern Instruments, Malvern-Instruments, Malvern, 2010, p. 8. [29] Malvern-Instruments, in: DLS Technical Note, Malvern-Instruments, Malvern, 2011. [30] F. doRosario, A.M.T. Louvisse, S.M. Saraiva, G. Gonzalez, J.F. de Oliveira, J. Colloid Interface Sci. 181 (1996) 1. [31] T.F. Tadros, in: T.F. Tadros (Ed.), Colloid Stability: The Role of Surface Forces – Part I, Wiley-VCH, Weinheim, 2007, p. 1.