Article Cite This: Environ. Sci. Technol. XXXX, XXX, XXX−XXX
pubs.acs.org/est
Inducing in Situ Crystallization of Vivianite in a UCT-MBR System for Enhanced Removal and Possible Recovery of Phosphorus from Sewage Jingbao Tian,† Xiang Cheng,*,†,‡ Shaoyu Deng,† Jiaqi Liu,† Bin Qiu,† Yan Dang,† Dawn E. Holmes,§ and Trevor David Waite‡ †
Beijing Key Laboratory for Source Control Technology of Water Pollution, Beijing Forestry University, Beijing 100083, China Water Research Center, School of Civil and Environmental Engineering, University of New South Wales, Sydney, NSW 2052, Australia § Department of Physical and Biological Sciences, Western New England University, 1215 Wilbraham Rd, Springfield, Massachusetts 01119, United States
Downloaded via UNIV OF SOUTHERN INDIANA on July 26, 2019 at 03:30:57 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
‡
S Supporting Information *
ABSTRACT: By mimicking iron(Fe)-based phosphorus (P) immobilization in natural environments, an Fe-retrofitted UCT-MBR involving in situ vivianite crystallization for removing and recovering P from sewage was developed, and its performance was examined in this work. We show that dosing of ferrihydrite, once biological P uptake reached its limit, enabled effective ongoing P removal; whereas conventional conditions in the anaerobic chamber of the University of Cape Town (UCT) system (i.e., a sludge retention time of hours and a completely mixed sludge phase) was insufficient for a satisfactory Fe(III) bioreduction, with the overaccumulation of Fe(III) as fine particles finally resulting in severe membrane fouling and collapse in P removal. The enhancement of reductive conditions in the anaerobic chamber by lowering agitation and adding biocarriers to favor Fe(III) reduction was found to be effective in enabling ongoing P removal and recovery. The average level of effluent P was as low as 0.18 mg/L for a period of 258 d under this condition. Using chemical and spectroscopic methods, the P product was identified as primarily vivianite: Fe3(PO4)2·8H2O. The in situ crystallization of vivianite as a sink for P enabled the UCT-MBR to continuously remove and recover sewage P with no need for sludge discharge.
1. INTRODUCTION Phosphorus (P) removal and recovery from urban sewage is highly desirable, not only for the protection of downstream aquatic ecosystems from eutrophication but also because the global phosphorus reserve is depleting.1,2 In municipal wastewater treatment plants (WWTPs), phosphorus can be removed from water by biological processes, chemical precipitation, or crystallization.3 By employing polyphosphate-accumulating organisms (PAOs), enhanced biological P removal (EBPR) processes were developed to remove P from water in a fairly cost-effective and environmentally sustainable way.4 The University of Cape Town (UCT) process is a successful example, whereby competition between heterotrophic PAOs and denitrifiers is reduced and satisfactory P removal is achieved. In recent years, the integration of biological nutrient removal processes with membrane bioreactors (MBR) has received considerable interest in view of the potential to further improve effluent quality and reduce sludge production and footprint of wastewater treatment systems.5 However, all the EBPR-centered processes merely © XXXX American Chemical Society
transfer phosphorus from the water phase to activated sludge (AS). For recovery purposes, additional treatments of the Prich AS are needed and frequently include a biological, physical, or chemical treatment for phosphate release and a following step of precipitation/crystallization.6,7 Chemical phosphate precipitation was introduced in the 1960s as one of the earliest measures to reduce the phosphorus load being discharged from wastewaters.8 Many WWTPs currently add iron or aluminum salts to the AS system to reduce P levels in order to comply with increasingly strict discharge limits.9 For example, the USEPA limits total P discharge in streams entering lakes and flowing waters to 0.05 and 0.1 mg/L, respectively.10 The precipitation of phosphate efficiently removes P from water, whereas separating and recovering P from the resulting chemically modified activated Received: Revised: Accepted: Published: A
February 17, 2019 May 16, 2019 June 28, 2019 June 28, 2019 DOI: 10.1021/acs.est.9b01018 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 1. Schematic of the vUCT-MBR system.
feasibility of inducing vivianite crystallization in the P-rich anaerobic AS unit. For this purpose, Fe(III) reduction in the system and the resulting variation in P distribution are investigated. The influences of iron amendments on the sludge properties and the overall performance of vUCT-MBR are then discussed.
sludge remains a major challenge. In order to solve this problem, crystallization of phosphate for obtaining purer, larger-sized, and thus separable grains was then widely studied and, in some places, demonstrated in pilot and full scales;11 the most popular product has been struvite (magnesium ammonium phosphate, MAP), and in some other cases, hydroxyapatite (calcium phosphate, HAP) was also considered.12,13 More recently, Cheng et al. investigated the crystallization of vivianite [Fe3(PO4)2·8H2O] as a product of phosphate recovery in anaerobic engineered environments such as septic systems and anaerobic sludge digesters.14,15 Compared with the MAP approach, vivianite-based P recovery would greatly save chemical costs by substituting Mg2+ with Fe2+, especially when Fe-containing industrial wastes can be used as Fe2+ sources. For industrial wastewaters with no ammonium or inadequate ammonium for MAP crystallization, the vivianite approach can be even more attractive. In fact, vivianite precipitates have been found in some iron-dosing WWTPs as one of the major P sinks.16,17 Iron salts in both Fe(II) and Fe(III) forms have been widely used in wastewater treatment for multiple reasons, e.g., to remove COD and P, control the odor of hydrogen sulfide (H2S), and increase sludge settling and dewatering properties.18−22 Iron has a significant influence on the cycling of elements in nature and is therefore expected to play an important role in the recovery of energy (C) and nutrients (e.g., N and P) from sewage in future WWTPs.23 For the recovery of P, the vivianite pathway would be a promising alternative which requires no chemical input or less additional chemical input (than alternatives in operation). Despite the fact that vivianite is recognized as a P sink in natural sediments and water treatment systems,24,25 research and practice of the vivianite pathway for P immobilization and recovery have been rare. The feasibility of incorporating vivianite crystallization into existing sewage biotreatment processes and the resulting effects on the systems is far from clear. In this study, an Fe(III)-amended UCT-type anaerobic/ anoxic/oxic-MBR system (vUCT-MBR with “v” describing vivianite as the expected P product) is developed to explore the
2. EXPERIMENTAL SECTION 2.1. Seed Sludge and Iron Source. Seed sludge was collected from the sludge recirculation line in Qinghe WWTP, Beijing, China. Ferrihydrite was used as an external iron source in this study with preparation following the method described by Cornell and Schwertmann.26 Briefly, KOH was added to 0.1 M of Fe(NO3)3 dropwise with vigorous stirring but avoiding pH >7.5. Iron content in the resulting slurry was determined after solubilizing the sample in 3 M HCl. 2.2. Experimental Setup. A bench-scale UCT-MBR system consisting of anaerobic (2 L), anoxic (2 L), and aerobic (2 L×4) chambers with UCT configuration, and a chamber (2 L) containing a submerged polyvinylidene fluoride (PVDF) hollow fiber membrane module (nominal pore size: 0.1 μm; membrane area: 0.2 m2) was set up in this study (Figures 1 and S1 of the Supporting Information (SI)). An oxygen depletion chamber (ODC, 2 L) was added to the mixed liquor recirculation line to allow oxygen stripping before the liquor entered the anoxic chamber for denitrification. The DO level in the aerated chambers increased from ∼3 to ∼6 mg/L along the AS flow. Coarse bubbling was provided in the membrane chamber by an air diffuser to control membrane fouling and to avoid the formation of bottom dead zones. By using a peristaltic pump (Longer, China), synthetic sewage was introduced to the UCT-MBR system as feed at a flow rate of 1 L/h, giving a total hydraulic retention time (HRT) of 14 h. The effluent was extracted from the membrane using a peristaltic pump in a mode of 9 min process−1 min relaxation. Trans-membrane pressure (TMP) was continuously measured by a recording system (Meacon, China) consisting of a pressure transducer (MIK-P300) and a data collector (MIK-RX200D). To maintain the permeate flux, the membrane was backwashed daily for 1 h (or longer if poor B
DOI: 10.1021/acs.est.9b01018 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
Figure 2. Performance of vUCT-MBR process: COD (a), TN (b), P (c), and TMP (d).
filtration performance was evident) by pumping a volume of permeate back through the membrane module from the effluent tank. During the start-up period, the UCT-MBR system was inoculated with 8 L of seed sludge for an initial level of suspended solids (SS) of 4800 mg/L. Synthetic sewage (composition given in SI Text S1) was prepared and used as the feed with a resulting COD of ∼400 mg/L, total nitrogen (TN) of 40 mg/L, and total phosphorus (TP) of 3.4 mg/L. The system was operated for 500 d that was divided into three periods (R1, R2, and R3; SI Table S1). Ferrihydrite was supplemented in periods R2 and R3 at 9.2 mg-Fe/L for an Fe/ P molar ratio in the feed of 1.5, which is the same as vivianite. The Fe-amended system for the purpose of inducing vivianite formation is then referred to as vUCT-MBR. In R3, 40 1 cm luffa cubes as biocarriers were added to and retained in the anaerobic chamber for the purpose of establishing a separated sludge phase with an extended sludge retention time (SRT). The presence of the luffa supports together with a decrease in agitation in the chamber by using a small blade (SI Figure S1) were expected to enhance the anaerobic bioreduction of Fe(III). During the entire experiment, no sludge was discharged except what was used for analysis. 2.3. Water Quality Measurement. Concentrations of COD, N species (TN, NH4+-N, NO2−-N, and NO3−-N), and P species (TP and PO4−P) were examined using EPA approved Hach testing reagents with a DR3900 spectrophotometer and a DRB-200 reactor when sample digestion was required (Hach, U.S.A.). Aqueous fractions were determined after filtering samples through 0.45 μm poly(ether sulfone) membranes (Membrana, Germany) and were marked “aq”, while the total amount in sludge mixture was marked “mx”. 2.4. Iron Measurement. The concentration of Fe(II) was measured using a Hach testing reagent based on Standard Methods.27 Total Fe (Fe(T)) was measured using the Hach
testing reagent that contained hydroxylamine hydrochloride for reducing Fe(III) to Fe(II). After sampling and filtration, water samples for aqueous Fe(II) measurement were immediately acidified to pH 3−5 to minimize Fe(II) oxidation. 2.5. Phosphorus Speciation. 2.5.1. SMT Method. The Standards, Measurements, and Testing protocol (SMT, by European commission) was adopted to determine five P fractions, namely, TP, inorganic phosphorus (IP), organic phosphorus (OP), apatite phosphate (AP), and nonapatite inorganic phosphate (NAIP).28 The procedure involved in the SMT analysis is provided in SI Text S2. 2.5.2. 31P NMR Analysis. Solution-state 31P NMR analysis was undertaken on P present in the AS. The sample preparation and P extraction procedure were modified from previous literature with the details given in SI Text S3.29 The NMR spectrum was obtained using a JNM-ECA600 spectrometer (JEOL, Japan) operated at 243 MHz at 25 °C. 2.6. Characterization of Sludge and Phosphorus Recovery Product. The morphology of sludge flocs was recorded using a Smart320 digital microscope (Optec, China). The precipitates collected from the anaerobic chamber of the vUCT-MBR were analyzed by stereomicroscopy (Leica S6D with Leica DFC450 camera). The suspended sludge and precipitates in the anaerobic chamber were further examined by scanning electron microscopy−energy dispersive spectroscopy (SEM−EDS) on a Merlin field emission microscope (Zeiss, Germany) equipped with an Oxford X-Max EDS detector to characterize the P products. The procedure of sample preparation for SEM examination is given in SI Text S4. Powder X-ray diffraction (XRD) data were collected for the precipitates using a Bruker D8 Advance diffractometer (Germany) with Cu Kα radiation (40 kV, 40 mA) at a scanning rate of 2°/min from 5°− 80°. Transmission 57Fe Mössbauer spectra were recorded at 4.2 and 300 K on a C
DOI: 10.1021/acs.est.9b01018 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology Topologic 500A system with a Rh(57Co) source. The spectra were then fitted using MossWinn 3.0i.
remove and recover P via Fe(III) reduction-induced vivianite crystallization. Trans-membrane pressure (TMP) in the reactor during the start-up period mostly stayed below 20 kPa (Figure 2d). When the SS in the reactor system was stabilized at ∼9000 mg/L in R1, TMP increase occurred more readily and later became irreversible. From the 110th d, the TMP increased to levels >40 kPa, which made it difficult to maintain the set flux and consequently the organic loading rate. The level of SS in the system started to decrease (Figure S2), which to some extent explains the fact that P release rather than removal was observed during this period. Iron(III) addition in R2 and R3 very effectively alleviated the membrane fouling problem; however, long-term addition resulting in Fe(III) build-up in R2 ultimately led to poor performance of the membrane with fine, red particles evident within the membrane module (Figures 2d and S6). Similar membrane fouling problems caused by high Fe(III) oxide content have been reported in earlier studies.9,36 The problem was solved in R3 when the rate and extent of Fe(III) reduction was increased in the anaerobic chamber, with this change inducing the precipitation/crystallization of Fe(II) minerals (dominated by vivianite as discussed in Fe(III) Reduction and P Distribution in vUCT-MBR System). 3.2. Fe(III) Reduction and P Distribution in vUCT-MBR System. As shown in Figure 3a, a conventional UCT-MBR
3. RESULTS AND DISCUSSION 3.1. Performance of vUCT-MBR System. Efficiencies of COD removal of >95% were achieved by the vUCT-MBR system with effluent COD levels below 20 mg/L for the entire experimental period (Figure 2a). This effective performance can be attributed to the high sludge level (9−14 g/L) in the system as revealed by suspended solids (SI Figure S2). TN removal was similarly efficient, averaging >90% as illustrated in Figure 2b. It is also worth noting that iron addition to vUCTMBR (R2 and R3), especially when Fe(III) was effectively reduced in R3, noticeably affected the profiles of COD and N removal, though the final efficiencies remained unchanged (SI Figures S3 and S4). The reactor system was unable to continuously remove phosphorus prior to ferrihydrite supplementation (in R1) as shown in Figure 2c. From day 101, P removal capacity dropped rapidly from 90% down to 0. Afterward, phosphorus flow actually reversed, that was transferring from the sludge to the aqueous phase as revealed by the negative removal efficiencies. P accumulated in the system represented 3.9% of the dry mass of the sludge after 101 d of operation, consistent with that observed in enhanced biological phosphorus removal (EBPR) systems.30 Because EBPR capacity can only be achieved with regular discharge of P-loaded activated sludge, this result suggests that the upper limit of biological P uptake was reached during this Fe-free period without sludge discharge. Phosphorus removal capacity recovered as ferrihydrite was dosed to the system in R2. Since most of the dosed Fe(III) was not reduced or dissolved (discussed in 3.2), the uptake of P from the water in R2 was likely a result of adsorption to ferrihydrite. The P removal capacity was maintained until day 215, after which a deterioration of P removal was again observed (Figure 2c). The negative efficiencies of P removal at that time indicate that some of the ferrihydrite-bound P was released back into the water. A possible reason for this release could be explained by an increase in ferrihydrite crystallinity, resulting from aging and phase transformations to less amorphous minerals (e.g., lepidocrocite and goethite), which exhibit substantially lower sorption capacities.31,32 In R3, enhancement of macro- and microscale anaerobic conditions in the anaerobic chamber by reducing agitation and adding porous carriers to improve Fe(III) reduction (discussed in Fe(III) Reduction and P Distribution in vUCT-MBR System) led to effective long-term P removal (258 d in this study) with high efficiencies. On average, PO4−P in the effluent was 0.18 mg/L, which complies with the current discharge limits of 0.2−0.5 mg P/L for WWTPs in many countries.33−35 The level of P in the aerobic chamber decreased significantly with Fe(III) reduction being improved in R3 (Figure S5), suggesting that the P immobilization was Fe(II)-induced (see further discussion in Fe(III) Reduction and P Distribution in vUCT-MBR System). The dynamics of the P bioremoval system resulting from iron addition are interesting, and in-depth investigations of PAO evolution and the effect of the separated sludge phase (in the anaerobic chamber) on anaerobic release of P from the recirculated suspended sludge are warranted. While we acknowledge the need for further investigation into these issues, we focus here on the ability of the novel vUCT-MBR process to continuously
Figure 3. Reduction of Fe(III) (a) and spatial distribution of P (b) in vUCT-MBR.
system (R2) with completely mixed AS as a single biophase was not able to reduce Fe(III) in the anaerobic chamber. This is because the short SRT of the anaerobic chamber in R2, equal to the HRT of 2 h, did not allow the biological reduction of Fe(III) to occur effectively based on the reaction kinetics.37 Adding porous luffa cubes as biocarriers and reducing the agitation strength to build a relatively immobilized AS phase in the anaerobic chamber was an effective way to improve Fe(III) D
DOI: 10.1021/acs.est.9b01018 Environ. Sci. Technol. XXXX, XXX, XXX−XXX
Article
Environmental Science & Technology
the crystallization and growth of vivianite in the sludge flocs, which would be expected to fall to the chamber bottom once a certain size had been reached (as discussed further in Characterization of Sludge and P Products). In addition, the vivianite formation-induced P flow from 1 M HCl-unextractable forms and AP to NAIP would also improve the efficiency of P recovery when the NAIP mostly present as vivianite crystals. Speciation of P in the sludge by 31P NMR suggests that the vUCT-MBR in R1 was a typical EBPR system with aqueous orthophosphate release in the anaerobic chamber (increases of 11%−15%, consistent with 25−30 mg/L in the measured PO4−Paq) and poly-P synthesis under anoxic (probably by denitrifying phosphate accumulating organisms)39,40 and aerobic conditions (Table 1 and Figure S8). However, this transformation of P species (poly-P hydrolysis and synthesis as the sludge goes through the chambers) was not clearly seen in R3. The TP build-up in the activated sludge during the continuous operation of vUCT-MBR (e.g., day 94 in R1:340 mg/L; day 396 in R3:516 mg/L) was mostly reflected by an increase in orthophosphate content (to >60%). As shown in Table 1 and SI Figure S5, the sludge ortho-P for all the chambers in R3 was at similar concentrations and mostly present in the solid phase (PO4−Paq in the anaerobic chamber decreased to 5.6 mg/L and in other chambers was
