Science of the Total Environment 442 (2013) 63–72

Contents lists available at SciVerse ScienceDirect

Science of the Total Environment

j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenv

Nutrient and dissolved organic carbon removal from natural waters using industrialby-products

Laura A. Wendling ⁎, Grant B. Douglas, Shandel Coleman, Zheng YuanCSIRO Land and Water, Centre for Environment and Life Sciences, Private Bag 5, Wembley WA 6913, Australia

H I G H L I G H T S

► The CaO-based WTR effectively removed DOC and P in column trials and had a high P sorption capacity.► Granular activated carbon attenuated organic contaminants but was unsuitable for P attenuation.► Fly ash removed a suite of both inorganic and organic nutrients and DOC from water but released Se.► By-products may mitigate eutrophication through targeted use in nutrient intervention schemes.

⁎ Corresponding author at: CSIRO Land andWater, PriAustralia. Tel.: +61 8 9333 6344; fax: +61 8 9333 621

E-mail address: Laura.Wendling@csiro.au (L.A. Wen

0048-9697/$ – see front matter. Crown Copyright © 20http://dx.doi.org/10.1016/j.scitotenv.2012.10.008

a b s t r a c t

a r t i c l e i n f o

Article history:Received 16 August 2012Received in revised form 1 October 2012Accepted 1 October 2012Available online 21 November 2012

Keywords:Water treatment residualEutrophicationDissolved organic carbonWater treatment

Attenuation of excess nutrients in wastewater and stormwater is required to safeguard aquatic ecosystems.The use of low-cost, mineral-based industrial by-products with high Ca, Mg, Fe or Al content as a solidphase in constructed wetlands potentially offers a cost-effective wastewater treatment option in areas with-out centralised water treatment facilities. Our objective was to investigate use of water treatment residuals(WTRs), coal fly ash (CFA), and granular activated carbon (GAC) from biomass combustion in in-situ watertreatment schemes to manage dissolved organic carbon (DOC) and nutrients. Both CaO- and CaCO3-basedWTRs effectively attenuated inorganic N species but exhibited little capacity for organic N removal. TheCaO-based WTR demonstrated effective attenuation of DOC and P in column trials, and a high capacity forP sorption in batch experiments. Granular activated carbon proved effective for DOC and dissolved organicnitrogen (DON) removal in column trials, but was ineffective for P attenuation. Only CFA demonstrated effec-tive removal of a broad suite of inorganic and organic nutrients and DOC; however, Se concentrations in col-umn effluents exceeded Australian and New Zealand water quality guideline values. Water treated byfiltering through the CaO-based WTR exhibited nutrient ratios characteristic of potential P-limitation withno potential N- or Si-limitation respective to growth of aquatic biota, indicating that treatment ofnutrient-rich water using the CaO-based WTR may result in conditions less favourable for cyanobacterialgrowth and more favourable for growth of diatoms. Results show that selected industrial by-products maymitigate eutrophication through targeted use in nutrient intervention schemes.

Crown Copyright © 2012 Published by Elsevier B.V. All rights reserved.

1. Introduction

Excess nutrient input to aquatic systems frequently results ineutrophication and ecosystem degradation. Attenuation of excessnutrients in industrial, urban, and agricultural wastewaters andstormwater is essential to safeguard aquatic ecosystems. Relativelylow initial capital investment coupled with low ongoing operatingand maintenance costs make constructed wetlands well-suited forwastewater treatment in developing countries, particularly inrural communities. Constructed wetlands for wastewater treatmentgenerally demonstrate effective attenuation of N via denitrification,

vate Bag 5, WembleyWA 6913,1.dling).

12 Published by Elsevier B.V. All rig

but are less efficient for P removal (Kivaisi, 2001). Whereas N removalis a microbially-mediated process, P removal is primarily due to sorp-tion or precipitation reactions with solid media; thus, incorporation ofmaterials with high P attenuation capacity into constructed treatmentwetlands may substantially enhance P removal.

The chemical speciation of nutrients in wastewater may substantiallyinfluence the rate andmechanism of nutrient removal as well as their ul-timate fate. Although inorganicN speciesmaybe expected to have amoreimmediate effect on phytoplankton growth, organic N can also impactaquatic productivity (Anderson et al., 2002; Seitzinger and Sanders,1999; Seitzinger et al., 2002). Organic forms of nutrients may comprisea substantial proportion of the total nutrient load in some environments(Kroeger et al., 2006; Scott et al., 2007; Petrone et al., 2009), and oftenprove more recalcitrant in water treatment schemes (Abe et al., 2008;Sattayatewa et al., 2010; Werker et al., 2002). Dissolved organic N

hts reserved.

64 L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

(DON) is frequently associated with dissolved organic carbon (DOC)transport and breakdown. The development of treatment systems withcapacity for removal of both inorganic and organic forms of nutrients aswell as DOC is required for treatment of wastewaters containingmultiplechemical species of nutrients.

The ratio of nutrients within a water body is a significant factor indetermining the potential for excessive algal growth. Excess P is oftenpresent in a water body during the process of eutrophication and Nbecomes the limiting nutrient. Many blue-green algae can fix atmo-spheric N, putting them at a competitive advantage when N is thegrowth limiting nutrient. The development of potentially toxicblooms of blue-green algae may be mitigated through substantial re-duction in P input and shifting of the aquatic system toward P- ratherthan N-limitation, concomitant with a sufficient reduction in total nu-trient concentrations.

In near-coastal environments, eutrophicationmay also be enhancedby Si deficiency through the growth inhibition of diatoms. Alterations tothe proportions of Si, N and P which result in a Si:N ratio less than 1:1may reduce diatom growth in favour of non-siliceous algae (Justić etal., 1995). Based on nutrient uptake kinetics, the following nutrient lim-itation criteria have been identified for diatom growth (Justić et al.,1995): potential N limitation where N:Pb10 and Si:N>1; potential Plimitation where Si:P>22 and N:P>22; and potential Si limitationwhere Si:Pb10 and Si:Nb1.

Constructed wetlands or similar structures which incorporate areactive mineral phase for enhanced P, DOC and DON attenuationmay be suitable for nutrient attenuation of wastewater in areas with-out access to centralised water treatment facilities. The use oflow-cost, mineral-based industrial by-products with high Ca, Mg, Feor Al content in constructed wetlands potentially offers a cost-effective wastewater treatment option. Industrial by-products havehistorically been viewed as unsuitable for use as environmentalamendments due to the potential for contamination by hazardousconstituents or to a lack of independent evidence indicating beneficialeffect. Depending on their physico-chemical characteristics and nutri-ent sorption capacities some mineral-based industrial by-productsmay be suitable for the attenuation of nutrients in wastewater, there-by facilitating their productive use whilst reducing the environmentalfootprint of the associated industries. Drinking water treatment resid-uals (WTRs), coal fly ash (CFA) and granular activated carbon (GAC)produced from industrial or agricultural residues are low-cost,mineral-based by-products which may be suitable for use as sorptivemedia in constructed wetlands, drain liners, permeable reactive bar-riers or similar applications (e.g. Chen et al., 2007; Dias et al., 2007;Drizo et al., 1999; Gibbons et al., 2009; King et al., 2010; Oguz, 2005).

Worldwide, large quantities of WTRs are generated each year as aresult of treatment processes to remove colour, turbidity and humicsubstances from drinking water. The coagulation agents used totreat water can include Fe and Al salts, organic polymers, CaO,Ca(OH)2 and/or Mg-containing materials. Several factors have led toincreased interest in re-use of WTRs including the implementationof increasingly restrictive environmental regulations for solid wastedisposal, increased disposal costs and decreased landfill capacities.

Similarly, there is substantial interest in the beneficial re-use ofCFA as a consequence of the enormous quantity of CFA generated an-nually and the associated disposal costs. Approximately 500 Mt ofCFA, a fine-textured predominantly mineral residue, are generatedeach year as a residue from fossil fuel combustion in coal-firedpower stations around the world (Ahmaruzzaman, 2010). On aver-age, only about 16% of the CFA generated globally is utilised in ameaningful way (Ahmaruzzaman, 2010); thus, the majority of theCFA generated each year is disposed of in landfills, which has the po-tential to cause significant environmental degradation (e.g. Quispe etal., 2012; Silva et al., 2012a,b).

Activated carbon filters are widely used in water and wastewatertreatment for the removal of synthetic organic chemicals and

naturally-occurring organic chemicals. Activated carbon is generatedwhen biomass is carbonized to charcoal and then activated withsteam. Some industries generate activated carbon through integratedprocesses, and a number of agricultural and industrial by-productsdemonstrate potential as precursors for the manufacture of activatedcarbon products (Dias et al., 2007; Ioannidou and Zabaniotou, 2007).

The objective of this study was to investigate the use of selectedlow value, mineral-based industrial by-products in in-situ watertreatment schemes to manage DOC and nutrients in contaminatedwaters. Mineral-based industrial by-products were investigatedusing a testing protocol which included physical, mineralogical, geo-chemical, and radiochemical characterisation. Nutrient and DOC at-tenuation by mineral-based industrial by-products was assessed inlaboratory column trials and batch tests were used to quantify the Psorption capacity of each by-product. Low-cost materials incorporatedinto the testing program included: a CaO-based groundwater treatmentresidual; an iron-rich CaCO3-based granular groundwater treatment re-sidual; CFA from a coal-fired power station; and pre-commercial gradeGAC resulting from wood combustion during electricity generation.Critical material performance indicators included nutrient and DOC up-take capacity, geochemical transformations and stabilisation.

2. Materials and methods

2.1. Column materials

Surface (0–15 cm) native Bassendean Sand for use in referencecolumns and as a nominally non-retentive phase in columnscontaining by-product mixtures was collected from the Swan CoastalPlain near Perth, Western Australia, and sieved to b2 mm to removeleaves and root fragments. The water treatment residuals used inthese experiments were CaO- and CaCO3-based materials used to re-move colour and odour from groundwaters. Both the CaO-based WTRand the iron-rich, CaCO3-based granularWTRwere obtained frommet-ropolitan groundwater treatment plants in Perth, Western Australia.Coal fly ash was sourced from a regional Western Australian coal-firedpower station. Granular activated carbon from biomass combustionwas obtained from an integrated eucalyptus tree processing plant inWestern Australia. The CFA, and granular CaCO3-based and CaO-basedWTRs were oven dried to a constant mass at 105 °C, and the CaO-basedWTR was ground to b63 μm particle diameter in a tungsten carbidering mill prior to use in column experiments. The biomass-based GACwas used on an as-received basis.

2.2. Material characterisation

For each sample, the electrical conductivity (EC) was determinedin a 1:5 solid to liquid (w/w) aqueous extract using deionisedwater, and by-product pH was determined in a 1:5 solid to liquidratio (w/w) of 0.01 M CaCl2 (Rayment and Higginson, 1992). Quanti-tative X-ray diffraction (XRD) analysis was used to characterise themineralogical composition of by-products. The XRD patterns wererecorded using Fe-filtered Co Kα radiation, a ¼° divergence slit, a½° anti-scatter slit and a Si strip detector. Diffraction patterns wererecorded in steps of 0.016°∙2Θ with 0.4 s counting time per step.X-ray fluorescence analyses for major elements (as oxides) andtrace elements were performed on fused glass disks and pressedpowder pellets of each material, respectively, analysed on a wave-length dispersive XRF system.

2.3. Laboratory column experiments

Experimental materials were contained within stainless steel col-umns (1.0 m length with 2.2 cm internal diameter) with glass wooland a fine stainless steel mesh filter placed within fittings at eachend of the columns to prevent blocking of column inlet or outlet.

65L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

Native Bassendean Sand (Playford et al., 1976) was employed as anominally non-reactive phase in all columns, and comprised 100%of triplicate reference columns. With the exception of GAC, the reac-tive phase was contained in test columns in a 50% (w/w) mixturewith native sand within the middle one-third of the column to simu-late the presence of a sorptive phase akin to that incorporated withina permeable reactive barrier or constructed wetland deployment. Dueto its low density, the GAC was contained within the middle one-thirdof the test column in a 50% (v/v) mixture with native sand. The topand bottom one-third of all experimental columns were comprisedentirely of native sand.

A mixture of DOC- and nutrient-rich surface water from WesternAustralia's Swan Coastal Plain was pumped through the columns ina saturated up-flow mode at a constant flow rate of ca. 0.2 mL/min.This flow rate was selected to facilitate an approximately 12-h resi-dence time for influent water within laboratory columns. Waterflow was controlled by a low volume peristaltic pump on the influentline. Influent water and effluents from experimental columns wereanalysed for pH and flow rate, alkalinity, nutrient and DOC content,and selected cations and anions, including: HCO3

−, NOx-N, NH3-N, N(total), soluble reactive P (SRP or PO4-P), P (total), Al, Ca, Cl−, Fe,Mg, Mn, K, Si, Na, and SO4

2−.Columns were operated to hydraulic failure, or until nutrient and

DOC sorption capacities could be determined. Experimental columnscontaining native Bassendean Sand, CaO-based WTR and GAC wereeach operated for approximately 180 d. Laboratory columns containingCaCO3-based WTR were terminated after 60 d operation when thematerial's P retention capacity was surpassed, whilst columnscontaining CFA were terminated after ca. 90 d operation due to detec-tion of Se in column effluents in excess of ANZECC/ARMCANZ (2000)guideline limits. Data reported herein have been adjusted to accountfor the addition or removal of nutrients and DOC by native BassendeanSand in experimental columns.

2.4. Determination of P sorption capacity

Triplicate 1.00±0.01 g subsamples of the native BassendeanSand and by-products were mixed thoroughly in 250 mL of 0.2%P (as K2HPO4) solution on a roller mixer. After 24 h samples werecentrifuged and the supernatants filtered to 0.22 μm. Soluble reac-tive phosphate in solution was quantified spectrophotometrically(Murphy and Riley, 1962).

2.5. Geochemical modelling

Interpretation of the solute geochemistry was carried out on efflu-ents from laboratory column trials using PHREEQC for WindowsV1.5.10 (Parkhurst, 1995) to examine geochemical transformationsvia geochemical modelling of the solute major ion (Na, K, Ca, Mg, Cl,HCO3, SO4) and Fe, Al and Mn chemistry. The PHREEQC calculationswere used to estimate the saturation index (SI) of mineral phases,in particular, those of Al, Fe and Mn minerals. The SI for a given min-eral is defined as: SI=log IAP/Ksp where IAP is the ion activity prod-uct of the chemical species in the reaction, and Ksp is the solubilityproduct for the designated mineral. All log Ksp values were sourcedfrom the PHREEQC database with the exception of that for struvite(MgNH4PO4·6H2O, log Ksp=13.2), which was taken from Scott etal. (1991).

3. Results

3.1. Influent water

The blend of surface waters used as the influent in column trialscontained relatively high concentrations of DOC (ca. 59.2±1.4 mg/L),DON (1.30±0.26 mg/L), and total P (0.69±0.05 mg/L) of which the

majoritywas present as PO4-P (0.52±0.11 mg/L). Concentrations of in-organic N species were low with only minor NOx-N (0.44±0.20 mg/L)and NH3-N (0.32±0.16 mg/L).

3.2. Material characterisation

Summaries of material mineralogical and geochemical character-istics are given in Tables 1 and 2, respectively. Quartz and mullitewere the dominant crystalline components in the CFA, along withtraces quantities of other crystalline minerals. The CFAwas comprisedof a high proportion of amorphous minerals, primarily derived fromreactions in the mineral matter of the coal at high temperatures dur-ing combustion. The occurrence of quartz in CFA is attributed to relictquartz grains in the coal not melted during the combustion process.Similar results have been reported in previous studies of CFA(Quispe et al., 2012; Silva et al., 2012a,b,c). Concentrations of Ag, I,Sb, Te and Yb in native Bassendean Sand, CaO- and CaCO3-basedWTRs, GAC and CFA were all below XRF limits of detection and arenot shown in Table 2. Mineralogical analysis indicated that the nativesand used in these experiments was comprised of >98% quartz (SiO2)with trace amounts of calcite (CaCO3), halite (NaCl) and hematite(α-Fe2O3; Table 1). The native sand exhibited an EC of 0.17 mS/cmand pH was 7.3; geochemical analysis confirmed that the sand wascomprised almost entirely of SiO2 (>98%) with minor Al2O3 (0.1%)and Fe2O3 (0.4%) (Table 2).

Quantitative XRD showed that the CaO-basedWTR was comprised ofapproximately 12% quartz, 39% calcite, 12% brucite (Mg(OH)2), 34%portlandite (Ca(OH)2) and 3% the hydrotalcite mineral pyroaurite (idealformula Mg6Fe2CO3(OH)16∙4H2O; Table 1). The EC of the CaO-basedWTR was 0.89 mS/cm, and pH was 9.6 (Table 2). Geochemical analysesindicated that the majority of the mineral phase, more than 45%, of theCaO-based WTR could be accounted for by CaO. Both the CaO- andCaCO3-basedWTR containedhigh levels of Ca; however, the CaCO3-basedWTRwas also relativelymore enrichedwith Fe (Table 2). Trace elementswhich displayed ≥2 times enrichment in the CaO-based WTR comparedto native sand included As, Ba, Ga, Ge, Rb, Sr, Tl, U, Y and Zn, whereas Br,Ce, Cr, Mo, Nb, Ni and Zr were relatively depleted in the CaO-basedWTR.

The EC of the CaCO3-based WTR residue was 0.04 mS/cm and pHwas 8.3 (Table 2). Quantitative XRD of the CaCO3-based WTR residueindicated a composition of ca. 1% quartz, 95% calcite, and 4% garnet[(Mg,Fe,Ca)3(Al,Fe,Cr)2Si3O12; Table 1], the latter which was used asa nucleation agent for carbonate crystallisation during production ofthe water treatment media. Geochemical analysis showed that CaOaccounted for nearly the entire CaCO3-based WTR mineral phase.Similar to the CaO-based WTR, trace elements which displayed ≥2times enrichment in the CaCO3-based WTR compared to native sandincluded As, Ba, Rb, Sr, Tl and U (Table 2). Several elements were de-pleted in the CaCO3-based WTR relative to the native sand, includingBr, Ce, Cr, Cu, Mo, Nb, Ni and Zr.

The pH and EC of CFA were 6.9 and 0.72 mS/cm, respectively. TheCFA was approximately 65% amorphous (e.g., little identifiable or-dered structure), and the ordered phase was comprised of 17% quartz,13% mullite (3Al2O3∙2SiO2), and 3% each hematite (α-Fe2O3) andmaghemite (γ-Fe2O3; Table 1). Fusion XRF indicated that nearly 50%of the CFA could be accounted for by SiO2, and that SiO2, Al2O3 andFe2O3 together comprised over 80% of the CFA solid phase (Table 2).Trace elements that displayed some enrichment (≥2 times) in theCFA when compared to native sand included As, Ba, Bi, Br, Cd, Ce,Co, Cs, Cu, Ga, Ge, Hf, La, Nb, Ni, Pb, Rb, Se, Sn, Sr, Th, Tl, U, V, Y, Znand Zr. A range of trace elements including As, Cr, Cu, Mn, Nb, Ni,Pb, Zn, Th Tl, V, U and Zr and the rare earth elements La, Ce, Nd, Smand Y were particularly enriched (>5 to >10+ times) relative tothe native sand. No major or trace elements were depleted (≥2times) in the CFA relative to native sand.

The GAC resulting from biomass combustion exhibited a pH of10.4 and an EC of 0.48 mS/cm. X-ray diffraction analysis showed

Table 1Chemical composition, pH and EC of native sand, coal fly ash (CFA), CaCO3- and CaO-based water treatment residuals (WTRs) and granular activated carbon (GAC) used inexperiments.

Mineral constituent Native sand CFA CaCO3-based WTR CaO-based WTR GAC

Quartz SiO2 >98% 16.8% 0.9% 12.0%Calcite CaCO3 b1% 95.2% 39.0% b5%Portlandite Ca(OH)2 33.8%Brucite Mg(OH)2 11.9%Mullite 3Al2O3 ∙2SiO2 12.5%Garnet (Mg,Fe,Ca)3(Al,Fe,Cr)2Si3O12 3.9%Pyroaurite Mg6Fe2CO3(OH)16 ∙4H2O 3.3%Hematite α-Fe2O3 b1% 3.0%Maghemite γ-Fe2O3 3.0%Halite NaCl b1%Amorphous – 64.7% >95%

66 L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

that the GAC was >95% amorphous, exhibiting only a small calcitepeak (Table 1). Geochemical analysis indicated that the measureableoxides were primarily Al2O3 and CaO (constituting only 3% of thetotal mass), with minor quantities of MgO, K2O, TiO2, MnO and P2O5

(Table 2). It was not possible to examine the trace element composi-tion of the GAC using XRF as the material was not amenable toanalysis.

3.3. Nutrient and DOC attenuation

Nutrient and DOC removal from or addition to inlet water by theBassendean sand, herein referred to as native sand, and industrialby-products are given in Table 3. Data have been corrected to accountfor the addition or removal of nutrients and DOC by native sand refer-ence material which was mixed with industrial by-products in exper-imental columns. The native sand had no sorptive capacity for DOC,and exhibited a release of approximately 155 mg/kg DOC shortlyafter initiation of flow through the column. Although DOC in sand ref-erence column effluents declined with time, the cumulative DOC inreference column effluents remained greater than the cumulativequantity of DOC in the influent water and resulted in a net releaseof DOC from the native sand (Table 3). The native sand also had nosorptive capacity for any N species. The mean cumulative content ofNOx-N in sand reference column effluents was approximately equalto cumulative NOx-N in influent water, showing no net change inaqueous NOx-N concentration as a result of passage through thesand. Cumulative NH3-N and DON in sand reference column effluentsincreased by 17% and 10%, respectively, compared to influent water. Amean overall 9% increase in total N was observed following passage ofwater through the native sand reference columns (Table 3).

Native sand reference column effluents exhibited slightly de-creased cumulative mean concentrations of PO4-P and total P com-pared to the influent water, resulting in attenuation of 6% PO4-P and8% total P, respectively, in influent water (Table 3, Fig. 1). The P sorp-tion capacity of the native sand was 160 mg/kg in batch experiments.

The CaO-based WTR showed moderate attenuation of DOC; refer-ence material-corrected values indicated removal of 38% of the totalDOC from influent water by CaO-basedWTR in laboratory column tri-als (Table 3). Total N removal by the CaO-based WTR was similar toDOC at ca. 35% but varied greatly among N species. Whilst dissolvedinorganic N (DIN) species NOx-N and NH3-N exhibited referencematerial-corrected removal of 92% and 73%, respectively, by theCaO-basedWTR only about 2% of DON in influent water was removed(Table 3). The CaCO3-based WTR also effectively attenuated DIN spe-cies in influent water, removing 97% and 61% of influent NOx-N andNH3-N, respectively (Table 3). In contrast to the CaO-based WTR,however, effluents from the CaCO3-based WTR column exhibited anet increase in both DOC (29% increase) and DON (58% increase),and total N attenuation of only 24%.

Approximately 98% of the PO4-P and total P in influent water wereeffectively removed by the CaO-based WTR (Table 3). In comparison,

the CaCO3-based WTR removed only 37% and 46% of PO4-P and totalP, respectively, from influent water. Batch experiments yielded a Psorption capacity of approximately 3,590 mg P/kg CaO-based WTR,and indicated that the CaCO3-based WTR released >400 mg P/kg.Due to low nutrient and DOC removal from influent water, the col-umn containing CaCO3-based WTR was terminated following passageof ca. 14.7 L water through the experimental column.

Approximately 27% of both DOC and total P, and 12% of PO4-P ininfluent water were removed by the CFA in laboratory column trials(Table 3). Batch experiments yielded a P sorption capacity of1833 mg P per kg CFA. Nearly half the total N was removed from in-fluent water by the CFA, most of which was DIN. Although not partic-ularly effective for DON attenuation (12% removal), the CFA removed86% and 84% of the NOx-N and NH3-N from influent water, respective-ly, in column trials (Table 3). Column experiments using CFA wereterminated following collection of ca. 26.7 L effluent from the exper-imental column due to detection of trace levels of As, Cd and Pb inCFA column effluents, and measurement of Se in column effluents atconcentrations greater than the Australian and New Zealand freshand marine water guidelines for recreational and aesthetic purposes(ANZECC/ARMCANZ, 2000).

Both N and DOC were effectively attenuated by the biomass-basedGAC in column trials (Table 3). Removal of 59% DOC from inlet waterby the GAC was observed following correction for DOC addition by na-tive sand. More than 42 g of DOC were attenuated per kg of GAC. Gran-ular activated carbon attenuation of DONwas similarly effective; 38% ofDON in influentwater was removed by the GAC (Table 3). The GAC alsoremoved DIN species from influent water, exhibiting greater attenua-tion of NH3-N (76%) than NOx-N (11%) and overall reference material-corrected 40% total N removal from influent water (Table 3). The GACexhibited no capacity for P attenuation in laboratory column experi-ments (Table 3); however, batch testing yielded a P sorption capacityof 2020 mg P per kg GAC.

For comparison between by-products, reference material-correctednutrient andDOC removal by each by-product and nutrient andDOC re-moval or addition by native sand reference material are shown in Fig. 1following passage of 20 L nutrient- and DOC-rich water through eachcolumn. Data for CaCO3-basedWTR in Fig. 1 are presented at time of col-umn termination (14.7 L).

3.4. Geochemical modelling of column effluents

Modelling of effluent geochemistry was undertaken to facilitatefurther interpretation of controls on major and trace element geo-chemistry in column effluents. Importantly, the estimated saturationindices provide a theoretical measure of a mineral saturation statewith respect to the co-existing solution and do not yield any inferenceas to kinetics of formation or dissolution. In addition, the modellingundertaken in this study did not take into account other solutessuch as natural humic substances as may be found in environmental

Table 2Mineralogical composition of native sand, coal fly ash (CFA), CaCO3- and CaO-based water treatment residuals (WTRs) and granular activated carbon (GAC) used in experiments.

Native sand CFA CaCO3-based WTR CaO-based WTR GAC

pH 7.3 6.9 8.3 9.6 10.4EC (mS/cm) 0.17 0.72 0.04 0.89 0.48SiO2 % 98.2 49.6 0.5 11.6 b0.005TiO2 % 0 1.2 b0.003 0.04 0.1Al2O3 % 0.1 22.4 0.3 0.9 1.4Fe2O3 % 0.4 9.7 0.5 0.3 0.1MnO % 0 0.1 0.02 0.001 0.1MgO % 0 1.1 0.6 6.8 0.4CaO % 0.07 1.6 49.7 45.4 1.7Na2O % 0 0.3 b0.002 0.03 0.02K2O % 0 0.8 b0.002 0.3 0.2P2O5 % 0.07 1.2 0.03 0.2 0.1SO3 % 0.02 0.3 0.09 0.5 0.05Cl mg/kg 0 149 275 672 32Sum % 98.9 88.3 51.8 66.2 4.1As mg/kg b2 17 4 8 NDa

Ba mg/kg 21 3741 254 77 NDBi mg/kg b3 3 b3 b3 NDBr mg/kg 2 4 b1 b1 NDCd mg/kg b4 4 b4 b4 NDCe mg/kg 39 407 24 b20 NDCo mg/kg b5 183 18 b5 NDCr mg/kg 287 170 7 66 NDCs mg/kg b11 19 17 b11 NDCu mg/kg 3 100 b1 2 NDGa mg/kg 1 45 b1 3 NDGe mg/kg b1 12 1 1 NDHf mg/kg b8 12 b8 b8 NDLa mg/kg 30 229 b18 b18 NDMo mg/kg 14 16 b1 1 NDNb mg/kg 8 66 3 4 NDNd mg/kg b11 158 b11 b11 NDNi mg/kg 16 283 b2 b2 NDPb mg/kg 3 53 b3 b3 NDRb mg/kg 2 62 5 15 NDSc mg/kg b7 30 b7 b7 NDSe mg/kg b2 11 b2 b2 NDSm mg/kg b11 15 b11 b11 NDSn mg/kg b3 9 b3 b3 NDSr mg/kg 7 2062 916 1411 NDTa mg/kg b7 b7 9 9 NDTh mg/kg 7 57 7 9 NDTl mg/kg 3 29 8 7 NDU mg/kg b1 28 6 16 NDV mg/kg 9 156 b7 8 NDY mg/kg 5 142 11 10 NDZn mg/kg 3 353 2 9 NDZr mg/kg 72 443 8 32 ND

a ND = not determined.

67L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

applications which may influence speciation and hence, mineralsaturation.

Iron, Al and Mn are important elements in by-product effluents dueto the potential neoformation of Al, Fe and Mn oxide/hydroxide min-erals and subsequent sorption of nutrients and DOC. Within the nativesand, CaO- and CaCO3-based WTR, CFA and GAC column effluents,ferrihydrite (Fe(OH)3), goethite (α-FeOOH), hematite (α-Fe2O3) andschwertmannite (Fe8O8(OH)6(SO4)·nH2O) were theoretically stableminerals (Table 4). Oversaturation of these mineral phases is likelylargely due to the presence of soluble Fe2+ in influentwater, with addi-tional Fe dissolution from Fe-bearing mineral phases in some experi-mental columns. Poorly crystalline ferrihydrite was most likely theprimary mineral phase controlling Fe solubility in column effluents,with potential for the formation of a more ordered Fe mineral such ashematite with time.

The Mn minerals manganite (MnOOH), pyrochroite [Mn(OH)2],pyrolusite (MnO2·H2O) and rhodochrosite (MnCO3) were undersatu-rated in all laboratory column effluents (Table 4). In contrast, alunite[KAl3(SO4)2(OH)6], basaluminite [Al4SO4(OH)10·5H2O] and gibbsite

[(Al(OH)3] were oversaturated in all column effluents, whilst poorlycrystalline Al [Al(OH)3(am)] was near saturation (Table 4). Aqueousgeochemistry of the Al2O3-SO3-H2O system is complex due to thelarge number of stable and metastable minerals that may form overa wide range of pH and sulphate concentrations. Basaluminite andAl(OH)3(am) were the mineral phases most likely controlling Al solu-bility in laboratory column effluents.

Calcium minerals were generally undersaturated in all column ef-fluents, with the exception of Ca-montmorillonite [Ca0.165Al2.33Si3.67O10(OH)2] which was slightly oversaturated. Effluents from nativesand, CaO- and CaCO3-basedWTR, CFA andGAC columnswere similarlyundersaturated with respect to dolomite [CaMg(CO3)2] and the Mgminerals brucite [Mg(OH)2], chlorite [Mg5Al2Si3O10(OH)8], chrysolite(Mg3Si2O5), sepiolite (Mg2Si3O7.5OH·3H2O) and talc [Mg3Si4O10(OH)2](Table 4).

The clay mineral illite was theoretically at or approaching satura-tion in all column effluents. Both kaolinite [Al2Si2O5(OH)4] andK-mica [KAl3Si3O10(OH)2] were oversaturated in the native sand,CaO- and CaCO3-based WTR, CFA and GAC column effluents (Table 4).

Table 3Cumulative nutrient and DOC removal from or enrichment of influent water by native sand, coal fly ash (CFA), CaCO3- and CaO-based water treatment residuals (WTRs) and gran-ular activated carbon (GAC) in laboratory column trials at time of column termination. The period of column operation is given in litres of column effluent. Materials are ranked inorder of increasing percent total P removal.

Column contents Effluent pH range Effluent volume (L) Cumulative nutrient and DOC removal (−) or enrichment (+)

NOx-N NH3-N DON TN PO4-P TP DOC

mg/kg % mg/kg % mg/kg % mg/kg % mg/kg % mg/kg % g/kg %

Influent water 6.9–7.4 – 0.44 mg/L 0.32 mg/L 1.3 mg/L 2.1 mg/L 0.52 mg/L 0.69 mg/L 59.2 g/LNative sand 7.0–8.2 43.3 +0.2a +0.7 +4 +17 +8 +10 +12 +9 −2 −6 −3 −8 +0.2 +6GAC 7.7–8.5 42.9 −86 −11 −579b −76 −876 −38 −1,540 −40 +278 +35 +91 +8 −42.19 −59CFA 7.7–8.8 26.7 −134 −86 −136 −84 −39 −12 −309 −49 −14 −12 −42 −27 −4.23 −27CaCO3-based WTR 7.2–8.1 14.7 −124 −97 −72 −61 +97 +58 −99 −24 −25 −37 −40 −46 +2.49 +29CaO-based WTR 8.9–13.0 43.3 −167 −92 −128 −73 −12 −2 −308 −35 −184 −98 −254 −98 −9.51 −38

a +xx=net increase as compared to inlet water.b Cumulative reductions of >50% as compared to inlet water in bold.

68 L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

The neoformation of crystalline aluminosilicate minerals such asmontomorillonite and kaolinite via precipitation of ions in aqueoussolution during the laboratory column trial is, however, highly un-likely due to the slow kinetics of phyllosilicate mineral formation.

Both Mg/Al and Mg/Fe hydrotalcite minerals were theoreticallyoversaturated in effluents from all laboratory columns (Table 4).Hydrotalcite minerals have been demonstrated to form via aging ofamorphous Al hydroxycarbonate gel coated on Mg hydroxide particleson a time scale of weeks to months, with the rate of mineral formationincreasing concomitantly with solution concentration and/or reactiontemperature (Vanderlaan et al., 1982). Hydrotalcite minerals arebased on a brucite structure inwhich some of the divalent cations in oc-tahedral positions within the hydroxide are replaced by trivalent cat-ions such as Al or Fe. The resulting layer charge is balanced by anions(e.g. CO3

2−, NO3−, SO4

2−, Cl−, etc.) in interlayer spaces. Hydrotalcite syn-thesis via co-precipitation from solution generally involves hydrother-mal treatment of the precipitate to enhance structure crystallinity(Brei et al., 2008). In our laboratory column trials, the residence timeof water within columns was 13.0±2.1 h (data not shown) and exper-iments were conducted at ambient temperature (ca. 25 °C). Thus,hydrotalcite minerals whilst theoretically oversaturated are unlikelyto have precipitated from solution in the experiment described herein.

Free Ca in solution resulting from the dissolution of Ca minerals maycontribute to PO4-P attenuation via Ca and P co-precipitation as hydroxy-apatite or co-sorption (Murphy and Stevens, 2010), whilst struvite pre-cipitation is a potential mechanism of P removal from waters rich in P,NH4 and Mg (Scott et al., 1991). Geochemical modelling of effluentsfrom laboratory column trials indicated that both hydroxyapatite andstruvite were theoretically undersaturated in all column effluents(Table 4).

Fig. 1. Nitrogen species removal from 20 L influent water (mg/kg) and cumulative percent tphosphorus (SUP) species removal from 20 L influent water (mg/kg) and cumulative percfrom 20 L influent water (g/kg) and cumulative percent DOC removal (right) in laboratory(14.7 L).

3.5. Material composition and effluent characteristics

Simple correlation coefficients indicate that the removal of PO4-Pfrom influent water by CaO- and CaCO3-based WTRs, CFA, and GAC incolumn trials was strongly associated with the Ca and Mg contents(r=0.70 and r=0.83, respectively) of the by-products (Table 5). Thisrelationship may be related to flocculation of humic acid- and/or fulvicacid-phosphate complexes in the presence of dissolved Ca and Mg(Sholkovitz, 1976); however, there was no apparent correlation be-tween DOC removal from influent water by the selected by-productsand their Ca or Mg content (Table 5). Terrestrially-derived humic sub-stances are known to coagulate in electrolyte solutions; dissolved organicmatter (DOM) in river water, which is comprised largely of humic sub-stances, exhibits rapid flocculation in sea water (e.g. Sholkovitz, 1976and references therein). Polyvalent cations such as Al, Fe and Ca facilitatecoagulation of DOMand its removal fromaqueous solution via the forma-tion of cation bridges between dissociated functional groups on the DOCand negatively charged surfaces. Coagulants–flocculants commonlyemployed in water treatment include salts of Fe and Al, lime [CaO orCa(OH)2] and various sources ofMg includingMgCO3,Mg(OH)2 orMgCl2.

Examination of the relationship between DOC, DON and PO4-P re-moval from influent water by CaO- and CaCO3-based WTRs, CFA, andGAC in column trials and the respective concentrations of Ca, Mg, FeandAl in column effluents yieldedmixed results (Table 5). Dissolved or-ganic carbon removal by both the CaO-basedWTR and GAC showed nocorrelationwith Ca orMg concentrations in column effluents; however,DOC removal from influent water by CFA was moderately correlatedwith the concentration of Ca in column effluents (r=0.59) whilst amoderate correlation was observed between DOC removal by theCaO-based WTR and effluent Mg concentration (r=0.60; Table 5).

otal nitrogen removal (left); soluble reactive phosphorus (SRP) and soluble unreactiveent total phosphorus removal (middle); and dissolved organic carbon (DOC) removalcolumn trials. Data for CaCO3-based WTR are presented at time of column termination

Table 4Mean saturation indices (SI, ±1 standard deviation from the mean) for effluents from laboratory columns containing native sand, coal fly ash (CFA), CaCO3- and CaO-based watertreatment residuals (WTRs) and granular activated carbon (GAC). Saturated minerals and those within ±0.5 SI for each effluent are shown in bold text.

Mineral Native sand reference CaO-based WTR CaCO3-based WTR Coal fly ash Granular activated C

Al(OH)3(am) −0.4±0.2 −1.0±0.8 −0.5±0.2 −0.4±0.2 −0.6±0.3Albite NaAlSi3O8 −3.8±0.6 −3.0±1.5 −3.6±0.3 −3.4±0.4 −3.9±0.4Alunite KAl3(SO4)2(OH)6 2.8±0.5 1.3±1.4 1.8±0.4 3.6±0.8 2.3±0.9Anorthite CaAl2Si2O8 −4.8±0.8 −5.0±2.0 −4.5±0.6 −4.3±0.5 −5.1±0.6Basaluminite Al4SO4(OH)10∙5H2O 2.9±0.7 0.9±2.1 2.1±0.6 4.1±0.9 2.3±1.1Brucite Mg(OH)2 −7.8±0.2 −7.8±1.9 −8.4±1.2 −7.9±0.1 −7.8±0.2Ca-montmorillonite Ca0.165Al2.33Si3.67O10(OH)2 1.9±0.8 1.9±3.0 1.9±0.5 2.7±0.7 1.6±0.7Calcite CaCO3 −1.7±0.3 −1.7±1.1 −1.4±0.2 −1.8±0.1 −1.7±0.1Chlorite Mg5Al2Si3O10(OH)8 −18.2±1.6 −18.2±7.5 −21.3±5.7 −17.9±0.8 −18.2±0.8Crysotile Mg3Si2O5 −15.4±0.8 −14.7±4.8 −17.1±3.5 −15.6±0.4 −15.3±0.5Dolomite CaMg(CO3)2 −3.6±0.4 −3.8±1.8 −3.9±1.1 −3.7±0.2 −3.5±0.2Ferrihydrite Fe(OH)3 1.4±0.2 0.3±0.3 1.6±0.2 1.4±0.3 1.1±0.4Gibbsite Al(OH)3 2.3±0.2 1.7±0.8 2.2±0.2 2.6±0.2 2.2±0.3Goethite FeOOH 7.3±0.2 6.2±0.3 7.5±0.2 7.3±0.3 7.0±0.4Gypsum CaSO4·2H2O −2.6±0.1 −2.7±0.3 −2.7±0.1 −2.7±0.1 −2.6±0.1Halite NaCl −6.3±0.1 −6.3±0.1 −6.3±0.1 −6.4±0.1 −6.4±0.1Hematite Fe2O3 16.6±0.3 14.4±0.7 17.0±0.5 16.6±0.5 15.9±0.8Hydrotalcite MgAl 19.7±1.6 18.6±10.6 16.0±6.9 19.9±0.7 19.8±0.9Hydrotalcite MgFe 11.5±1.6 6.6±6.1 8.4±6.8 11.2±1.0 11.1±1.2Hydroxyapatite Ca5(PO4)3OH −6.8±1.2 −13.1±4.7 −6.6±0.8 −7.3±0.4 −6.6±0.8Illite K0.6 Mg0.25Al2.3Si3.5O10(OH)2 0.2±0.9 0.3±2.4 0.1±0.3 0.9±0.7 −0.1±0.8Jarosite-K KFe3(SO4)2(OH)6 −1.6±0.4 −4.8±1.1 −1.8±0.6 −1.6±0.9 −2.6±1.3Jarosite-Na NaFe3(SO4)2(OH)6 −4.2±0.4 −7.3±1.1 −4.3±0.6 −4.1±0.9 −5.2±1.2Jurbanite AlSO4OH −2.3±0.2 −3.3±2.6 −2.9±0.1 −2.0±0.3 −2.5±0.4K-feldspar KAlSi3O8 −2.6±0.7 −1.9±1.5 −2.4±0.3 −2.3±0.5 −2.7±0.5K-mica KAl3Si3O10(OH)2 7.6±1.0 7.1±2.8 7.5±0.6 8.4±0.8 7.2±1.0Kaolinite Al2Si2O5(OH)4 4.7±0.6 4.2±2.2 4.5±0.4 5.3±0.5 4.4±0.6Manganite MnOOH −9.9±0.6 −96.4±18.0 −9.4±0.5 −72.4±43.1 −9.5±0.50Pyrochroite Mn(OH)2 −9.6±0.6 −96.4±18.0 −9.3±0.5 −72.3±43.2 −9.3±0.5Pyrolusite MnO2∙H2O −16.0±0.9 −96.6±16.9 −15.3±0.6 −74.3±40.1 −15.6±0.5Quartz SiO2 −0.4±0.1 0.0±0.5 −0.3±0.1 −0.3±0.1 −0.4±0.1Rhodochrosite MnCO3 −2.2±0.6 −96.1±19.5 −2.0±0.4 −70.0±46.8 −1.9±0.5Sepiolite Mg2Si3O7.5OH∙3H2O −11.8±0.7 −10.7±2.6 −12.9±2.3 −11.8±0.4 −11.8±0.4SiO2(am) −1.6±0.1 −1.3±0.5 −1.6±0.1 −1.6±0.1 −1.7±0.1Schwertmannite Fe8O8(OH)6(SO4)·nH2O 16.3±1.2 7.7±2.7 17.5±1.8 16.6±2.1 13.7±3.2Struvite MgNH4PO4·6H2O −34.2±0.9 −39.6±12.6 −35.3 0.9 −59.7±33.1 −45.4±24.8Talc Mg3Si4O10(OH)2 −13.3±1.0 −11.8±4.1 −14.9±3.5 −13.4±0.6 −13.3±0.6

69L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

Effluent Al concentration was strongly correlated (r=0.78) with DOCremoval by the CaO-basedWTR, but there was no relationship betweeneffluent Al concentration and DOC removal observed in the otherby-product columns.

The relationship betweenDON removal from influentwater and cat-ion concentrations in each column effluent differed from that observedfor DOC (Table 5). Potential mechanisms of DOC andDON removal fromwater include precipitation, physical adsorption and biodegradation.Previous work has shown that the DOC in column influent water waslargely comprised of humic substances or naturally-occurring organicmatter, which is refractory and thus not highly susceptible to biodegra-dation (Petrone et al., 2009; Petrone, 2010). A large proportion of DONin column influentwater was associatedwith the non-humic fraction ofDOC (Petrone et al., 2009), comprised of lowermolecularweight organ-ic acids and less recalcitrant organic and aliphatic acids as well as labileand N-rich compounds. Attenuation of DOC in laboratory column trialslikely occurred via amixture of sorption and precipitation reactions. En-demic bacteria in the unsterilized influent water and native sands usedin this experiment may have contributed to biodegradation of DON incolumn influent, whilst precipitation and sorption are likely additionalDON removal mechanisms.

3.6. Nutrient ratios of treated effluents

Ratios of nutrients within influent water and effluents from labo-ratory columns containing native sand and industrial by-products in-dicate that only post-treatment effluents from the column containingCaO-based WTR consistently exhibited a shift in nutrient ratios awayfrom zones of potential N- and Si-limitation and towards a zone of

potential P-limitation (Fig. 2). Whilst effluents from the columnscontaining CaCO3-based WTR, CFA, and GAC generally showed littlechange in potential N- or P-limitation compared to influent water,all post-treatment effluents exhibited a relative enrichment in Siand subsequent shift away from potential Si-limitation. The substan-tial concomitant decreases in N and P observed in effluents from the lab-oratory column containing CaO-based WTR is reflected in substantiallyaltered nutrient ratios of column effluents (Fig. 2). Post-treatment efflu-ents from the column containing CaO-based WTR exhibited nutrient ra-tios characteristic of potential P-limitation with no potential N- orSi-limitation respective to growth of aquatic biota, indicating that treat-ment of nutrient- and DOC-rich water using the CaO-basedWTRmay re-sult in conditions less favourable for cyanobacterial growth and morefavourable for growth of diatoms.

4. Discussion

Initial increases in DOC and nutrient concentrations in the nativesand reference column effluent relative to inlet water, in addition tosustained DOC release throughout the column trial likely occurred dueto flushing of loosely-held organicmaterial ormaterial dislodged duringsoil collection and column packing, as well as possible breakdown ofparticulate interstitial organic material within the sand-dominatedmatrix.

Nutrient removal efficiencies varied by analyte and by-product, butwere >85% in some cases (e.g. NOx-N attenuation using CFA, andCaCO3- and CaO-based WTR; PO4-P and TP attenuation by CaO-basedWTR; Table 3). The GAC examined herein was particularly effectivefor DOC attenuation, exhibiting >4% (w/w) DOC retention. The P

Fig. 2. Potential nutrient limitation diagrams showing the distribution of influent waterand effluents from native sand reference columns and experimental columnscontaining coal fly ash, CaCO3- and CaO-based water treatment residuals (WTRs) andgranular activated carbon relative to the respective zones of potential N limitation (in-dicated by –N, top), P limitation (indicated by –P, middle) and Si limitation (indicatedby –Si, bottom).

Table 5Simple correlation coefficients (r) between DOC, DON and PO4-P removal (mg/kg)from influent water by CaO- and CaCO3-based water treatment residuals, coal fly ash,and biomass-derived granular activated carbon in column trials and material composi-tion, and between concentrations (mg/L) of DOC, DON and PO4-P removed from influ-ent water and major cations in column effluents.

CaO (%) MgO (%) Fe2O3 (%) Al2O3 (%)

Across all treatmentsDOC removed (mg/kg) −0.44 0.21 0.14 0.22PO4-P removed (mg/kg) 0.70a 0.83 0.07 0.04DON removed (mg/kg) −0.72 −0.22 0.06 0.14

Ca (mg/L) Mg (mg/L) Fe (mg/L) Al (mg/L)

CaO-based WTRDOC removed (mg/L) 0.29 −0.13 −0.34 0.78PO4-P removed (mg/L) 0.23 0.37 −0.27 0.25DON removed (mg/L) −0.62 0.22 −0.05 0.25

CaCO3-based WTRDOC removed (mg/L) −0.52 0.60 −0.25 −0.36PO4-P removed (mg/L) 0.77 −0.87 0.34 0.39DON removed (mg/L) −0.67 0.65 −0.31 −0.36

Coal fly ashDOC removed (mg/L) 0.59 −0.41 −0.73 −0.04PO4-P removed (mg/L) 0.67 −0.33 −0.66 −0.07DON removed (mg/L) 0.63 −0.39 −0.61 −0.06

Granular activated carbonDOC removed (mg/L) 0.08 0.01 −0.80 −0.32PO4-P removed (mg/L) −0.48 −0.15 0.65 0.47DON removed (mg/L) 0.01 0.33 −0.48 −0.56

a Simple correlation (r)≥0.70 shown in bold text.

70 L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

sorption capacities determined for the selected by-products in batch ex-periments ranged from 1.8 mg/g (CFA) to 3.6 mg/g (CaO-based WTR),which is greater than many phyllosilicate minerals and approximatelyequal to the 3.5 mg P/g sorption capacity reported for the strongly Psorbing mineral goethite (α-FeOOH; He et al., 1994).

Several studies have demonstrated the potential for productiveuse of CFA for P removal (e.g. Drizo et al., 1999; Oguz, 2005; Chenet al., 2007); however, a major barrier to CFA utilisation in environ-mental applications is the often substantial enrichment of a range ofpotentially toxic trace elements in the residue following coal combus-tion. Thus, potentially toxic trace elements in CFA column effluentswere closely monitored during this study. The column trial was ter-minated when Se in CFA column effluents exceeded fresh and marinewater quality guideline concentrations for aesthetic and recreationalpurposes (ANZECC/ARMCANZ, 2000). Concentrations of As and Cdin CFA column effluents exceeded Australian and New Zealand guide-line concentrations for drinking water quality. These findings do notnecessarily preclude CFA from use in environmental applications,but indicate that elutriation tests and environmental toxicity screen-ing should be conducted prior to field application. The trace elementleaching observed herein is consistent with numerous other studiesreporting the mobilisation of potentially toxic trace elements suchas As, Cr, Mo, and V from CFA (e.g. Phung et al., 1979; Qafoku et al.,1999; Quispe et al., 2012; Vassilev and Vassileva, 2007); however,CFA zeolitisation may reduce the lability of trace elements (Querolet al., 2002). Heat and/or alkali treatment of CFA can be used toform zeolitised fly ash (ZFA) which has been demonstrated to effec-tively attenuate both NH4-N and PO4-P in wastewater (Murayamaet al., 2003; Wu et al., 2006).

Worldwide, an increase has been observed in the use of activated car-bon for water treatment, particularly for removal of organic chemicalsand natural organicmatter (Enecon, 2001; Dias et al., 2007). The GAC ex-amined herein proved effective for DOC and DON removal from influentwater in column trials. This result is supported by a study conducted byEnecon (2001) demonstrating similar or superior removal of phenol,

tannin, methylisoborneol (MIB), microcystin and atrazine from waterby the GAC used in this study in comparison to commercially-availableactivated carbon (Enecon, 2001). Pore size distribution, surface chemis-try andmineral content are characteristics which affect adsorption of or-ganic compounds by activated carbon. The demonstrated ability of theGAC used in this study to remove both MIB, a small organic molecule,and large tannic acid molecules (Enecon, 2001) is indicative of broadpore size distribution and ability to remove awide range of contaminantsfrom water.

Although nutrient concentrations are largely responsible for stim-ulating algal growth, the total nutrient load of an aquatic ecosystemdetermines the biomass of aquatic plants. Transformation processeswhich occur in a body of water can release additional nutrients, forexample from sediments and suspended particles. Attenuation ofDOC in wastewaters is also important because it has a substantial in-fluence on the concentration of dissolved oxygen. The dissolved oxy-gen concentration in a water body is dependent upon the balancebetween the flux of biologically available organic carbon and therate of heterotrophic respiration, and the diffusion of atmospheric ox-ygen and oxygen generated via photosynthesis by macrophytes andphytoplankton. The GAC used in this column trial demonstrated ef-fective removal of DOC and TN from influent water, largely due to at-tenuation of DON and NH3-N (Table 3). The biomass-derived GAC

71L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

used in this study did not attenuate P in column trials, but removed>2 g P/kg in batch experiments, indicating that an increased contacttime or greater mixing than occurred in column trials may enhance Premoval from water by the GAC.

Water treatment residuals have demonstrated utility as soilamendments to mitigate off-site transport of applied P (Dayton etal., 2003; Ippolito et al., 2003, 2011), and as sorptive media for usein constructed wetlands designed to treat high-P waters (Babatundeet al., 2009). In column trials, the granular CaCO3-based WTR wasless effective for nutrient and DOC removal from influent waterthan the ground CaO-basedWTR, likely at least partially due to differ-ences in the availability of reactive surface area. The reactive surfacearea of the CaCO3-based WTR, which had a physical morphology of1–2 mm granules or beads, was substantially less than that of thepowdered (b63 μm particle size) CaO-based WTR.

Iron deposition on CaCO3 WTR granule surfaces which occurredduring treatment of Fe-rich groundwater accounts for the Fe contentof the CaCO3-based WTR. The poorly crystalline, metastable hydratediron oxide ferrihydrite forms via the rapid hydrolysis of Fe3+ salts orrapid oxidation of solubilised Fe2+, and is the precursor ofmore crystal-line iron oxides such as goethite and hematite. The Fe-rich CaCO3-basedWTRwas expected to attenuate PO4

2− in nutrient-rich water due to thehigh selectivity of poorly crystalline Fe oxides for phosphate anions;however, the CaCO3-basedWTR exhibited little capacity for PO4

2− sorp-tion in column trials and a net release of PO4

2− from CaCO3-based WTRwas observed in batch experiments. This is likely due to exhaustion ofthe P sorption capacity of the CaCO3-based WTR during groundwatertreatment, limiting the potential re-use of this material for nutrient at-tenuation. Geochemical modeling did not indicate saturation of eitherhydroxyapatite or iron phosphate minerals in effluents from theCaCO3-based WTR column.

At pH 10–11, crystalline hydroxyapatite has been shown to formfrom amorphous Ca phosphate in approximately 24 h (Liu et al.,2001); however, despite the high Ca content of the CaO-based WTRused in this study, the initially high pH of CaO-based WTR column ef-fluents and the high proportion of P removed from influent water,geochemical modeling indicated that hydroxyapatite remained theo-retically undersaturated throughout the column trial and thus wasunlikely to have been a mechanism of P removal.

Although the CaO-based WTR demonstrated effective attenuationof inorganic N species, DOC and P in column trials, and a high capacityfor P sorption in batch experiments, it proved unsuitable for DON re-moval from water. Both the CaO- and CaCO3-based WTRs exhibited anet release of DON in the first 20 and 14.7 L of column effluents, re-spectively (Fig. 1). As the column trial continued, a slight net decreasein DON in CaO-based WTR column effluents was observed as com-pared to influent water, likely due to DON mineralisation in situ.Thus, while not a net source of DON the CaO-basedWTR was not suit-ed for targeted attenuation of organic N species in wastewater.

5. Conclusions

Of the industrial by-products examined herein, only CFA demon-strated effective removal of a broad suite of both inorganic and organ-ic nutrients and DOC; however, the CaO-based WTR exhibited a highproportion of P removal from nutrient-rich water and substantial Psorption capacity, whilst the GAC was particularly effective for DONand DOC removal from water. Despite its demonstrated efficacy fornutrient and DOC removal, comprehensive elutriation testing of CFAand its derivatives (e.g. ZFA) should be conducted prior to use ofthis by-product or derivatives in nutrient intervention schemes dueto the potential for leaching of trace elements. The CaO-based WTRinvestigated here demonstrated substantial potential for use in nutri-ent intervention schemes designed for DOC, P and/or inorganic Nattenuation, but proved unsuitable for DON attenuation. In contrast,the GAC exhibited suitability for removal of organic contaminants,

including organic N species, from water but limited applicability forP attenuation. It may be possible to accomplish broad-scale nutrientand DOC removal using a combination of CaO-based WTR and GAC;however, further investigation of possible synergistic or antagonisticeffects is required.

The CFA, GAC, and CaO-based WTR investigated in this study dem-onstrate potential to mitigate eutrophication of freshwater systemsthrough their targeted application within nutrient intervention struc-tures such as constructed wetlands, permeable reactive barriers or sim-ilar deployments where the characteristics of influent wastewater areknown and matched to sorbent properties of individual by-products.In particular, effluents from the column containing CaO-based WTRexhibited nutrient ratios characteristic of potential P-limitation withno potential N- or Si-limitation respective to growth of aquatic biota.This result indicated that treatment of nutrient- and DOC-rich waterusing the CaO-based WTR may result in conditions less favourable forcyanobacterial growth and more favourable for growth of diatoms.Where acidity derived from anthropogenic sources or acid sulphatesoils is also present, incorporation of alkaline materials such as CaO-based WTR or GAC may function to attenuate acidity in wastewaterswhilst attenuating nutrients and DOC.

Acknowledgements

The authors gratefully acknowledge the Western Australian Gov-ernment for support provided through the Water Foundation andthe Department of Water.

References

Abe K, Komada M, Ookuma A. Efficiency of removal of nitrogen, phosphorus and zincfrom domestic wastewater by a constructed wetland system in rural areas: acase study. Water Sci Technol 2008;58:2427–33.

Ahmaruzzaman M. A review on the utilization of fly ash. Prog Energy Combust2010;36:327–63.

Anderson DM, Glibert PM, Burkholder JM. Harmful algal blooms and eutrophication:Nutrient sources, composition, and consequences. Estuaries 2002;25:704–26.

ANZECC/ARMCANZ. Australian and New Zealand guidelines for fresh and marine waterquality. National Water Quality Management Strategy no. 4A. October 2000.Canberra, Australia: Australian and New Zealand Environment and ConservationCouncil and the Agriculture and Resource Management Council of Australia andNew Zealand; 2000.

Babatunde AO, Zhao YQ, Burke AM, Morris MA, Hanrahan JP. Characterization ofaluminium-based water treatment residual for potential phosphorus removal inengineered wetlands. Environ Pollut 2009;157:2830–6.

Brei VV, Melezhyk OV, Starukh GM, Oranskaya EI, Mutovkin PA. Organic precursor syn-thesis of Al-Mg mixed oxides and hydrotalcites. Microporous Mesoporous Mater2008;113:411–7.

Chen J, Kong H, Wu D, Hu Z, Wang Z, Wang Y. Phosphate immobilization from aqueoussolution by fly ashes in relation to their composition. J Hazard Mater 2007;B139:293–300.

Dayton EA, Basta NT, Jakober CA, Hattey JA. Using water treatment residuals to reducephosphorus in agricultural runoff. J Am Water Works Assoc 2003;95:151–9.

Dias JM, Alvim-Ferraz MCM, Almeida MF, Rivera-Utrilla J, Sánchez-Polo M. Waste ma-terials for activated carbon preparation and its use in aqueous-phase treatment: areview. J Environ Manage 2007;85:833–46.

Drizo A, Frost CA, Grace J, Smith KA. Physico-chemical screening of phosphate-removingsubstrates for use in constructed wetland systems. Water Res 1999;33:3595–602.

Enecon. Integrated Tree Processing of Mallee Eucalypts. Publication No. 01/160, projectno. OIL-3A. Barton Australia: Rural Industries Research and development Corpora-tion; 2001.

Gibbons MK, Mortola MM, Gagnon GA. Phosphorus adsorption on water treatment re-sidual solids. J Water Supply Res Technol 2009;59:1-10.

He ZL, Yang X, Yuan KN, Zhu ZX. Desorption and plant-availability of phosphate sorbedby some important minerals. Plant Soil 1994;162:89–97.

Ioannidou O, Zabaniotou A. Agricultural residues as precursors for activated carbonproduction – a review. Renew Sustain Energy Rev 2007;11:1966–2005.

Ippolito JA, Barbaric KA, Heil DM, Chandler JP, Redente EF. Phosphorus retention mech-anisms of a water treatment residual. J Environ Qual 2003;32:1857–64.

Ippolito JA, Barbarick KA, Elliott HA. Drinking water treatment residuals: a review of re-cent uses. J Environ Qual 2011;40:1-12.

Justić D, Rabalais NN, Turner RE, Dortch Q. Changes in nutrient structure ofriver-dominated, coastal waters: stoichoimetric nutrient balance and its con-sequences. Estuar Coast Shelf Sci 1995;40:339–56.

King KW, McDonald J, Moore JF, Agrawal SG, Fischer EN, Balogh JC. Nutrient and pesticideremoval from laboratory-simulated tile drainage discharge. Trans ASABE 2010;53:769–77.

72 L.A. Wendling et al. / Science of the Total Environment 442 (2013) 63–72

Kivaisi AK. The potential for constructed wetlands for wastewater treatment and reusein developing countries: a review. Ecol Eng 2001;16:545–60.

Kroeger KD, Cole ML, Valiela I. Groundwater-transported dissolved organic nitrogenexports from coastal watersheds. Limnol Oceanogr 2006;51:2248–61.

Liu C, Huang Y, Shen W, Cui J. Kinetics of hydroxyapatite precipitation at pH 10 to 11.Biomaterials 2001;22:301–6.

Murayama N, Yoshida S, Takami Y, Yamamoto H, Shibata J. Simultaneous removal ofNH4

+ and PO43− in aqueous solution and its mechanism by using zeolite synthe-

sized from coal fly ash. Sep Sci Technol 2003;38:113–29.Murphy J, Riley JP. A modified single solution method for the determination of phos-

phate in natural waters. Anal Chim 1962;27:31–6.Murphy PNC, Stevens RJ. Lime and gypsum as source measures to decrease phosphorus

loss from soils to water. Water Air Soil Pollut 2010;212:101–11.Oguz E. Sorption of phosphate from solid/liquid interface by fly ash. Colloids Surf A

Physicochem Eng Asp 2005;262:113–7.Parkhurst DL. Users guide to PHREEQC — a computer program for speciation,

reaction-path, advective transport, and inverse geochemical calculations. USGSWater Research Investigation Report 95–4277. Washington, D.C.: United StatesGeological Survey; 1995

Petrone KC. Catchment export of carbon, nitrogen, and phosphorus across an agro-urbanland use gradient, Swan-Canning River system, southwestern Australia. J GeophysRes 2010;115:G01016.

Petrone KC, Richards JS, Grierson PF. Bioavailability and composition of dissolved or-ganic carbon and nitrogen in a near coastal catchment of south-western Australia.Biogeochemistry 2009;92:27–40.

Phung HT, Lund LJ, Page AL, Bradford GR. Trace elements in fly ash and their release inwater and treated soils. J Environ Qual 1979;8:171–5.

Playford PE, Cockbain AE, Low GH. Geology of the Perth Basin Western Australia. Perth:Geological Survey of Western Australia; 1976.

Qafoku NP, Kukier U, Sumner ME, Miller WP, Radcliffe DE. Arsenate displacement fromfly ash in amended soils. Water Air Soil Pollut 1999;114:185–98.

Querol X, Moreno N, Umaña JC, Alastuey A, Hernández E, López-Soler A, et al. Synthesisof zeolites from coal fly ash: an overview. Int J Coal Geol 2002;50:413–23.

Quispe D, Pérez-López R, Silva LFO, Nieto JM. Changes in mobility of hazardous ele-ments during coal combustion in Santa Catarina power plant (Brazil). Fuel2012;94:495–503.

Rayment GE, Higginson FR. Australian Laboratory Handbook of Soil and Water Chemi-cal Methods. Melbourne: Inkarta Press; 1992.

Sattayatewa C, Pagilla K, Sharp R, Pitt P. Fate of organic nitrogen in four biological nu-trient removal wastewater treatment plants. Water Environ Res 2010;82:2306–15.

Scott WD, Wrigley TJ, Webb KM. A computer model of struvite solution chemistry.Talanta 1991;38(8):889–95.

Scott D, Harvey J, Alexander R, Schwarz G. Dominance of organic nitrogen from head-water streams to large rivers across the conterminous United States. GlobBiogeochem Cycles 2007;21:GB1003.

Seitzinger SP, Sanders RW. Atmospheric inputs of dissolved organic nitrogen stimulateestuarine bacteria and phytoplankton. Limnol Oceanogr 1999;44:721–30.

Seitzinger SP, Sanders RW, Styles R. Bioavailability of DON from natural and anthropo-genic sources to estuarine plankton. Limnol Oceanogr 2002;47:353–66.

Sholkovitz ER. Flocculation of dissolved organic and inorganic matter during themixing of river water and seawater. Geochim Cosmochim Acta 1976;40:831–45.

Silva LFO, DaBoit K, Sampaio CH, Jasper A, Andrade ML, Kostova IJ, et al. The occurrenceof hazardous volatile elements and nanoparticles in Bulgarian coal fly ashes andthe effect on human health exposure. Sci Total Environ 2012a;416:513–26.

Silva LFO, Jasper A, Andrade ML, Sampaio CH, Dai S, Li X, et al. Applied investigation onthe interaction of hazardous elements binding on ultrafine and nanoparticles inChinese anthracite-derived fly ash. Sci Total Environ 2012b;419:250–64.

Silva LFO, Sampaio CH, Guedes A, Fdez-Ortiz de Vallejuelo S, Madariaga JM.Multianalytical approaches to the characterisation of minerals associated withcoals and the diagnosis of their potential risk by using combined instrumentalmicrospectroscopic techniques and thermodynamic speciation. Fuel 2012c;94:52–63.

Vanderlaan RK, White JL, Hem SL. Formation of hydrotalcite in mixtures of aluminiumhydroxycarbonate and magnesium hydroxide gels. J Pharm Sci 1982;71:780–6.

Vassilev SV, Vassileva CG. A new approach for the classification of coal fly ashes basedon their origin, composition, properties, and behaviour. Fuel 2007;86:1490–512.

Werker AG, Dougherty JM, McHenry JL, Van Loon WA. Treatment variability for wet-land wastewater treatment design in cold climates. Ecol Eng 2002;19:1-11.

Wu D, Zhang B, Li C, Zhang Z, Kong H. Simultaneous removal of ammonium andphosphate by zeolite synthesized from fly ash as influenced by salt treatment.J Colloid Interface Sci 2006;304:300–6.