Phosphorus removal by wollastonite: A constructed wetland substrate

Ecological Engineering 15 (2000) 121–132

Phosphorus removal by wollastonite: A constructed wetlandsubstrate

Andrea S. Brooks a, Melissa N. Rozenwald a, Larry D. Geohring a,Leonard W. Lion b, Tammo S. Steenhuis a,*

a Department of Agricultural and Biological Engineering, 226 Riley-Robb Hall, Cornell Uni6ersity, Ithaca, NY 14853, USAb Department of Ci6il and En6ironmental Engineering, Cornell Uni6ersity, Ithaca, NY 14853, USA

Received 7 August 1998; received in revised form 17 September 1999; accepted 22 October 1999

Abstract

Wollastonite, a calcium metasilicate mineral mined in upstate New York, is an ideal substrate for constructedwetland ecosystems for removing soluble phosphorus from secondary wastewater. Design parameters, required fordesigning a full-scale constructed wetland, were measured in vertical upflow columns with hydraulic residence timesvarying from 15 to 180 h. Secondary wastewater was pumped vertically upward through eleven soil columns, 1.5 min length and 15 cm in diameter and influent and effluent concentrations of soluble phosphorus were monitored forup to 411 days. Greater than 80% removal (up to 96%) was observed in nine out of 11 columns and effluentconcentrations of soluble phosphorus ranged from 0.14 to 0.50 mg/l (averaging 0.28 mg/l) when the residence timewas \40 h. Columns with a decreased residence time averaged 39% removal. A direct relationship between residencetime and soluble phosphorus removal was established. © 2000 Elsevier Science B.V. All rights reserved.

Keywords: Tertiary wastewater treatment; Wastewater; Constructed wetlands; Phosphorus removal; Sewage; Water quality;Secondary sewage effluent; Loading rate

www.elsevier.com/locate/ecoleng

1. Introduction

Excessive phosphorus loading from manmadesources is commonly linked to eutrophication oflakes and streams (Hillbricht-Ilkowska et al.,1995). Phosphorus is contributed by agriculturalrunoff and erosion from agricultural land, urbanrunoff and the most easily manipulated source

and focus of this paper, wastewater effluent. Al-though most communities in the USA have pri-mary and secondary treatment, in the near futuremany of them will be required to have tertiarytreatment, to reduce wastewater effluent concen-trations of phosphorus (P) to levels as low as 1.0mg/l (EPA, 1987). Very few ecosystems exist, thatcan reduce P to those levels.

The use of both natural and created wetlands astools in the treatment of polluted waters, is in-creasing in popularity as an ecological engineeringalternative to conventional, chemical based meth-

* Corresponding author. Tel.: +1-607-2552489; fax: +1-607-2554080.

E-mail address: [email protected] (T.S. Steenhuis)

0925-8574/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.

PII: S 0925 -8574 (99 )00056 -7

A.S. Brooks et al. / Ecological Engineering 15 (2000) 121–132122

ods (Mitsch and Jørgensen, 1989; Kadlec andKnight, 1996). Wetland ecosystems have the ad-vantages of being self-perpetuating, low-mainte-nance and cost-efficient treatment systems(Fennessy and Mitsch, 1989). Constructed wet-land ecosystems, offer better opportunities forwastewater treatment than natural wetlands (Ver-hoeven and Meuleman, 1999). In treating sewagewater for small to medium sized communities,they can be used for reducing BOD and sus-pended solids from secondary wastewater (Reedand Brown, 1992; Kern and Idler, 1999; Verho-even and Meuleman, 1999). P removal has beenvariable at best (Watson et al., 1989; Mann, 1990;Jenssen et al., 1993; Greenway and Woolley,1999) because substrates are often selected basedon local availability and particle size for reducedclogging, without consideration for their capacityfor P removal (Watson et al., 1989; Mann, 1990).One of the ecotechnological means of attainingsufficient low P concentration, is the use of spe-cialized substrates with physical and chemicalproperties conducive to P removal, while main-taining sufficient permeability (House et al.,1994). Specialized substrates used for P removalin wetlands, are pumice, sand, LECA (a reactiveporous media) and industrial wastes such as blastfurnace slag and fly ash (Jenssen et al., 1993;Mann and Bavor, 1993; Nur Onar et al., 1996).Wollastonite, a calcium metasilicate mined innorthern New York state, is abundant and thetailings left after the production of pure wollas-tonite are inexpensive. Preliminary studies haveshown promise for efficient P removal with Wol-lastonite tailings (Corrigan, 1993; Geohring et al.,1995). The mineral composition of wollastonite isreported to be close to 50% CaO (Buckley, 1997).P Removal in mediums, with high calcium con-tent is typically attributed to either adsorption(Iannou et al., 1994; Nur Onar et al., 1996) orchemical precipitation of calcium phosphates(Hundal, 1988; House et al., 1994; Agbenin,1996). Chemical precipitation of P by wollastoniteis likely, as a stable mineral Hydroxyapatite(Ca5(PO4)3(OH)(s)). The equilibrium concentrationof this compound can be as low as 5.9×10−4

mg/l as phosphorus (Brooks, 1997). Adsorption isalso possible through either ion or ligand ex-change at soil surfaces.

The use of wollastonite in constructed wetlandecosystems for P removal, requires the quantify-ing of the mechanism(s) facilitating P removal.Because these types of ecosystems are so complex(Mitsch and Jørgensen, 1989) the quantificationof their reactions is initially best done, under wellcontrolled conditions in the laboratory beforefull-scale implementation. Our first objective was,therefore, to obtain design parameter (such asrequired retention for optimal P removal). Be-cause horizontal flow wetlands, may have over-land flow complicating the interpretation of theresults, we used a vertical upflow system. Solublephosphorus removal was monitored over a widerange of residence times. A second objective ofthis research was to come to a more comprehen-sive understanding and characterization of themechanism involved in P sorption from wollas-tonite, through the use of controlled laboratorybatch experiments. Previous research has indi-cated that a lengthy residence time (days toweeks) facilitates high P removal in wetland sys-tems (Jenssen et al., 1993). In order to maximize Premoval, hydraulic residence times on a scale ofdays were included in this study.

2. Material and methods

2.1. Batch experiments

Simple batch experiments were performed, pri-marily to characterize the phosphorus removalmechanism of wollastonite. Time series batch ex-periments were conducted by mechanically agitat-ing 125 ml sample bottles with a 20:1 solution/soilratio (40 ml phosphate solution to 2 g of purewollastonite) for designated periods of time rang-ing from 0.1 to 72 h and measuring the residualconcentration of phosphate in solution. This wasdone at initial concentrations of both 5 and 10mg/l. The soluble phosphorus was measured bythe ascorbic acid method (APHA et al., 1985).

A second set of time series experiments wasdone, to determine the effects of calcium phos-phate seeding on P removal. This was done bymixing 10 mg/l phosphate solution with 2 g ofwollastonite and agitating the mixture for 3 days.

A.S. Brooks et al. / Ecological Engineering 15 (2000) 121–132 123

The wollastonite was then filtered out of thesample, using a 0.45 mm filter and then dried in anoven at 40°C for 24 h. The filtrate was analyzedfor residual P concentration. The dried wollas-tonite was then used as a calcium phosphate ‘seed’in kinetic experiments. One gram of the driedwollastonite and 1 g of ‘new’ wollastonite werecombined and then mixed with a 5 mg/l phospho-rus solution and the phosphorus concentrationwas monitored over 3 days.

2.2. Column experiments

Phosphorus removal by wollastonite was stud-ied in vertical upflow columns of 1.5 m length.Columns were constructed with 15 cm diameterpoly-vinylchloride (PVC) tubes, cut to 1.5 mlengths, with sampling ports located 10 cm fromthe top of each column. The columns were cappedat the base with a circular piece of PVC with aport in the center, that served as the influentconnection. Fig. 1 shows the general layout of a

filled vertical upflow column. All experimentswere run in a constant “temperature room” (18–21°C) in the Department of Agricultural and Bio-logical Engineering at Cornell University.

The columns were filled in the following order:A wire screen was placed on the bottom of thecolumn, followed by 15 cm of washed gravel, asecond screen and then the column was filled withthe wollastonite substrate up to :2 cm below theeffluent sampling port.

The vertical upflow system consisted of peri-staltic pumps and tubing, that ran from a waste-water source tank, through a sand filter and intothe bottom port of a column. Secondary waste-water was pumped up through the wollastonitematrix to :2 cm below the effluent port, wherethe water was allowed to pond to obtain averageeffluent concentrations from the sampling port.

The inflow system consisted of irrigation tubingthat fed directly into each column individuallyfrom a 100 l tank, filled with secondary wastewa-ter in an adjacent room. Influent wastewater camefrom the Cayuga Wastewater Treatment Plant inIthaca, NY. Wastewater was picked up weekly–biweekly in a 1100 l tank and pumped into a 600l storage tank, with the residual wastewater storedat RT until it was needed. The secondary effluentwas then pumped from the 600 l holding tank,through a sand filter to remove suspended solidsand into the previously mentioned 100 l tank, thatwas the source tank for the influent tubing. Influ-ent concentrations of soluble reactive phosphoruswere measured before and after the sand filter.The sand filter was box shaped and constructedwith PVC sheets, with the following dimensions:0.6×0.6×1.2 m. The filter was filled with :7.5cm of gravel followed by a geo-textile sheet and38.1 inches of fine blasting sand (effective diame-ter 0.45–0.55 mm) and 38.1 inches of anthracitecoal (effective diameter 0.9–1.0 mm).

Effluent and influent samples of the columnsfor phosphorus analysis, were taken from thecolumn every 3 days. The effluent flow rate wasmeasured each time samples were taken, as weremeasurements of the pressure differential betweeninflow and outflow. Samples were either analyzedfor soluble reactive phosphorus using the ascorbicacid method on the date they were taken, orFig. 1. Schematic of vertical upflow column.


Fig. 2. Time dependence of phosphorus removal.

stored overnight at 4°C for measurement the fol-lowing day. If it was not possible to measure thesamples within 2 days, the samples were storedfrozen until they could be analyzed.

Tap water inflow with a high phosphorus con-centration (:5–12 mg/l), was made by addingpotassium phosphate dihydrate (KH2PO4) to tapwater, fed into two of the columns and sampledas described above.

Eight columns were filled with an 1:1 mixture ofwollastonite tailings, the by-product of the miningprocess and pure wollastonite powder, two werefilled with a 3:1 tailings/pure wollastonite mixtureand one with tailings alone. Columns were moni-tored for up to 411 days. The mixture of tailingsand pure wollastonite, was created by addingequal volumes of each substrate into a cementmixer and agitating for 15 min. Six columns had asecondary wastewater influent source and two hadtap water with added phosphorus (KH2PO4). Theinitial residence time (RT1) was \40 h in all buttwo of the columns. The residence time was re-duced in some columns after close to 50 days ofoperation (RT2) to monitor changes in theeffluent concentrations of soluble reactive phos-phorus based as a function of residence time. Theaverage residence time was based on measuredeffluent flow rates using the following equation:

Trave=

pr2hf

Qave

(1)

where Traveis the average residence time (h); Qa6e

is the average measured flow rate (ml/h); r is theradius of the column (cm); h is the height of thesubstrate in the column (cm); f is the porosity ofthe substrate.

The RT1 and RT2 of each column, as well asother attributes of the individual columns, areshown in Table 1.

3. Results


Time series experiments were done, by mechan-ically agitating known concentrations of phos-phate solution with wollastonite for designatedperiods of time and then measuring the dissolvedphosphate left in solution. Fig. 2 shows the timedependence of phosphorus removal in batch ex-periments. Ninety percent removal was observedin the first 12 h of the experiment and 98–100%removal was achieved after 72 h of mechanicalshaking (data not shown). The data points areaverages obtained from three experiments donewith initial concentrations of 5 and 10 mg/l andthe error bars represent91 S.D.

A.S

.B

rookset

al./E

cologicalE

ngineering15

(2000)121

–132

125

Table 1Vertical columns experimental summarya

C4 C5 C6 C7 C8 C9 C10 C11C2Column C3C1

3:1 T+P 1:1 T+P 1:1 T+P 1:1 T+P 1:1 T+P 1:1 T+P 1:1 T+P TailingsSubstrate 1:1 T+P1:1 T+P 3:1 T+P0.54 0.54 0.54 0.54 0.540.54 0.34Porosity 0.470.470.540.54

52 78 67 70RT1 (h) 8054 4059 22 22 9046 161 161 130 21346 4646 46Days of Operation 95 141

17160 17 228 17 2319 15 45RT2 (h)Days of Operation 115213 83 115115 115 115 115

SW SW SW TW TWSW SWSWInfluent Source SW SW SW

a T+P, tailings/pure wollastonite powder; SW, secondary wastewater; TW, tap water/phosphate solution; RT1,2, initial and secondary residence times.


The addition of a crystal seed of calcium phos-phate was accomplished, by adding wollastonitethat had been previously agitated with a phos-phate solution for 72 h (98% measured removal)and dried overnight. Time dependence of phos-phorus removal in experiments carried out withthe seeding process are shown in Fig. 3. After 2h, 27% of the phosphorus had been removed inthe experiment with seeding, versus 16% in thetime experiments without seeding. However, netremoval in both seeded and non seeded experi-ments was the same, 80% after 8 h and 90% after12 h.


The percent decrease of soluble reactive phos-phorus (SP) in secondary wastewater, treated invertical upflow columns at RT1 and RT2 as calcu-lated by Eq. (1), is given in Table 2. The averageeffluent concentration, S.D. and percent removalare averages over the number of days of opera-tion shown for RT1 and RT2. The results shown,permit comparisons of residence time, as well asinfluent source.

A maximum average removal of 96% was ob-

served in column C9 over a period of 83 days.The longest period of successful removal (\85%), was observed in column C1. Fig. 4 showsthe influent and effluent concentrations of solublereactive phosphorus in column C1. The sec-ondary wastewater inflow varied between 0.16and 9.8 mg/l (average is 3.4 mg/l) and the averageeffluent concentration over a period of 411 dayswas 0.3 mg/l. Column C1 removed, on average,89% over the first 95 days of operation and thenremoved 90% over the next 213 days. The columnshowed no signs of a decrease in effectivenessover the duration of this experiment. Removalsof soluble reactive phosphorus \80% wereachieved in nine out of 11 vertical upflowcolumns, with average effluent concentrations ofSP ranging from 0.14 to 0.50 mg/l (averaging0.28 mg/l).

Fig. 5 shows the influent and effluent concen-tration of soluble reactive phosphorus in the ver-tical column, filled only with wollastonite tailings,without the addition of pure powder wollastonite.The tailings column was started at an RT1 of 40h. The residence time was decreased to 30 h after50 days of operation and then increased after:170 days of operation to over 100 h. The

Fig. 3. Time dependence of phosphorus removal with seeding.

A.S

.B

rookset

al./E

cologicalE

ngineering15

(2000)121

–132

127

Table 2Soluble reactive phosphorus (SP) removal in vertical flow columnsa

C6 C7 C8 C9Column C10C1 C11C2 C3 C4 C5

1:1 T+P 1:1 T+P 1:1 T+P 1:1 T+P 1:1 T+P3:1 T+P Tailings1:1 T+P3:1 T+P1:1 T+P1:1 T+PSubstrate22 91 49 71 63 84 93 40RT1 (h) 5854 26

0.15 0.28 0.31 0.29 0.180.14 0.190.38Average SP (mg/l) 0.680.650.24(0.05)(0.1) (0.06) (0.48) (0.59) (0.21) (0.12) (0.21)(0.11) (0.28) (0.33)(S.D.)

89 90 89 86 85 85 87 8491 53SP removal (%) 5118 179 1517 30RT2 (h) 451519143

2.17 0.25 0.57Average SP (mg/l) 1.410.21 1.86 2.54 2.07 2.23(1.7) (0.10) (0.23)(1.80) (1.09)(S.D.) (1.27)(1.47)(1.5)(0.10)

43 39 40 96 66SP removal (%) 90 49 36SW SW SW TW TWSW SWSWInfluent source SW SW SW

a T+P, tailings/pure wollastonite powder; SW, secondary wastewater; TW, tap water/phosphate solution; RT1,2, initial and secondary residence times.


Fig. 4. Effluent concentrations of soluble reactive phosphorus in column C1.

Fig. 5. Effluent concentrations of soluble reactive phosphorus in column C11, wollastonite tailings.

effluent concentrations, reflect these changes inthe column residence time (Fig. 5). Similar resultswere observed in all other wastewater influentcolumns. Columns receiving tap water influent(columns C9 and C10) maintained removal per-centages \80% regardless of residence time(Table 2).

4. Discussion


Phosphorus removal in batch experiments hasbeen attributed to adsorption in studies wheresignificant removal by pumice rock was observed


in 60 min (Nur Onar et al., 1996) and whereremoval by river sediments occurred within 10min (Fox et al., 1989). The removal reactionproceeded much more slowly in this experiment,although significant removal was observed on ascale of h. P removal associated with complexa-tion and precipitation of hydroxyapatite, occurredin B24 h in an experiment by Mann and Bavor(1993). Although 10% removal of P was observedin B2 h in this experiment, the most significantfractional removal occurred between 2 and 8 h.Hundal (1988) studied phosphorus removal by aNarrabi medium clay loam and found resultssimilar to this study, describing the removal in thefirst 6 h as nucleation of calcium phosphate onsoil surfaces and associated the second reaction(up to 24 h) with precipitation of amorphouscalcium phosphate over the nucleated sites. Basedon the kinetics of the reaction(s) taking place inthis experiment (Fig. 2, Fig. 3), it does not appearthat removal can be fully attributed to rapidadsorption reactions, but may indicate either acombination of both adsorption and a more com-plex reaction such as surface precipitation or asimply mineral dissolution and subsequent precip-itation of phosphate without the influence ofadsorption.

Ferguson (1971) showed that the time delaybefore phosphorus removal, occuring in a super-

saturated solution can be reduced, with the addi-tion of a crystal seed of precipitated calciumphosphate. The increased rate of calcium phos-phate precipitation is attributed to the eliminationof the induction period of crystal growth. Theincreased rate of P removal in the seeding experi-ments in this study (Fig. 3) indicates that precipi-tation may be occurring. Furthermore, anincreased rate of removal after seeding would notbe expected in a purely adsorptive process unlessthe specific surface area were somehow greatlyincreased by this process.


It has been shown repeatedly that phosphorusremoval from soil exhibits both rapid and slowstages (Ryden and Pratt, 1980; Nur Onar et al.,1996). This is visible in the batch experiments(Fig. 2). A similar trend was found in the columnexperiments after a data point at 72 h is includedin the graph (Fig. 6). The relationship betweenresidence time and percent removal in the verticalcolumns is shown in Fig. 7. The data shownrepresent average removal at RT1 and RT2 inseven different columns. The residence time wasnot changed in two columns and, therefore, thereis only an RT1 for each of those, resulting in atotal of 16 data points. The data from the tap

Fig. 6. Data from Fig. 2 plotted on a log scale with a data point at 72 h included.


Fig. 7. Comparison of residence time and percent removal of soluble reactive phosphorus.

water columns is not included. Percent removalincreases linearly with residence time until :43 h.The fractual removal appears to level off beyondthat time.

The initial rapid reaction is typically described asadsorption, but can also be attributed to theformation of calcium phosphates of high solubility(Frossard et al., 1995). The subsequent slowerreaction has been described as either chemicalprecipitation or incorporation of the adsorbate intothe solid matrix (McGlathery et al., 1994). Sposito(1984) indicated that the rapid phase usually lastsfor B50 h and the slow stage can continue beyond50 days. This is in agreement with the data fromthis experiment, as removal continued at a rapidrate up to :43 h (the first stage) and then appearedto level off (Fig. 6). The results of this experimentindicate that, in order to achieve maximum re-moval, the residence time in the vertical columnsmust be long enough to achieve at least the second(slow) stage of phosphorus removal.

The observed result, that the column filled withwollastonite tailings by themselves, followed thesame removal trend as the tailings/pure powdermixtures (Fig. 5) has significant economical conse-quences. The cost of the pure wollastonite powderis $308/ton compared to $3/ton for the tailings bythemselves. Future experiments will include repli-cates of columns filled with wollastonite tailings, atlengthy residence times (\40 h).

The relationship between increasing influent con-centrations of SP and percent removal in verticalcolumn C1 is shown in Fig. 8. The residence timein the column was initially 54 h, but was increasedto 160 h just before a period of increased influentconcentrations of soluble reactive phosphorus.While the effluent concentration increased in allother columns (residence times varying from 17 to78 h) with an increase in influent concentration, theaverage effluent concentration in column C1 actu-ally decreased by 0.13 mg/l. The same relationshipwas observed in tap water influent columns, butwas not seen in any of the other wastewater influentcolumns.

Fig. 8. Relationship between increasing influent concentrationand percent removal of SP.


The excellent removal observed in tap watercolumns (regardless of residence time) may beattributable, in part, to the form of phosphate insolution. The chemical phosphorus, added to thetap water influent source was potassium phos-phate monobasic (KH2PO4). This salt is a strongelectrolyte and will dissociate in solution produc-ing H2PO4

−, HPO4−2 and PO4

−3 ions immediatelyavailable for reaction. The wastewater phospho-rus will be a combination of ortho- and polyphos-phates, as well as organic phosphorus. Prolongedcontact with micro-organisms in wastewater, willcause polyphosphates to hydrolyze to orthophos-phates and that, along with competition withother compounds in the wastewater, may cause adelay in the removal process. An alternate possi-bility is that biofilm coatings formed in columnsreceiving wastewater input altered the wollas-tonite surface adsorptive properties and/or de-creased the rate at which wollastonite dissolved toproduce the Ca2+ that reacted with the phosphatespecies from a precipitate. Although the rapid andslow stages of removal observed in wastewaterinfluent columns were not observed in tap watercolumns, it is possible that they still exist, but atresidence times shorter than studiedexperimentally.

5. Conclusion

Vertical flow columns filled with a 1:1 mixtureof wollastonite powder and wollastonite tailingsare shown to be an efficient means of achievingsoluble reactive phosphorus removal, from sec-ondary municipal wastewater. Wollastonite tail-ings by themselves also show promise as a moreeconomic means of obtaining removal in con-structed wetland ecosystems, although further re-search would be necessary to verify the resultsfound in this study. Residence time is clearly asignificant factor in optimizing phosphorus re-moval in the design of full-scale wetland systems.The highest percent removal was achieved at resi-dence times \40 h. The relationship betweenpercent removal and residence time was exhibitedas two phases of soluble reactive phosphorus re-moval. The first suggests a rapid reaction, that is

attributed to adsorption or the formation of cal-cium phosphates of high solubility and the sec-ond, slower reaction, suggests the result fromeither chemical precipitation or incorporation ofthe adsorbate into the solid matrix. Phosphorusremoval by wollastonite substrate with a tap wa-ter inflow, was not as sensitive to residence timeas with wastewater. This difference was attributedto interference with other compounds found inwastewater or to biofilm modification of thecolumn substrate. Batch experiments showed 90%removal of phosphorus after 12 h of agitation andprecipitation of calcium phosphates was indicatedby an increased removal rate in seeding experi-ments. In summary, wollastonite tailings are anattractive ecotechnological alternative to chemicaltreatment methods for removing phosphorus froma secondary treated system. Other possible eco-engineering uses of wollastonite are as a substratefor dairy milk house wetland systems and forindividual leach field application, in areas withshallow soils near open water bodies.

References

Agbenin, J.O., 1996. Phosphorus sorption by three cultivatedSavanna alfisols as influenced by pH. Fert. Res. 44, 107–112.

APHA, AWWA and WPCP, 1985. Standard Methods forExamination of Water and Wastewater, 16th ed. AmericanPublic Health Association, Washington, DC.

Brooks, A.S., 1997. Phosphorus Removal by Wollastonite: AConstructed Wetland Substrate. Unpublished MS Thesis.Department of Agricultural and Biological Engineering,Cornell University, Ithaca, NY.

Buckley, M., 1997. NYCO Minerals Inc., 124 Mountain ViewDrive, Willsboro, NY 12996. Personal communications.

Corrigan, N., 1993. Selection Of Substrate For PhosphorusAdsorption In Artificially Constructed Wetlands. Unpub-lished National Science Foundation/Research Experiencefor Undergraduates Report. Department of Agriculturaland Biological Engineering, Cornell University, Ithaca,NY.

EPA, 1987. Phosphorus Removal: Design Manual. Environ-mental Protection Agency, Washington DC.

Fennessy, M.S., Mitsch, W.J., 1989. Design and use of wet-lands for renovation of drainage from coal mines. In:Mitsch, W.J., Jørgensen, S.E. (Eds.), Ecological Engineer-ing. An Introduction to Ecotechnology. Wiley, NY.

Ferguson, J.F., 1971. Effect of crystal seeding on rate ofcalcium phosphate precipitation. Water Res. 5, 369–381.


Fox, I., Malati, M.A., Perry, R., 1989. The adsorption andrelease of phosphate from sediments of a river receivingsewage effluent. Water Res. 23, 725–732.

Frossard, E., Brossard, M., Hedley, M.J., Metherell, A., 1995.Reactions controlling the cycling of P in soils. In: Tiessen,H. (Ed.), Phosphorus in the Global Environment. Wiley,Chichester, UK.

Geohring, L.D., Steenhuis, T.S., Corrigan, N., et al., 1995.Specialized substrates for phosphorus removal with con-structed wetlands. In: Campbell, K.L. (Ed.), Versatility ofWetlands in the Agricultural Landscape. ASAE Publica-tion 70IP0995. American Society of Agricultural Engi-neers, St. Joseph, MI, pp. 607–617.

Greenway, M., Woolley, A., 1999. Constructed wetlands inQueensland: Performance efficiency and nutrient bioaccu-mulation. Ecol. Eng. 12, 39–56.

Hillbricht-Ilkowska, A., Ryszkowski, L., Sharpley, A.N., 1995.Phosphorus transfers and landscape structure: ripariansites and diversified land use patterns. In: Tiessen, H. (Ed.),Phosphorus in the Global Environment. Wiley, Chichester,U.K.

House, C.H., Broome, S.W., Hoover, M.T., 1994. Treatmentof nitrogen and phosphorus by a constructed upland-wet-land wastewater treatment system. Water Sci. Tech. 29,177–184.

Hundal, H.S., 1988. A mechanism of phosphate adsorption onNarrabri medium clay loam soil. J. Agr. Sci. 111, 155–158.

Iannou, A., Dimirkou, A., Doula, M., 1994. Phosphate sorp-tion by calcium-bentonite as described by commonly usedisotherms. Comm. Soil Sci. Plant Anal. 25, 2299–2313.

Jenssen, P.D., Maehlum, T., Krogstad, T., 1993. Potential useof constructed wetlands for wastewater treatment in north-ern environments. Water Sci. Tech. 28, 149–157.

Kadlec, R.H., Knight, R.L., 1996. Treatment Wetlands. LewisPublishers, Boca Raton, FL.

Kern, J., Idler, C., 1999. Treatment of domestic and agricul-tural wastewater by reed systems. Ecol. Eng. 12, 13–26.

Mann, R.A., 1990. Phosphorus removal by constructed wet-lands: substratum adsorption. In: Cooper, P.F., Findlater,B.C. (Eds.), Constructed Wetlands in Water PollutionControl: Advances in Water Pollution Control. PergamonPress, Oxford, pp. 97–105.

Mann, R.A., Bavor, H.J., 1993. Phosphorus removal in con-structed wetlands using gravel and industrial waste sub-strata. Water Sci. Tech. 27, 107–113.

McGlathery, K.J., Marino, R., Howarth, R.W., 1994. Variablerates of phosphate uptake by shallow marine carbonatesediments: Mechanisms and ecological significance. Bio-geochemistry 25, 127–146.

Mitsch, W.J., Jørgensen, S.E. (Eds.), 1989. Ecological Engi-neering. An Introduction to Ecotechnology. Wiley, NY.

Nur Onar, A., Balkaya, N., Akyuz, T., 1996. Phosphateremoval by adsorption. Env. Tech. 17, 207–213.

Reed, S.C., Brown, D.S., 1992. Constructed wetland design:the first generation. Water Env. Res. 64, 776–781.

Ryden, J.C., Pratt, P.F., 1980. Phosphorus removal fromwastewater applied to land. Hilgardia 48, 1–36.

Sposito, G., 1984. The Surface Chemistry of Soils. OxfordUniversity Press, New York, NY.

Verhoeven, J.T.A., Meuleman, A.F.M., 1999. Wetlands forwastewater treatment: opportunities and limitations. Ecol.Eng. 12, 5–12.

Watson, J.T., Reed, S.C., Kadlec, R.H., Knight, R.L., White-house, A.E., 1989. Performance expectations and loadingrates for constructed wetlands. In: Hammer, D.A. (Ed.),Constructed Wetlands for Waste Water Treatment. LewisPublishers, Chelsea, MI, pp. 319–351.

.

Documents

Phosphorus removal by wollastonite: A constructed wetland substrate