Journal of Colloid and Interface Science 342 (2010) 437–444

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

Surface reactivity of a-Al2O3 and mechanisms of phosphate sorption: In situATR-FTIR spectroscopy and f potential studies

Mirella Del Nero a,*, Catherine Galindo a, Rémi Barillon a, Eric Halter b, Benoit Madé b

a Institut Pluridisciplinaire Hubert Curien, UMR 7178 CNRS/UdS, 23 rue du Loess, BP 28, 67037 Strasbourg Cedex 2, Franceb Ecole Nationale Supérieure des Mines, Mines ParisTech, Centre de Géosciences, 35 rue Saint Honoré, 77305 Fontainebleau Cedex, France

a r t i c l e i n f o a b s t r a c t

Article history:Received 30 June 2009Accepted 23 October 2009Available online 27 October 2009

Keywords:Alumina–solution interfacePhosphate ionsAdsorptionSurface precipitationIn situ spectroscopyf potentialSurface charge

0021-9797/$ - see front matter � 2009 Elsevier Inc. Adoi:10.1016/j.jcis.2009.10.057

* Corresponding author. Fax: +33 3 88 10 64 31.E-mail address: mireille.delnero@ires.in2p3.fr (M.

We have investigated the effect of solution parameters on the adsorption of phosphate ions and oncharges and structures, i.e., on the nature of species, at the a-Al2O3 colloid/solution interface by usingthe batch method, f potential measurements, and in situ ATR-FTIR spectroscopy. The uptake of phosphatedecreases with the extent of surface deprotonation (i.e., pH), imparts negative charges to the colloid sur-face, and induces IEP shifts showing chemical sorption. Use of complementary techniques provides evi-dence that phosphate is sorbed at low pH (3.3) by a combination of surface reactions of complexation andprecipitation, whose relative contributions depend on phosphate loading. Surface complexation includesfast reactions of ligand exchange with single coordinated hydroxyls, and electrostatic attraction of H2PO�4ions at positively charged surface sites. This is supported by experiments at low coverage showing sharpand linear decrease of f potential (i.e., surface charge) with amount of phosphate sorbed. At high cover-age, f potential values are low and independent of phosphate loading. Formation of surface precipitates ofAl-phosphate is inferred from the assignment of the ATR-FTIR absorption band at 1137 cm�1, whoseintensity increases with phosphate solution content and reaction time, to the P–O-stretching vibrationmode for phosphate sorbed at high concentrations on a-Al2O3. In situ ATR-FTIR spectroscopy reveals alsostructural reorganizations of surface hydroxyls with time, due to surface hydration and to surface precip-itation continuing over extended periods along alumina dissolution.

� 2009 Elsevier Inc. All rights reserved.

1. Introduction charged hydroxyls, through attractive mechanisms such as hydro-

Sorption of phosphate ligands on minerals and colloids receivesmuch attention in environmental chemistry. Phosphate has beenrecognized as an agricultural contaminant of surface waters andaquifers, due to the overapplication of P-containing fertilizers.Sorption reactions onto common soil minerals, such as iron andaluminum (hydrous) oxides, regulate the bioavailability and themobility of the phosphate anion and can retard significantly its dis-semination in natural waters. Phosphate ions also play an impor-tant role in the migration behavior of metals in ecosystems. Thebinding of phosphate ions on Fe-(hydr)oxide or Al-(hydr)oxide sur-faces is indeed expected to control the surface reactivity of manysoils and sediments toward dissolved metals. Identifying phos-phate species at oxide/solution interfaces is thus crucial for pre-dicting accurately the geo-cycling of phosphate and metals.

Much research work has been done on the adsorption mecha-nisms of phosphate ions on iron and aluminum (hydr)oxides. Stud-ies of the macroscopic phosphate sorption suggested inner-spheresurface complex (ISC) formation at surface hydroxyl groups (e.g.,[1]), outer-sphere surface complex (OSC) formation at positively

ll rights reserved.

Del Nero).

gen bonding (e.g., [2]) or electrostatic adsorption (e.g., [3]) ofH2PO�4 ions, and surface precipitation [4]. The phosphate surfacespeciation on Fe-(hydr)oxides has been widely investigated atthe molecular level via ex situ and in situ Fourier transform infrared(FTIR) spectroscopic analysis. The FTIR studies have shown that dif-ferent complexes of phosphate can (co)exist at the surface and thatthe speciation is highly dependent on pH, phosphate loading, andreactivity and concentration of surface groups. A bidentate binu-clear surface complex, which is protonated at low pH, was identi-fied at surfaces of goethite [5] and ferrihydrite [6]. Evidence wasfound that a monodentate complex also forms on goethite, at highpH and low phosphate concentration [5]. Elzinga and Sparks [7]showed that a mixture of two phosphate complexes existed atthe hematite surface at pH 3.5–7, with one species coordinatedas a monodentate binuclear (bridging) complex and the secondspecies as a monodentate mononuclear complex. They suggestedthat, alternatively, the bridging complex is a monodentate mono-nuclear complex exhibiting strong hydrogen bonding to an adja-cent site, and the second species a monodentate complexexhibiting limited hydrogen bonding. Regarding aluminum(hydr)oxides, surface studies by ex situ techniques have suggestedthe formation of both phosphate surface complexes and Al-phos-phate precipitates. Investigations by 31P solid-state nuclear mag-

438 M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444

netic resonance indicated that phosphate is bound on gibbsite as amixture of different species, i.e., of monodentate inner-sphere sur-face complex and Al-phosphate precipitates, with the latter speciesbeing dominant at low pH and at high phosphate concentration [8].Laiti et al. [9,10] used ex situ FTIR spectroscopy to study phosphatespecies at the c-alumina surface. They suggested that phosphateforms a monodentate, inner-sphere surface complex whose degreeof protonation is dependent on pH. They proposed that the com-plex is the precursor to the (surface) precipitation of a AlPO4 phase,in c-Al2O3 suspensions prepared at low pH and high phosphate/surface site ratios. However, the dynamics of phosphate ions atAl-(hydr)oxide/solution interfaces is unknown, since these systemshave only been investigated via ex situ spectroscopic analysis.Transition from surface complexation to precipitation, and forma-tion of precipitates, must be substantiated by in situ investigationsof structural changes occurring at the interface during the sorptionof phosphate and the dissolution of mineral.

The aim of the present study was to identify the adsorptionmechanisms of phosphate (denoted here as P) at the surface ofa-Al2O3 colloids by means of in situ techniques, i.e., attenuated to-tal reflectance (ATR)-FTIR spectroscopy and zeta (f) potential mea-surements. The effects of reaction time (tR) and phosphate loading([P]S) were investigated, with particular focus on the acidic pHwhich is reported to promote various mechanisms of phosphatesorption, including surface precipitation [8–10] and electrostaticattraction of H2PO�4 ions at positively charged surface groups[2,3]. Measurements of f potential variation with key parameterswere performed to determine the charges imparted to a surfaceby the adsorption of phosphate, which information is useful to dis-tinguish between different sorption species [4]. The phosphate sur-face speciation on a-alumina was investigated during the processof sorption by ATR-FTIR spectroscopy, using a method that allowedthe in situ recording of FTIR spectra of the colloid/aqueous phos-phate interface as function of time. This study provides in situ spec-troscopic data on the surface speciation at the a-Al2O3/phosphatesolution interface, during adsorption of phosphate ions and disso-lution of alumina colloids.

0

5

10

15

20

25

30

35

3 4 5 6 7 8 9final pH

Phos

phat

e so

rbed

, [P

]S

(µm

ol /

g A

l2 O3 )

8µM - 3 d.33µM - 3 d.45µM - 3 d.100µM - 3 d.400µM - 14 d.

[P]I,AQ - tR :

Fig. 1. Effect of pH on the sorption of phosphate in a-Al2O3/0.01 M NaClsuspensions, for different initial concentrations ([P]I,AQ) of aqueous phosphate (tR:reaction time).

2. Experimental section

The alumina colloids used in this study were Alfa Aesar’s a-Al2O3 crystallites (chemical purity: 99.95%, surface area: 7.6 m2/g) having the XRD lines characteristics of corundum. The particlesize was 380 ± 20 nm in 0.01 M NaCl suspensions at pH 3.3, asdetermined by applying the Contin algorithm to photon correlationspectroscopy measurements. Solutions were prepared by using re-agent grade chemicals (HCl, NaCl, NaHPO4, NaOH) and ultrapurewater. Batch experiments were performed to investigate the phos-phate sorption as a function of pH, for different initial concentra-tions of aqueous phosphate (8 lM 6 [P]I,AQ 6 400 lM). A series ofexperiments were carried out to obtain sorption isotherms underacidic conditions, for different reaction times (tR). The experimentswere performed at pH 3.3, for which: (i) the percentage of adsorp-tion was high and (ii) no formation of bayerite, gibbsite, or boehm-ite occurred during experiments, as final Al concentrations ofexperimental solutions were lower than those expected at equilib-rium with regard to these solid phases. The batch experimentswere conducted as follows. Suspensions of nonpretreated a-Al2O3

(2.5 g/L) in 0.1, 0.01, or 0.002 M NaCl electrolyte (preequilibratedwith atmospheric CO2) were prepared at desired values of pHand initial aqueous phosphate concentration in individual HDPEtubes. The tubes were gently shaken at 298 K. After the desiredperiod of equilibration (shaking), the pH was measured and thesuspensions were centrifuged at 8000 rpm for 2.5 h (cutoff:15 nm). The centrifuged solutions were analyzed by inductively

coupled plasma-mass spectrometry and by UV–visible spectropho-tometry (using a phosphate–molybdate complex at 700 nm) todetermine aqueous concentrations of Al and phosphate, respec-tively. f potential of phosphated alumina was determined by mea-suring, with a Malvern NanoZS apparatus, the electrophoreticmobility (EM) of the dispersed particles in suspensions. The EMwas converted into f potential value by using the Smolukowskiequation. The isoelectric point (IEP) of the a-Al2O3 colloids andIEP of phosphated a-Al2O3 colloids were determined from EM mea-surements performed along base titration of alumina suspensions(initial pH: 3.3, [P]I,AQ = 0, 50, 100, 400, or 900 lM; electrolyte solu-tion: 0.01 M NaCl). EM measurements were also performed todetermine: (i) f potential values using suspensions at pH 3.3 ob-tained from the batch experiments at different [P]I,AQ and tR values,and (ii) f potential variations due to successive additions of aqueousphosphate to a-Al2O3 suspensions at pH 3.3 which were previouslyaged for 3 days without phosphate (tpre-eq: 3 days). In situ ATR-FTIRexperiments were conducted by using a Bruker Equinox IFS 55equipped with horizontal Ge crystal and MCT detector. Before theexperiment, the crystal surface was coated with a stable layer ofa-Al2O3 colloids: (i) by letting particles from CO2-free alumina sus-pensions (at pH 3.3 or 6) to settle on crystal overnight, (ii) by rinsingthe coating with electrolyte and drying it under N2 gas flow, and (iii)by repeating the procedure three times. The maximum thicknesswas estimated to be �1.2 lm by using coherence probe micros-copy. The in situ ATR-FTIR experiments were conducted at low pH(3.3) where the high solubility of Al-oxyhydroxides prevents theformation of secondary precipitates such as boehmite, bayerite, orgibbsite, whose OH-bending vibration bands may overlap stronglywith P–O-stretching vibration bands of phosphate in the region900–1200 cm�1. The alumina-coated crystal was brought in contactwith 0.01 M NaCl (pH 3.3) in the ATR cell and an FTIR spectrum wastaken as reference baseline. In some experiments, aqueous phos-phate was added after a desired period of colloid–electrolyte equil-ibration. FTIR spectra (2000 scans/spectrum, resolution of 4 cm�1)were collected every 20 min during the experiment.

3. Experimental results and discussion

3.1. The pH dependence of phosphate sorption

Fig. 1 reports data on the pH dependency of removal of aqueousphosphate at different initial concentrations (8 lM 6 [P]I,AQ 6

M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444 439

400 lM) in alumina suspensions ([a-Al2O3] = 2.5 g L�1 in 0.01 MNaCl electrolyte, tR = 3 or 14 days). Fig. 2 presents data on the pHdependency of f potentials, for different [P]I,AQ values. In the ab-sence of ions determining surface potential (w0) other than H+/OH�, the colloids exhibited a high value (8.2) of isoelectric point,i.e., of pH at which surface carries no net electrical charge.

At low and moderate concentrations of aqueous phosphate([P]I,AQ 6 100 lM), adsorption envelopes show two main zones ofpH with regard to the phosphate behavior (Fig. 1). In the range ofpH 3–6, adsorption was maximal and independent of pH. Theamount of phosphate sorbed ([P]S) increased with an increasingof [P]I,AQ, whereas the percentage of sorption decreased (e.g., thepercentage of sorption was equal to �75 and �40% for [P]I,AQ

values equal to 45 and 100 lM, respectively). This suggested thatdifferent types of binding sites of H2PO�4 and/or different types ofphosphate species were present at the alumina surface. In therange of pH 6–9, there was observed a decrease of phosphate up-take when pH was increased (Fig. 1). Moreover, the value of [P]S

at given pH remained constant in the [P]I,AQ range 30–100 lM;i.e., (high-affinity) sites available for phosphate binding were pres-ent in limited amounts. The dependency of phosphate sorption onpH or [P]I,AQ and the nature of the sorption species have beenwidely related to the aqueous phosphate speciation and to the ex-tent of (de-)protonation of surface functional groups, e.g., [3]. Thereduction in sorption of phosphate on a-Al2O3, observable whenpH increased from 6 to 9 (Fig. 1), can be related to an equilibriumshift between aqueous phosphate species, i.e., to a change of mainsolution species from H2PO�4 to HPO2�

4 . The HPO2�4 ion is indeed

unfavorable to adsorption by electrostatic attraction or by hydro-gen bonding to surface hydroxyls [2]. The reduction in phosphatesorption with pH in the range 6–9 (Fig. 1) was also concomitantwith a decrease of the extent of surface protonation (Fig. 2). Surfacedeprotonation can affect phosphate sorption by: (i) lowering theelectrostatic attraction exerted by a-Al2O3 surface on phosphateanions, (ii) reducing formation of outer-sphere surface complexes,reported to transform slowly into inner-sphere ones [3], and (iii)changing surface speciation of a-Al2O3 from highly reactive, posi-tively charged sorption sites to less reactive, neutral sites [11].Fig. 2 shows that an increase of [P]I,AQ shifted the IEP of a-Al2O3 to-ward lower and lower pH values. This is strong evidence that phos-phate ions were w0 – determining ions in a-Al2O3 suspensions andthat they reacted in coordination with aluminol groups on colloids.

-60

-40

-20

0

20

40

60

80

3 4 5 6 7 8 9 10

Zeta

pot

entie

l (m

V)

0µM50µM100µM400µM900µM

[P]I,AQ :

Final pH

Fig. 2. pH dependence of f potential of a-Al2O3 colloids in 0.01 M NaCl suspensions,for different initial concentrations of aqueous phosphate ([P]I,AQ).

Indeed, formation of OSC of phosphate can decrease f potentials,whereas only formation of IS species that takes place in the Sternlayer of DL leads to f potential reversals and IEP shift. Thus, IEPshift and reversal of f potentials (e.g., 6.0 6 pH 6 8.2,[P]I,AQ = 100 lM) are evidence of the chemical sorption of phos-phate ions. Inner-sphere phosphate surface species were likely ma-jor phosphate species forming at alumina surface, in the [P]I,AQ

range 10–100 lM.The shape of the adsorption curve obtained for a high concen-

tration of aqueous phosphate ([P]I,AQ = 400 lM), and a high reac-tion time (tR = 14 days) was different from those observed forlower concentrations (Fig. 1). The amount of phosphate sorbed(at [P]I,AQ = 400 lM) decreased significantly from pH 3.5 to 4.5,and showed a slight pH dependence in the pH range 4.5–7.5. At agiven pH, an increase of [P]I,AQ from 100 to 400 lM increased theamount of phosphate sorbed, whereas an increase of [P]I,AQ at mod-erate aqueous phosphate concentration (30–100 lM) had no effecton the value of [P]S at pH above 6. Thus, the changes observed inthe macroscopic behavior of phosphate with the level of concen-tration of aqueous phosphate indicated major changes in theadsorption mechanisms. Fig. 2 shows that, when the aqueousphosphate concentration was increased from moderate (100 lM)to high (900 lM), the sorption of phosphate shifted the IEP of a-Al2O3 toward a lower pH value and took place on a negativelycharged surface within a wide range of pH (pH > 5.5). These fea-tures provided evidence for the chemisorption of phosphate ona-Al2O3, at high aqueous phosphate concentrations. The possiblesorption mechanisms were the formation of SC at poorly reactivesurface sites and/or the surface precipitation.

3.2. The macroscopic phosphate sorption at pH 3.3

Fig. 3 reports data on the kinetics of removal of aqueous phos-phate at different initial concentrations in (2.5 g L�1) a-Al2O3/0.01 M NaCl suspensions at pH 3.3. Each data point was obtainedfrom an individual batch experiment performed at defined [P]I,AQ

and tR values. Data series correspond to different [P]I,AQ values.Phosphate sorption followed typical time dependence with atwo-step trend observable in other studies of anion retention onAl- and Fe-oxide (e.g., [2,11,12]). During the first hour, the uptakeof phosphate was rapid and reached �80–90% of phosphate re-moved over the sorption period. This suggested fast reactions ofion exchange or ligand exchange with surface sites [11]. After therapid step, the phosphate uptake continued slowly for hours or

0%

20%

40%

60%

80%

100%

0 50 100 150 300 350Reaction time (hours)

Phos

phat

e so

rbed

(%)

10 µM 30 µM 60 µM 100 µM 200 µM 400 µM[P]I,AQ :

Fig. 3. Effect of time on sorption of phosphate at different initial aqueousconcentrations ([P]I,AQ) in a-Al2O3/0.01 M NaCl suspensions at pH 3.3. Lines areguides for eyes.

0

10

20

30

40

50

60

0 200 400 600 800 1000Initial aqueous phosphate, [P]I,AQ (µM)

Phos

phat

e so

rbed

, [P]

s (µm

ol/g

)

0.01 3 h.0.01 3 d.0.01 14 d.0.01 51 d.0.01 105 d.0.1 3 h.0.002 3 h.0.01 (qe)

I (M) tR

Batch experiment conditions

Fig. 5. Isotherms of sorption of aqueous phosphate at varying initial concentrations([P]I,AQ) in a-Al2O3/NaCl suspensions at pH 3.3 (I, ionic strength; tR, reaction time).

440 M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444

days, possibly due to the decrease of available sorption sites [11],or diffusion [12], and/or secondary processes such as surface pre-cipitation and structural reorganization of surface (complexes). Inorder to gain insights into the rate-limiting steps, several kineticmodels, e.g., Elovich, intraparticle diffusion, and pseudo-second or-der (cf. [13]), were applied to each of our data series. Best fits withcorrelation coefficients higher than 0.99 are reported in Fig. 4a.They were obtained by fitting our data (as plots of t/qt vs. t) withthe pseudo-second-order rate equation [14] whose linear form ist=qt ¼ 1=kq2

e þ t=qe (k, overall rate constant in g mol�1 s�1; qt andqe, amount of sorbate at sorbent surface, in mol g�1, at time t andat equilibrium, respectively). Conformation of our data to thismodel provided evidence of the chemical sorption of phosphateon a-Al2O3, as a rate-limiting step, irrespective of [P]I,AQ. Fig. 4b re-ports the [P]I,AQ dependency of the model parameters qe and k. Theincrease of the sorption capacity qe with the initial phosphate con-tent of solutions over the entire [P]I,AQ range indicated that sorp-tion was not limited by saturation of a limited amount of highlyreactive sites on a-Al2O3. The two-step trend in qe and k variationsindicated instead multiple sites and/or reactions of chemical sorp-tion. Fig. 5 compares values of parameter qe with phosphateadsorption isotherms. It shows that sorption was at near equilib-rium after a 3-day equilibration time, within analytical error.Applying the Langmuir and Freundlich equations (cf. [13]) indi-cated that the near-equilibrium sorption of phosphate on a-Al2O3

was best described (R2 > 0.99) by the Frendlich isotherm (datanot shown). Conformation of data to the Freundlich isotherm sup-ported surface site heterogeneity, and multisite and/or multispe-cies sorption. Fig. 4b shows that the highest sorption rateconstants were observed at the lowest sorption capacities(qe 6 13 lmol/g) and were dependent on [P]I,AQ (660 lM). Inter-estingly, adsorption isotherms showed that a decrease of ionicstrength increased (slightly) the sorption of phosphate at low solu-tion concentrations (Fig. 5). The effect was consistent with the for-

0

10

20

30

0 100 200 300 400[P]I,AQ (µM)

q e(µ

mol

/g)

10

100

1000

10000

k (g

/(mol

.s))

0.E+00

1.E+07

2.E+07

3.E+07

0 100 200 300t (h)

107 . t

/ q t

(h.g

/mol

)

(a)

(b)

Fig. 4. Kinetics modeling of phosphate sorption: (a) fitting of the data series shownin Fig. 3 to the pseudo-second-order rate equation and (b) values of fittedparameters, i.e., sorption capacity, qe, and sorption rate constant, k.

mation of an outer-sphere surface complex of phosphate [6]. Thus,experimental features suggested that sorption of phosphate at lowcoverage occurred mainly via a fast ligand-exchange process,together with electrostatic attraction at highly reactive andprotonated sites. By contrast, sorption rate constants were at lowvalues in the [P]I,AQ range 100–400 lM and were independent ofphosphate solution content, unlike sorption capacity. This sug-gested that the increase of phosphate sorption with [P]I,AQ in therange 100–400 lM was due to a single, slow process such as theformation of surface precipitate or of SC at less reactive sites.

3.3. Dependency of f potential of a-Al2O3 on phosphate sorption at pH3.3

Change in f potential of a-Al2O3 colloids with increasing phos-phate adsorption at pH 3.3 was investigated by f potential mea-surements at varying concentrations of aqueous phosphate.Measurements were made to determine: (i) f potential valuesusing suspensions obtained from batch experiments at different[P]I,AQ and tR, and (ii) f potential variations due to successive addi-tions of aqueous phosphate to a-Al2O3 suspensions which werepreviously aged for 3 days without aqueous phosphate (tpre-eq:3 days). Fig. 6 reports f potential data as a function of phosphatesorbed on a-Al2O3. The linear and two-step trend dependence off potential on phosphate coverage is a major finding of this study.At near-equilibrium sorption (tR: 3 days), f potential decreasedsteeply with an increase of phosphate sorbed at low coverage(i.e., up to �18 lmol/g), while it was only slightly dependent on[P]S at higher coverage (up to �50 lmol/g). Reactions at low cover-age were fast because data points, obtained from experimentswhere aqueous phosphate was added at high rate (r: 25 lM/min)to preequilibrated suspensions (tpre-eq: 3 days), fell on thestraight-line f potential vs. [P]S obtained at near-equilibrium sorp-tion (tR: 3 days). By contrast, there observed a deviation from thisstraight line for data from batch experiments at short equilibrationtime (tR: 3 h), which suggested structural reorganizations withtime of hydroxyls at interface. At high surface coverage, f potentialdecreased slightly with an increasing of aqueous phosphate/a-Al2O3 reaction time.

3.4. Discussion on data of f potential variation with [P]S

Linear variations in f potential (i.e., in surface charge) and breakof slope, observable in Fig. 6, provided evidence for two main (setsof) phosphate sorption reactions at low and high coverage, respec-

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0 10 20 30 40 50 60

ζ /

ζ0

Batch. tR : 3 h. tR : 3 d. tR : 105 d.

Phosphate added. r : 25µM / h. tpre-eq : 3 d.r : 25µM / 20 min. tpre-eq : 3 d.

Experiment :

Phosphate sorbed, [P]S (µmol/g)

Fig. 6. Relative f potential variation of a-Al2O3 colloids as function of phosphatesorbed ([P]S) at pH 3.3: effects of reaction time (tR) of batch experiments or of rate ofphosphate addition (r) to 3-days aged alumina suspensions (f0: f potential ofnonphosphated a-Al2O3).

10081037

1091

1113

1134 1216

(a)

(b)

50 min.

90 min.

2 h.

4 h.tpre-eq. :

1174

950 1000 1050 1100 1150 1200 1250

950 1000 1050 1100 1150 1200 1250

Wavenumber (cm-1)

Rel

ativ

e ab

sorb

ance

Fig. 7. In situ ATR-FTIR spectra recorded during interaction between a-Al2O3

colloids and a 0.01 M NaCl solution at pH 3.3: (a) effect of equilibration time (tpre-eq)and (b) curve fitting of spectrum obtained at tpre-eq = 4 h (colloid coating depositedon ATR crystal from suspensions at pH 6 and dried before experiment).

M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444 441

tively. f Potential and sorption data at low coverage supported theoccurrence of fast, high-affinity reactions of ligand exchange withhydroxyls (likely singly coordinated AlOHþ2 sites) that greatly re-duced a-Al2O3 surface charge and that were favored by electro-static attraction of H2PO�4 ions (reaction (1)). The latterhypothesis is consistent with data (Fig. 5) showing an effect of io-nic strength on phosphate sorption at low surface coverage. Possi-ble reactions of ligand exchange included formation ofdeprotonated ISC, monodentate and bidentate binuclear species(reactions (2) and (3), respectively), and of protonated ISC of phos-phate (reactions (4) and (5)).

AlOHþ2 þH2PO�4 $ AlOHþ2 � � �H2PO�4 ð1ÞAlOHþ2 þH2PO�4 $ AlOPO2�

3 þH2Oþ 2Hþ ð2Þ2AlOHþ2 þH2PO�4 $ Al2O2PO�2 þ 2H2Oþ 2Hþ ð3ÞAlOHþ2 þH2PO�4 $ AlOPO3H� þH2OþHþ ð4Þ2AlOHþ2 þH2PO�4 $ Al2O2PO2Hþ 2H2OþHþ ð5Þ

Surface charge lowering along inner-sphere sorption might de-crease the electrostatic attraction exerted by surface on H2PO�4 ionsand the formation of OSC. f potential data also suggested that phos-phate sorption occurred at structurally reorganizing interfaces, atshort reaction times. Because changes in the nature of surfacehydroxyls with reaction time are of major interest in adsorptionstudies, they were investigated here by in situ ATR-FTIR spectros-copy analyzes of a-Al2O3 surface (cf. Section 4.1). Break of slopein f potential (surface charge) variation observable in Fig. 6 canbe interpreted by formation of ISC of phosphate with less reactiveneutral sites and/or by major change in the sorption process dueto saturation of the singly coordinated sites. Particularly, increasingformation of surface precipitate of aluminum phosphate with [P]S

would impart to the surface a much smaller negative charge thanreactions (2)–(5). Indeed, multilayer sorption (of Al and phosphate)induces no drastic change in the character of the surface [4], and thesurface charge should closely approach with time that of AlPO4(s).The slight decrease of f potential with reaction time, observed athigh phosphate coverage (Fig. 6), is consistent with the multilayersorption. Hypothesis of surface precipitation is also supported bythe decrease of the dissolved Al content with [P]S (data not shown).Aqueous Al concentrations are equal to 30, 18, and 10 lM at [P]S

values of 0, �30, and �50 lmol/g, respectively. Such a trend could,however, result from a decrease of the a-Al2O3 dissolution rate,because phosphate surface complexes may block dissolution-active

surface sites [1]. Again, in situ ATR-FTIR spectroscopy was needed togain insights into sorption species structures, and into their varia-tion with reaction time and phosphate sorption.

4. In situ ATR-FTIR spectroscopy results

4.1. Spectra of nonphosphated alumina at pH 3.3

Fig. 7a reports the time dependency of ATR-FTIR spectra, in theregion 950–1250 cm�1, of a nonphosphated alumina colloidal layerin contact with 0.01 M NaCl at pH 3.3. Prior to use in experiment atpH 3.3, the layer was deposited by using CO2-free suspensions atpH 6 and was dried under N2 atmosphere in order to avoid contam-ination by CO2 and adsorption of carbonate ions. Deposition at pH6 led to optimal coverage (�70%) of the surface area of the ATRcrystal, and to formation of a homogenous layer of a maximumthickness of 1.2 lm. Fig. 7a shows that contacting the electrolyteat pH 3.3 with the layer resulted in OH-bending vibration bands,whose intensity increased with time, tpre-eq. Since the first hour,there observed the simultaneous growth of well-defined bands at1008 and 1037 cm�1—latter band position being characteristic ofbayerite (Table 1)—and broad absorbance at 1050–1175 cm�1.

Decomposition of spectrum (at tpre-eq: 4 h) in Fig. 7b showedclearly the individualization of an intense band at �1091 cm�1

and of two minor vibrations (1113 and 1134 cm�1) from the broadabsorption region. An additional strong band appeared (at tpre-eq:90 min) at 1174 cm�1, with a shoulder at 1216 cm�1. The strongbands at 1091 cm�1 and at 1174 cm�1 are tentatively ascribed toOH-bending vibration of poorly crystalline boehmite (Table 1).Anyway, the experiment evidenced structural reorganizations atthe solid–solution interface due either to surface (re)hydration at

(a)

950 1000 1050 1100 1150 1200

1154113010771045

1012

from 1 to 6 : [P]I,AQ (µM) - tR

50 - 20 min. 40 min. 1.5 h. 2 h. 100 - 3.5 h. 4 h.

1

23456

Rel

ativ

e ab

sorb

ance

(b)

Table 1Infrared active band positions, in region 800–1200 cm�1, of Al-(hydr)oxides, phosphated Al-(hydr)oxides, and aqueous phosphate species.

Mineral/species Vibration mode and position (cm�1) Reference

Al-(hydr)oxides Al–O stretching (antisymmetric) OH bendingBayerite 1030 990 810 [15]Alumina gel 1030 990 810 [16]Synthetic gibbsite 1060 1024 969 [17]Natural gibbsite (1090) 1060 1020 970 940 915Corundum 1072 [17]Boehmite 1160 1070 [18]Poorly cr boehmite 1153 1067 [19]

Phosphated Al-(hydr)oxides P–O stretchingAlumina gel 1130–1140 1040–1050 [16]Aged c-Al2O3 1115 (m3) 900 (m1) [9]

Aqueous species P–O stretching (m3)

PO3�4 ðaqÞ 1006 (850: m1) [6]

HPO2�4 ðaqÞ 1077 989

H2PO�4 ðaqÞ 1160 1074 940 870H3PO4ðaqÞ 1179 1006 (888: m1)

442 M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444

pH 3.3 of the a-Al2O3 colloids (that resulted in ‘‘bayeritization/boehmitization” at particle surface, cf. [20]) or to Al-hydroxidesformed previously in suspensions at pH 6 by dissolution/precipita-tion processes. Thus, ATR-FTIR analyzes were performed on colloi-dal layers deposited by using nonphosphated alumina suspensionsat pH 3.3, although the pH condition led to low coverage (�40%) ofthe surface area of the ATR crystal. Poorly intense but well-definedbands at 1008 and 1037 cm�1, whose intensity stabilized after�3 h, were visible in some spectra of nonphosphated layers in con-tact with 0.01 M NaCl at pH 3.3. This was evidence of a ‘‘bayeriti-zation” process occurring to a limited extent at the particlesurface at pH 3.3, and reorganization of surface hydroxyl groups,leading to different surface states of alumina [20] that may be de-picted as:

AlOH < Al2OH < Al2O !hydrationAl2OH < Al2O

< AlOH !hydrationAlOH < AlðOHÞ2

These IR evidences, i.e., bayeritization and reorganization of surfacehydroxyl groups, might explain deviation from a straight line, ob-servable in Fig. 6, for data-point f potential vs. [P]S obtained frombatch experiments at short equilibration time (tR: 3 h).

950 1000 1050 1100 1150 1200

Wavenumber (cm-1

)

950 1000 1050 1100 1150 1200

1033

1009 1137

1084

1

5

(c)

Fig. 8. In situ ATR-FTIR spectra recorded during phosphate sorption at pH 3.3 on a-Al2O3 colloids: (a) effect of reaction time (tR) after addition of aqueous phosphate([P]I,AQ) to a-Al2O3–0.01 M NaCl solution system preequilibrated for 3 h, and (b andc) curve fittings of spectra 1 and 5, respectively (colloid coating deposited on ATRcrystal from suspensions at pH 3.3).

4.2. Spectra of phosphated alumina at pH 3.3

ATR-FTIR spectra obtained by adding low phosphate concentra-tions ([P]I,AQ < 50 lM) to an electrolyte in contact with colloidallayers deposited at pH 3.3 showed poorly intense absorbancethroughout the region 1000–1200 cm�1, making it difficult to carryany spectra analysis. Fig. 8a reports spectra recorded after two suc-cessive additions of 50 lM aqueous phosphate to an electrolytethat was previously equilibrated for �3 h (tpre-eq: 3 h) with colloidsdeposited at pH 3.3. The spectra were very different from that ofaqueous H2PO�4 species, whose P–O-stretching vibration showswell-resolved m3 bands at 1159, 1074, and 940 cm�1 (Table 1).The absorbance observable in Fig. 8a was univocally related tophosphate sorption and it increased with aqueous phosphate/col-loids reaction time (tR) or aqueous phosphate concentration. Itsshape, too, changed with an increase of reaction time at a [P]I,AQ va-lue of 50 lM, and with an increase of aqueous phosphate concen-tration (from 50 to 100 lM). Spectra decomposition suggested thatthe broad absorption signals were due to contribution of multiplevibration bands for both P–O stretching and OH bending,positioned at low reaction times at �1012, 1038–1045, �1077,�1130, and �1154 cm�1 (Fig. 8b). Vibration bands at �1033 and

950 1000 1050 1100 1150 1200 1250

Rela

tive

abso

rban

ce950 1000 1050 1100 1150 1200 1250

Wavenumber (cm-1)

30 min.1 h.2 h.

4 h.

tR : 7-10 h.

1032

1082

1131

Experimental conditions: − α−Al2O3 coated on ATR crystal- pre-equilibration with 0.01 M NaCl at pH 3.3 (tpre-eq: 3h.)- phosphate addition [P]I,AQ : 200 µM

0 min.Phosphate addition

Fig. 9. In situ ATR-FTIR spectra recorded during phosphate sorption at pH 3.3 on a-Al2O3 colloids, for an aqueous phosphate concentration of 200 lM: (a) effect ofreaction time (tR) and (b) curve fitting of the spectrum obtained at tR = 7 h. See Fig. 8for experimental conditions (spectra obtained after subtraction of spectrum at tR: 0).

M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444 443

1137 cm�1 showed high relative contributions to absorbance forspectra recorded at the highest [P]I,AQ and tR values investigated,where a strong band at 1084 cm�1 and a minor band at1009 cm�1 were also resolved by decomposition (Fig. 8c). It mustbe noted that the band at 1077 cm�1 (and at 1154 cm�1) observa-ble in Fig. 8b cannot be ascribed to H2PO�4 aqueous species, whoseconcentration in experiment was under the detection limit(800 lM) of ATR-FTIR techniques. Assignment of the bands in theregion 1000–1080 cm�1 to defined surface species was notstraightforward though. On the one hand, the bands can be as-cribed to OH-bending vibration modes only, because their posi-tions agreed with those reported for boehmite or bayerite(Table 1) or were observed in experiments using nonphosphatedalumina deposited at pH 3.3 (at 1008 and 1037 cm�1, Fig. 7). Var-iation in band intensity (e.g., at 1010 cm�1) and position (i.e., from1077 to 1084 cm�1, and from 1045 to 1033 cm�1) observable inFig. 8b and c suggested that reactions of phosphate sorption in-duced structural reorganizations of surface hydroxyls. Surfacecharge decrease with phosphate sorption (Fig. 6) might indeed in-duce transfer of proton to Al2O surface sites, leading to Al2OH sur-face species and favoring surface hydration. On the other hand,contribution of P–O-stretching vibration for sorbed phosphate, inaddition to OH-bending vibration, cannot be ruled out. Indeed,bands at 1077 and 1157 cm�1 could be ascribed to OSC of phos-phate forming at low reaction times. Moreover, variations in bandposition (i.e., 1077–1084 and 1045–1033 cm�1) and in band inten-sity, observable in Fig. 8b and c, could relate to formation of ISC ofphosphate. Band positions for the P–O-stretching mode of phos-phate surface complexes, and structure of surface complexes, haveindeed been shown to depend strongly on surface coverage andphosphate solution content [5,7]. Thus, many questions remain be-cause phosphate species on alumina showed poorly intense andunresolved absorption bands, unlike on Fe-oxides, so that OH-bending vibrations of surface hydroxyls overlapped strongly withthose for P–O stretching in the region 1000–1080 cm�1. By con-trast, the absorption band at �1130–1137 cm�1, whose relativeintensity increased with [P]I,AQ (Fig. 8), was characteristic of phos-phate sorbed in experiment, i.e., at high phosphate concentration.

Fig. 9a shows the time dependency of ATR-FTIR spectra re-corded after addition of 200 lM aqueous phosphate to an electro-lyte in contact with colloids deposited at pH 3.3 (tpre-eq: 3 h). Therewas observed an increase in absorbance with reaction time (up totR: 10 h) in the region 1000–1200 cm�1. Spectra decompositionshowed that the same bands at 1032, 1082, and 1131 cm�1 con-tributed to absorbance, irrespective of reaction time (Fig. 9b). Onlythe relative proportion between former bands and that positionedat 1131 cm�1 varied with tR, the latter vibration being of highestintensity. This experiment confirmed that the relative intensity ofthe band at �1130 cm�1 increased with [P]I,AQ, and that its positionwas characteristic of P–O stretching for phosphate sorbed at highconcentration.

Because its intensity increased with phosphate solution contentand with reaction time, whereas its position was insensitive to[P]I,AQ (unlike for surface complexes), we assign the band at1130–1137 cm�1 to P–O stretching for surface precipitates of alu-minum phosphate. Moreover, its position was consistent with pre-vious studies showing formation of AlPO4 (surface) precipitates.Nanzyo [16] reported absorption bands assigned to P–O-stretchingmode for phosphate sorbed on alumina gel at positions of 1130–1140 cm�1 (as a strong band) and 1040–1050 cm�1 (as a smallpeak), which were similar to those for Al-phosphate gel. Laitiet al. [9,10] reported that FTIR spectra of aged c-Al2O3 suspensioncontaining an excess of phosphate showed characteristics in agree-ment with those of AlPO4(s), with a strong broad m3 band around1115 cm�1 and a weaker m1 band at 900 cm�1, and were differentthan spectra at low phosphate concentrations. They interpreted

the IR results by the fast formation of surface complexes followedby the transformation, at high phosphate/AlOH ratios, of aluminaphase into secondary Al-phosphate precipitates whose formationrate depended on the aqueous phosphate to surface site ratio.ATR-FTIR features described here also supported that phosphatewas sorbed at high coverage on a-Al2O3 via fast formation of Al-phosphate surface precipitates that followed surface complexationand that continued over extended periods along colloiddissolution.

5. Summary and conclusion

This study shows that using the traditional batch method, f po-tential measurements, and in situ ATR-FTIR spectroscopy providedvaluable insights into the nature of surface species that formed inlow pH a-Al2O3 suspensions, along colloid–solution interactionsand phosphate sorption processes. All techniques concurred withthe conclusion that phosphate was sorbed at low pH (3.3) on a-Al2O3 via a combination of surface reactions of complexation andprecipitation, whose relative contributions depended on surfacecoverage and reaction time. High-affinity reactions of ligand ex-change with single coordinated hydroxyls (e.g., reactions (2)–(5)),and formation of an outer-sphere complex at the protonated sites(reaction (1)), accounted for experimental features observed withincreasing phosphate sorbed at low coverage, as:

(i) steep and linear decrease of f potential, i.e., of surfacecharge, at near equilibrium,

(ii) decrease of sorption rate constants due to a decrease ofavailability of AlOH sites.

A major finding on the kinetics and reactions of phosphate sur-face complexation came from confrontation between f potentialsand ATR-FTIR spectra. Reactions of surface complexation of

444 M. Del Nero et al. / Journal of Colloid and Interface Science 342 (2010) 437–444

phosphate were fast but occurred at a colloid/solution interfacewhose structures and charges were rearranging for hours alongsurface hydration. At high phosphate coverage, surface processeswere even continuing for days, as evidenced by ATR-FTIR analyzesof the a-Al2O3/phosphate solution interface. Indeed, absorptionband characteristics for OH-bending mode and the band at1137 cm�1 assigned to P–O stretching for phosphate sorbedshowed increasing intensities with time. Spectral features, as wellas low f potentials showing poor dependence on [P]S and on reac-tion time, supported the hypothesis of the formation of Al-phos-phate surface precipitates at high coverage. This study provides apicture of an a-Al2O3/solution interface that evolves over extendedperiods, due to surface precipitation leading to significant reorga-nizations of surface hydroxyls and continuing all along aluminadissolution.

Acknowledgments

This work was supported by REALISE, the ‘‘Alsace Region Re-search Network in Environmental Sciences and Engineering” andthe Alsace Region. We also thank GNR PARIS for financial support.We are grateful to an anonymous reviewer for manuscriptimprovement.

References

[1] W. Stumm, Colloids Surf. A 120 (1997) 143–166.[2] N.I. Chubar, V.A. Kanibolotskyy, V.V. Strelko, G.G. Gallios, V.F. Samanidou, T.O.

Shaposhnikova, V.G. Milgrandt, I.Z. Zhuravlev, Colloids Surf. A 255 (2005) 55–63.

[3] M.-X. Zhu, K.-Y. Ding, S.-H. Xu, X. Jiang, J. Hazard. Mater. 165 (2009) 645–651.[4] L. Li, R. Stanforth, J. Colloid Interface Sci. 230 (2000) 12–21.[5] M.I. Tejedor-Tejedor, M.A. Anderson, Langmuir 6 (1990) 602–611.[6] Y. Arai, D.L. Sparks, J. Colloid Interface Sci. 241 (2001) 317–326.[7] E.J. Elzinga, D.L. Sparks, J. Colloid Interface Sci. 308 (2007) 53–70.[8] T.J. Van Emmerik, D.E. Sandström, O.N. Antzurkin, M.J. Angove, B.B. Johnson,

Langmuir 23 (2007) 3205–3213.[9] E. Laiti, P. Persson, L.-O. Ohman, Langmuir 12 (1996) 2969–2975.

[10] E. Laiti, P. Persson, L.-O. Ohman, Langmuir 14 (1998) 825–831.[11] O.R. Harvey, R.D. Rhue, J. Colloid Interface Sci. 322 (2008) 384–393.[12] C. Luengo, M. Brigante, J. Antelo, M. Avena, J. Colloid Interface Sci. 300 (2006)

511–518.[13] M.B. McGechan, D.R. Lewis, Biosyst. Eng. 82 (1) (2002) 1–24.[14] Y.S. Ho, G. McKay, Proc. Biochem. 34 (1999) 451.[15] H. Elderfield, J.D. Hem, Mineral. Mag. 39 (1973) 89–96.[16] M. Nanzyo, J. Soil Sci. 35 (1984) 63–69.[17] R.I. Frost, J.T. Kloprogge, S.C. Russell, J.L. Szetu, Appl. Spectrosc. 53 (1999) 423–

434.[18] R.I. Frost, J.T. Kloprogge, S.C. Russell, J.L. Szetu, Appl. Spectrosc. 53 (1999) 572–

582.[19] S.A. Dickie, A.J. McQuillan, Langmuir 20 (2004) 11630–11636.[20] X. Yang, Z. Sun, D. Wang, W. Forsling, J. Colloid Interface Sci. 308 (2007) 395–

404.