Investigation of the Surface Properties of Solid-PhaseHydrous Aluminum Oxide Species in Simulated

Wastewater Using Atomic Force Microscopy

Anselm Omoike, Guoliang Chen, Gary W. Van Loon, and J. Hugh Horton*

Department of Chemistry, Queen’s University, Kingston, Ontario, Canada K7L 3N6

Received February 3, 1998. In Final Form: May 27, 1998

Hydrous aluminum oxide particles precipitated from simulated wastewater were examined using atomicforce microscopy in order to understand their structural and coating properties. Three types of particleswere examined: aluminum oxides coprecipitated by adding alum in the presence of a solution of phosphatesand tannic acid; postprecipitated particles formed by adding phosphates and tannic acid to already formedaluminum oxides; and a control case consisting of particles precipitated in the absence of either phosphateor organic component. Using tapping mode and phase imaging atomic force microscopy, it was found thatthe postprecipitated particles had distinctly different viscoelastic properties than either of the other twoparticle types and also varied markedly in particle size and morphology. These observations are consistentwith a model in which the postprecipitated particles are coated with an organic coating of tannic acid. Theresults are discussed in the context of the relative effectiveness of these aluminum oxides in removingphosphates and other contaminants from wastewater during sewage treatment.

Introduction

In addition to producing real space images of surfacetopography, therecentadvances inatomic forcemicroscopy(AFM) have allowed imaging of the chemical propertiesof surfaces on the nanometer scale.1 Among thesetechniques, the use of phase imaging in intermittentcontact or tapping mode AFM has been shown to producenanometer scale images of variations in surface viscoelas-tic properties. In the phase imaging mode, the phase shiftof the oscillating cantilever is measured as a function oftip position on the surface. The observed phase shift ∆φis related to the tip-sample force interactions, albeit ina complex fashion.2,3 However, under the correct imagingconditions, this technique can not only map out thetopography across the sample surface but also allow achemical identification of surface features. Much of theprevious work on phase imaging AFM data, and particu-larly its interpretation, has concentrated on polymersamples.4-6 In this paper, we show how the techniquecan be applied to a quite different chemical system; here,we have investigated the surface-coating properties of aseries of hydrous aluminum oxide particles derived fromthe hydrolysis of aluminum during wastewater treatment.

Alum [Al2(SO4)3‚nH2O] is the most widely used coagu-lant in water and wastewater treatment. The use of alumrequires that sufficient alkalinity be present in thewastewater to produce solid hydrous aluminum oxidespecies. These species remove orthophosphates present

in the wastewater by forming insoluble aluminum hy-droxyphosphate or other complexes. In an activatedsludge plant, there are several points7 at which alum canbe added, but two common addition points are (1) beforethe commencement of biological treatment and (2) afterthe aeration chamber prior to final clarification. In theformer case, it is possible that the orthophosphate ionsare coprecipitated with alum in the presence of dissolvedorganic matter. In the latter postprecipitation case, theorthophosphate and dissolved organic concentrations inthe effluent have been reduced and recycling of sludgefrom this system is synonymous to the use of prehydrolyzedalum in treating wastewater. Depending upon wherealum is added during the wastewater treatment process,the reaction between aluminum oxides and componentsdissolved in wastewater produces solid-phase productsthat couldbedifferent inchemicalandsurfacecomposition.

This work presents results probing (using AFM) thechanges in the solid-phase reaction products formed onthe addition of alum to simulated wastewater-containingphosphates. We use tannic acid in the simulated waste-water as a surrogate for the dissolved organic componentsin the real case. Tannic acid is well characterized andcontains the same functional groupsssalicylic, carboxylic,and phenolicsas are found in the humic material thatforms the largest component of dissolved organic materialin wastewater. Three different solid phases of hydrousaluminum oxide species were prepared and examined,simulating various stages in the wastewater treatmentprocess: coprecipitated particles that correspond to theaddition of alum before or at the aerator, postprecipitatedparticles that represent the situation where solid hydrousaluminum oxide species are recycled in the biologicalsludge and encounter phosphate in the aerator, and finallya control experiment in which hydrous aluminum oxide(Al(OH)3) was simply precipitated from a NaHCO3solution in the absence of phosphate or organic compo-nents. A second control consisting of tannic acid dispersedon a mica substrate was also imaged. Using phase-

* To whom correspondence should be addressed: tel (613)-545-2379; fax (613)-545-6669; e-mail hortonj@chem.queensu.ca.

(1) Wiesendanger, R. In Scanning Probe Microscopy and Spectros-copy: Methods and Applications; Cambridge University Press: NewYork, 1994.

(2) Winkler, R. G.; Spatz, J. P.; Sheiko, S.; Moller, M.; Reineker, P.;Marti, O. Phys. Rev. B 1996, 54, 8908.

(3) Burnham, N. A.; Kulik, A. J.; Gremaud, G.; Gallo, P.-J.; Oulevey,F. J. Vac. Sci. Technol., B 1996, 14, 794.

(4) Bar, G.; Thomann, Y.; Brandsch, R.; Cantow, H. J.; Whangbo, M.H. Langmuir 1997, 13, 3807.

(5) Overney, R. M.; Meyer, E.; Frommer, J.; Guntherodt, H.-J.;Fujihira, M.; Takhano, H.; Gotoh, Y. Langmuir 1994, 10, 1281.

(6) Hoper, R.; Gesang, T.; Possart, W.; Hennemann, O. D.; Boseck,S. Ultramicroscopy 1995, 60, 17.

(7) Bowker, R. P. G.; Stensel, H. D. In Phosphorus Removal FromWastewater; Noyes Data Corporation: New Jersey, USA, 1990.

4731Langmuir 1998, 14, 4731-4736

S0743-7463(98)00130-9 CCC: $15.00 © 1998 American Chemical SocietyPublished on Web 07/30/1998

imaging AFM, we demonstrate the presence of an organiccoating on the postprecipitated particles that modifies thetip-sample interaction as well as causing considerablechanges in particle morphology. These results are usedto explain the varying effectiveness of these particles inthe wastewater treatment process.

Experimental SectionParticle preparation was carried out as follows. The synthetic

wastewater was made up of a solution of NaHCO3 (0.208 molL-1), KHPO4 (3.70 mol L-1), and tannic acid, C76H52O46 (0.100mol L-1). All the chemicals used were reagent grade obtainedfrom BDH except the alum, which was commercial grade,obtained from the Kingston Water Purification Plant, Kingston,Ontario. Distilled deionized water was used for all samplepreparations. A mixer equipped with a four-blade propeller anda variable speed control was used for mixing the components inthe simulated wastewater. The coprecipitated particles weresynthesized by adding alum (430 mg as Al L-1) to the syntheticwastewater, and the components were allowed to age for 5 minat a stirring speed of 380 rpm. To synthesize the postprecipitatedparticles, the alum (430 mg as Al L-1) was added to the syntheticwastewater in the absence of phosphates and tannic acid andthe mixture was allowed to age, with stirring, for 5 min beforethe subsequent addition of the phosphate and tannic acidcomponents and then aged for a further 5 min. Finally, the controlAl(OH)3 particles were synthesized by adding alum (430 mg L-1

as Al) to a NaHCO3 (0.208 mol L-1) solution.Samples were prepared for AFM imaging in the following

manner. Aqueous dispersions of all three hydrous aluminumoxide samples were produced by placing approximately 0.35 gL-1 slurries of the particles in an ultrasonic bath for 60 min. A30 µL portion of the dispersion was syringed onto a freshly cleavedmica substrate 1 cm2 in area. The sample was then spun at 4000rpm on a spin coater for 60 s to ensure an even distribution ofparticles over the substrate surface. The sample was allowed toair-dry at for 1 h and then imaged in the AFM. The controlsample consisting of tannic acid deposited on mica was preparedfrom a 3.8 × 10-8 mol L-1 solution of tannic acid in methanol.This solution was dispersed and spin coated onto a mica substratein a fashion similar to the alumina particles.

All AFM data shown were acquired using a PicoSPM operatedin MAC mode (Molecular Imaging, Tempe, AZ), using a Nano-scope IIE controller (Digital Instruments, Santa Barbara, CA).The MAC mode is essentially the same as tapping mode, exceptthat the cantilever is magnetically coated and is driven by anexternal oscillating magnetic field.8 The cantilevers had a forceconstant of ∼0.5 N m-1 and a resonance frequency of ∼100 kHz.All images were acquired under ambient conditions, at thefundamental resonance frequency of the Si cantilevers. Heightand phase shift data were all recorded simultaneously, as afunction of both cantilever oscillation amplitude (Ao) and set pointratio rsp ) Asp/Ao. Images were recorded at scan rates of 1-2lines/s using a 30 µm × 30 µm scanner.

Results

AFM Images. Figure 1 shows the postprecipitatedparticles dispersed on the mica substrate, acquired atvarious values of both cantilever oscillation amplitude(Ao) and set point r. The left image is height mode data,while the right image is phase imaging data. The imagesize and z scale (i.e., height or phase shift) are the samefor each image. In all cases, dark regions correspond tolower values of height and phase shift, while brighterregions correspond to higher values. Larger scale images,up to 20 µm square, demonstrated that the pattern seenin Figure 1 is typical across the surface. Several observa-tions are immediately apparent. In all cases, the contrastbetween height and phase shift data is reversed. However,the range of contrast varies considerably, in both heightand phase imaging data, with the most contrast visible

at large Ao and intermediate set points and with the leastat low Ao and large set points. Images were also acquiredat intermediate set point values at eachAo and show trendsin contrast intermediate to those images shown here. Setpoints of r < 0.5 generally produced images of poor qualityand often showed large tip oscillations, so are not presentedhere. Another interesting observation is seen in parts cand d of Figure 1. Here we see two images of the samearea, taken one after another. The surface featuresindicated with the circles in these figures change slightlyfrom one image to another. The changes suggest thatthere is a tip-sample interaction taking place. While thetwo images are a particularly good example of this effect,it was observed in a number of images, some taken oneafter another at exactly the same set point and amplitude.Imaging a small region (500 nm square) of the sample incontact mode did not show any sign of the particles on thesurface. Reimaging the same area in MAC mode showedthat the particles had been swept out in the contact modeimage region. A similar effect was observed with thetannic acid sample (Figure 4). Changes in particle shapeor other evidence of tip-sample interaction of a similarnature were not observed in the case of the coprecipitatedor Al(OH)3 control particles discussed below.

Figure 2 shows the coprecipitated particles depositedon the mica substrate. These images were acquired atthe same set point, 0.65, but three different oscillationamplitudes of 116, 58, and 29 nm (similar to the amplitudesfor Figure 1, above). Note that the Z-range in the heightmode is 10 times that of Figure 1, but the Z-range of thephase imaging data is the same. Images at different setpoints were acquired, ranging from 0.9 to 0.25 and weresimilar to those seen here. Figure 3 shows the controlAl(OH)3 particles on the mica substrate at an intermediateoscillation amplitude of 58 nm and set point of 0.5. Imagesof the control particles were also acquired at set pointsranging from 0.9 to 0.15 and amplitudes ranging from 20to 105 nm. In all cases, the height and phase imageslooked very similar to those shown here and, indeed, tothe images of the coprecipitated particles in Figure 2, andso are not repeated. Again, images taken over larger scanareas, up to 20 µm square, show that the features seenin Figures 2 and 3 are typical of the surface. Besides thegeneral lack of dependence of image contrast on bothamplitude and set point, several other observations areof note. First, the particles are in both cases roughlycircular, with diameters ranging from 40 to 500 nm. Crosssectional profiles also reveal that the particles range inheight from 10 to 150 nm. The coprecipitated and controlparticles are similar in both size and shape and show amarked contrast in morphology to the postprecipitatedparticles, which are extremely flat and irregular in shape.We also emphasize the observation that, regardless ofsize, the coprecipitated particles show little contrast withrespect to the substrate in the phase imaging mode, andcontrol particles show virtually none, other than someslight shading or brightening at the particle boundaries.This latter effect is probably simply due to the failure ofthe feedback loop to adequately follow the rapid changesin sample height as the tip tracks across the particles.

Figure 4 shows images of the mica sample on whichtannic acid was deposited from a methanol solution. Againwe present images at various set points and oscillationamplitudes comparable to those obtained in the previousexperiments. The Z-range in height mode (10 nm) issimilar to that for the postprecipitated particles in Figure1, while the phase imaging Z-range is the one-half thatin the other figures. It proved possible to acquire higherresolution images with this sample than with the previous(8) Han, W.; Lindsay, S. M.; Jing, T. Appl. Phys. Lett. 1996, 69, 4111.

4732 Langmuir, Vol. 14, No. 17, 1998 Omoike et al.

ones. All the images in Figure 4 are 500 nm across. Wenote that the tannic acid tends to form small agglomeratesabout 15 nm in diameter which can exist as isolatedparticles or can in turn form larger agglomerates some300-600 nm in diameter in which the smaller agglomer-ates can be resolved. We also observed that by reducingthe tannic acid concentration in the methanol solution,the size and frequency of the larger agglomerates de-creased over the surface. There are similarities in thephase contrast behavior to postprecipitated particles inFigure 1. Again, the contrast is reversed between heightand phase imaging. Here, the contrast also undergoes areversal in the height and phase images themselves aswe go from set points of 0.9-0.3 at Ao ) 106 nm. Like thepostprecipitated particle images in Figure 1, it provedimpossible to acquire useful images at set points lowerthan about 0.7 for tip oscillation amplitudes less thanabout 100 nm. We note that for the lower Ao images inparts c and d of Figure 4 that there is considerably lesscontrast, especially in the phase images.

Other Techniques for Particle Characterization.We report briefly here on the results for some othertechniques used for characterizing the chemical composi-

tion and activity of these aluminum oxide particles. Thesurface reactivities of particles were measured using aferron complexation procedure.9 Aluminum, present asa component of a solid, reacts with ferron reagent to forma colored complex, which absorbs radiation at 370 nm. Tocombine with ferron, Al-OH, Al-PO4, or Al-organicmatter bonds at the interface between the hydrolyticproduct and the adjacent solution must be broken.Aluminum that is present in weakly bonded labile formstherefore would be expected to react rapidly, whereasaluminum that is strongly bound and inert in the solidwould react much more slowly. An amorphous precipitatewith large surface area would also be expected to reactmore rapidly than a more crystalline product, due to thehigh surface concentration of aluminum ions able to reactreadily with the ferron reagent in solution. Measuringthe rate of reaction between aluminum species in the solidsand the reagent therefore gives an indication of the surfacespeciation and morphology of the aluminum-containingsolid. As a simple estimate of the reaction rate, the time

(9) Duffy, S. J.; Van Loon, G. W. Environ. Sci. Technol. 1994, 28,1950.

Figure 1. Height (left) (Z range ) 20 nm) and phase imaging (right) MAC mode AFM images of the postprecipitated aluminumoxide particles dispersed on a mica substrate. All images are 2 µm square and were acquired at the following tip oscillationamplitudes and set points: (a) Ao ) 104 nm, rsp ) 0.90; (b) Ao ) 104 nm, rsp ) 0.60; (c) Ao ) 46 nm, rsp ) 0.90; (d) Ao ) 46 nm,rsp ) 0.75; (e) Ao ) 22 nm, rsp ) 0.90; (f) Ao ) 22 nm, rsp ) 0.70.

Properties of Aluminum Oxide Species Langmuir, Vol. 14, No. 17, 1998 4733

required to recover 50% of the solid-phase aluminum,designated as t50, was determined. Increasing t50 valuesare indicative of decreasing reactivity of aluminum in the

solid phases. The t50 values obtained for solids used inthe AFM experiments were in the order control <coprecipitated , postprecipitated. The t50 value obtainedfor the postprecipitated particles was 6.9 times greaterthan the value determined for the coprecipitated particles.

High-resolution solid-state magic angle spinning (MAS)27Al NMR was also performed on each of the particles.The chemical shift for the control particles was 5.18 ppm,for the coprecipitated particles was -5.32 ppm, and forthe postprecipitated particles was -1.19 ppm. Theposition of all these peaks is consistent with the reportedrange of -10 to 20 ppm assigned to octahedral aluminum10

in aluminum oxides. The downfield shift observed in thelatter two solids is indicative of the presence of phosphorusin the samples with the coprecipitated particles having ahigher phosphorus concentration and hence a greater shiftdownfield.

Finally, we determined the extent of tannic acidincorporation into the particles. Tannic acid has a strongadsorption band at 278 nm. The UV-visible absorptionspectrum was obtained for the solution before additionboth of sodium bicarbonate and of the filtrate afterprecipitation of the aluminum oxides. The adsorptionspectrum of the filtrate indicated that essentially all ofthe tannic acid had been incorporated into the precipitate.

Discussion

Any analysis of phase imaging data must be consideredwith particular care, as a number of factors can affect theimage contrast. The phase shift angle ∆φ can be shownto be a function of several different factors. Here, we follow

(10) Duffy, S. J.; Van Loon, G. W. Can. J. Chem. 1995, 73, 1645.

Figure 2. Height (left) (Z range ) 200 nm) and phase imaging(right) MAC mode AFM images of the coprecipitated aluminumoxide particles dispersed on a mica substrate. All images are2 µm square and were acquired at the following tip oscillationamplitudes and set points: (a) Ao ) 116 nm, rsp ) 0.65; (b) Ao) 58 nm, rsp ) 0.65; (c) Ao ) 29 nm, rsp ) 0.65.

Figure 3. Height (left) (Z range ) 200 nm) and phase imaging(right) MAC mode AFM images of the control aluminum oxideparticles dispersed on a mica substrate. The images are 2 µmsquare and were acquired at a tip oscillation amplitude of 58nm and set point 0.50.

Figure 4. Height (left) (Z range ) 10 nm) and phase imaging(right) MAC mode AFM images of tannic acid adsorbed on amica substrate. All images are 500 nm square and were acquiredat the following tip oscillation amplitudes and set points: (a)Ao ) 106 nm, rsp ) 0.90; (b) Ao ) 106 nm, rsp ) 0.30; (c) Ao )52 nm, rsp ) 0.90; (d) Ao ) 26 nm, rsp ) 0.90.

4734 Langmuir, Vol. 14, No. 17, 1998 Omoike et al.

the derivation of Whangbo.4,11 Assuming the cantileveris at its resonance frequency in the free state, the phaseshift angle is given by

where Q and k are the quality factor12 and force constantof the cantilever, respectively. Both are constants in thiscase. The term σ is the overall force derivative experiencedby the cantilever: that is, the sum of all the forcederivatives acting on the cantilever

where z is the tip-sample separation. When the net forceacting on the tip is attractive (i.e., van der Waals forcespredominate) σ and the phase shift are negative, while ifthe net force is repulsive (indentation forces predominate),σ and the phase shift are positive. The problem, then, isto determine under what conditions each of these twoeffects will dominate the tip-sample interaction.

At low values of Ao, the tip does not penetrate far intothe surface, and thus little tip-sample repulsion isexperienced: attractive van der Waals and capillary forcespredominate, mainly arising from water on the surface ofthe sample. Practically, it has been shown for polymersurfaces that this region is important for about Ao < 40nm and rsp > 0.8. It would appear that such behavior issimilar for this surface. This is typified by parts a, e, andf of Figure 1, where the phase imaging data show littlecontrast, particularly in the case of Figure 1e where bothrequirements are met. Alternatively, at high values of Aoand moderate values of rsp, the tip will now penetrate farinto the surface, and tip-sample repulsion will dominatethe interaction. This is typified in Figure 1b, which wasacquired at both high values of Ao and intermediate setpoint. Here we see the largest contrast in the phaseimaging data of the coprecipitated particles.

The first problem, of course, is to determine whichobjects, the raised or lowered ones in the height imagesof Figure 1, correspond to the particles and which to thesubstrate. By the arguments above, it should be apparentthat at high values of the set point, the height mode shouldmost closely approximate the actual sample topography.At lower set points soft samples in particular could bedeformed considerably: this would lead to the tip pen-etrating far into the sample and the feedback loop couldthus be “fooled” into showing the softer regions on thesurface as a negative excursion in the height profile. Sincethe raised objects in parts a, c, and e of Figure 1 (high setpoints) all appear raised in the remaining images, thisindicates that these are indeed the real particles. Sincethey appear as raised features in the all the height images,it would appear that the particle height is sufficientlylarge that any deformation effect is canceled out at lowersetpoints. Having identified thepostprecipitatedparticlesin the height images, we turn our attention to the phaseimaging data. We first note that the data demonstratethat what we are seeing is truly a single, incomplete layerof particles on the mica substrate. If we were seeing twoor more layers of the same particles on top of one anotherwe might expect there to be little contrast in the phaseimaging data, as everything present would have the same

viscoelastic properties. We see instead that the particlesshow a negative excursion in the relevant phase imagingdata consistent with a particle whose surface propertiesare quite different from the underlying mica substrate. Itshould also be noted that at very low values of rsp weshould observe a contrast inversion in the phase imagingdata. This is because at low rsp, the tip now penetratesfar into the surface, increasing the tip contact radius, ⟨a⟩,and hence φ4. Contrast inversions have indeed beenobserved on polymer surfaces, but generally at lower rspthan we were able to image with the postprecipitatedparticles. We found that the feedback loop becameunstable at low set points on the postprecipitated samples,which is presumably due to very strong tip-sampleinteraction as the tip penetrates the surface layer. Thiseffect was not repeated in the case of the other samples.Figure 4 shows that in the case of the tannic acid adsorbedon mica, a constrast reversal does indeed occur betweenr ) 0.9 and r ) 0.3.

In the case of the coprecipitated and control Al(OH)3particles, we observe few changes in the phase imagingor height images as a function of Ao and rsp. In particular,at large to intermediate values of Ao and moderate setpoints, both the particle and substrate show similar phaseshifts. This demonstrates that both particle and substrateare similar in their viscoelastic properties. This isconsistent with the fact that both the aluminum oxideparticles, and the aluminosilicate mica substrate havesimilar surface structural groups present. The Young’smoduli E of mica (biotite) and amorphous Al2O3, of 40GPa13 and 280 GPa,14 respectively, are of similar ordersof magnitude. Comparing these phase shift results forthe control Al(OH)3 and coprecipitated particles to thosefor the postprecipitated particles clearly demonstrates thatthe latter are characterized significantly different vis-coelastic properties. In addition, the tannic acid imagesin Figure 4 are characterized by large changes in contrastboth in the phase and height data as a function of setpoint and tip osciallation amplitude, which are morereminiscent of the images seen in Figure 1 of thepostprecipitated particles. This is good evidence that thesedifferences may be accounted for by tannic acid coatinglayer on the postprecipitated particles. The tannic acidcould be bound to the particle surface via an Al-O-Cbond to form a tannate ligand or could simply be stronglyphysisorbed on the hydrous aluminum oxide particles. Inany case, since tannic acid is a large organic molecule15

(molecular weight of 1701.23 g mol-1), it might be expectedto form a fairly compliant organic coating.

Other than their viscoelastic properties, these particlesalso clearly demonstrate a wide variation in morphology.First, of course, the coprecipitated and control particlesare generally spherical in shape and vary widely in size.The postprecipitated particles also vary widely in theirlateral dimensions but are remarkably uniform in height.A line profile across Figure 1a shows a particle height ofabout 2 nm. This height measurement is taken underconditions of high set point and tip oscillation amplitudewhich should give the closest approximation to the actualheight of surface features. Nonetheless, we shouldemphasize that this probably underestimates the actualheight of the features in the image. In addition, we haveseen evidence of tip-sample interaction in the postpre-

(11) Magonov, S. N.; Elings, V.; Whangbo, M. H. Surf. Sci. 1997, 375,L385.

(12) Chen, G. Y.; Warmack, R. J.; Thundat, T.; Allison, D. P.; Huang,A. Rev. Sci. Instrum. 1994, 65, 2532.

(13) Touloukian, Y. S.; Ho, C. Y. In Physical Properties of Rocks andMinerals; McGraw-Hill: New York, 1981.

(14) Cottrell, A. H. In The Mechanical Properties of Matter; JohnWiley & Sons: New York, 1964.

(15) The Aldrich Library of FT-IR Spectra, 1st ed.; Aldrich ChemicalCompany, Inc.: 1984; Vol. 1, p 544c.

∆φ ) Qσ/k (1)

σ ) ∑i

∂Fi

∂z(2)

Properties of Aluminum Oxide Species Langmuir, Vol. 14, No. 17, 1998 4735

cipitated particle images (Figure 1). Given that the phaseimaging data show that these particles are coated witha tannic acid layer, this observation at least is notsurprising. Since the hydrophilic end (carboxylic acidgroup) of the tannic acid is presumably bound to theparticle surface, this leaves a hydrophobic overlayersurrounding the particle. This would not be expected tobind strongly to the hydrophilic mica substrate, certainlyless so than the oxide or hydroxyl-terminated coprecipi-tated and control particles which themselves appear to bemuch more strongly bound to the substrate. Theseobservations suggest that we are probably not seeingindividual postprecipitated particles in these images butrather aggregated particles of relatively uniform size thatare loosely bound to the substrate and to one another.The postprecipitated particles are thus probably consid-erably smaller in size than either of the other two particletypes.

These observations are in agreement with other workon these chemical systems, all of which are concernedwith hydrous aluminum oxide particles that are precipi-tated in the presence of organic acids such as tannic acid.16

In these situations, the solid has been shown (usingtitration methods) to have a large surface area and (usingtransmission electron microscopy) to consist of many smallparticles, sometimes aggregated together.12 The AFMobservations are also consistent with the ferron agentcomplexation results. All else being equal, the AFMresults would indicate that the reaction time t50 for theferron complex with the postprecipitated particles shouldbe less than that for the coprecipitated particles since theformer have been shown to have a considerably largersurface area in the AFM images. Instead, the oppositebehavior is observed, with the postprecipitated particleshaving much longer t50 times than either the coprecipitatedor control particles. The presence of a tannic acid coating

on the postprecipitated solid would prevent the ferronreagent from directly contacting the aluminum atoms atthe surface and lead to longer t50 times.

Ourobservationsmayalsoprovideonereasonphosphateremoval is limited in the case of solids where the aluminumhas been prehydrolyzed in the presence of organic matter.In this situation, the coating on the surface of the inorganicprecipitate could prevent entry of the phosphate to regionswhere it is able to form a specific bond with aluminumatoms.

Conclusions

Using the AFM, we have imaged a series of hydrousaluminum oxides particles derived from the hydrolysis ofalum during a model wastewater treatment process. Theuse of phase imaging data has demonstrated that hydrousaluminum oxide particles formed in a postprecipitationprocess (that is, in which the particles precipitate in theabsence of organic components and are subsequently agedin the presence of tannic acid and phosphates) havemarkedly different viscoelastic properties than particlescoprecipitated in the presence of the organic and phosphatecomponents, or indeed control Al(OH)3 particles. In thelatter case, these particles show considerably less contrastwith the mica substrate in phase imaging mode. Theseobservations are attributed to the incorporation of thetannic acid as a coating layer on the postprecipitatedparticles. In addition, the hydrous aluminum oxideparticles show considerable variations in morphology,depending on the preparation process. These results canexplain the relative effectiveness of these particles inremoving the phosphate components from wastewater.

Acknowledgment. We thank the Natural Sciencesand Engineering Research Council of Canada for financialsupport.

LA980130+(16) Ng, K. F.; Kwong, K.; Huang, P. M. Geoderma 1981, 26, 179.

4736 Langmuir, Vol. 14, No. 17, 1998 Omoike et al.