Mechanical and Surface Chemical Properties of SomeSolid-Phase Hydrous Aluminum Oxide/Tannic Acid

Particles Investigated Using Scanning Probe Methods

Anselm Omoike and J. Hugh Horton*

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

Received July 9, 1999. In Final Form: September 20, 1999

Hydrous aluminum oxide particles precipitated from simulated wastewater were dispersed on micasubstrates and examined using atomic force microscopy, adhesion force measurements, interfacial forcemicroscopy (IFM), and zeta potential methods in order to understand their structural and coating properties.Results on IFM of the mica substrate are also reported. Three types of particles were examined: aluminumoxides coprecipitated by adding alum in the presence of a solution of phosphates and tannic acid;postprecipitated particles formed by adding phosphates and tannic acid to already formed aluminumoxides; a control case consisting of particles precipitated in the absence of either phosphate or organiccomponent. In all cases, the particles all had similar values for the reduced modulus of the tip-samplesystem, although in the case of the postprecipitated particles there was evidence for the presence of acompliant organic coating. The adhesive force and zeta potential measurements were also consistent withthis observation. The results are discussed in the context of the relative effectiveness of these aluminumoxides in removing phosphates and other contaminants from aqueous systems.

Introduction

Aluminum oxides are an important component of manyimportant systems, including soils,1,2 ore processing, andwater filtration systems.3 In the latter case, hydrousaluminum oxides are added to the water to perform twofunctions: to remove phosphorus via the formation ofinsoluble hydroxy aluminum phosphates; to more rapidlyeffect the sedimentation of organic components by theformation of large flocs of aluminum hydroxides whichcan entrap solid components in the wastewate and allowthem to more rapidly settle to the bottom of the sedi-mentation tank. Clearly, the nature of these particle’smorphology and their chemical properties depends verymuch on the conditions under which they are formed andthe chemical species found on the particle surface.4

We have recently reported5 on a study hydrous alumi-num oxide particles precipitated from alkaline solutionsof alum (Al2(SO4)3‚nH2O) in the presence of phosphatesand tannic acid (which simulates the humic organiccompounds found wastewater).1,6 This work was carriedout using tapping mode atomic force microscopy (AFM)in order to determine the morphology of these particlesand to elucidate the chemical identity of their surfacecoating. The particles were dispersed on a mica substratefor ease of imaging. We found that so-called “postpre-cipitated” particles, that is, hydrous aluminum oxidesprecipitated from solution and subsequently exposed toa mixture of tannic acid and phosphate ion, demonstratedlarge changes in contrast with the mica substrate in thephase imaging mode as a function of set point and tiposcillation amplitude. These changes in the phase imageswere not observed either with control particles of alumi-num oxide or with “coprecipitated” particles in which

aluminum oxides were precipitated out of solution in thepresence of phosphates and tannic acid. We could ten-tatively conclude that the postprecipitated particlesdemonstrated large changes in their surface viscoelasticproperties which must be associated with the presence ofa compliant organic coating on the particle surface thatis absent in the case of the coprecipitated particles.

The present paper undertakes to quantify the natureof this change in the viscoelastic properties of the surfaceby using a combination of scanning probe techniques.These include tapping-mode AFM,7 measurements ofadhesive forces between tip and sample as a function ofpH,8 and interfacial force microscopy9 (IFM) for measuringthe contact modulus between tip and sample, essentiallya measure of the hardness of the particles. In addition,zeta potential measurements10 on these particles arepresented. Using these methods, we show that thepresence of a tannic acid layer bound to the aluminumoxide particles has important effects on both the mor-phology and the chemical reactivity of these materials.

Experimental SectionFull details of the chemical synthesis of the hydrous aluminum

oxide particles used in this study have been described else-where.5,11 Briefly, particles of Al(OH)3 were formed by additionof alum to alkaline solutions of NaHCO3. Control particles wereprecipitated directly from this solution. Postprecipitated particleswere allowed to age for 5 min in solution followed by additionof tannic acid and phosphates to the mixture. Coprecipitatedparticles were precipitated directly from alkaline solutions

* To whom correspondence should be addressed: Tel (613)-533-2379. FAX (613)-533-6669. E-mail: hortonj@chem.queensu.ca.

(1) Goh, T. B.; Violante, A.; Huang, P. M. Soil Sci. Soc. Am. J. 1986,50, 820.

(2) Goh, T. B.; Huang, P. M. Clays Clay Miner. 1986, 34, 37.(3) Bowker, R. P. G.; Stensel, H. D. Phosphorus Removal From

Wastewater; Noyes Data Corp.: New Jersey, 1990.

(4) Schofield, R. K.; Taylor, A. W. J. Chem. Soc. 1954, 4445.(5) Omoike, A.; Chen, G.; Van Loon, G.; Horton, J. H. Langmuir

1998, 14, 4731.(6) Dempsey, B. A.; Ganho, R. M.; O’Melia, C. R. J. AWWA 1984, 76,

141.(7) Putman, C. A. J.; van der Werf, K. O.; de Grooth, B. G.; Van

Hulst, N. K.; Greve, J. Appl. Phys. Lett. 1994, 64, 2454.(8) van der Werf, K. O.; Putman, C. A. J.; de Grooth, B. G.; Greve,

J. Appl. Phys. Lett. 1994, 65, 1195.(9) Houston, J. E.; Michalske, T. A. Nature 1996, 356, 266.(10) Shaw, D. J. Introduction to Colloid and Surface Chemistry;

Butterworth-Heinemann, Ltd.: Oxford, U.K., 1992.(11) Omoike, A. I.; VanLoon, G. W. Water Res., in press.

1655Langmuir 2000, 16, 1655-1661

10.1021/la990896p CCC: $19.00 © 2000 American Chemical SocietyPublished on Web 12/15/1999

containing phosphate and tannic acid and allowed to age for afurther 5 min. The particles were then filtered from solution andallowed to air-dry.

All the particles were deposited on to freshly cleaved micasubstrates approximately 1 cm × 1 cm square. The particleswere first dispersed into distilled water by placement in anultrasonic bath for 60 min. The concentration of the dispersionwas 0.35 g L-1, and 30-100 µL of the dispersion was placed onthe mica substrate, spin-coated at 4000 rpm for 60 s, andsubsequently dried in air for 1 h. The amounts of dispersion usedwere chosen in order to obtain an even distribution of particlesof approximately 50% coverage on the substrate for the imagingexperiments. Larger volumes of the solution, resulting in 100%coverage of the surface with particles, were required whenpreparing samples on which force curve data was acquired.

AFM data were acquired using a PicoSPM operated in MACmode (Molecular Imaging, Tempe, AZ), using a Nanoscope IIEcontroller (Digital Instruments, Santa Barbara, CA). The MACmode is essentially the same as tapping mode, except that thecantilever is magnetically coated and is driven by an externaloscillating magnetic field.12 The cantilevers had a force constantof ∼1.0 N m-1 and a resonance frequency of ∼100 kHz. Imageswere acquired under ambient conditions, as well as undersolutions of various pH, at the fundamental resonance frequencyof the Si cantilevers. The solution-damped frequency of thecantilevers was ∼30 kHz. Height and phase shift data were allrecorded simultaneously, as a function of both cantilever oscil-lation amplitude (Ao) and set point ratio rsp) Asp/Ao. Images wererecorded at scan rates of 1-2 lines/s using a 30 µm × 30 µmscanner.

Force-distance curves for measurement of adhesion forcesbetween tip and sample8 were obtained using the same apparatus,in this case employing a tip with force constant of 0.1 N m-1.Curves were acquired under freshly prepared unbuffered NaOHor HCl solutions of pH ranging from 3 to 10. Unbuffered solutionswere used in order to prevent any unwanted interactions betweenthe surfaces and ions in solution. Some 300-500 curves wereobtained for each sample and were analyzed by finding thedifference between the value of the maximum tip deflection atthe bottom of the adhesive well and the zero of tip deflection atlarge tip-sample separations.13 The reported values of theadhesive interaction are an average of all the force curves obtainedwhile the reported errors reflect the standard deviation of thedata.

Nanomechanical data on the particles was acquired using aninterfacial force microscope (IFM).14 This differs from a conven-tional AFM in that the cantilever is replaced by a differentialcapacitance force sensor. The instrument used here has beenpreviously described15 and allows imaging of the sample surface(albeit at lower resolution than the tapping mode AFM), selectionof the features of interest, and then measurement of theirmechanical properties with nanonewton force resolution. Theforce sensor was calibrated using a thiol-covered gold standard.The force-distance curves were acquired by bringing a tungstentip into contact with the sample up to a preset repulsive load,followed by withdrawal from the surface. The repulsive interac-tion can be analyzed according to Hertzian theory to yield a valuefor the reduced modulus, E*, for this system. The relationshipbetween the applied force on the tip, F, and the tip displacement,D, is given by

where R is the reduced radius of curvature (essentially equal tothe tip radius of curvature in this case). R for the tips used inthis study was 80-100 nm, as determined from electronmicroscope images.

Zeta potential measurements on the particles were carriedout using a Laser Zee electrophoresis cell. Measurements weretaken again at a pH range of 3-10, with a cell potential of 150V.

ResultsTo act as a standard, IFM data were obtained for mica

surfaces of the type used as a substrate for our dispersedparticles as shown in Figure 1. The graph shows therestoring force applied to the sensor assembly as a functionof tip displacement: larger values of the displacementindicate a closer approach of the tip toward the surface.Two curves are shown: the first (dashed line) obtainedabout 15 min after the mica sample had been cleaved andplaced in the IFM; the second (solid line) obtained severalhours later. The curves are offset slightly from one anotherfor clarity. In both cases a complete cycle of tip approachto the surface followed by withdrawal is indicated by thearrows. Far from the surface, there is zero restoring forceon the IFM sensor, indicating that there is no surface-tipinteraction. In the first case, as we approach the surfacewe note a strong repulsive (positive) interaction as the tipis embedded in the surface. Just before the repulsiveinteraction there might be a very slight (<0.05 µN)attractive well. Upon withdrawal from the surface, wenote a somewhat larger, but still slight, adhesive interac-tion. In the second curve, acquired some hours later at apoint nearby on the surface, we see a much more definiteadhesive interaction on approach of some 0.05 µN. Thisis followed by a similar repulsive interaction as beforeand finally a much larger adhesive interaction of 0.26 µNcan be observed as the tip leaves the surface.

The reduced modulus for this system was determinedto be 15 ( 2 GPa using Hertzian contact theory and fittingthe data to eq 1 as described above. This value wasdetermined from the average E* from 15 curves. The fittedcurves are reproduced as the dotted lines in Figure 1.

AFM and IFM data were also acquired on the controlAl(OH)3 particles. The inset to Figure 2a shows an imageof these particles obtained using the IFM. As might beexpected14 the resolution in these images is much poorerthan that which can be obtained using tapping-mode AFMas demonstrated by the image shown in Figure 2b. Forcecurves were obtained on a number of particle sites, bothnear the center and edges of the particles. Points on thesurface at which force data were acquired are indicatedby crosses on the inset image. The particles did not appearto be significantly perturbed by the tip during acquisi-

(12) Han, W.; Lindsay, S. M.; Jing, T. Appl. Phys. Lett. 1996, 69,4111.

(13) Han, T.; Williams, J. M.; Beebe, T. P. Anal. Chim. Acta 1995,307, 365.

(14) Joyce, S. A.; Houston, J. E. Rev. Sci. Instrum. 1991, 62, 710.(15) Warren, O. L.; Graham, J. F.; Norton, P. R. Rev. Sci. Instrum.

1997, 68, 4124.

F ) 43

E*R1/2D3/2 (1)

Figure 1. Force-distance curves obtained using the interfacialforce microscope on a mica surface in air for 20 min (dashedline) and 3 h (solid line) after cleaving the sample. Approachand retraction portions of the curves are indicated using arrows.The thicker line on the second curve indicates the curve fittingof eq 1 to the data as described in the text.

1656 Langmuir, Vol. 16, No. 4, 2000 Omoike and Horton

tion of the force curves, as indicated by before-and-afterimages of the probed site. The curves were quite repro-ducible; a typical force curve for the control particles isshown in Figure 2a. As can be observed, there is littleevidence for any adhesion during either the loading orunloading portion of the curve. A reduced modulus forthis system was determined as with the mica sample givingan E* of 12 ( 3 GPa.

A tapping mode AFM image acquired for the copre-cipitated particles, in which the aluminum hydroxide isprecipitated directly from a solution containing tannicacid, is shown in Figure 3a. As we have previouslyobserved,5 these particles do not look very different fromthe control aluminum hydroxides. IFM force-distancecurves were obtained again on a number of particle sites.It was found that, unlike the previous cases, these particlescould be perturbed by the tip during acquisition of theforce curves. This was usually characterized by eithermovement of the particles about on the surface overdistances of up to 500 nm or, in some cases, the formationof indentations within the particle. A typical curve is shownin Figure 3b. This curve shows evidence for a strongadhesive interaction between tip and sample of some 0.5µN. E* values for the coprecipitated particles weredetermined to be 18 ( 2 GPa. The fitted curves are shownas dotted lines in Figure 3b.

An image of the postprecipitated particles is shown inFigure 4a. As has been previously observed, these particlesdiffer significantly in morphology from either the control

Al(OH)3 particles or the coprecipitated particles. They aremuch smaller in size and appear to form floc-typestructures on the surface. The IFM force curve (Figure4b) also differs significantly from the other cases. At firstglance, it seems fairly similar and fitting the data givesan E* value of 15 ( 4 GPa. However, closer examinationof the approach portion of the curve, as seen in the insetto Figure 4b, shows there is a region 4 nm wide in whichthe curve shows a much shallower slope than in the mainportion of the curve. This could be reproduced in a numberof different points on the sample. There is also a smalladhesive interaction seen on retraction from the surface.

Figure 5 shows a typical data set collected in themeasurement of the adhesive interaction between thesilicon nitride tip and a mica surface covered with, in thiscase, the postprecipitated particles at a pH of 8.3. Datasets of similar quality were acquired on the mica surfaceand for the coprecipitated and control particles at variouspH values. The results are summarized in the four graphsof Figure 6 which show the measured adhesive force forthe mica surface and all three particles as a function ofpH. While the (attractive) adhesive force between tip andsample can be obtained from the retraction portion of theforce-distance curve, the approach portion of the curvealso contains information on the forces between tip andsample. In this case, as in Figure 5, the approach portionof the curve often showed a repulsive interaction betweentip and sample which could not be reliably measured.Occasionally an attractive interaction was observed. Thisis indicated on the graphs for the control particles and the

Figure 2. (a) Force-distance curve obtained using theinterfacial force microscope on control Al(OH)3 particles dis-persed on a mica substrate. Approach and retraction portionsof the curves are indicated using arrows. The dotted lineindicates the curve fitting of eq 1 to the data to acquire thereduced modulus value for this system as described in the text.The inset shows an image of the particle surface in which thecrosses indicate the positions on the sample on which forcecurves were obtained. (b) AFM image of the control particlesin height (left) and phase imaging (right) mode. The images are2 µm square and acquired at a set point of 0.85 and tip oscillationamplitude of 10 nm. The z-range in the height images is 20.4nm and in the phase images 19°.

Figure 3. (a) AFM image of the coprecipitated particles inheight (left) and phase imaging (right) mode. The images are2 µm square and acquired at a set point of 0.85 and tip oscillationamplitude of 10 nm. The z-range in the height images is 20.4nm and in the phase images 19°. (b) Force-distance curveobtained using the interfacial force microscope on coprecipitatedparticles dispersed on a mica substrate. Approach and retractionportions of the curves are indicated using arrows. The dottedline indicates the curve fitting of eq 1 to the data to acquire thereduced modulus value for this system as described in the text.

Surface Properties of Aluminum Oxide Particles Langmuir, Vol. 16, No. 4, 2000 1657

mica surface as either an “A” for an attractive interactionon approach or an “R” indicating a repulsive interactionon approach. As can be observed, in the case of the controlsample the attractive regime is limited to a narrow regionaround pH 6-8. The boundaries of this attractive regionare roughly indicated by the vertical dashed lines on thegraphs. On mica, an attractive regime was seen below apH of 6. On the postprecipitated and coprecipitatedparticles, a definitive attractive potential was not obsevedon tip approach (and hence is not indicated on the figure),although on some curves, a small attractive well wasobserved at pH values below 5 in the approach curve.Certainly, below a pH of 5 no repulsive interactions wereobserved on approach of these two particles.

Zeta potential measurements on the three particles aresummarized in Figure 7. In the case of the control particleswe see an isoelectronic point as indicated by a zetapotential of zero at a pH of 4.7. In the other two cases, thesign of the zeta potential is negative throughout the pHrange examined here. For the control and coprecipitatedparticles, the zeta potential is a fairly linear function ofpH with similar slopes in the two cases. However, thecoprecipitated particles have a zeta potential which is 20mV lower at any given pH than that of the control particles.The postprecipitated particles show a different behaviorwith pH. Here, the zeta potential drops from -20 to -45mV over a pH range of 3-4.5. The zeta potential thenplateaus at -45 mV, and then drops fairly precipitouslyto -65 mV at pH values above 8.

DiscussionWe first consider the IFM data acquired on the mica

surface. The measured reduced modulus of this system of15 GPa is lower than the bulk modulus observed for biotitemica of 40 GPa.16 We would propose that this discrepancyarises because mica is an isotropic substance withsignificantly different mechanical properties in the direc-tions parallel and perpendicular to the aluminosilicatesheets which make up the crystal structure.17,18 In theliterature, as with our system, the measurement is takenin a direction normal to the sheet structure, which hashigher strength due to strong covalent bonds. The sheetsthemselves are bound to one another by only by weakelectrostatic interactions, and the modulus of the systemin the direction parallel to the sheets is presumably muchlower, probably by several orders of magnitude, althoughno measurements of this value appear to have been made.Imaging the surface after indentation shows that insteadof a simple indentation crater, the surface has beenfractured horizontally with mica layers piled along thesides of the indentation region. This demonstrates thatalthough the tip exerts a contact force mainly normal tothe surface, there is a component parallel to the surface.Thus, the overall measured reduced modulus must be aweighted average of the relative contributions of thesetwo components. Since the modulus in the perpendiculardirection is much smaller, the net result is a reducedmodulus lower than expected.

(16) Carmichael, R. S. Handbook of Physical Properties of Rocks;CRC Press: Boca Raton, FL, 1984; Vol. III.

(17) Bailey, S. W. Rev. Mineral. 1984, 13, 1.(18) Tang, H.; Joachim, C.; Devillers, J. J. Vac. Sci. Technol., B 1994,

12 2179.

Figure 4. (a) AFM image of the postprecipitated particles inheight (left) and phase imaging (right) mode. The images are2 µm square and acquired at a set point of 0.85 and tip oscillationamplitude of 10 nm. The z-range in the height images is 14.4nm and in the phase images 6.4°. (b) Force-distance curveobtained using the interfacial force microscope on postprecipi-tated particles dispersed on a mica substrate. Approach andretraction portions of the curves are indicated using arrows.The dotted line indicates the curve fitting of eq 1 to the datato acquire the reduced modulus value for this system asdescribed in the text. The inset shows the region of the samecurve just prior to the onset of the strong repulsive interactionson a smaller scale.

Figure 5. (a) A typical force-distance curve obtained usingthe conventional AFM and a Si3N4 cantilever of force constant0.1 N/m. The curve was obtained on the postprecipitatedparticles at pH 8.3. (b) Typical distribution of adhesive (pull-off) forces observed. In this case 350 observations were made,and the average and standard deviation of the pull-off force areindicated.

1658 Langmuir, Vol. 16, No. 4, 2000 Omoike and Horton

The reduced moduli observed for the three differentparticles as well as the mica surface are essentially thesame within experimental error. The Young’s modulus ofAl2O3 (corundum) has been reported variously as 250-280 GPa.16,19 The reduced modulus (E*) of the tip samplesystem used here is given as

where Es and Et are the Young’s modulus of sample andtip, respectively, and ν is the Poisson ratio. Since theYoung’s modulus of tungsten, Et, is on the order of 380GPa,20 for many substrates where Es is much smaller thanEt the reduced modulus of the system is essentially thesame as the Young’s modulus of the sample. Here this isnot the case, and using values of 250 GPa for the Young’smodulus of Al2O3

16 and values of 0.3 for both νs16 and νt,20

we arrive at a theoretical reduced modulus for this systemof 166 GPa, still much larger than that observed here.

The fact that the observed reduced modulus values forthe particles are similar to that of the mica could suggestof course that we are not actually acquiring force curvemeasurements on the particles themselves but insteadon the substrate, either because we were indenting on thewrong area on the surface or because the particles weremoving out from underneath the tip during the indentationprocess. The first possibility seems very unlikely as beforeand after images of the surface showed that indentationwas occurring on top of the particles, while the particlesthemselves were at high coverages (>50%) on the surface.

The possibility of particles moving during indentationmust be taken more seriously. However, the only systemin which we saw evidence in the imaging of particles havingmoved during the indentation process was with thecoprecipitated particles. In this case not only was thereduced modulus of 18 ( 2 GPa the highest observed, butwe also saw indentations within the particles themselves.Also, the adhesion characteristics of the four systems werevery different, which suggests we are observing differentsurfaces. It is also possible of course that if the particlesare significantly harder than the substrate, then the tipwill push them into the substrate surface, and themeasured quantity will be the substrate modulus.

The reduced modulus measurement then suggests thatthe particles themselves are fairly similar to one anotherin terms of their bulk mechanical properties and are quitea bit softer than corundum. In all probability the particleshave been extensively hydrolyzed to Al2O3‚2H2O or Al-(OH)3. While Young’s modulus and Poisson ratio data forAl(OH)3 do not appear to have been published, presumablythe modulus of this material must be lower than that ofcorundum, leading to the much lower reduced modulusvalue observed here.

A significant observation is the shallow slope seen onthe approach curve on the postprecipitated particles inFigure 4b. In this case, we have hypothesized that thesystem consists of a thin overlayer of tannic acid adsorbedon the particle surface. Other similar systems which havebeen examined include stearic acid monolayers on aluminausing AFM21,22 and IFM studies of alkanethiol monolayersadsorbed on a gold film23-25 In this latter case, force-

(19) Cotrell, A. H. The Mechanical Properties of Matter; John Wiley& Sons: New York, 1964.

(20) Grigoriev, I. S.; Meilikhov, E. Z. Handbook of Physical Quantities;CRC Press: Boca Raton, FL, 1997.

(21) Burnham, N. A.; Dominguez, D. D.; Mowery, R.L.; Colton, R. J.Phys. Rev. Lett. 1990, 64, 1931.

(22) Burnham, N. A.; Colton, R. J.; Pollock, H. M. J. Vac. Sci. Technol.,A 1991, 9, 2548.

Figure 6. Plots of the adhesive (pull-off) forces observed as a function of pH on a mica surface and on the various particles examinedhere. The system studied is indicated on each graph. The notations R and A denote pH regions in which there was either a repulsiveor attractive tip-sample interaction upon the approach portion of the force-distance curve.

E* ) (1 - υs2

Es+

1 - υt2

Et)-1

(2)

Surface Properties of Aluminum Oxide Particles Langmuir, Vol. 16, No. 4, 2000 1659

distance curves showed a fairly shallow repulsive interac-tion followed by a stiffer interaction as the indentationproceeded. In accordance with eq 1, the slope of the curveshould decrease as the value of the reduced modulus ofthe system decreases, hence indicating that the shallowregion corresponds to a soft overlayer. No adhesive orattractive interaction was seen. The lack of these interac-tions was expected since in this case the monolayerconsisted of a hydrophobic alkyl layer. In our case, wealso see this shallow repulsive interaction before the onsetof a much stiffer interaction with the particle itself. Thisis a good indication that what we see here is a soft layerof tannic acid a few nanometers thick adsorbed on the

surface overlaying the aluminum hydroxide particlebeneath it. There are some differences, however, with thealkanethiol system in that here there appears to be adiscontinuity between the hard and soft regions, as wellas a small adhesive interaction upon retraction. Thissuggests that perhaps the tip has “punched through” thetannic acid layer to encounter the aluminum oxide particlebeneath. Certainly the adhesive interaction suggests somekind of hydrophilic surface is present, either the aluminumoxide surface itself or hydrophilic phenolic or ester linkageson the tannic acid molecule. Whether adhesion arises dueto a true interaction between tip and sample or becauseof a capillary effect due to water adsorption cannot,however, be discerned from the IFM data for reasonsdescribed below.

IFM measurements were carried out in air using atungsten tip. As such, a major contribution to the attractiveinteractions observed in these data must arise fromcapillary forces between tip and sample. For example, inthe case of the mica sample we see in Figure 1 thatimmediately after cleaving we see little adhesive interac-tion. After longer exposure times to air, the adhesionincreases. After several hours at even the relatively low(20%) humidity in which experiments were conducted,mica adsorbs several monolayers of water.26 Any adhesiveinteractions observed in the IFM data on the particlesamples must have considerable contribution due tocapillary effects,27 and thus these observations are notparticularly instructive in trying to understand thechemical functionality of the surface. We thus considerfurther the data summarized in Figure 6, in whichadhesive forces were measured under more controlledconditions under solution of varying pH.

In this case, the probe consisted of a Si3N4 tip. Such tipshave been used to examine tip-sample interactionsbetween a variety of different substrates, including Si3N4,28

mica,29 corundum (Al2O3),30 and silica,31 as a function ofpH. Silicon nitride undergoes hydrolysis upon exposureto aqueous solution to form various silanol and silylaminesurface groups. This tip is thus amphoteric with anisoelectronic point being reported variously as pH 5.5-730 or 6-8.5.28

We start by looking at the results for the control (Al-(OH)3) particles as seen in the upper left graph in Figure6. We observe that on approach, there is an attractiveregion between pH 5.2 and 6.7 (as indicated by the A anddashed line boundaries on the figure), while at other pHvalues the approach curve shows a repulsive interaction.This is consistent with previous observations on acorundum (Al2O3) surface and Si3N4 tip and can beexplained on the basis of surface charge effects alone.30,31

The isoelectronic point of a corundum surface is at pH )9.32 Thus at pH values above 9, the surface will benegatively charged while below pH ) 9 the surface is

(23) Joyce, S. A.; Houston, J. E.; Michalske, T. A. Appl. Phys. Lett.1992, 60, 1175.

(24) Joyce, S. A.; Thomas, R. C.; Houston, J. E.; Michalske, T. A.;Crooks, R. M. Phys. Rev. Lett. 1992, 68, 2790.

(25) Thomas, R. C.; Houston, J. E.; Michalske, T. A.; Crooks, R. M.Science 1993, 259, 1883.

(26) Hu, J.; Xiao, X.-D.; Ogeltree D. F.; Salmeron, M. Surf. Sci. 1995,327, 358.

(27) Weisenhorn, A. L.; Maivald, P.; Butt, H. J.; Hansma, P. K. Appl.Phys. Lett. 1989, 54, 2651.

(28) Senden, T. J.; Drummond, C. J. Colloids Surf., A 1995, 94, 29.(29) Weisenhorn, A. L.; Maivald, P.; Butt, H.-J.; Hansma, H.-J. Phys.

Rev. B 1992, 45, 11226.(30) Feldman, K.; Fritz, M.; Hahner, G.; Marti A.; Spencer, N. D.

Tribol. Int. 1998, 31, 99.(31) Marti, A.; Hahner, G.; Spencer, N. D. Langmuir 1995, 11, 4632.(32) Hahner, G.; Marti, A.; Spencer, N. D. Tribol. Lett. 1997, 3, 359.

Figure 7. Zeta potentials of the various particles as a functionof pH. The particle studied is indicated on each graph. Thecurves are to be used as a guide to the eye.

1660 Langmuir, Vol. 16, No. 4, 2000 Omoike and Horton

positively charged. Similarily, the tip will be negativelycharged at pH values above its isoelectronic point andpositively charged below. The net result is that the surfaceand tip will have opposite charges and thus attract at pHvalues intermediate to their respective isoelectronic pointsand have like charges and a repulsive interaction willoccur at pH outside these values. The data in Figure 6suggest then an isoelectronic point for the tip of about 6.7,consistent with the range of values quoted for Si3N4 in theliterature and an isoelectronic point of about pH ) 5.2 forthe control particles. This latter value is roughly consistentwith the zeta potential measurements on the controlparticles seen in Figure 7 which show that the zetapotential becomes zero, and hence the isoelectronic pointis reached at a pH of 4.7. This is quite different from theisoelectronic point of pH ) 9 reported for the corundumsurface. It suggests that, as might be expected, the controlparticles are hydrolyzed to a considerable extent and areprobably closer to Al(OH)3 which, since the surface chargemust arise from protonation/deprotonation of danglingAl-O- species, will be expected to have quite a differentisoelectronic point from that of corundum. The result isalso, of course, consistent with the lower values than wouldbe expected for Al2O3 obtained in the reduced modulusmeasurement on these particles.

While the isoelectronic point was not the same, theadhesive interaction observed on the control Al(OH)3particles is qualitatively consistent with previous obser-vations on a corundum surface. In that case, both theadhesive interaction and the lateral force were found toreach at maximum at a pH intermediate to the twoisoelectronic points. We see a similar trend in the adhesiondata in Figure 6, with the adhesive interaction reachinga maximum at pH 5.7.

The mica surface is known to exhibit a net negativecharge under all pH conditions,33 although charge reversalis known to occur under conditions of high ionic strength.On the mica surface, we note an attractive potential upontip approach at pH values of less than 5, as seen in Figure6. This is roughly consistent with the observations on thecontrol particles which indicated that the isoelectronicpoint of the tip is near a pH of 6.7. Below this point, thetip should be positively charged, and hence we observe anattractive interaction on approach. In this case, there doesnot appear to be any dramatic change in the tip-sampleadhesive interaction as a function of pH. No interactionexceeded 5 nN. This is consistent with previous observa-tions on this system28 although in that case no attemptwas made to quantify the adhesive interaction. It isimportant to note that in the case of the experiments onthe particles themselves, the adhesive interaction behavioras a function of pH differs considerably from that of themica substrate, demonstrating that our force distancecurves are truly being acquired on particle sites on thesurface rather than on the substrate.

The adhesive interaction between tip and sample differsconsiderably as a function of pH on the postprecipitatedand coprecipitated samples, and indeed from the controlcase. The postprecipitated particles show a broad maxi-mum of about 5 nN in the adhesive interaction at pHvalues ranging from 5 to 8. The coprecipitated particleson the other hand, show a large increase in the adhesiveinteraction between tip and sample above pH ) 8, withthe interaction becoming as high as 20 nN, much largerthan that observed on any other particles. Clearly thesurface chemistry of these two particles must be quitedifferent.

The coprecipitated particles contain a number of dif-ferent -OH terminated sites, all of which will tend to beionized at higher pH. Presumably this large concentrationof ionized sites leads to the adhesive interaction at highpH. This is also indicated by the zeta potential measure-ments, which show a more negative zeta potential thanthat of the control particles at any given pH. Presumablythis is due to the higher number of ionizable sites presentarising from phosphates or carboxylic and phenolic sites(from tannic acid) present in the coprecipitated particlesin addition to, of course, Al-OH sites available on bothparticles.

We finally note that on the postprecipitated particles,there is little or no adhesive interaction, suggesting a fairlylow density of hydroxyl sites on the surface, at leastcompared to the coprecipitated particles. Since the zetapotential of these particles is also less negative, this isalso an indication that there are fewer hydroxyl sites.Indeed, the zeta potential of the postprecipitated particlesplateaus around a pH of 4.5 and does not begin to dropagain until a pH of 8. This indicates that unlike the othertwo cases, there are no sites on the postprecipitatedparticle’s surface which can be ionized over this pH range.Once a pH of 8 is reached, however, the zeta potentialdrops. This is the pH condition at which we would expectionization of phenoxy OH groups from the tannic acid tobegin, as the pKa of the phenoxy group is in the range of8-9. This observation, then, is consistent with a surfacecoating of tannic acid on the postprecipitated particles.

ConclusionsA series of aluminum hydroxyl-based particles dispersed

on a mica substrate have been examined by scanning probemethods. Zeta potential measurements have also beenmade on the particles in solution as a function of pH.These particles are important as synthetic versions of thecolloidal particles present during wastewater treatmentprocesses. Tapping mode AFM images of aluminumhydroxide particles precipitated from alkaline solutionsof Al2(SO4)3 showed that the particles were spherical andseveral hundred nanometers in diameter. IFM measure-ments on the particles gave a reduced modulus of thesystem of 12 ( 3 GPa, while force microscopy and zetapotential measurements showed the particles had anisoelectronic point at a pH of about 5, consistent with themain surface features being Al-OH sites. When similarparticles were coprecipitated in the presence of phosphatesand tannic acid, the morphology and elastic properties ofthe system were little changed, although the zeta potentialof the system was lower, consistent with more ionizableOH sites on the surface arising from phosphate andcarboxcylic acid groups. Particles precipitated from solu-tion and then exposed to tannic acid (so-called postpre-cipitated) showed very different morphological charac-teristics. AFM images show much smaller particles, aswell as very high contrast with the mica substrate in thephase-imaging mode. IFM and zeta potential measure-ments were consistent with the particles being coated ina layer of tannic acid. This model for the particles explainsthe relative differences in chemical reactivity of thepostprecipitated and coprecipitated aluminum oxidestoward the removal of phosphates and other contaminantsfrom aqueous systems.

Acknowledgment. We acknowledge the NaturalSciences and Engineering Research Council of Canadafor financial support. We also acknowledge John Grahamfor his help in the acquisition of the IFM data.

LA990896P(33) Kekicheff, P.; Marcelija, S.; Senden, T. J.; Shubin, V. E. J. Chem.

Phys. 1993, 99, 6098.

Surface Properties of Aluminum Oxide Particles Langmuir, Vol. 16, No. 4, 2000 1661