- Email: [email protected]
Chemical Force Microscopy Investigation of Phosphate Adsorption on the Surfaces of Iron(III) Oxyhydroxide Particles
Journal of Colloid and Interface Science 254, 205–213 (2002) doi:10.1006/jcis.2002.8575
Chemical Force Microscopy Investigation of Phosphate Adsorption on the Surfaces of Iron(III) Oxyhydroxide Particles D. I. Kreller, G. Gibson, G. W. vanLoon, and J. H. Horton1 Department of Chemistry, Queen’s University, Kingston, Ontario, Canada K7L 3N6 Received January 25, 2002; accepted July 9, 2002
Phosphate-modified AFM tips were prepared by the deposition of self-assembled monolayers (SAMs) of bis(11-thioundecyl) phosphate on Au-coated silicon nitride cantilevers. The properties of the PO2 H-terminated SAMs were investigated by studying the pH-dependent force interactions of the tips with phosphateand carboxylic acid-terminated SAM control surfaces. The PO2 H functional groups had a pKa of ≈5.0. A chemical force microscopy (CFM) study was conducted on the interactions between the probes and the surfaces of hydrous ferric oxide particles prepared in our laboratory by hydrolytic precipitation from FeCl3 . The forces between PO2 H probes and the hydrous ferric oxide surfaces were seen to exhibit a strong pH dependence, with maximum attractive forces occurring for pH values between 5 and 8. The effects of postprecipitation of the hydrous ferric oxide colloids with orthophosphate, H2 PO− 4 , dimethylphosphate, (CH3 O)2 PO2 H (DMP), and tannic acid (TA) on the adhesive interactions between the PO2 H tips and the solids were also investigated. Attenuated total reflectance infrared spectroscopy (ATR-IR) was used to verify the presence of surfaceadsorbed species and zeta potentiometric measurements to determine surface charge on the colloids. We show that the method of chemical force titration using phosphate-terminated tips can differentiate between these various colloids and that it shows promise as a general method for studying this environmentally important class of compounds. C 2002 Elsevier Science (USA) Key Words: self-assembled monolayers; phosphate; iron(III) oxyhydroxide; colloid; atomic force microscopy; chemical force microscopy.
INTRODUCTION
The behavior of phosphate and natural organic matter (NOM) in soils and aqueous media is widely known to be strongly influenced by their association with surfaces of colloidal mineral oxide and oxyhydroxide particles of iron and aluminum (1, 2). Strong retention of phosphate and NOM on such mineral surfaces controls the mobility, reactions, and bioavailability of these nutrients. Competitive adsorption occurs between phosphate and organic species at mineral surfaces. For example, phosphate does not adsorb as effectively to a mineral surface to which an organic 1 To whom correspondence should be addressed. Fax: 613-533-6669. E-mail: [email protected].
205
material is already complexed as it does to an otherwise uncoated mineral surface. Inhibitory binding relationships among such compounds have been studied in the context of soils and water research (3, 4). Orthophosphate species (PO3− and protonated versions 4 therof) are present in low concentrations in wastewater, but its release in wastewater effluent can be of environmental significance because it can be a contributing factor to eutrophication of the receiving water body. To reduce the levels of phosphorus in wastewater effluents, the high affinity of phosphate for Fe and Al based oxyhydroxides is used to advantage in the process of chemical coagulation. Salts such as FeCl3 or Al2 (SO4 )3 are added to the wastewater stream and the M(III) ions undergo alkaline hydrolysis to generate colloidal oxyhydroxide “floc” particles. Coprecipitation of phosphate or its adsorption onto the reactive and initially very high surface area precipitates leads to incorporation into insoluble solids and its removal from solution. The competition of organic materials with phosphate ions for binding sites on the surfaces of flocculant particles can have important implications for the design of wastewater treatment plants. For example, phosphate removal may be enhanced when at least a portion of the coagulant is added at the outlet of the aerator, where the concentration of organic matter is reduced (5). The combined techniques of atomic force microscopy (AFM) and chemical force microscopy (CFM) enable topographical imaging and the investigation of chemical interactions at the surfaces of nonconducting substrates. This feature makes them, in principle, applicable to a broad range of environmentally relevant surface interactions (6). These techniques can be used to augment information derived via other chemical methods, i.e., solution adsorption/desorption, and surface analysis techniques such as infrared spectroscopy. Atomic force microscopy is used primarily as an imaging method. Chemical force microscopy has been introduced as a method for probing the various adhesive and repulsive interactions at surfaces (7–10). These include van der Waal’s, hydrogen-bonding, and electrostatic charge interactions. CFM allows both the imaging of surfaces with chemical specificity (11) and the quantitative measurement of sitespecific interactions between the tip and sample (12), in principle with nanometer-scale resolution. Quantitative measurements of Hamaker constants for tip/sample systems in various solvents 0021-9797/02 $35.00
C 2002 Elsevier Science (USA)
All rights reserved.
206
KRELLER ET AL.
can be made. Also, as will be explored further in this work, CFM allows the method of force titration in which the tip–sample interaction is measured as a function of some property, in this case pH, of an aqueous solution (13). Force pH titrations allow the determination of ionization constants, i.e., pK a , of functional groups on the interacting surfaces and the pH dependence of adhesive attractions between them. While early workers examined the interaction of unmodified silicon nitride tips with surfaces, more recent work has concentrated on probe tips which have been modified to be terminated with various end groups, including alkylsilanes (14), alkylthiols (15), chiral molecules (16), DNA and proteins (17, 18), and silica particles (19). In the present work, we are interested in the process of phosphate adsorption on colloidal particles; hence, a tip functionalized with a phosphate end group is required. The preparation of alkylthiol-based phosphate self-assembled monolayers and phosphate CFM has attracted surprisingly little attention given that, in addition to the importance of phosphate in environmental systems, phospholipid-type molecules are also of interest in biological systems, including cell membranes and DNA (Hansma). The synthesis and AFM characterization of phospholipid bilayers deposited using Langmuir– Blodgett and related techniques (20–23) have been more thoroughly explored. Preparation of a phosphate surface via deposition of a phosphate-terminated alkylthiol compound on Au is attractive because the self-assembly of ordered arrays of alkylthiols on Au is a well-established general procedure, and these monolayer systems have been quite well characterized. The synthesis and self-assembly on Au of the bis(11-thioundecyl)phosphate compound used here has been reported (24, 25) and the interaction of this phospholipid monolayer with methylviologen and Ca2+ ions has been explored using electrochemical methods. More recently, CFM studies of phosphate monolayers self-assembled from 11-thioundecyl-1-phosphonic acid (i.e., the monoalkylsubstituted version of the phosphate explored in this paper) have been carried out (26–28). We have previously reported (29, 30) on a study of hydrous aluminum oxides, a class of colloids closely related to the iron oxyhydroxides reported on here. The hydrous aluminum oxide particles were prepared by hydrolytic precipitation from alkaline solutions of alum, (Al2 (SO4 )3 · nH2 O) in the presence of phosphates and tannic acid. This work was carried out using a combination of tapping mode atomic force microscopy, chemical force microscopy, and interfacial force microscopy to determine both the morphology of these particles and to elucidate the chemical identity of their surface coating. The force microscopy data in this previous work were acquired, however, using unmodified silicon nitride tips. In the present paper, we report on the preparation of phosphate-terminated AFM probes and their application in the study of phosphate adsorption to hydrous iron oxide colloids. The compound bis(11-thioundecyl)phosphate, (HSC11 H22 O)2 PO2 H, was synthesized and subsequently deposited by self-assembly of an alkylthiol monolayer on a gold
film to form a phosphate-terminated surface. The force interactions of PO2 H-terminated AFM probes were studied as a function of pH, first with a series of control test surfaces and subsequently with a series of Fe3+ -based hydrous oxide colloids including “bare” Fe oxyhydroxide colloids as well as particles postprecipitated with orthophosphate (PO3− 4 ), dimethylphosphate (CH3 O)2 PO2 H (DMP), and tannic acid (TA, a surrogate for natural organic matter). Attenuated total reflectance infrared spectroscopy (ATR-IR) was used to verify the presence of surface-adsorbed species via their characteristic stretching frequencies. While chemical force measurements using phosphate-coated tips have been previously reported, notably the work of van der Vegte and Hadziioannou (11, 27) and Smith and co-workers (26, 28), these are the first results involving tips terminated with the dialkylphosphate (e.g., monoprotic) compound, as opposed to the diprotic tips used in the above work. More importantly, we apply the chemical force methods to a real world system, namely, metal oxide colloidal particles, as opposed to the model systems of self-assembled monolayers on Au explored in the previous work. We show that the method of chemical force titration using phosphate-terminated tips can readily distinguish between the surfaces of these various colloids. This technique shows promise as a general method for studying this important class of compounds. MATERIALS AND METHODS
Synthetic Methods The synthesis of bis(11-thioundecyl)phosphate was carried out using the method of Nakashima and Taguchi (24, 25). The various hydrous iron oxide colloids substrates used in the force titration experiments were synthesized at room temperature (293 K) using room temperature reagents. It is important to note that only a fraction of the product was ultimately used for the preparation of the AFM samples. Also, large excesses of the compounds which were used to variously coat the surfaces of particles (i.e., tannic acid, orthophosphate, dimethylphosphate) were introduced into the solutions to ensure complete surface coverage of the particles; hence, significant quantities of unadsorbed compound generally remained in solution. (i) Native hydrous iron oxides: 50 µL of 41% w/v ferric chloride solution (Fanchem Ltd. Oakville, Ontario) was added by micropipette to 1 L of distilled water to form a 7 ppm Fe solution. The pH was adjusted to a value of 6 using a 0.50 M NaHCO3 solution. For uncoated hydrous oxide colloids, the precipitate was allowed to age for 20 min with 300 rpm stirring, and then a portion was collected on a freshly cleaved mica slide (Grade V-5, SPI Supplies). To do this, a slide was placed on a 0.45-µm filter paper (Osmonics) in a filter funnel, and the hydrolysis solution was transferred into it under suction. As the filtrate was removed, the solids settled and a portion was collected on the mica slide. The deposited solids were allowed to dry at room temperature for 24–48 h prior to analysis. (ii) Hydrous iron hydroxides postprecipitated with orthophosphate: to the solution containing 20-min-aged
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE
precipitate as in part (i), phosphate as HPO2− 4 (Fisher Scientific) from a stock solution was added in a molar ratio of P : Fe = 7 : 1. The solution was allowed to age a further 15 min and the solids were then similarly collected for analysis. (iii) Hydrous iron hydroxides postprecipitated with dimethylphosphate: to the solution containing 20-min-aged precipitate as in part (i), DMP (Across) was added in a molar ratio of DMP : Fe = 7 : 1. The solution was aged a further 15 min and then collected as above. (iv) Hydrous iron hydroxides postprecipitated with tannic acid: to the solution containing 20-min-aged precipitate as in part (i), tannic acid (TA, Aldrich) was added in a molar ratio of TA : Fe = 1.4 : 1, the solution was aged for a further 10 min, and then the solids were collected as above. The TA-coated colloid displayed low adhesion on the mica substrate, presumably due to the presence of the organic overlayer on the precipitate. This required the use of double-stick tape applied to the mica before colloid deposition. Instrumentation Infrared data were acquired for bis(11-thioundecyl)phosphate in transmission mode using a KBr disk. Infrared data for the colloids and bis(11-bisthioundecyl)phosphate-modified Au surface were acquired in attenuated total reflectance (ATR) mode using a ZnSe crystal. AFM data were acquired using a PicoSPM (Molecular Imaging, Tempe, Arizona), using a Nanoscope IIE controller (Digital Instruments, Santa Barbara, CA). Images were acquired under ambient conditions as well as in solutions of various pH. Force– distance curves were obtained using the same apparatus. For the acquisition of force curves, the CFM tips and samples were immersed in freshly prepared unbuffered aqueous NaOH or HCl solutions of pH ranging from 3 to 10. The pH of the solutions was checked before and after the experimental run to ensure that the pH had not changed. Some 300–500 force curves were obtained for each sample. The reported values of the adhesive interaction are an average of all the curves while the reported errors reflect the standard deviation of the data. The AFM tips were standard silicon nitride cantilevers (Digital Instruments, Santa Barbara, CA) coated with 25 nm of Au. The force constants of the cantilevers ranged from 0.2 to 1.0 Nm−1 as determined by the method of Hutter and Bechhoefer (31, 32). Experimental runs for a pH series were performed using a single tip on a given sample. In addition, several checks were made to ensure the reproducibility of the data. In cases where samples consisted of self-assembled monolayers or bare gold, the tip and sample end groups were interchanged, yielding similar results. Repeating the colloid CFM experiments yielded curves that were identical in profile as a function of pH, in which the adhesive force varied in magnitude by about 20%. This variation can be attributed to changes in tip radius, which will affect the tip–sample contact area and hence the number of tip–sample bonds formed and the measured adhesive force. Generally, it is believed that between 40 and 100 bonds are formed between tip and sample, depending on tip radius (13). As we note in the discussion, it
207
is the profile of the curves as a function of pH, which is the most important factor that we take into account in the analysis of our data. In a comparison of the results on different tip/sample pairs, it is only in cases where the change in overall magnitude is greater than 50% that we attribute any significance to the data. In addition, the runs could be repeated from either high to low or low to high pH values without markedly affecting the results. The Au-coated tips were functionalized with PO2 H SAMs by immersing them for 48 h in a 1.0 × 10−3 mol L−1 solution of bis(11-thioundecyl)phosphate in n-octanol. Substrate SAM surfaces of bis(11-thioundecyl)phosphate were similarly prepared by the immersion of mica substrates (previously coated with 25 nm of Au and subsequently hydrogen-flame annealed) in the same thiol solution. Tips coated with CO2 H end groups were prepared by immersing them in a 1 mM solution of 16thiohexadecanoic acid (Aldrich, 95%) in methanol for 24 h. These deposition conditions were chosen to ensure that both tips and samples were saturated with a complete monolayer of alkylthiol. RESULTS AND DISCUSSION
Infrared Spectra Figure 1a shows the infrared spectra of bis(11-thioundecyl) phosphate in transmission mode IR and an ATR of the same molecule after it has been adsorbed to a Au surface from noctanol solution. Using a combination of theory and experiment, previous workers (33) have studied the IR and Raman spectra of a similar molecule, dimethylphosphate, (CH3 O)2 PO2 H). The IR spectrum of bis(11-thioundecyl)phosphate appears to be similar to DMP in most respects. The band of interest here for bis(11-thioundecyl)phosphate lies at 2548 cm−1 and is associated with the S–H stretching vibration. This peak disappears in the spectrum when the compound is adsorbed on Au, demonstrating the formation of a covalent thiol linkage to the surface. The spectra are also very similar to those previously published for 11-thioundecyl-1-phosphonic acid (26). Figure 1b shows the FTIR-ATR spectra of two iron oxide colloids, the first being the unmodified colloid and the second being precipitated at a pH of 6 in the presence of the orthophosphate species. The two spectra have similar peaks below 2000 cm−1 . However, the colloid precipitated in the presence of phosphate has two additional bands at 1009 and 1086 cm−1 associated with P–O stretching vibrations. Infrared spectra of phosphate ion in solution and adsorbed on similar iron oxides such as goethite have been studied extensively. The phosphate IR peak positions and intensities appear to depend strongly on such factors as pH and the method of preparation of the colloid (34, 35). Our results are in close agreement with these previous studies, although other workers have found up to five phosphate-related peaks in this spectral range. The new peaks have been variously attributed to both monodentate and bidentate complexes of phosphate ion adsorbed on the surface in a variety of ionization states, depending on pH. Tejedor-Tejedor and Anderson (35) assign peaks at
208
KRELLER ET AL.
literature IR assignments appear to remain somewhat controversial and are also for the crystalline form of iron oxyhydroxide, goethite, while our colloids are seen (below) to be more amorphous in nature. At this stage, the IR analysis does not allow a more precise assignment other than to demonstrate that there is phosphate ion adsorbed on the surface of the iron colloid. We must look to other methods to offer insight into the precise configuration. AFM Images and Force Titrations
FIG. 1. (a) FTIR spectra of bis(11-thioundecyl)phosphate: (i) ATR spectrum of compound adsorbed on an annealed Au film and (ii) transmission spectrum of the compound. Note the disappearance of the S–H stretching peak at 2450 cm−1 . (b) FTIR-ATR spectra of (i) iron colloid particles precipitated in the presence of orthophosphate ion and (ii) unmodified iron colloid particles.
1123, 1006, and 982 cm−1 to the protonated bidentate complex, ≡Fe2 HPO4 while the deprotonated bidentate complex (nomi−1 nally ≡Fe2 PO− 4 ) adsorbs at 1096 and 1044 cm . Persson and co-workers (34) dismiss the possibility of bidentate complexes being present at the surface and instead assign all their observed IR bands using a monodentate binding model. At a pH of 6, the conditions under which our colloids were prepared, they observed four distinct peaks in their spectra. They assign the 1001and 1178-cm−1 bands to a fully protonated phosphate species, i.e., ≡Fe–O–PO3 H2 , and the bands at 1049 and 1122 cm−1 to the deprotonated, i.e., ≡Fe–O–PO3 H− , species. On the whole, the IR peaks seen in the present study seem to be more consistent with the band assignments given for a combination of protonated and deprotonated bidentate complexes on the surface, but do not unambiguously fit either model. Many of the
Figure 2 shows a typical pair of AFM images of the iron oxyhydroxide particles postprecipitated with DMP and deposited on the mica slide. The images in Fig. 2 were acquired in contact mode in solution, working at a pH >10. The particle morphology appeared similar to that of the unmodified iron oxyhydroxide material as well as the material that had been postprecipitated with orthophosphate. Since it is known that, upon aging, such particles can take on a crystalline form (36), they were also examined by powder X-ray diffraction. This showed that the particles used here were completely amorphous. We did find, however, that the particles tended to cluster in one of two ways on the surface: either in a highly disordered arrangement as indicated in the top image or in regions where the particles appear to be layered and “plate-like” in appearance, as shown in the lower image. However, a higher resolution image of such a region shows that even in these more ordered arrangements, the particles appear to remain amorphous, consistent with the X-ray results. Higher resolution images showed little sign of structure. While uninformative about the details of the chemical termination on the surface, the images are useful in the sense that they serve as a diagnostic to ensure the presence of a reasonably even coating of particles on the surface at the location at which the chemical force titration experiments are to be made. Generally, force titrations were carried out in the more ordered regions of the sample to ensure that we were working in a reasonably flat region of the sample. Figures 3–5 show a series of chemical force titrations in which the adhesive interaction between tip and sample is plotted as a function of the pH of the solution. Data in Figures 3 and 5 were acquired under low ionic strength conditions: i.e., the only ions in solution were derived from the acid or base (HCl or NaOH) used to adjust the solution to its desired pH value. In Fig. 4, the data were obtained in pH-adjusted 1.0 mol L−1 NaCl solutions. Unbuffered solutions were used to avoid competitive adsorption or other possible complicating effects due to buffer ions. We also show, in the inset to Fig. 4, a histogram of the pull of forces observed during a typical experiment with the phosphate tip on an iron colloid sample. Since the dialkylphospate SAM has not been examined previously using chemical force methods, and we wish to probe its behavior with a set of less well-defined iron oxide colloids, it is important to first characterize the tip on well-controlled systems. In curve (i) of Fig. 3, we see the titration curve of the bis(11thioundecyl)phosphate coated tip on a substrate surface coated
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE
209
FIG. 2. AFM images of the iron colloid particles deposited on the mica substrate. Plate (a) shows a 10.1 × 10.1 µm image of a disordered region (see text) on the surface. Plate (b) shows the same scale and shows a region where the particles have formed a more orderly layered array on the surface.
with the same substance. This is in effect a phosphate vs phosphate, or PO2 H vs PO2 H, titration. A maximum is seen at a pH of approximately 5.0. Analogous to similar experiments previously reported with CO2 H-coated (11, 27) or PO(OH)2 -coated (26, 28) tip–sample pairs, this maximum must correspond to the pK a of phosphate groups of this dialkylphosphate surface. Such a pK a is similar, but slightly higher than the second pK a of 4.2 observed for monolayers of 11-thioundecyl-1-phosphonic acid (26). At a pH value corresponding to the pK a of the dialkylphosphate
under examination, we expect half the sites on the tip and sample to be ionized. Under these conditions, the maximum opportunity for H-bonding occurs between a neutral, i.e., unionized, P–OH group on one surface and an ionized P–O− group on the other. As the pH is increased above the pK a value, the majority of functional groups on both surfaces become deprotonated, and the adhesive interactions are reduced. At pH values greater than 8.5, essentially all of the P–OH groups would be deprotonated, resulting in a strong repulsive interaction.
210
KRELLER ET AL.
FIG. 3. Chemical force titrations performed under low ionic strength conditions (i) using a bis(11-thioundecyl)phosphate-terminated tip on a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate; (ii) using a 16thiohexadecanoic acid-terminated tip on a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate; and (iii) using a bis(11-thioundecyl) phosphate-terminated tip on an annealed Au surface.
The phosphate groups of the alkylthiol SAM were further characterized by chemical force titration of a PO2 H control surface on Au-coated mica against a CFM probe coated in 16thiohexadecanoic acid. This compound is also well known to self-assemble on Au to form a CO2 H-terminated surface. In curve (ii) of Fig. 3, the chemical force titration of a CO2 Hterminated tip against the PO2 H-terminated Au surface is shown. Similar results were obtained in the opposite configuration, i.e., with a phosphate-coated tip and carboxylate-coated sample. A large maximum, similar in magnitude to that seen for the PO2 H tip on the PO2 H surface in curve (i) of Fig. 3, can be observed, this time centered at a pH of 4. This result is consistent with the titration of a mixed-component system and may be understood with reference to the distribution diagram in Fig. 6. At a pH of 4, close to the pK a of the carboxylate tip, we would expect the tip surface sites to be terminated with equal fractions of COO− and CO2 H groups. The sample surface sites, meanwhile, will still be substantially protonated at this pH and terminated with PO2 H end groups. Under these conditions, H-bonding is maximized between the surfaces. At pH values below 4, the tip carboxylate groups also become protonated, and while H-bonding still occurs, it is less strong than when the negatively charged sites are available. Above pH 4 while the surface of the tip is substantially
ionized, the PO2 H-terminated surface has also become partially ionized, resulting in increased electrostatic repulsion between the two. Curve (iii) in Fig. 3 shows the result of force titration of a PO2 H-coated tip against an unmodified annealed Au surface. In this case, there is a relatively strong interaction over a wide pH range. In Fig. 3b, force curves for the phosphate tip/Au sample are shown for three different pH values. As can be seen, there is little change in the shape of the force curves with pH, other than the magnitude of the interaction. Such curve shapes were typical for these types of samples. Again, reversal of tip and sample surfaces yielded the same titration curve. The drop in adhesive forces at high and low pH values is readily explained by the adsorption of ions in the electrical double layer at the relatively high ionic strengths required to maintain the extremes of pH, leading to electrostatic repulsion between the like-charged tip and sample. The large adhesive interactions at the mid-pH range are less easy to understand, as there would be expected to be little in the way of specific phosphate–Au bonding that could give rise to the strong adhesive interaction. They may instead be due to a hydrophobic interaction between the gold and a disordered self-assembled monolayer. The infrared results (Fig. 1a) do not, however, show any evidence of disorder in the phosphate SAM, which would generally be indicated by a broadening of the C–H stretching peaks at 2900 cm−1 . Indeed, the infrared results are essentially identical to those previously described for the similar 11-thioundecyl-1-phosphonic acid (26). If the S could change its bonding site by migrating from the tip to the sample (and vice versa), the bidentate nature of the bis(11-thioundecyl)phosphate could allow the molecule to bridge between tip and sample, leading to a strong adhesive interaction. This would seem to be unlikely, however, since the Au–S bond is known to be very strong and would be expected to have a large barrier to exchange. Nor is it supported by the Force–distance curves, which show little change with pH, demonstrating that there are no measurable changes in configuration of the phosphate monolayer with pH. Another possibility is that the phosphate headgroup itself acts as a ligand toward the Au. However, phosphate ion is known to act as only a weak ligand toward Au (37). To examine the effect of ionic strength on the tip–sample interaction, several experiments were carried out in 1.0 mol L−1 NaCl solutions. Results from these experiments are shown in Fig. 4. As can be seen, a comparison of the force scale in Fig. 4 with those in Figs. 3 or 5 shows that the adhesive interaction is much reduced for the two systems studied under these conditions. This is expected, as under the high ionic strength conditions an electrical double layer will form as the headgroups are ionized, leading to a strong repulsive potential. This is also consistent with the reduction in the magnitude of the interaction between the phosphate tip and the unmodified Au surface seen in curve (ii) of Fig. 3, which otherwise has much the same shape as the analogous curve acquired under lower ionic strength conditions. The results for the titration of the PO2 H tip with a PO2 Hterminated surface under high ionic strength conditions are shown in Fig. 3(i). Here, the shape is much the same as in
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE
FIG. 4. Chemical force titrations performed under high ionic strength conditions (1.0 M NaCl) all using a bis(11-thioundecylphosphate)-terminated tip on (i) a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate and (ii) an annealed Au surface. The inset to Fig. 4 shows a typical histogram of the adhesion forces observed. This example is for the bis(11-thioundecylphosphate)terminated tip on the dimethylphosate-coated iron colloid at a pH of 3.7.
the case of low ionic strength, and the maximum is again seen to appear at a pH of about 4.5, although the maximum is less well-defined. This result is somewhat intermediate to those observed at low ionic strength and to the results seen at high ionic strength observed by Zhang et al. (26, 28) when titrating an 11thioundecylphosphite tip on an 11-thioundecylphosphite surface (i.e., diprotic PO(OH)2 -terminated tip and sample). In that case, the curve took the form of a series of sigmoidal step functions as the OH sites were ionized at two different pK a values. These values are similar to the pK a values for a free molecule in solution, but are somewhat lower, due to various effects such as intramolecular H-bonding on the surface (26). In Fig. 5, we show the results for the phosphate-terminated tip titrated against a series of iron oxyhydroxide colloids precipitated under a variety of conditions. Curve 5(i) is a repeat of curve 3(i), showing the PO2 H vs PO2 H titration for compari-
FIG. 5. Chemical force titrations performed under low ionic strength conditions, all using a bis(11-thioundecyl)phosphate-terminated tip on (i) a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate, (ii) iron hydroxide colloids deposited on mica, (iii) iron hydroxide colloids postprecipitated in the presence of orthophosphate ion, and (iv) iron hydroxide colloids postprecipitated in the presence of dimethylphosphate.
211
FIG. 6. Proposed distribution diagrams for the various surface species. Panels (i) and (ii) are for the ionization states of phosphate and carboxylate end groups of alkylthiol SAMs, respectively, while panel (iii) represents the ionization state distribution of the hydroxyl groups of a hydrous iron oxide surface.
son purposes with the oxyhydroxide titration curves. In curve 5(ii) we show the titration of the phosphate tip on the unmodified iron colloid. In this case, there is a broad peak, maximizing around a pH of 6, but with still significant (and roughly constant) adhesive interactions occurring at pH values below this. The data appear more scattered than those obtained in the standard samples, probably reflecting the inhomogeneous nature of the substrate surface. The broad and relatively large maximum in adhesive interaction, however, is quite striking when compared to the limited pH ranges in which adhesion occurs for the phosphate tip in conjunction with other surfaces and appears to be in agreement with the expected high specific affinity of phosphate for these iron colloids. Zeta potential measurements conducted on the iron colloid showed that the colloid has an isoelectric point at a pH of 5.5, near the pH at which tip–sample adhesive interaction maximizes. As indicated by the distribution diagram for iron oxide in Fig. 6, this is also the start of the pH range where the surface should be dominated by neutral Fe–OH groups. At high pH values, the adhesive interaction also remains significant up to a pH of about 12. Presumably at pH values higher than this, ionization of the majority of groups on the colloid surface and phosphate tip leads to a repulsive interaction between them. Titration curves obtained from modified colloid surfaces show a number of interesting features. Most striking is curve 5(iii) that shows the titration of the phosphate tip on an iron colloid postprecipitated in the presence of orthophosphate ion. Zeta potential measurements showed that the phosphate-coated iron oxyhydroxide colloids are more acidic than the uncoated colloids with the isoelectric pH shifting from 5.5 to 4.3. The force titration curve is virtually identical in appearance to that obtained for the phosphate on phosphate titration previously shown in Fig. 5(i) or Fig. 3(i). This evidently suggests that, from the viewpoint of the tip–sample interaction, an iron colloid postprecipitated with orthophosphate ion has very similar surface properties to that of the alkyl phosphate SAM on gold. The phosphate of the SAM is bound to the Au in a geometry that exposes mainly the monoprotic phosphate end group on the surface; this would suggest that
212
KRELLER ET AL.
the orthophosphate ion binds in such a way that a similar group is exposed on the surface of the colloid. The obvious way for this to occur is for the phosphate to bind in a bidentate fashion to the colloid surface, i.e., ≡Fe2 O2 PO2 H, as has been suggested by previous workers (35). In both cases a –PO2 H end group is exposed on the surface, leading to a similar interaction with the tip. The observation of attractive interactions between the PO2 H tip and the phosphate postprecipitated colloid also supports the possibility of the adsorption of phosphate in multiple layers on these minerals, as observed by Nooney et al. (38). The hypothesis of a bidentate mode of phosphate surface binding was tested by synthesizing an iron oxyhydroxide postprecipitated with dimethylphosphate, (CH3 O)2 PO2 H. In this case, DMP must bind to the iron colloid surface through its PO2 H functionality and leave two exposed methoxy groups on the surface with which the phosphate tip would have little interaction. The presence of DMP was found to have a negligible effect on the colloid zeta potential. As can be seen in the force titration profile in Fig. 5(iv), while the general form of the pH profile of the force interactions (maximum at pH 6) of the PO2 H tip and the DMP coated surface is the same as that with the unmodified colloid, the tip–sample attractive interactions are substantially reduced. The interpretation of this result is that surface-adsorbed DMP has the effect of consuming a fraction of the surface sites with which the phosphate groups of the tip can bind, but does not introduce any other additional mode of attractive interaction with the probe. The final modified colloid to be prepared was one in which the iron oxyhydroxide was postprecipitated with tannic acid. Zeta potential measurements on this colloid showed a large shift in isoelectric point, down to a pH of about 2.5. No adhesive wells were observed for the interaction of the phosphate-coated tip on these colloids over the entire pH range explored (ca. 3 to 12) (40). This result is consistent with the observation that the postprecipitated TA forms a coating on the surface that inhibits the adsorption of phosphate ions onto the iron oxyhydroxide colloid (5). SUMMARY
Phosphate-terminated surfaces were prepared by the selfassembly of bis(11-thioundecyl)phosphate monolayers on Aucoated mica and on Au-coated AFM tips. The surfaces were characterized using FTIR spectroscopy and chemical force microscopy. Force curves for the PO2 H-terminated tips were obtained as a function of pH, and the adhesive interaction for the PO2 H-terminated tip on the PO2 H-terminated Au surface demonstrated that the surface pK a of the phosphate monolayer was 5.0. The phosphate tip was further characterized by its interaction with both –CO2 H terminated and bare Au surfaces. The tips were then used to characterize a series of iron oxyhydroxide colloids using force titration measurements. These colloids showed the expected strong affinity for phosphate ion, as demonstrated by the large adhesive tip–colloid interaction
observed. Colloids modified by postprecipitation with phosphate ion gave titration curves that strongly resembled those obtained for a phosphate-coated tip on a phosphate-coated sample, suggesting that orthophosphate ion binds to the colloid surface in a bidentate fashion. Further experiments on colloids postprecipitated with dimethylphosphate or tannic acid demonstrated that these species block the interaction of phosphate with iron surface sites, in the case of tannic acid, quite dramatically. These experiments demonstrate the potential of this method for characterizing and distinguishing between the surfaces of this important class of compounds. ACKNOWLEDGMENTS We acknowledge the Natural Sciences and Engineering Research Council of Canada for financial support. The Ravensview Water Pollution Control Plant, Kingston, Ontario, is acknowledged for the donation of a sample of ferric chloride solution.
REFERENCES 1. Dolfin, J., Chardon, E. J., and Japenga, J., Soil Sci. 164, 171 (1999). 2. Liu, C., and Huang, P. M., Can. J. Soil Sci. 80, 445 (2000). 3. Hawke, D., Carpenter, P. D., and Hunter, K. A., Environ. Sci. Technol. 23, 187 (1989). 4. Lopez-Hernandez, D., Siegert, G., and Rodriguez, J. V., Soil Sci. Soc. Am. J. 50, 1460 (1986). 5. Omoike, A. J., and vanLoon, G. W., Wat. Res. 33, 3617 (1999). 6. Eggleston, C. M., Higgins, S. R., and Maurice, P. A., Environ. Sci. Technol. 32, 456 (1998). 7. Weisenhorn, A. L., Maivald, P., Butt, H. J., and Hansma, P. K., Phys. Rev. B 45, 11226 (1992). 8. Burnham, N. A., Colton, R. J., and Pollock, H. M., J. Vac. Sci. Technol. A 9, 2548 (1991). 9. Butt, H. J., Biophys. J. 60, 1438 (1991). 10. Burnham, N. A., Dominguez, D. D., Mowery R. L., and Colton, R. J., Phys. Rev. Lett. 64, 1931 (1990). 11. van der Vegte E. W., and Hadziioannou, G., Langmuir 13, 4357 (1997). 12. Han, T., Williams, J. M., and Beebe, T. P., Jr., Anal. Chim. Acta 307, 365 (1995). 13. Vezenov, D. V., Noy, A., Rozsnyai, L. F., and Lieber, C. M., J. Am. Chem. Soc. 119, 2006 (1997). 14. Ito, T., Namba, M., Buhlmann, P., and Umezawa, Y., Langmuir 13, 4323 (1997). 15. Ulman, A., “An Introduction to Ultrathin Organic Films.” Academic Press, Boston, 1991. 16. McKendry, R., Theoclitou, M. E., Rayment, T., and Abell, C., Nature 391, 566 (1998). 17. Lioubashevski, O., Patolsky, F., and Willner, I., Langmuir 17, 5134 (2001). 18. Hansma, H. G., Ann. Rev. Phys. Chem. 52, 71 (2001). 19. Bowen, W. R., Hilal, N., Lovitt, R. W., and Wright, C. J., Colloids Surf. A 157, 117 (1999). 20. Kim, K., Kim, C., and Byun, Y., J. Biomed. Mater. Res. 52, 836 (2000). 21. Tang, Z., Jing, W., and Wang, E., Langmuir 16, 1696 (2000). 22. Dufrene, Y. F., Boland, T., Schneider, J. W., Barger, W. R., and Lee, G. U., Faraday Discuss. 111, 79 (1998). 23. Yang, X. M., Xiao, D., Xiao, S. J., and Wei, Y., Appl. Phys. A 59, 139 (1994). 24. Nakashima, N., and Taguchi, T., Colloid Surf. A 103, 159 (1995). 25. Nakashima, N., and Taguchi, T., in “Interfacial Design and Chemical Sensing” (T. E. Mallouk and D. J. Harrison, Eds.), pp. 145–154. ACS Symp. Ser. 561, Am. Chem. Soc., Washington, DC, 1994.
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE 26. Zhang, J., Kirkham, J., Robinson, C., Wallwork, M. L., Smith, D. A., Marsh, A., and Wong, M., Anal. Chem. 72, 1973 (2000). 27. van der Vegte, E. W., and Hadziioannou, G. J., Phys. Chem. B 101, 9563 (1997). 28. Smith, D. A., Wallwork, M. L., Zhang, J., Kirkham, J., Robinson, C., Marsh, A., and Wong, M., J. Phys. Chem. B 104, 8862 (2000). 29. Omoike, A., and Horton, J. H., Langmuir 16, 1655 (2000). 30. Omoike, A., Chen, G., vanLoon, G. W., and Horton, J. H., Langmuir 14, 4731 (1998). 31. Hutter, J. L., and Bechhoefer, J., Rev. Sci. Instrum. 64, 1868 (1993). 32. Hutter, J. L., and Bechhoefer, J., Rev. Sci. Instrum. 64, 3342 (1993). 33. Guan, Y., Choy, G. S.-C., Glaser, R., and Thomas, G. J., Jr., J. Phys. Chem. 99, 12054 (1995).
213
34. Persson, P., Nilsson, N., and Sj¨oberg, S., J. Colloid Interface Sci. 177, 263 (1996). 35. Tejedor-Tejedor, M. I., and Anderson, M. A., Langmuir 6, 602 (1990). 36. Duffy, S. J., and vanLoon, G. W., Environ. Sci. Technol. 28, 1950 (1994). 37. Paunovic, M., and Sambucetti, C., Proc. Electrochem. Soc. 31, 34 (1994). 38. Nooney, M. G., Murrell, T. S., Corneille, J. S., Rusert, E. I., Hossner, L. R., and Goodman, D. W., J. Vac. Sci. Technol. A 14, 1357 (1996). 39. Cornell, R. M., and Schwertmann, U., “The Iron Oxides: Structure, Properties, Occurences and Uses.” VCH, Weinheim, 1996. 40. Kreller, D. I., Gibson, G., Novak, W., vanLoon, G. W., and Horton, J. H., submitted to Colloid Surf. A.
Chemical Force Microscopy Investigation of Phosphate Adsorption on the Surfaces of Iron(III) Oxyhydroxide Particles D. I. Kreller, G. Gibson, G. W. vanLoon, and J. H. Horton1 Department of Chemistry, Queen’s University, Kingston, Ontario, Canada K7L 3N6 Received January 25, 2002; accepted July 9, 2002
Phosphate-modified AFM tips were prepared by the deposition of self-assembled monolayers (SAMs) of bis(11-thioundecyl) phosphate on Au-coated silicon nitride cantilevers. The properties of the PO2 H-terminated SAMs were investigated by studying the pH-dependent force interactions of the tips with phosphateand carboxylic acid-terminated SAM control surfaces. The PO2 H functional groups had a pKa of ≈5.0. A chemical force microscopy (CFM) study was conducted on the interactions between the probes and the surfaces of hydrous ferric oxide particles prepared in our laboratory by hydrolytic precipitation from FeCl3 . The forces between PO2 H probes and the hydrous ferric oxide surfaces were seen to exhibit a strong pH dependence, with maximum attractive forces occurring for pH values between 5 and 8. The effects of postprecipitation of the hydrous ferric oxide colloids with orthophosphate, H2 PO− 4 , dimethylphosphate, (CH3 O)2 PO2 H (DMP), and tannic acid (TA) on the adhesive interactions between the PO2 H tips and the solids were also investigated. Attenuated total reflectance infrared spectroscopy (ATR-IR) was used to verify the presence of surfaceadsorbed species and zeta potentiometric measurements to determine surface charge on the colloids. We show that the method of chemical force titration using phosphate-terminated tips can differentiate between these various colloids and that it shows promise as a general method for studying this environmentally important class of compounds. C 2002 Elsevier Science (USA) Key Words: self-assembled monolayers; phosphate; iron(III) oxyhydroxide; colloid; atomic force microscopy; chemical force microscopy.
INTRODUCTION
The behavior of phosphate and natural organic matter (NOM) in soils and aqueous media is widely known to be strongly influenced by their association with surfaces of colloidal mineral oxide and oxyhydroxide particles of iron and aluminum (1, 2). Strong retention of phosphate and NOM on such mineral surfaces controls the mobility, reactions, and bioavailability of these nutrients. Competitive adsorption occurs between phosphate and organic species at mineral surfaces. For example, phosphate does not adsorb as effectively to a mineral surface to which an organic 1 To whom correspondence should be addressed. Fax: 613-533-6669. E-mail: [email protected].
205
material is already complexed as it does to an otherwise uncoated mineral surface. Inhibitory binding relationships among such compounds have been studied in the context of soils and water research (3, 4). Orthophosphate species (PO3− and protonated versions 4 therof) are present in low concentrations in wastewater, but its release in wastewater effluent can be of environmental significance because it can be a contributing factor to eutrophication of the receiving water body. To reduce the levels of phosphorus in wastewater effluents, the high affinity of phosphate for Fe and Al based oxyhydroxides is used to advantage in the process of chemical coagulation. Salts such as FeCl3 or Al2 (SO4 )3 are added to the wastewater stream and the M(III) ions undergo alkaline hydrolysis to generate colloidal oxyhydroxide “floc” particles. Coprecipitation of phosphate or its adsorption onto the reactive and initially very high surface area precipitates leads to incorporation into insoluble solids and its removal from solution. The competition of organic materials with phosphate ions for binding sites on the surfaces of flocculant particles can have important implications for the design of wastewater treatment plants. For example, phosphate removal may be enhanced when at least a portion of the coagulant is added at the outlet of the aerator, where the concentration of organic matter is reduced (5). The combined techniques of atomic force microscopy (AFM) and chemical force microscopy (CFM) enable topographical imaging and the investigation of chemical interactions at the surfaces of nonconducting substrates. This feature makes them, in principle, applicable to a broad range of environmentally relevant surface interactions (6). These techniques can be used to augment information derived via other chemical methods, i.e., solution adsorption/desorption, and surface analysis techniques such as infrared spectroscopy. Atomic force microscopy is used primarily as an imaging method. Chemical force microscopy has been introduced as a method for probing the various adhesive and repulsive interactions at surfaces (7–10). These include van der Waal’s, hydrogen-bonding, and electrostatic charge interactions. CFM allows both the imaging of surfaces with chemical specificity (11) and the quantitative measurement of sitespecific interactions between the tip and sample (12), in principle with nanometer-scale resolution. Quantitative measurements of Hamaker constants for tip/sample systems in various solvents 0021-9797/02 $35.00
C 2002 Elsevier Science (USA)
All rights reserved.
206
KRELLER ET AL.
can be made. Also, as will be explored further in this work, CFM allows the method of force titration in which the tip–sample interaction is measured as a function of some property, in this case pH, of an aqueous solution (13). Force pH titrations allow the determination of ionization constants, i.e., pK a , of functional groups on the interacting surfaces and the pH dependence of adhesive attractions between them. While early workers examined the interaction of unmodified silicon nitride tips with surfaces, more recent work has concentrated on probe tips which have been modified to be terminated with various end groups, including alkylsilanes (14), alkylthiols (15), chiral molecules (16), DNA and proteins (17, 18), and silica particles (19). In the present work, we are interested in the process of phosphate adsorption on colloidal particles; hence, a tip functionalized with a phosphate end group is required. The preparation of alkylthiol-based phosphate self-assembled monolayers and phosphate CFM has attracted surprisingly little attention given that, in addition to the importance of phosphate in environmental systems, phospholipid-type molecules are also of interest in biological systems, including cell membranes and DNA (Hansma). The synthesis and AFM characterization of phospholipid bilayers deposited using Langmuir– Blodgett and related techniques (20–23) have been more thoroughly explored. Preparation of a phosphate surface via deposition of a phosphate-terminated alkylthiol compound on Au is attractive because the self-assembly of ordered arrays of alkylthiols on Au is a well-established general procedure, and these monolayer systems have been quite well characterized. The synthesis and self-assembly on Au of the bis(11-thioundecyl)phosphate compound used here has been reported (24, 25) and the interaction of this phospholipid monolayer with methylviologen and Ca2+ ions has been explored using electrochemical methods. More recently, CFM studies of phosphate monolayers self-assembled from 11-thioundecyl-1-phosphonic acid (i.e., the monoalkylsubstituted version of the phosphate explored in this paper) have been carried out (26–28). We have previously reported (29, 30) on a study of hydrous aluminum oxides, a class of colloids closely related to the iron oxyhydroxides reported on here. The hydrous aluminum oxide particles were prepared by hydrolytic precipitation from alkaline solutions of alum, (Al2 (SO4 )3 · nH2 O) in the presence of phosphates and tannic acid. This work was carried out using a combination of tapping mode atomic force microscopy, chemical force microscopy, and interfacial force microscopy to determine both the morphology of these particles and to elucidate the chemical identity of their surface coating. The force microscopy data in this previous work were acquired, however, using unmodified silicon nitride tips. In the present paper, we report on the preparation of phosphate-terminated AFM probes and their application in the study of phosphate adsorption to hydrous iron oxide colloids. The compound bis(11-thioundecyl)phosphate, (HSC11 H22 O)2 PO2 H, was synthesized and subsequently deposited by self-assembly of an alkylthiol monolayer on a gold
film to form a phosphate-terminated surface. The force interactions of PO2 H-terminated AFM probes were studied as a function of pH, first with a series of control test surfaces and subsequently with a series of Fe3+ -based hydrous oxide colloids including “bare” Fe oxyhydroxide colloids as well as particles postprecipitated with orthophosphate (PO3− 4 ), dimethylphosphate (CH3 O)2 PO2 H (DMP), and tannic acid (TA, a surrogate for natural organic matter). Attenuated total reflectance infrared spectroscopy (ATR-IR) was used to verify the presence of surface-adsorbed species via their characteristic stretching frequencies. While chemical force measurements using phosphate-coated tips have been previously reported, notably the work of van der Vegte and Hadziioannou (11, 27) and Smith and co-workers (26, 28), these are the first results involving tips terminated with the dialkylphosphate (e.g., monoprotic) compound, as opposed to the diprotic tips used in the above work. More importantly, we apply the chemical force methods to a real world system, namely, metal oxide colloidal particles, as opposed to the model systems of self-assembled monolayers on Au explored in the previous work. We show that the method of chemical force titration using phosphate-terminated tips can readily distinguish between the surfaces of these various colloids. This technique shows promise as a general method for studying this important class of compounds. MATERIALS AND METHODS
Synthetic Methods The synthesis of bis(11-thioundecyl)phosphate was carried out using the method of Nakashima and Taguchi (24, 25). The various hydrous iron oxide colloids substrates used in the force titration experiments were synthesized at room temperature (293 K) using room temperature reagents. It is important to note that only a fraction of the product was ultimately used for the preparation of the AFM samples. Also, large excesses of the compounds which were used to variously coat the surfaces of particles (i.e., tannic acid, orthophosphate, dimethylphosphate) were introduced into the solutions to ensure complete surface coverage of the particles; hence, significant quantities of unadsorbed compound generally remained in solution. (i) Native hydrous iron oxides: 50 µL of 41% w/v ferric chloride solution (Fanchem Ltd. Oakville, Ontario) was added by micropipette to 1 L of distilled water to form a 7 ppm Fe solution. The pH was adjusted to a value of 6 using a 0.50 M NaHCO3 solution. For uncoated hydrous oxide colloids, the precipitate was allowed to age for 20 min with 300 rpm stirring, and then a portion was collected on a freshly cleaved mica slide (Grade V-5, SPI Supplies). To do this, a slide was placed on a 0.45-µm filter paper (Osmonics) in a filter funnel, and the hydrolysis solution was transferred into it under suction. As the filtrate was removed, the solids settled and a portion was collected on the mica slide. The deposited solids were allowed to dry at room temperature for 24–48 h prior to analysis. (ii) Hydrous iron hydroxides postprecipitated with orthophosphate: to the solution containing 20-min-aged
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE
precipitate as in part (i), phosphate as HPO2− 4 (Fisher Scientific) from a stock solution was added in a molar ratio of P : Fe = 7 : 1. The solution was allowed to age a further 15 min and the solids were then similarly collected for analysis. (iii) Hydrous iron hydroxides postprecipitated with dimethylphosphate: to the solution containing 20-min-aged precipitate as in part (i), DMP (Across) was added in a molar ratio of DMP : Fe = 7 : 1. The solution was aged a further 15 min and then collected as above. (iv) Hydrous iron hydroxides postprecipitated with tannic acid: to the solution containing 20-min-aged precipitate as in part (i), tannic acid (TA, Aldrich) was added in a molar ratio of TA : Fe = 1.4 : 1, the solution was aged for a further 10 min, and then the solids were collected as above. The TA-coated colloid displayed low adhesion on the mica substrate, presumably due to the presence of the organic overlayer on the precipitate. This required the use of double-stick tape applied to the mica before colloid deposition. Instrumentation Infrared data were acquired for bis(11-thioundecyl)phosphate in transmission mode using a KBr disk. Infrared data for the colloids and bis(11-bisthioundecyl)phosphate-modified Au surface were acquired in attenuated total reflectance (ATR) mode using a ZnSe crystal. AFM data were acquired using a PicoSPM (Molecular Imaging, Tempe, Arizona), using a Nanoscope IIE controller (Digital Instruments, Santa Barbara, CA). Images were acquired under ambient conditions as well as in solutions of various pH. Force– distance curves were obtained using the same apparatus. For the acquisition of force curves, the CFM tips and samples were immersed in freshly prepared unbuffered aqueous NaOH or HCl solutions of pH ranging from 3 to 10. The pH of the solutions was checked before and after the experimental run to ensure that the pH had not changed. Some 300–500 force curves were obtained for each sample. The reported values of the adhesive interaction are an average of all the curves while the reported errors reflect the standard deviation of the data. The AFM tips were standard silicon nitride cantilevers (Digital Instruments, Santa Barbara, CA) coated with 25 nm of Au. The force constants of the cantilevers ranged from 0.2 to 1.0 Nm−1 as determined by the method of Hutter and Bechhoefer (31, 32). Experimental runs for a pH series were performed using a single tip on a given sample. In addition, several checks were made to ensure the reproducibility of the data. In cases where samples consisted of self-assembled monolayers or bare gold, the tip and sample end groups were interchanged, yielding similar results. Repeating the colloid CFM experiments yielded curves that were identical in profile as a function of pH, in which the adhesive force varied in magnitude by about 20%. This variation can be attributed to changes in tip radius, which will affect the tip–sample contact area and hence the number of tip–sample bonds formed and the measured adhesive force. Generally, it is believed that between 40 and 100 bonds are formed between tip and sample, depending on tip radius (13). As we note in the discussion, it
207
is the profile of the curves as a function of pH, which is the most important factor that we take into account in the analysis of our data. In a comparison of the results on different tip/sample pairs, it is only in cases where the change in overall magnitude is greater than 50% that we attribute any significance to the data. In addition, the runs could be repeated from either high to low or low to high pH values without markedly affecting the results. The Au-coated tips were functionalized with PO2 H SAMs by immersing them for 48 h in a 1.0 × 10−3 mol L−1 solution of bis(11-thioundecyl)phosphate in n-octanol. Substrate SAM surfaces of bis(11-thioundecyl)phosphate were similarly prepared by the immersion of mica substrates (previously coated with 25 nm of Au and subsequently hydrogen-flame annealed) in the same thiol solution. Tips coated with CO2 H end groups were prepared by immersing them in a 1 mM solution of 16thiohexadecanoic acid (Aldrich, 95%) in methanol for 24 h. These deposition conditions were chosen to ensure that both tips and samples were saturated with a complete monolayer of alkylthiol. RESULTS AND DISCUSSION
Infrared Spectra Figure 1a shows the infrared spectra of bis(11-thioundecyl) phosphate in transmission mode IR and an ATR of the same molecule after it has been adsorbed to a Au surface from noctanol solution. Using a combination of theory and experiment, previous workers (33) have studied the IR and Raman spectra of a similar molecule, dimethylphosphate, (CH3 O)2 PO2 H). The IR spectrum of bis(11-thioundecyl)phosphate appears to be similar to DMP in most respects. The band of interest here for bis(11-thioundecyl)phosphate lies at 2548 cm−1 and is associated with the S–H stretching vibration. This peak disappears in the spectrum when the compound is adsorbed on Au, demonstrating the formation of a covalent thiol linkage to the surface. The spectra are also very similar to those previously published for 11-thioundecyl-1-phosphonic acid (26). Figure 1b shows the FTIR-ATR spectra of two iron oxide colloids, the first being the unmodified colloid and the second being precipitated at a pH of 6 in the presence of the orthophosphate species. The two spectra have similar peaks below 2000 cm−1 . However, the colloid precipitated in the presence of phosphate has two additional bands at 1009 and 1086 cm−1 associated with P–O stretching vibrations. Infrared spectra of phosphate ion in solution and adsorbed on similar iron oxides such as goethite have been studied extensively. The phosphate IR peak positions and intensities appear to depend strongly on such factors as pH and the method of preparation of the colloid (34, 35). Our results are in close agreement with these previous studies, although other workers have found up to five phosphate-related peaks in this spectral range. The new peaks have been variously attributed to both monodentate and bidentate complexes of phosphate ion adsorbed on the surface in a variety of ionization states, depending on pH. Tejedor-Tejedor and Anderson (35) assign peaks at
208
KRELLER ET AL.
literature IR assignments appear to remain somewhat controversial and are also for the crystalline form of iron oxyhydroxide, goethite, while our colloids are seen (below) to be more amorphous in nature. At this stage, the IR analysis does not allow a more precise assignment other than to demonstrate that there is phosphate ion adsorbed on the surface of the iron colloid. We must look to other methods to offer insight into the precise configuration. AFM Images and Force Titrations
FIG. 1. (a) FTIR spectra of bis(11-thioundecyl)phosphate: (i) ATR spectrum of compound adsorbed on an annealed Au film and (ii) transmission spectrum of the compound. Note the disappearance of the S–H stretching peak at 2450 cm−1 . (b) FTIR-ATR spectra of (i) iron colloid particles precipitated in the presence of orthophosphate ion and (ii) unmodified iron colloid particles.
1123, 1006, and 982 cm−1 to the protonated bidentate complex, ≡Fe2 HPO4 while the deprotonated bidentate complex (nomi−1 nally ≡Fe2 PO− 4 ) adsorbs at 1096 and 1044 cm . Persson and co-workers (34) dismiss the possibility of bidentate complexes being present at the surface and instead assign all their observed IR bands using a monodentate binding model. At a pH of 6, the conditions under which our colloids were prepared, they observed four distinct peaks in their spectra. They assign the 1001and 1178-cm−1 bands to a fully protonated phosphate species, i.e., ≡Fe–O–PO3 H2 , and the bands at 1049 and 1122 cm−1 to the deprotonated, i.e., ≡Fe–O–PO3 H− , species. On the whole, the IR peaks seen in the present study seem to be more consistent with the band assignments given for a combination of protonated and deprotonated bidentate complexes on the surface, but do not unambiguously fit either model. Many of the
Figure 2 shows a typical pair of AFM images of the iron oxyhydroxide particles postprecipitated with DMP and deposited on the mica slide. The images in Fig. 2 were acquired in contact mode in solution, working at a pH >10. The particle morphology appeared similar to that of the unmodified iron oxyhydroxide material as well as the material that had been postprecipitated with orthophosphate. Since it is known that, upon aging, such particles can take on a crystalline form (36), they were also examined by powder X-ray diffraction. This showed that the particles used here were completely amorphous. We did find, however, that the particles tended to cluster in one of two ways on the surface: either in a highly disordered arrangement as indicated in the top image or in regions where the particles appear to be layered and “plate-like” in appearance, as shown in the lower image. However, a higher resolution image of such a region shows that even in these more ordered arrangements, the particles appear to remain amorphous, consistent with the X-ray results. Higher resolution images showed little sign of structure. While uninformative about the details of the chemical termination on the surface, the images are useful in the sense that they serve as a diagnostic to ensure the presence of a reasonably even coating of particles on the surface at the location at which the chemical force titration experiments are to be made. Generally, force titrations were carried out in the more ordered regions of the sample to ensure that we were working in a reasonably flat region of the sample. Figures 3–5 show a series of chemical force titrations in which the adhesive interaction between tip and sample is plotted as a function of the pH of the solution. Data in Figures 3 and 5 were acquired under low ionic strength conditions: i.e., the only ions in solution were derived from the acid or base (HCl or NaOH) used to adjust the solution to its desired pH value. In Fig. 4, the data were obtained in pH-adjusted 1.0 mol L−1 NaCl solutions. Unbuffered solutions were used to avoid competitive adsorption or other possible complicating effects due to buffer ions. We also show, in the inset to Fig. 4, a histogram of the pull of forces observed during a typical experiment with the phosphate tip on an iron colloid sample. Since the dialkylphospate SAM has not been examined previously using chemical force methods, and we wish to probe its behavior with a set of less well-defined iron oxide colloids, it is important to first characterize the tip on well-controlled systems. In curve (i) of Fig. 3, we see the titration curve of the bis(11thioundecyl)phosphate coated tip on a substrate surface coated
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE
209
FIG. 2. AFM images of the iron colloid particles deposited on the mica substrate. Plate (a) shows a 10.1 × 10.1 µm image of a disordered region (see text) on the surface. Plate (b) shows the same scale and shows a region where the particles have formed a more orderly layered array on the surface.
with the same substance. This is in effect a phosphate vs phosphate, or PO2 H vs PO2 H, titration. A maximum is seen at a pH of approximately 5.0. Analogous to similar experiments previously reported with CO2 H-coated (11, 27) or PO(OH)2 -coated (26, 28) tip–sample pairs, this maximum must correspond to the pK a of phosphate groups of this dialkylphosphate surface. Such a pK a is similar, but slightly higher than the second pK a of 4.2 observed for monolayers of 11-thioundecyl-1-phosphonic acid (26). At a pH value corresponding to the pK a of the dialkylphosphate
under examination, we expect half the sites on the tip and sample to be ionized. Under these conditions, the maximum opportunity for H-bonding occurs between a neutral, i.e., unionized, P–OH group on one surface and an ionized P–O− group on the other. As the pH is increased above the pK a value, the majority of functional groups on both surfaces become deprotonated, and the adhesive interactions are reduced. At pH values greater than 8.5, essentially all of the P–OH groups would be deprotonated, resulting in a strong repulsive interaction.
210
KRELLER ET AL.
FIG. 3. Chemical force titrations performed under low ionic strength conditions (i) using a bis(11-thioundecyl)phosphate-terminated tip on a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate; (ii) using a 16thiohexadecanoic acid-terminated tip on a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate; and (iii) using a bis(11-thioundecyl) phosphate-terminated tip on an annealed Au surface.
The phosphate groups of the alkylthiol SAM were further characterized by chemical force titration of a PO2 H control surface on Au-coated mica against a CFM probe coated in 16thiohexadecanoic acid. This compound is also well known to self-assemble on Au to form a CO2 H-terminated surface. In curve (ii) of Fig. 3, the chemical force titration of a CO2 Hterminated tip against the PO2 H-terminated Au surface is shown. Similar results were obtained in the opposite configuration, i.e., with a phosphate-coated tip and carboxylate-coated sample. A large maximum, similar in magnitude to that seen for the PO2 H tip on the PO2 H surface in curve (i) of Fig. 3, can be observed, this time centered at a pH of 4. This result is consistent with the titration of a mixed-component system and may be understood with reference to the distribution diagram in Fig. 6. At a pH of 4, close to the pK a of the carboxylate tip, we would expect the tip surface sites to be terminated with equal fractions of COO− and CO2 H groups. The sample surface sites, meanwhile, will still be substantially protonated at this pH and terminated with PO2 H end groups. Under these conditions, H-bonding is maximized between the surfaces. At pH values below 4, the tip carboxylate groups also become protonated, and while H-bonding still occurs, it is less strong than when the negatively charged sites are available. Above pH 4 while the surface of the tip is substantially
ionized, the PO2 H-terminated surface has also become partially ionized, resulting in increased electrostatic repulsion between the two. Curve (iii) in Fig. 3 shows the result of force titration of a PO2 H-coated tip against an unmodified annealed Au surface. In this case, there is a relatively strong interaction over a wide pH range. In Fig. 3b, force curves for the phosphate tip/Au sample are shown for three different pH values. As can be seen, there is little change in the shape of the force curves with pH, other than the magnitude of the interaction. Such curve shapes were typical for these types of samples. Again, reversal of tip and sample surfaces yielded the same titration curve. The drop in adhesive forces at high and low pH values is readily explained by the adsorption of ions in the electrical double layer at the relatively high ionic strengths required to maintain the extremes of pH, leading to electrostatic repulsion between the like-charged tip and sample. The large adhesive interactions at the mid-pH range are less easy to understand, as there would be expected to be little in the way of specific phosphate–Au bonding that could give rise to the strong adhesive interaction. They may instead be due to a hydrophobic interaction between the gold and a disordered self-assembled monolayer. The infrared results (Fig. 1a) do not, however, show any evidence of disorder in the phosphate SAM, which would generally be indicated by a broadening of the C–H stretching peaks at 2900 cm−1 . Indeed, the infrared results are essentially identical to those previously described for the similar 11-thioundecyl-1-phosphonic acid (26). If the S could change its bonding site by migrating from the tip to the sample (and vice versa), the bidentate nature of the bis(11-thioundecyl)phosphate could allow the molecule to bridge between tip and sample, leading to a strong adhesive interaction. This would seem to be unlikely, however, since the Au–S bond is known to be very strong and would be expected to have a large barrier to exchange. Nor is it supported by the Force–distance curves, which show little change with pH, demonstrating that there are no measurable changes in configuration of the phosphate monolayer with pH. Another possibility is that the phosphate headgroup itself acts as a ligand toward the Au. However, phosphate ion is known to act as only a weak ligand toward Au (37). To examine the effect of ionic strength on the tip–sample interaction, several experiments were carried out in 1.0 mol L−1 NaCl solutions. Results from these experiments are shown in Fig. 4. As can be seen, a comparison of the force scale in Fig. 4 with those in Figs. 3 or 5 shows that the adhesive interaction is much reduced for the two systems studied under these conditions. This is expected, as under the high ionic strength conditions an electrical double layer will form as the headgroups are ionized, leading to a strong repulsive potential. This is also consistent with the reduction in the magnitude of the interaction between the phosphate tip and the unmodified Au surface seen in curve (ii) of Fig. 3, which otherwise has much the same shape as the analogous curve acquired under lower ionic strength conditions. The results for the titration of the PO2 H tip with a PO2 Hterminated surface under high ionic strength conditions are shown in Fig. 3(i). Here, the shape is much the same as in
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE
FIG. 4. Chemical force titrations performed under high ionic strength conditions (1.0 M NaCl) all using a bis(11-thioundecylphosphate)-terminated tip on (i) a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate and (ii) an annealed Au surface. The inset to Fig. 4 shows a typical histogram of the adhesion forces observed. This example is for the bis(11-thioundecylphosphate)terminated tip on the dimethylphosate-coated iron colloid at a pH of 3.7.
the case of low ionic strength, and the maximum is again seen to appear at a pH of about 4.5, although the maximum is less well-defined. This result is somewhat intermediate to those observed at low ionic strength and to the results seen at high ionic strength observed by Zhang et al. (26, 28) when titrating an 11thioundecylphosphite tip on an 11-thioundecylphosphite surface (i.e., diprotic PO(OH)2 -terminated tip and sample). In that case, the curve took the form of a series of sigmoidal step functions as the OH sites were ionized at two different pK a values. These values are similar to the pK a values for a free molecule in solution, but are somewhat lower, due to various effects such as intramolecular H-bonding on the surface (26). In Fig. 5, we show the results for the phosphate-terminated tip titrated against a series of iron oxyhydroxide colloids precipitated under a variety of conditions. Curve 5(i) is a repeat of curve 3(i), showing the PO2 H vs PO2 H titration for compari-
FIG. 5. Chemical force titrations performed under low ionic strength conditions, all using a bis(11-thioundecyl)phosphate-terminated tip on (i) a Au-coated mica surface terminated with bis(11-thioundecyl)phosphate, (ii) iron hydroxide colloids deposited on mica, (iii) iron hydroxide colloids postprecipitated in the presence of orthophosphate ion, and (iv) iron hydroxide colloids postprecipitated in the presence of dimethylphosphate.
211
FIG. 6. Proposed distribution diagrams for the various surface species. Panels (i) and (ii) are for the ionization states of phosphate and carboxylate end groups of alkylthiol SAMs, respectively, while panel (iii) represents the ionization state distribution of the hydroxyl groups of a hydrous iron oxide surface.
son purposes with the oxyhydroxide titration curves. In curve 5(ii) we show the titration of the phosphate tip on the unmodified iron colloid. In this case, there is a broad peak, maximizing around a pH of 6, but with still significant (and roughly constant) adhesive interactions occurring at pH values below this. The data appear more scattered than those obtained in the standard samples, probably reflecting the inhomogeneous nature of the substrate surface. The broad and relatively large maximum in adhesive interaction, however, is quite striking when compared to the limited pH ranges in which adhesion occurs for the phosphate tip in conjunction with other surfaces and appears to be in agreement with the expected high specific affinity of phosphate for these iron colloids. Zeta potential measurements conducted on the iron colloid showed that the colloid has an isoelectric point at a pH of 5.5, near the pH at which tip–sample adhesive interaction maximizes. As indicated by the distribution diagram for iron oxide in Fig. 6, this is also the start of the pH range where the surface should be dominated by neutral Fe–OH groups. At high pH values, the adhesive interaction also remains significant up to a pH of about 12. Presumably at pH values higher than this, ionization of the majority of groups on the colloid surface and phosphate tip leads to a repulsive interaction between them. Titration curves obtained from modified colloid surfaces show a number of interesting features. Most striking is curve 5(iii) that shows the titration of the phosphate tip on an iron colloid postprecipitated in the presence of orthophosphate ion. Zeta potential measurements showed that the phosphate-coated iron oxyhydroxide colloids are more acidic than the uncoated colloids with the isoelectric pH shifting from 5.5 to 4.3. The force titration curve is virtually identical in appearance to that obtained for the phosphate on phosphate titration previously shown in Fig. 5(i) or Fig. 3(i). This evidently suggests that, from the viewpoint of the tip–sample interaction, an iron colloid postprecipitated with orthophosphate ion has very similar surface properties to that of the alkyl phosphate SAM on gold. The phosphate of the SAM is bound to the Au in a geometry that exposes mainly the monoprotic phosphate end group on the surface; this would suggest that
212
KRELLER ET AL.
the orthophosphate ion binds in such a way that a similar group is exposed on the surface of the colloid. The obvious way for this to occur is for the phosphate to bind in a bidentate fashion to the colloid surface, i.e., ≡Fe2 O2 PO2 H, as has been suggested by previous workers (35). In both cases a –PO2 H end group is exposed on the surface, leading to a similar interaction with the tip. The observation of attractive interactions between the PO2 H tip and the phosphate postprecipitated colloid also supports the possibility of the adsorption of phosphate in multiple layers on these minerals, as observed by Nooney et al. (38). The hypothesis of a bidentate mode of phosphate surface binding was tested by synthesizing an iron oxyhydroxide postprecipitated with dimethylphosphate, (CH3 O)2 PO2 H. In this case, DMP must bind to the iron colloid surface through its PO2 H functionality and leave two exposed methoxy groups on the surface with which the phosphate tip would have little interaction. The presence of DMP was found to have a negligible effect on the colloid zeta potential. As can be seen in the force titration profile in Fig. 5(iv), while the general form of the pH profile of the force interactions (maximum at pH 6) of the PO2 H tip and the DMP coated surface is the same as that with the unmodified colloid, the tip–sample attractive interactions are substantially reduced. The interpretation of this result is that surface-adsorbed DMP has the effect of consuming a fraction of the surface sites with which the phosphate groups of the tip can bind, but does not introduce any other additional mode of attractive interaction with the probe. The final modified colloid to be prepared was one in which the iron oxyhydroxide was postprecipitated with tannic acid. Zeta potential measurements on this colloid showed a large shift in isoelectric point, down to a pH of about 2.5. No adhesive wells were observed for the interaction of the phosphate-coated tip on these colloids over the entire pH range explored (ca. 3 to 12) (40). This result is consistent with the observation that the postprecipitated TA forms a coating on the surface that inhibits the adsorption of phosphate ions onto the iron oxyhydroxide colloid (5). SUMMARY
Phosphate-terminated surfaces were prepared by the selfassembly of bis(11-thioundecyl)phosphate monolayers on Aucoated mica and on Au-coated AFM tips. The surfaces were characterized using FTIR spectroscopy and chemical force microscopy. Force curves for the PO2 H-terminated tips were obtained as a function of pH, and the adhesive interaction for the PO2 H-terminated tip on the PO2 H-terminated Au surface demonstrated that the surface pK a of the phosphate monolayer was 5.0. The phosphate tip was further characterized by its interaction with both –CO2 H terminated and bare Au surfaces. The tips were then used to characterize a series of iron oxyhydroxide colloids using force titration measurements. These colloids showed the expected strong affinity for phosphate ion, as demonstrated by the large adhesive tip–colloid interaction
observed. Colloids modified by postprecipitation with phosphate ion gave titration curves that strongly resembled those obtained for a phosphate-coated tip on a phosphate-coated sample, suggesting that orthophosphate ion binds to the colloid surface in a bidentate fashion. Further experiments on colloids postprecipitated with dimethylphosphate or tannic acid demonstrated that these species block the interaction of phosphate with iron surface sites, in the case of tannic acid, quite dramatically. These experiments demonstrate the potential of this method for characterizing and distinguishing between the surfaces of this important class of compounds. ACKNOWLEDGMENTS We acknowledge the Natural Sciences and Engineering Research Council of Canada for financial support. The Ravensview Water Pollution Control Plant, Kingston, Ontario, is acknowledged for the donation of a sample of ferric chloride solution.
REFERENCES 1. Dolfin, J., Chardon, E. J., and Japenga, J., Soil Sci. 164, 171 (1999). 2. Liu, C., and Huang, P. M., Can. J. Soil Sci. 80, 445 (2000). 3. Hawke, D., Carpenter, P. D., and Hunter, K. A., Environ. Sci. Technol. 23, 187 (1989). 4. Lopez-Hernandez, D., Siegert, G., and Rodriguez, J. V., Soil Sci. Soc. Am. J. 50, 1460 (1986). 5. Omoike, A. J., and vanLoon, G. W., Wat. Res. 33, 3617 (1999). 6. Eggleston, C. M., Higgins, S. R., and Maurice, P. A., Environ. Sci. Technol. 32, 456 (1998). 7. Weisenhorn, A. L., Maivald, P., Butt, H. J., and Hansma, P. K., Phys. Rev. B 45, 11226 (1992). 8. Burnham, N. A., Colton, R. J., and Pollock, H. M., J. Vac. Sci. Technol. A 9, 2548 (1991). 9. Butt, H. J., Biophys. J. 60, 1438 (1991). 10. Burnham, N. A., Dominguez, D. D., Mowery R. L., and Colton, R. J., Phys. Rev. Lett. 64, 1931 (1990). 11. van der Vegte E. W., and Hadziioannou, G., Langmuir 13, 4357 (1997). 12. Han, T., Williams, J. M., and Beebe, T. P., Jr., Anal. Chim. Acta 307, 365 (1995). 13. Vezenov, D. V., Noy, A., Rozsnyai, L. F., and Lieber, C. M., J. Am. Chem. Soc. 119, 2006 (1997). 14. Ito, T., Namba, M., Buhlmann, P., and Umezawa, Y., Langmuir 13, 4323 (1997). 15. Ulman, A., “An Introduction to Ultrathin Organic Films.” Academic Press, Boston, 1991. 16. McKendry, R., Theoclitou, M. E., Rayment, T., and Abell, C., Nature 391, 566 (1998). 17. Lioubashevski, O., Patolsky, F., and Willner, I., Langmuir 17, 5134 (2001). 18. Hansma, H. G., Ann. Rev. Phys. Chem. 52, 71 (2001). 19. Bowen, W. R., Hilal, N., Lovitt, R. W., and Wright, C. J., Colloids Surf. A 157, 117 (1999). 20. Kim, K., Kim, C., and Byun, Y., J. Biomed. Mater. Res. 52, 836 (2000). 21. Tang, Z., Jing, W., and Wang, E., Langmuir 16, 1696 (2000). 22. Dufrene, Y. F., Boland, T., Schneider, J. W., Barger, W. R., and Lee, G. U., Faraday Discuss. 111, 79 (1998). 23. Yang, X. M., Xiao, D., Xiao, S. J., and Wei, Y., Appl. Phys. A 59, 139 (1994). 24. Nakashima, N., and Taguchi, T., Colloid Surf. A 103, 159 (1995). 25. Nakashima, N., and Taguchi, T., in “Interfacial Design and Chemical Sensing” (T. E. Mallouk and D. J. Harrison, Eds.), pp. 145–154. ACS Symp. Ser. 561, Am. Chem. Soc., Washington, DC, 1994.
PHOSPHATE ADSORPTION ON HYDROUS IRON(III) OXIDE 26. Zhang, J., Kirkham, J., Robinson, C., Wallwork, M. L., Smith, D. A., Marsh, A., and Wong, M., Anal. Chem. 72, 1973 (2000). 27. van der Vegte, E. W., and Hadziioannou, G. J., Phys. Chem. B 101, 9563 (1997). 28. Smith, D. A., Wallwork, M. L., Zhang, J., Kirkham, J., Robinson, C., Marsh, A., and Wong, M., J. Phys. Chem. B 104, 8862 (2000). 29. Omoike, A., and Horton, J. H., Langmuir 16, 1655 (2000). 30. Omoike, A., Chen, G., vanLoon, G. W., and Horton, J. H., Langmuir 14, 4731 (1998). 31. Hutter, J. L., and Bechhoefer, J., Rev. Sci. Instrum. 64, 1868 (1993). 32. Hutter, J. L., and Bechhoefer, J., Rev. Sci. Instrum. 64, 3342 (1993). 33. Guan, Y., Choy, G. S.-C., Glaser, R., and Thomas, G. J., Jr., J. Phys. Chem. 99, 12054 (1995).
213
34. Persson, P., Nilsson, N., and Sj¨oberg, S., J. Colloid Interface Sci. 177, 263 (1996). 35. Tejedor-Tejedor, M. I., and Anderson, M. A., Langmuir 6, 602 (1990). 36. Duffy, S. J., and vanLoon, G. W., Environ. Sci. Technol. 28, 1950 (1994). 37. Paunovic, M., and Sambucetti, C., Proc. Electrochem. Soc. 31, 34 (1994). 38. Nooney, M. G., Murrell, T. S., Corneille, J. S., Rusert, E. I., Hossner, L. R., and Goodman, D. W., J. Vac. Sci. Technol. A 14, 1357 (1996). 39. Cornell, R. M., and Schwertmann, U., “The Iron Oxides: Structure, Properties, Occurences and Uses.” VCH, Weinheim, 1996. 40. Kreller, D. I., Gibson, G., Novak, W., vanLoon, G. W., and Horton, J. H., submitted to Colloid Surf. A.