Adsorption of Organic Acids on TiO2 Nanoparticles: Effects of pH

Jun 7, 2008 - Nanoparticle Size, and Nanoparticle Aggregation. John M. Pettibone,† David M. Cwiertny,‡,§ Michelle Scherer,‡ and Vicki H. Grassi...
2 downloads 0 Views 185KB Size
Download PDF
Langmuir 2008, 24, 6659-6667

6659

Adsorption of Organic Acids on TiO2 Nanoparticles: Effects of pH, Nanoparticle Size, and Nanoparticle Aggregation John M. Pettibone,† David M. Cwiertny,‡,§ Michelle Scherer,‡ and Vicki H. Grassian*,†,§ Department of Chemical and Biochemical Engineering, Department of CiVil and EnVironmental Engineering, and Department of Chemistry, UniVersity of Iowa, Iowa City, Iowa 52242 ReceiVed December 20, 2007. ReVised Manuscript ReceiVed March 25, 2008 In this study, the adsorption of two organic acids, oxalic acid and adipic acid, on TiO2 nanoparticles was investigated at room temperature, 298 K. Solution-phase measurements were used to quantify the extent and reversibility of oxalic acid and adipic acid adsorption on anatase nanoparticles with primary particle sizes of 5 and 32 nm. At all pH values considered, there were minimal differences in measured Langmuir adsorption constants, Kads, or surface-area-normalized maximum adsorbate-surface coverages, Γmax, between 5 and 32 nm particles. Although macroscopic differences in the reactivity of these organic acids as a function of nanoparticle size were not observed, ATR-FTIR spectroscopy showed some distinct differences in the absorption bands present for oxalic acid adsorbed on 5 nm particles compared to 32 nm particles, suggesting different adsorption sites or a different distribution of adsorption sites for oxalic acid on the 5 nm particles. These results illustrate that molecular-level differences in nanoparticle reactivity can still exist even when macroscopic differences are not observed from solution phase measurements. Our results also allowed the impact of nanoparticle aggregation on acid uptake to be assessed. It is clear that particle aggregation occurs at all pH values and that organic acids can destabilize nanoparticle suspensions. Furthermore, 5 nm particles can form larger aggregates compared to 32 nm particles under the same conditions of pH and solid concentrations. The relative reactivity of 5 and 32 nm particles as determined from Langmuir adsorption parameters did not appear to vary greatly despite differences that occur in nanoparticle aggregation for these two different size nanoparticles. Although this potentially suggests that aggregation does not impact organic acid uptake on anatase particles, these data clearly show that challenges remain in assessing the available surface area for adsorption in nanoparticle aqueous suspensions because of aggregation.

Introduction Titanium dioxide is a manufactured nanomaterial that is used in many different applications, including photocatalysts, solar cells, biomaterials, memory devices, and environmental catalysts.1–11 At the nanoscale, particle properties, such as electronic band gaps, magnetic moments, specific heats, and melting points, as well as particle morphology and surface reactivity, become size-dependent,12 especially for particles with diameters less than 10 nm. In addition, nanoparticles are often thermodynamically less stable because of the high ratio of surface to bulk atoms.13 For example, because of differences in surface free energies, anatase is the more stable phase of titanium dioxide for * Author to whom correspondence should be addressed. E-mail: [email protected]. † Department of Chemical and Biochemical Engineering. ‡ Department of Civil and Environmental Engineering. § Department of Chemistry. (1) Liu, G. H.; Wang, J. L.; Zhu, Y. F.; Zhang, X. R. Phys. Chem. Chem. Phys. 2004, 6, 985. (2) Nakamura, R.; Imanishi, A.; Murakoshi, K.; Nakato, Y. J. Am. Chem. Soc. 2003, 125, 7443. (3) Tan, B.; Wu, Y. Y. J. Phys. Chem. B 2006, 110, 15932. (4) Oregan, B.; Gratzel, M. Nature 1991, 353, 737. (5) Zhang, H. Z.; Penn, R. L.; Hamers, R. J.; Banfield, J. F. J. Phys. Chem. B 1999, 103, 4656. (6) Wang, C. Y.; Groenzin, H.; Shultz, M. J. J. Am. Chem. Soc. 2005, 127, 9736. (7) Martyanov, I. N.; Uma, S.; Rodrigues, S.; Klabunde, K. J. Chem. Commun. 2004, 2476. (8) Martin, S. T.; Kesselman, J. M.; Park, D. S.; Lewis, N. S.; Hoffmann, M. R. EnViron. Sci. Technol. 1996, 30, 2535. (9) Chen, X.; Mao, S. S. Chem. ReV. 2007, 107, 2891. (10) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (11) Gong, X. Q.; Selloni, A.; Vittadini, A. J. Phys. Chem. B 2006, 110, 2804. (12) Lucas, E.; Decker, S.; Khaleel, A.; Seitz, A.; Fultz, S.; Ponce, A.; Li, W. F.; Carnes, C.; Klabunde, K. J. Chem.sEur. J. 2001, 7, 2505. (13) Cao, G. Nanostructures & Nanomaterials: Synthesis, Properties & Applications; Imperial College Press: London, 2004.

nanoparticles below 15 nm in diameter,14 whereas for larger particles of bulk TiO2, rutile is the most thermodynamically stable phase. Relatively few studies have probed nanoparticle size effects related to the surface adsorption properties of TiO2, and there is yet to be a clear consensus on the impact of particle size on such processes. Zhang et al.5 investigated the adsorption of organic acids onto 6-16 nm anatase nanoparticles. From thermodynamic considerations, smaller nanoparticles were predicted to show enhanced adsorption, as determined by the Langmuir coefficient for adsorption, Kads, to minimize the surface free energy. Solution phase adsorption studies with a series of organic acids seemed to confirm this prediction, with as much as a 70-fold increase in Kads value reported for the smaller particles. In another study, Wang et al.6 used sum frequency generation to show that smaller TiO2 nanoparticles were also more reactive as determined by the relative amounts of dissociative versus molecular adsorption of methanol from the gas phase with dissociative adsorption increasing as a function of decreasing nanoparticle size. In another recent study using solution phase approaches, the surface adsorption of Cd2+ was found to decrease with decreasing TiO2 particle size, a result counter to the aforementioned studies of Zhang et al. and Wang et al.15 Thus, it remains unclear as to whether a general trend of increasing interfacial reactivity with decreasing particle size exists for TiO2 suspensions or if other factors are at play that may result in the different size-dependent trends previously reported for adsorption on TiO2 nanoparticle surfaces. (14) Barnard, A. S.; Zapol, P.; Curtiss, L. A. J. Chem. Theor. Comput. 2005, 1, 107. (15) Gao, Y.; Wahi, R.; Kan, A. T.; Falkner, J. C.; Colvin, V. L.; Tomson, A. B. Langmuir 2004, 20, 9585.

10.1021/la7039916 CCC: $40.75  2008 American Chemical Society Published on Web 06/07/2008

6660 Langmuir, Vol. 24, No. 13, 2008

An important factor that has not been given too much consideration in the aqueous phase nanoparticle adsorption studies is the role that nanoparticle aggregation can play in the actual amount of surface area available for adsorption and reaction. Studies have shown that aggregation of TiO2 nanoparticles is important in nanoparticle behavior due to the nature and size of the aggregates (i.e., the packing density of the nanoparticles) and aggregation can potentially have an impact on nanoparticle reactivity,16–18 nanoparticle-cellular interactions, and nanoparticle toxicity.19 In a noteworthy study, Dunphy-Guzman et al. investigated the aggregation of different polymorphs of TiO2 nanoparticles as a function of pH.20 The results of their study showed very stable, dispersed suspensions of TiO2 nanoparticles at pH 1 and 12 due to the presence of an electric double layer, as these pH values are far from the point of zero charge, pHpzc, for TiO2.When the pH was altered so that it approached the pHzpc of the phases considered, the repulsive forces between nanoparticles decreased, causing the TiO2 nanoparticles to aggregate. Dunphy-Guzman et al. also found that nanoparticle aqueous suspensions were more stable at extremely acidic pH values rather than extremely basic pH values, which they interpreted as evidence that TiO2 nanoparticles are less capable of holding negative surface charge. On the basis of such considerations, the influence of pH in controlling the stability of TiO2 suspensions will also likely impact the surface adsorption and reactivity of nanoparticles. In the study described herein, we have taken an integrated approach to investigating both adsorption and aggregation of nanoparticles of different size. Although investigations of size dependent adsorption and aggregation have been previously performed, there is an additional need for a greater understanding as to how surface adsorption and reactivity are affected by nanoparticle aggregation. Here we quantitatively measure the extent of oxalic and adipic acid adsorption on 5 and 32 nm TiO2 particles while also exploring how changes in pH, and therefore nanoparticle aggregation, affect surface reactivity and surface properties. We used macroscopic solution phase adsorption studies (i.e., batch systems) and traditional equilibrium adsorption models to quantitatively compare organic acid uptake in suspensions of 5 and 32 nm particles. Complementary light scattering measurements allowed the aggregation state of these suspensions as a function of pH to be considered, whereas ATR-FTIR spectroscopy was employed to probe if different surface sites for organic acid adsorption could be identified for nanoparticles of different size. By integrating experimental results like the ones shown here, there is the potential to gain a greater understanding of the effects that primary particle size, aggregate formation, and pH have on nanoparticle reactivity, nanoparticular-cellular interactions and any adverse health effects that may be associated with nanoparticles.

Experimental Section Reagents. The majority of experiments were conducted with two sizes of TiO2 nanoparticles that were purchased from commercial sources. Nanoparticles supplied by Nanostructured and Amorphous Materials, Inc. (Houston, TX) were reported by the vendor to have an average primary particle size of 5 nm and to consist entirely of (16) Liu, Z. Y.; Sun, D. D.; Guo, P.; Leckie, J. O. Chem.sEur. J. 2007, 13, 1851. (17) Tseng, Y. H.; Lin, H. Y.; Kuo, C. S.; Li, Y. Y.; Huang, C. P. React. Kinet. Catal. Lett. 2006, 89, 63. (18) Rachel, A.; Subrahmanyam, M.; Boule, P. Appl. Catal., B 2002, 37, 301. (19) Grassian, V. H.; Adamcakova-Dodd, A.; Pettibone, J. M.; O’Shaughnessy, P. T.; Thorne, P. S. Nanotoxicology 2007, 1, 211. (20) Guzman, K. A. D.; Finnegan, M. P.; Banfield, J. F. EnViron. Sci. Technol. 2006, 40, 7688.

Pettibone et al. anatase. Nanoparticles from Alfa Aesar (Ward Hill, MA) were reported to have a primary particle size of 32 nm and were 99.9% anatase. In addition, a select number of experiments were conducted with rutile particles with a primary particle diameter of 30 nm that were purchased from MTI Corp (Richmond, CA). Solution phase adsorption measurements were conducted using solutions of 2-(Nmorpholino)ethanesulfonic acid (MES; Sigma Aldrich; g99%), 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES; Sigma Aldrich, g99.5%), or 0.01 N hydrochloric acid (HCl; Fisher Scientific; certified ACS plus). Solutions also contained sodium chloride (NaCl; Fisher Scientific; certified ACS), which was used to poise ionic strength. The aqueous solutions of oxalic (Aldrich, 98%) and adipic acid (Alfa Aesar, 99%) and all stock solutions and reactors were made using Optima water (Fisher Scientific). Nanoparticle Characterization. Powder X-ray diffraction (XRD) was used to determine the bulk crystalline phase of each TiO2 nanoparticle size. Diffraction patterns were collected using a Rigaku Miniflex II diffractometer with a Co source. As the phase stability of titania nanoparticles has been suggested to depend strongly on solution pH,21 diffraction patterns were collected not only for powders as received but also for powders that had been suspended in the aqueous solutions used in batch adsorption studies. For XRD analysis of processed TiO2 particles, suspensions were prepared in either pH 2 or pH 6.5 solutions (the lowest and highest pH values used in our experiments, respectively) and were vigorously mixed for approximately 24 h in an amber vial. After this mixing step, the nanoparticles in the pH 6.5 solution were allowed to gravitationally settle out of suspension, the solution was decanted off the solid particles, and the particles were then dried in the oven at 40 °C. On the other hand, pH 2 suspensions were essentially stable over very long time scales. Therefore, a small amount of 5 N NaOH was added to the suspension to quickly raise its pH to a near-neutral value, thereby destabilizing the suspension. The particles were then allowed to settle, decanted, and dried as described for the pH 6.5 particles. After drying, both sets of particles were ground using a mortar and pestle and sieved prior to XRD analysis. The specific surface area of each particle size was determined from seven-point N2-BET adsorption isotherm measurements performed on a Quantachrome 4200e surface area analyzer. Prior to surface area analysis, samples were degassed for 3 h at 300 °C. In addition to experimentally measured BET specific surface areas, the geometric surface area of each particle size was estimated using a TiO2 density of 4.23 g/cm3 and by assuming a spherical nanoparticle shape. Solution Phase Adsorption Studies. Solution phase adsorption studies were conducted in aqueous batch reactors at pH 2, 5.5, and 6.5. Solutions at pH 2 consisted of 0.01 N HCl, whereas pH 5.5 and 6.5 solutions consisted of 25 mM MES and HEPES buffer, respectively. All solutions were adjusted to an ionic strength of 0.02 M through the addition of NaCl. Although experiments at pH 2 were conducted with oxalic and adipic acid, experiments at pH 5.5 were limited to adipic acid, while experiments at pH 6.5 only considered oxalic acid uptake. The pKa values for oxalic acid are pKa1 ) 1.27 and pKa2 ) 4.28 along with adipic acid having values of pKa1 ) 4.43 and pKa2 ) 5.41. It should be noted that, for simplicity, we will refer to the organic adsorbate as an acid, even though at certain pH values considered herein both species will be in their conjugate base form. Additional experiments with each organic acid were also conducted in electrolyte solutions in which no attempt was made to adjust or control system pH. For these experiments without pH control, an organic acid solution was added to a 0.02 M NaCl solution. These systems without pH control were specifically conducted to mimic the experimental protocols conducted in an earlier study of organic acid adsorption on TiO2 nanoparticles.5 Adsorption experiments were performed using the following protocol. Stock solutions of each acid with concentrations of 10 and 100 mM were prepared from pure acid form. An appropriate volume of electrolyte or buffer solution was added to 8 mL glass vials prior (21) Finnegan, M. P.; Zhang, H. Z.; Banfield, J. F. J. Phys. Chem. C 2007, 111, 1962.

Organic Acids on TiO2 Nanoparticles to the addition of the organic acid. Preceding the addition of solid, a 1 mL sample was taken from each vial for quantification of the initial organic acid concentration in the system, leaving 6 mL of buffer and organic acid as the reaction volume. A variety of suspension concentrations were used, with nanoparticle suspension concentrations ranging from 0.5 to 6 g/L and initial organic acid concentrations ranging from 0.1 to 2.2 mM. After construction, the vials were crimp sealed, covered in foil to inhibit photoinduced chemistry, and mixed end-over-end on a circular rotator (Cole-Parmer) for at least 2 h. At the conclusion of each experiment, the pH of each suspension was measured using a microelectrode, thus allowing accurate measurements of pH in low suspension volumes, and aqueous samples were taken so that the dissolved concentration of organic acid at equilibrium could be quantified. Approximately 1 mL of suspension was withdrawn from the reactor with a disposable syringe, and the sample was subsequently passed through a 0.2 µm syringe driven filter unit (Xpertec) to separate the solution from the TiO2 solid. Filtered samples were analyzed using an Agilent HPLC that was equipped with a diode array UV-vis detector and used a mobile phase of 0.01 N H2SO4. Oxalic and adipic acid were analyzed at 254 and 210 nm, respectively. Instrument response measured at these wavelengths was converted to aqueous phase concentrations using standards of oxalic and adipic acid that were prepared daily in an appropriate aqueous solution. The concentration of adsorbed organic acid was then determined from the difference in initial and final concentrations measured in the experimental systems. Experiments were also conducted to explore whether the uptake of oxalic acid on each size of TiO2 particle was reversible. After an adsorption study conducted at pH 6.5, the particles were separated into six 1-mL aliquots and centrifuged for 10 min. The supernatant was decanted off the solid, and the solid was subsequently resuspended in fresh HEPES buffer. The reactor was then allowed to mix for at least 2 h (the typical duration of our adsorption experiments), at which time a filtered sample was taken and analyzed using HPLC to measure the amount of acid present in solution after reaching equilibrium in the new solution. It should be noted that during our characterization and adsorption studies, we detected an impurity on the surface of the manufactured 32 nm particles. According to the manufacturer, these particles were synthesized using a chemical vapor decomposition (CVD) method, and it is not uncommon to have organic material left on the surface of metal oxides particles generated in this way due to the incomplete reaction of precursor materials.22 The clearest evidence of the surface impurity was found during HPLC analysis of the filtered samples taken from our batch adsorption experiments. During HPLC analysis, a large peak with a very short retention time was observed, suggesting that this impurity was leaching from the surface during our batch studies. This feature was only observed during analysis of solutions that had been in contact with 32 nm particles, as a similar peak was not observed for samples from 5 nm particle suspensions. This material was assumed to be organic in nature. However, we were able to remove it from the particles via repeated washing with deionized (DI) water. To assess the impact of this surface impurity on our adsorption results, we conducted a small number of batch studies with these washed particles. The washing process involved suspending an appropriate mass of 32 nm particles in deionized water and stirring the suspension overnight. The suspension was allowed to settle out and the solution was subsequently decanted off the particles. The process was repeated, and the particles were then dried at 40 °C, ground using a mortar and pestle, and sieved. These processed particles exhibited essentially identical reactivity to 32 nm particles that were used as received. This result serves as an example of the possible limitations of using commercially available nanomaterials and the need for the rigorous characterization of such particles if they are to be used in fundamental studies of nanoparticle reactivity and toxicity. Nanoparticle Aggregation Studies. The aggregation of 5 and 32 nm TiO2 nanoparticles in aqueous solutions was examined using (22) Buzby, S.; Franklin, S.; Shah, S. I. Synthesis, Properties and Applications of Oxide Nanoparticles; John Wiley & Sons, Inc.: Hoboken, NJ, 2007.

Langmuir, Vol. 24, No. 13, 2008 6661 two different light-scattering techniques. A commercial dynamic light scattering (DLS) instrument (Malvern Zetasizer Nano ZS) equipped with a green laser at 532 nm was used to obtain relative aggregate sizes before and after reaction with oxalic acid. Suspensions were prepared approximately 24 h prior to DLS analysis using aqueous solutions that had been passed through a 0.2 µm syringedriven filter to minimize the influence of dust on the size measurements. All suspensions were allowed to sit overnight so that the size of aggregates generated was assumed to be at steady state. DLS size measurements were made at pH 2 and 6.5 and a constant ionic strength of 0.02 M, identical conditions to those used in batch sorption studies. A range of the TiO2 concentrations was analyzed (0.002, 0.01, 0.02, 0.05, 0.1, and 0.2 g/L), and for solutions containing oxalic acid, a ratio of 1 g/L TiO2 per mM acid was used (e.g., 0.1 g/L TiO2with an oxalic acid concentration 0.1 mM). Sedimentation experiments, which used a UV-vis spectrometer to monitor changes in scattered light in suspensions as a function of time, followed previous analytical techniques23 with slight modifications.24 Briefly, suspensions of TiO2 or TiO2 with oxalic acid were placed in a 1-cm path length cuvette and agitated so as to form a uniform suspension. The cuvette was placed in the UV-vis instrument and the amount of transmitted light at 508 nm was measured over time. At 508 nm wavelength, a linear relationship between scattered light and the TiO2 suspension concentrations was observed for nanoparticle concentrations of 0.1-0.5 g/L used in the sedimentation studies. ATR-FTIR Spectroscopy. Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy was used to compare the adsorption of adipic and oxalic acid on the two different sizes of anatase nanoparticles. A suspension of 1.5 mg of TiO2 nanoparticles in 1 mL of Optima water was placed on a ZnSe horizontal ATR cell (Pike Technologies, Inc.) and allowed to dry overnight, creating a thin film that evenly coated the crystal. The crystal was flushed with a delicate stream of water to rinse any loosely bound particles. Acid solution (adipic or oxalic acid) at constant pH was added every 10 min to the horizontal ATR cell to allow time for equilibration of the solution with the thin film. The solution was added until equilibrium was reached, which correlated to the time when the integrated peak area of the oxalic acid peaks remained constant with time or upon addition of a new fresh solution at that concentration. A large concentration range for the two organic acids was investigated from 0.001 to 100 mM. However, for surface adsorption experiments only low concentration data are shown where bands to adsorbed species dominate and, due to the weakness of the signal at these low concentrations, there is no contribution from solution phase acid. For a few select experiments, D2O was used instead of H2O and to remove any contributions to the spectrum from the water-bending mode (1642 cm-1), which can typically, but not always, be easily subtracted out of the spectrum, so as to confirm assignments of adsorbed oxalic and adipic acid absorption bands (vide infra).

Results and Discussion Nanoparticle Characterization. A summary of the physical characterization of the 5 and 32 nm TiO2 nanoparticles is provided in Table 1, and XRD patterns for both particle sizes are shown in Figure 1. Diffraction patterns are presented both for TiO2 nanoparticles characterized as received, as well as for particles that had been suspended in the aqueous buffer solutions used in the batch experiments and subsequently dried prior to XRD analysis. For the as received TiO2 nanoparticles, diffraction patterns for the 5 nm particles revealed that they consist entirely of anatase, whereas the 32 nm particles are primarily anatase, but also contain a small amount of rutile. The observation of a small amount of rutile in the 32 nm particles is not unexpected on the basis of (23) Phenrat, T.; Saleh, N.; Sirk, K.; Tilton, R. D.; Lowry, G. V. EnViron. Sci. Technol. 2007, 41, 284. (24) Cwiertny, D. M.; Handler, R. M.; Schaefer, M. V.; Grassian, V. H.; Scherer, M. M. Geochim. Cosmochim. Acta 2007, 72, 1365.

6662 Langmuir, Vol. 24, No. 13, 2008

Pettibone et al.

Figure 1. Powder X-ray diffraction patterns of 5 and 32 nm TiO2 particle that were analyzed as received from the commercial vendor, as well as after suspension in aqueous solutions at pH values used in aqueous phase adsorption studies. Reference diffraction patterns for the common TiO2 polymorphs rutile and anatase are shown for comparison with all diffraction patterns normalized. Table 1. Summary of TiO2 Nanoparticle Properties: Size, Surface Area, and Crystalline Phase Nanostructured and Amorphous Materials, Inc. primary particle size, nm BET surface area, m2/g (measureda) geometric surface area, m2/g (calcd) crystalline phase a

Alfa Aesar

5

32

219 ( 3

41 ( 2

280

45

anatase (100%)

anatase/rutile( 3). At pH 2, both the 5 and 32 nm particles aggregate extensively, as the hydrodynamic diameters estimated by DLS analysis are considerably greater than the primary particle sizes. Furthermore, the sizes of these aggregates were observed to increase with increasing solid concentration (see Figure 3), although minimal variation in the size of aggregates was observed for TiO2 concentrations above approximately 0.05 g/L. Of note, the 32 nm particles always generated smaller aggregates than the 5 nm TiO2 primary particles in suspensions with the same mass concentration at pH 2, suggesting that the extent of aggregation was greatest for the smaller primary particle size.

Organic Acids on TiO2 Nanoparticles

Langmuir, Vol. 24, No. 13, 2008 6665

Figure 4. Intensity-normalized particle size distributions obtained from DLS measurements performed on pH 2 suspensions of 5 and 32 nm TiO2 primary particles that had been equilibrated with oxalic acid. For 5 nm particles, a solids concentration of 0.05 g/L was used with an initial oxalic acid concentration of 0.05 mM. For 32 nm particles, a solids concentration of 0.1 g/L was used with an initial oxalic acid concentration of 0.1 mM. Data are shown for triplicate measurements

DLS measurements were also taken at pH 6.5 (data not shown). Consistent with the results of previous studies,20 the aggregate size at this higher pH value, which is closer to the zero-point of charge of bulk TiO2 particles (pHzpc for bulk TiO2 is 5.9), was greater than that observed at pH 2. Note, however, that the pH of the zero point of charge for TiO2 nanoparticles has a broad range of 4.8-8.1 depending on phase, size, and preparation.9 In contrast to the pH 2 data, the size of aggregates at pH 6.5 in both 5 and 32 nm primary particle suspensions were roughly the same size, but the presence of sedimenting particles at higher concentrations made the analysis of the DLS studies in these highly aggregated systems invalid as gravitational settling becomes prevalent in the suspensions. Figure 4 shows the influence of oxalic acid on the size distribution of aggregates in pH 2 suspensions of TiO2 as determined by DLS measurements. In Figure 4, a systematic and reproducible shift in the size of aggregates in both 5 and 32 nm TiO2 suspensions is observed upon the addition of oxalic acid. In fact, the relative size of the aggregates in the presence of oxalic acid and solid concentrations is roughly the same in both 5 and 32 nm primary particle suspensions. Although the size of the aggregates at these observed solid and acid concentrations was similar, this small data set does not infer that similar aggregate sizes for the two nanoparticles with adsorbed oxalic acid will follow that trend. These results only suggest that the uptake of oxalate on the surface of TiO2 nanoparticles further destabilizes the suspension, resulting in increased aggregation. Results from sedimentation studies with 5 and 32 nm particle suspensions are shown in Figure 5. It is important to remember that sedimenting occurs due to greater gravitational forces on the aggregates than the opposing buoyancy forces and the relative rates of sedimentation are affected by cross-sectional area and/or the density of aggregates in each solution. One advantage of sedimentation studies is that they can provide qualitative,

Figure 5. Sedimentation plots for 0.1 g/L suspensions of 5 and 32 nm TiO2 particles at pH 2 and 6.5. Data are shown for suspensions with and without 0.1 mM oxalic acid.

complementary information on the relative size of aggregates in suspensions with higher solution concentrations that are more representative of those used in the solution phase adsorption studies. At pH 2 (Figure 5a), the 32 nm particles did not settle and the suspension remained stable over a long period of time, even though DLS measurements show that aggregation is indeed occurring in the system. The aggregates are assumed to have strong repulsive electric double layers that most likely keep the suspension stable, in some cases over a time frame as long as 1 month.20 The 5 nm particles behaved similarly to the 32 nm particles in that they also were stable over relatively long periods of time, but the rate of sedimentation was slightly greater in the 5 nm suspension, suggesting that its aggregates are either larger or more dense. This is in agreement with the DLS data that showed the 5 nm primary particles grew into larger aggregates than the aggregates of 32 nm primary particles at all solid concentrations considered. When oxalic acid is present in suspensions at pH 2 (Figure 5a), sedimentation rates increase in both 5 and 32 nm primary particle suspensions. This destabilizing influence of oxalic acid at pH 2 is consistent with the increase in aggregate hydrodynamic diameter measured by DLS analysis (see Figure 4). Interestingly, the resulting aggregates produced from 5 and 32 nm particles behave similarly, with similar rates of settling observed in each system. This suggests that the aggregates generated from 5 and 32 nm primary particles in the presence of oxalic acid have similar buoyancy that could be related to similar aggregate size or have different particle densities and packing that contribute to similar settling. Sedimentation experiments conducted at pH 6.5 revealed a more complicated aggregate settling behavior than that observed

6666 Langmuir, Vol. 24, No. 13, 2008

Pettibone et al.

Figure 6. ATR-FTIR spectra for 5 and 32 nm TiO2 particles exposed to (a) oxalic and (b) adipic acid. In panel a, spectra are shown for 0.05 and 0.1 mM concentrations of oxalic acid at pH 6.5 on 5 and 32 nm particles, respectively. The solution-phase spectrum for 100 mM oxalic acid is also shown for comparison. In panel b, spectra for adipic acid were collected at pH 5.5. Concentrations of 0.1 and 0.5 mM were used for showing the adsorbed adipic acid on the 5 and 32 TiO2 surfaces. The solution-phase adipic acid spectrum is shown for comparison. *Asterisks denote peaks due to adsorption on the ZnSe ATR crystal.

at pH 2 (Figure 5b). For example, there appears to be multiple regimes of particle growth and sedimentation occurring in both 5 and 32 nm primary particle suspensions, with 32 nm particles settling out of solution more rapidly, at least at early time periods. Despite these initial differences in sedimentation rates, a greater fraction of particles had settled in the 5 nm primary particle suspensions after approximately 60 min, suggesting that the distributions of aggregate size and density are relatively broad in both 5 and 32 nm primary particle suspensions at pH 6.5. We also found that there was little or no effect on the sedimentation rate of particles due to the adsorption of the oxalic acid on the surface at pH 6.5. The observation of extensive aggregation in both 5 and 32 nm primary particle suspensions over the range of pH values considered raises questions regarding the appropriate nanoparticle characteristic to use as the basis of comparison for size-dependent reactivity. In particular, the similar degree of particle aggregation in 5 and 32 nm systems could be put forth as a possible explanation for the different size-dependent trend in particle reactivity we observed relative to the study of Zhang et al.5 We note, however, that our adsorption experiments and aggregate characterization studies were conducted over a range of pH values that overlapped with the pH values used in the experimental studies of Zhang et al.5 We would expect, therefore, a similar influence of aggregation on particle reactivity to be observed in both studies. On the basis of BET surface area, we observed comparable reactivity of 5 and 32 nm suspensions both at pH 2 and at higher pH values (5.5 and 6.5 for adipic and oxalic acid, respectively). Because the extent of 5 and 32 nm particle aggregation is less at pH 2 compared to these higher pH values, we would expect the relative reactivity of 5 and 32 nm particles to change over this pH range if aggregation were exerting a large influence on the surface area available for organic acid uptake, as was seen in previous studies of Fe(II) adsorption on goethite particles.24 In particular, our investigations with oxalic acid indicate that the value of Γmax(5 nm)/Γmax(32 nm) was approximately equal to 1 (32) Diebold, U.; Ruzycki, N.; Herman, G. S.; Selloni, A. Catal. Today 2003, 85, 93. (33) Hug, S. J.; Sulzberger, B. Langmuir 1994, 10, 3587.

for all pH conditions, and hence aggregation states, explored. This result would support the conclusion that nanoparticle aggregation in solution does not diminish the amount of surface area available for organic acid uptake if the calculated surface site density of anatase was similar to the maximum coverage observed in the adsorption studies. A first approximation calculation based on a spherical particle and a reported surface site density for anatase (101)32 along with competitive oxalate adsorption with water onto the anatase surface33 shows nearly identical surface coverage to what is observed experimentally. Step edges were considered equivalent to the more planar (101) face based on theoretical calculations from Gong et al.34 ATR-FTIR Studies of Oxalic and Adipic Acid Adsorption on TiO2. On the basis of results presented from the solution phase measurements, the 5 and 32 nm particles behave similarly when comparing their macroscopic reactivity toward oxalic and adipic acid in aggregated systems. To examine whether there are any molecular level differences for the interaction of these acids with TiO2 surfaces, ATR-FTIR spectroscopy was used to investigate the adsorption of these organic acids from solution. In Figure 6a, ATR-FTIR spectra are presented for the adsorption of oxalic acid on 5 and 32 nm particles coated on a ZnSe crystal in a horizontal ATR flow cell. Spectra are shown for oxalic acid adsorption at pH 6.5 along with the pure solution phase spectrum at the same pH. Several interesting points are worth noting from these spectra. First, the solution phase spectra show that at pH 6.5 oxalic acid is completely deprotonated to yield oxalate, C2O42-, as seen by the characteristic absorptions at 1577 and 1309 cm-1, which is consistent with pKa values and previous results.33,35–37 The broader band above 1600 cm-1 is due to poor subtraction of the water bending mode, which seemed problematic in the liquid phase spectra. Additionally above 1600 (34) Gong, X. Q.; Selloni, A.; Batzill, M.; Diebold, U. Nat. Mater. 2006, 5, 665. (35) Duckworth, O. W.; Martin, S. T. Geochim. Cosmochim. Acta 2001, 65, 4289. (36) Johnson, S. B.; Yoon, T. H.; Slowey, A. J.; Brown, G. E. Langmuir 2004, 20, 11480. (37) Dobson, K. D.; McQuillan, A. J. Spectrochim. Acta, Part A 1999, 55, 1395.

Organic Acids on TiO2 Nanoparticles

cm-1, there is a small contribution due to adsorption of oxalic acid onto the ZnSe ATR crystal. (These bands are noted by an asterisk.) Second, a comparison of the spectra for oxalic acid adsorption on 5 versus 32 nm particles shows that there are some differences in peak frequencies, as well as an additional absorption band present in the spectrum for oxalic acid adsorption on the 5 nm TiO2 particles. In particular, the spectrum for adsorption on the 32 nm particles shows characteristic oxalic acid peaks near 1287, 1431, 1695, and 1719 cm-1, which have been previously assigned in the literature to the symmetric stretch of the carboxylate group, νs(CO2), a combination band, and the asymmetric stretch, νas(CO2).33,37 The spectrum for oxalic acid adsorption on the 5 nm particles shows that these same absorption bands are present but some are shifted in frequency (for example, the band at 1703 cm-1 is now centered at 1695 cm-1 and the band at 1287 cm-1 is now at 1300 cm-1). Of greater significance is the broad band near 1630 cm-1, which is either completely absent in the spectra for oxalic acid adsorption on 32 nm particles or relatively weak compared to the other absorption bands in the spectrum. This additional peak has been seen in previous ATR-FTIR experiments with small nanometer-sized anatase and rutile particles. In some cases, the peak was assigned to hydration changes or changes in intensity due to hydrogen bonding.38 In other studies it has also been assigned to νas(CO2).39 In our studies, additional time course experiments with 5 nm particles showed that the 1630 cm-1 peak was the first to appear in the spectrum; thus, it appears to correspond to a unique adsorbed species on 5 nm particle surfaces and not just changes in the hydrogen bonding. Furthermore, this species is there when D2O is used instead of H2O, showing in fact that it is not a water absorption band but an oxalate absorption band. Our data are more consistent with assigning this peak to a red-shifted asymmetric stretch, νas(CO2), for oxalic acid adsorption. With the appearance of the additional peak for oxalate on the 5 nm particles, there most likely is a lower energy binding site available for adsorption that is populated first and may be associated with edge and corner sites, which will be in greater abundance for the 5 nm particles compared to the larger 32 nm particles. Figure 6b shows the ATR-FTIR spectra for adipic acid adsorption on 5 and 32 nm TiO2 at pH 5.5. At pH 5.5, there are three characteristic peaks for both solution and adsorbed phases with frequencies near 1411, 1468, and 1550 cm-1. These peaks are characteristic of the deprotonated form of the acid. These spectra show that, at pH 5.5, adipate is the predominant solution and adsorbed phase species. Furthermore, unlike oxalic acid adsorption, there are no additional absorption bands in the spectrum for adsorption on 5 nm TiO2 particles compared to the 32 nm particles. Thus while adsorption of oxalic acid on the smaller nanoparticles showed distinct adsorption sites, this is not the case for adipic acid. There are several possible reasons for this difference between adipic and oxalic acid. Adipic acid is a six carbon dicarboxylic acid and is known to form monodentate complexes on metal oxides, which differs from the smaller oxalic acid that forms bidenatate ring structures.35 The longer monodentate ligands could cause more strain, thereby blocking available surface sites that oxalic acid is able to bind to due to charging effects33 or transannular strain if ring structures are formed.35 The fact that for oxalic acid the protonated form of the acid is present on the surface may also play a role in the differences between the fact that unique adsorption sites are observed for oxalic but not adipic (38) Hug, S. J.; Bahnemann, D. J. Electron Spectrosc. Relat. Phenom. 2006, 150, 208. (39) Mendive, C. B.; Bredow, T.; Blesa, M. A.; Bahnemann, D. W. Phys. Chem. Chem. Phys. 2006, 8, 3232.

Langmuir, Vol. 24, No. 13, 2008 6667

acid adsorption on the 5 nm TiO2 particles compared to the 32 nm particles.

Conclusions For the reversible uptake of adipic and oxalic acid on 5 and 32 nm anatase particles at a variety of pH values, the adsorption coefficients and maximum surface coverages were similar when normalized to specific surface area values determined experimentally by BET measurements or via estimates based on particle geometry. DLS analysis and experimentally measured sedimentation rates indicate that both 5 and 32 nm particles aggregate extensively over the range of pH values considered, that under certain solution-phase conditions 5 nm particles can aggregate to a greater extent than the 32 nm anatase particles, and that the presence of organic acids diminishes TiO2 suspension stability. Although the relative size of 5 and 32 nm particle aggregates varied considerably as a function of pH, the relative reactivity of these particle suspensions was essentially equivalent at all pH values considered, which suggests that aggregation does not influence the uptake of organic acids on TiO2 surfaces. Even though we did not observe macroscopic differences in TiO2 nanoparticle reactivity with size from solution-phase adsorption studies, ATR-FTIR studies suggest different surface complexes for oxalic acid adsorbed on the surface of 5 compared to 32 nm anatase particles. In addition to demonstrating the importance of pH control in experiments exploring interfacial reactivity, the current study illustrates the benefit of an integrated experimental approach that links the macroscopic reactivity of nanoparticles in batchscale systems to the molecular-level processes taking place at nanoparticle interfaces. Furthermore, our data also serve as an example that particle aggregation can vary with respect to particle size and that the adsorbate will also influence suspension stability. Although nanoparticle aggregation appears to have minimal impact on organic acid adsorption on the TiO2 particles considered herein, this generally may not be the case for other nanoparticle suspensions. The tendency for nanoparticles to aggregate, both in the presence and absence of adsorbed species, raises a number of questions and issues regarding the appropriate basis for comparing the size-dependent reactivity of metal oxide nanoparticles. There remains a need, therefore, to better quantify reactive surface area in aqueous nanoparticle suspensions and to continue to evaluate the applicability of surface areas determined from analysis of dry powders (e.g., BET or geometric surface areas) as a measure of reactive surface area in these suspensions. Acknowledgment. Although the research described in this article has been funded in part by the Environmental Protection Agency through grant number EPA RD-83171701-0 to VHG, it has not been subjected to the Agency’s required peer and policy review and therefore does not necessarily reflect the views of the Agency and no official endorsement should be inferred. This material is based upon work partially supported by the National Science Foundation under Grants No. EAR0506679 and CHE0639096. Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. Supporting Information Available: Figure S1 shows isotherms at pH 6.5 scaled to TiO2 mass. Figure S2 shows isotherms at pH 2 scaled to BET surface area. This material is available free of charge via the Internet at http://pubs.acs.org. LA7039916