Aqueous Solution Interface

Mar 2, 1999 - alkaline solution wash of the TiO2 films was shown to remove contaminants ... structure which are similar to those observed from concent...
0 downloads 0 Views 162KB Size
Download PDF
2402

Langmuir 1999, 15, 2402-2408

Infrared Spectroscopy of the TiO2/Aqueous Solution Interface Paul A. Connor, Kevin D. Dobson, and A. James McQuillan* Department of Chemistry, University of Otago, P.O. Box 56, Dunedin, New Zealand Received July 10, 1998. In Final Form: November 23, 1998 An innovative approach has been used to probe the molecular nature of the metal oxide/aqueous solution interface. Internal reflection spectroscopy of thin colloidal TiO2 films, under aqueous solutions of pH 11.7-2.3, has been used to obtain differential in situ infrared spectra related to interfacial species. An alkaline solution wash of the TiO2 films was shown to remove contaminants arising from the sol evaporationfilm deposition process. Specific infrared absorptions have been assigned to terminal Ti-OH and TiOH2+, adsorbed water Ti-OH2, and bridging Ti-OH+-Ti species from the pH dependence of spectra and from deuteration experiments. These surface species determine the pH-dependent surface charge and the enhanced interfacial ionic concentrations observed in our previously published STIRS results. The enhanced interfacial ionic concentrations were also observed to have spectral effects related to interfacial water structure which are similar to those observed from concentrated aqueous solutions containing these ions. The proposed interfacial species have long been components of models of hydrous oxide surfaces, but this is the first in situ vibrational spectroscopic analysis aimed at their identification. Scheme 1

Introduction 1

Metal oxides are found widely in nature and in technology.2 They frequently occur in aqueous environments and have significant roles in soil science,1 catalysis,3 detergency,3 and hydrometallurgy.4 The behavior of metal oxides is often determined by surface processes, but the understanding of their surface chemistry in aqueous environments is somewhat limited. Oxide surface properties also influence the surface behavior of metals in biological and other aqueous media.5,6 Studies of metal oxide/aqueous solution interfaces have been largely confined to macroscopic properties typified by surface charge7 and adsorption5 measurements. Such measurements have been used to produce molecular models of surface species,8 often with a minimum of direct molecular evidence to test these models. In aqueous solutions metal oxides have pH-dependent surface charge which can be explained by the existence of equilibria9 such as in Scheme 1, where M represents a surface metal ion. Models of surface charge behavior have usually been based on two of these equilibria, i.e., (a) and (b) or (b) and (c). At sufficiently low pH an oxide surface is expected to be protonated giving a positive surface charge. At high pH the surface will be deprotonated with a negative charge. At some intermediate pH the surface has net zero charge, and this pH is called the point of zero charge (PZC) or isoelectric point. A surface * Corresponding author. E-mail: [email protected]. (1) Greenland, D. J., Hayes, M. H. B., Eds. The Chemistry of Soil Processes; John Wiley and Sons: New York, 1981. (2) Kirk-Othmer Encyclopedia of Chemical Technology, 3rd ed; WileyInterscience: New York, 1983; Vol. 23, pp 138-152. (3) Gratzel, M. Heterogeneous Photochemical Electron Transfer; CRC: Boca Raton, FL, 1989. (4) Marsden, J.; House, I. The Chemistry of Gold Extraction; Ellis Horwood: London, 1992. (5) Boehm, H. P. Faraday Discuss. Chem. Soc. 1971, 52, 264. (6) Sundgren, J.-E.; Bodo¨, P.; Lundstro¨m, I. J. Coll. Interface Sci. 1986, 110, 9. (7) Yates, D. E.; Healy; T. W. J. Chem. Soc., Faraday Trans. 1980, 76, 9. (8) Rodrı´guez, R.; Blesa, M. A.; Regazzoni, A. E. J. Colloid Interface Sci. 1996, 177, 122 (9) Hunter, R. J. Foundations of Colloid Science; Oxford University Press: New York, 1989; Vol. 1, Chapter 6.

metal ion may have a residual charge which depends on the surface group and the crystal face. For simplicity, we have represented M in Scheme 1 as uncharged. Titanium dioxide exists as the minerals rutile, anatase, and brookite.10 Both rutile and anatase have been well studied by conventional surface chemical methods.11 From infrared (IR) studies of the TiO2-gas interface,12-15 it is known that the surface of the metal oxide is hydroxylated when exposed to water vapor, with several types of hydroxyl species present. It has been proposed12 that there is one chemically active face (001 on anatase and 110 on rutile) which contains both terminal and bridging OH groups. The other major crystal faces contain metal ions which are believed to only weakly bind water molecules.16 The bridging and terminal hydroxides have been observed via magic angle spinning NMR.17 IR spectroscopy has long been used as a structurally sensitive technique for the detection of species at the metal oxide-gas interface,11-14 but the strong absorptions of water have hindered its application to the metal oxidewater interface. However, internal reflection techniques facilitate IR spectroscopy of aqueous solutions by providing (10) Jones, A. J., Ed. Titanium: From Mining to Biomaterials; Department of Industry, Technology and Commerce: Australia, 1991. (11) Hadjiivanov, K. I.; Klissurski, D. G. Chem. Soc. Rev. 1996, 25, 66. (12) Primet, M.; Pichat, P.; Mathieu, M.-V. J. Phys. Chem. 1971, 75, 5, 1216. (13) Yates, D. J. C. J. Phys. Chem. 1961, 65, 746. (14) Jones, P.; Hockey, J. A. Trans. Faraday. Soc. 1971, 67, 2669. (15) Jones, P.; Hockey, J. A. J. Chem. Soc., Faraday Trans. 1 1972, 68, 907. (16) Yates, D. E.; James, R. O.; Healy, T. W. J. Chem. Soc., Faraday Trans. 1 1980, 76, 1. (17) Crocker, M.; Herold, R. H. M.; Wilson, A. E.; Mackay, M.; Emeis, C. A.; Hoogendoorn, A. M. J. Chem. Soc., Faraday Trans. 1996, 92, 2791.

10.1021/la980855d CCC: $18.00 © 1999 American Chemical Society Published on Web 03/02/1999

TiO2/Aqueous Solution Interface

short and repeatable sample path lengths. While the earliest in situ IR spectroscopic studies of the surfaces of metal oxides in aqueous solutions were carried out on suspensions,18-20 more recent studies have found that deposited particle films21,22 provide greater stability and sensitivity of spectral data. Some metal oxide particle films have been prepared by evaporation of aqueous colloid solutions onto internal reflection elements.22 We have used this approach to study adsorption of ions and ligands on metal oxides22-25 and other semiconductors.26,27 A previously reported in situ infrared study, addressing the nature of surface groups and interfacial water structure at a metal oxide/aqueous solution interface, is that of Tejedor-Tejedor and Anderson.19 They assigned absorptions from aqueous goethite suspensions to surface hydroxyl groups and to changes in interfacial structure with pH and ionic strength. Recent IR studies of the electrode-aqueous solution interface28-30 have also addressed similar issues. These studies have shown that interfacial water structure is dependent on both surface charge29 and electrolyte species,28 and similar considerations are expected to be significant at the metal oxideaqueous solution interface. The purpose of the present paper is to examine the infrared spectral evidence for the nature of the TiO2 surface groups which determine its surface charge and adsorption properties. The sensitivity of the infrared difference spectra we have been able to obtain from the metal oxide colloid films suggested that some features of the spectra may be associated with particular surface groups. We have recently developed the surface titration by internal reflectance spectroscopy (STIRS) technique24 to monitor surface charge and adsorption at metal oxide surfaces in aqueous solutions with infrared spectroscopy. However, we have not hitherto considered the spectroscopic evidence for the surface groups. In this paper we provide a preliminary molecular interpretation of infrared spectra obtained by internal reflection spectroscopy of thin TiO2 colloid films immersed in aqueous solutions of widely ranging pH.

Langmuir, Vol. 15, No. 7, 1999 2403 Sampling Method. TiO2 sol-gel films were prepared by overnight drying in air of aqueous TiO2 sol onto the surface of a single reflection 45° ZnSe internal reflection prism (Harrick). A hemispherical glass chamber to contain the aqueous solutions was sealed to the TiO2-coated prism surface with an O-ring. The prism assembly was mounted in a Digilab FTS-60 spectrometer with a Prism Liquid Cell accessory (Harrick). All solutions were allowed to equilibrate with the TiO2 films for at least 20 min before spectra were collected. The spectrometer was purged with dried air. Variable-intensity bands at 2350 cm-1 and at ∼2900 cm-1, present in all spectra, are due to fluctuations in the levels of CO2(g) and trace hydrocarbons in the purge air, respectively. All spectra were collected using 64 scans at 4 cm-1 resolution. Internal Reflection Infrared Spectra of TiO2 Sol-Gel Films. Most of the spectra presented in the Results are differential absorbance spectra from a change in composition of the solution adjacent to the film, when the film is present for both the sample and background data collection. Any solution pH change induces surface chemical reactions which result in a change in surface charge and a concomitant change in the ionic distribution within the interfacial electrical double layer. Thus, the difference spectra may contain contributions from changes in any of these parts of the system. Any species that increases in concentration with solution composition change will have bands with positive absorbance, whereas any species that is lost will result in bands with negative absorbance. In general, the bulk metal oxide and the bulk solution can be assumed to make no contribution to the pH-induced spectral changes, provided these phases are stable over the pH range of interest and the solution concentrations are kept low. STIRS Solutions and Procedures. In the STIRS technique, where the infrared spectrum of interfacial groups is monitored during an acid/base titration of a solid film, surface charge is indicated by a surface excess concentration of infrared-active tetramethylammonium (TMA+) or perchlorate (ClO4-) ions.24 Both sets of STIRS data in this paper, which were based on TMAP and on the infrared inactive NaCl, were collected in the same manner. A series of solutions of constant ionic strength (5 × 10-3 mol dm-3) was produced from various concentrations of TMAOH/TMAP/HClO4 or NaOH/NaCl/HCl, to give a pH range from 11.7 to 2.3. A 5 × 10-3 mol dm-3 NaOH solution was used both as an initial surface wash and for a background spectrum. Other solutions, in order of decreasing pH, were then flowed over the film at 4.5 cm3 min-1 allowing a contact time of 20 min prior to the collection of spectra.

Materials and Methods Materials. Solutions were prepared using either 18 MΩ cm-1 MilliQ water or 99.9 atom % D2O (Aldrich). The D2O was distilled under nitrogen and stored in Nalgene HDPE bottles prior to use. DCl (Aldrich, 99.5% atom %) and NaOD (Aldrich, 99.9 atom %) were used in the D2O solutions. NaCl (Fluka), HCl (BDH), NaOH (Merck), tetramethylammonium perchlorate (TMAP) (ICN Pharmaceuticals), and tetramethylammonium hydroxide (TMAOH) (BDH) were used as received. Aqueous TiO2 sols were prepared by hydrolysis of TiCl4 (Aldrich) as previously described.22 The sols are substantially amorphous TiO2 with some crystalline anatase.31 (18) Biber, M. V.; Stumm, W. Environ. Sci. Technol. 1994, 28, 763. (19) Tejedor-Tejedor, M. I.; Anderson, M. A. Langmuir 1986, 2, 203. (20) Tunesi, S.; Anderson, M. A. Langmuir 1992, 8, 487. (21) Hug, S. J.; Sulzberger, B. Langmuir 1994, 10, 3587. (22) Connor, P. A.; Dobson, K. D.; McQuillan, A. J. Langmuir 1995, 11, 4193. (23) Duffy, N. W.; Dobson, K. D.; Gordon, K. C.; Robinson, B. H.; McQuillan, A. J. Chem. Phys. Lett. 1997, 266, 451. (24) Dobson, K. D.; Connor, P. A.; McQuillan, A. J. Langmuir 1997, 13, 2614. (25) Dobson, K. D.; McQuillan, A. J. Langmuir 1997, 13, 3392. (26) Awatani, T.; McQuillan, A. J. J. Phys. Chem. 1998, 102, 2B, 4110. (27) Kelso, G.; Peng, G.; McQuillan, A. J. To be published. (28) Bergstro¨m, P.-A.; Lindgren, J.; Kristiansson, O. J. Phys. Chem. 1991, 95, 8575. (29) K Ataka, K.; Yosuyanagi, T.; Osawa, M. J. Phys. Chem. 1996, 100, 10664. (30) Isawata, Y.; Xia, X. J. Electoanal. Chem. 1996, 41, 95. (31) Moser, J.; Gratzel, M. J. Am. Chem. Soc. 1983, 105, 6547.

Results Infrared Spectra of TiO2 Sol-Gel Films in Air. Figure 1a shows the internal reflection infrared spectrum of water recorded with the ZnSe internal reflection prism. This spectrum32 is dominated by the broad band between 3700 and 2900 cm-1 of the strongly absorbing hydroxyl stretching vibrations (νOH), the relatively sharp but weaker hydroxyl bending mode (δOH) at 1637 cm-1, and the librational modes which absorb strongly below 1000 cm-1. Figure 1b shows the infrared spectrum of a TiO2 solgel film recorded against a background spectrum of the ZnSe prism in air. By comparison with Figure 1a, this spectrum is dominated by the significant absorptions of bulk H2O trapped in the gel matrix. The most prominent of the additional absorptions is a band, with peak at 750 cm-1, which arises from the lattice vibrations33 of TiO2. Several weak bands are detectable in the 1500-1000 cm-1 region. Those at ∼1550, 1430, 1290, and 1058 cm-1 can be assigned as surface carbonate species by comparison with spectra of adsorbed carbonates on TiO213 and ZrO2 sol-gels.25 These are formed by the adsorption of atmo(32) Walfren, G. E. In Water-A Comprehensive Treatise; Franks, F., Ed.; Plenum Press: New York, 1972; Chapter 5. (33) Farmer, V. C., Ed. The Infrared Spectra of Minerals; Mineral Society: London, 1974.

2404 Langmuir, Vol. 15, No. 7, 1999

Connor et al.

Figure 2. Internal reflection infrared spectra of (a) D2O and (b) TiO2 film in D2O. The background spectra were of (a) the clean, dry prism and (b) D2O on a clean prism.

Figure 1. Internal reflection infrared spectra of (a) water, (b) TiO2 sol-gel film in air, and (c) TiO2 film in water. The background spectra for (a) and (b) were of the clean, dry prisms. The background spectrum for (c) was of water on a clean prism.

spheric CO2. Weak bands at 1121 and 1100 cm-1 are detectable and are also observed on dry ZrO2 sol-gel films.25 The spectrum of the same TiO2 film in water with respect to that of water alone on the ZnSe prism is shown in Figure 1c. This spectrum of the film under water exposes the underlying absorptions associated with the TiO2/H2O interface. This differential spectrum slightly overcompensates for the infrared absorption bands of bulk H2O, but this is barely evident. The spectrum in Figure 1c has similarities to that of Figure 1b but has some notable differences. The absorption in the νOH region has a different band shape from that of bulk H2O, with a prominent peak at 3100 cm-1 and a shoulder at 3500 cm-1. The removal of the bulk H2O bending mode absorption reveals a band at 1610 cm-1. A single band at 1385 cm-1 has replaced the 1430 cm-1 band of Figure 1b, while the three weak bands in the 1200-1000 cm-1 region are unchanged. The removal of bulk water absorptions has resolved a band at 890 cm-1 from the TiO2 lattice vibrations at 750 cm-1. Figure 2a shows the infrared spectrum of D2O on a bare ZnSe prism. The νOD band occurs between 2700 and 2100 cm-1, the δOD mode is at 1204 cm-1, and the librational absorptions appear below 800 cm-1. The bands in this spectrum are the same as those in Figure 1a, but are shifted due to the isotopic exchange. Note that the νOD region overlaps the absorption bands of gaseous CO2 about 2350 cm-1, leading to distortion of the νOD band shape. Figure 2b shows the spectrum of D2O on TiO2 with a background of D2O on the uncoated prism. The absence of absorptions in the νOH region indicates that all the

bulk water trapped in the film and any surface OH species on the TiO2 have exchanged with solution D2O within the time frame of the experiment. In the νOD region of Figure 2b, the absorption profile is similar to that of the νOH region of Figure 1c with some slight differences in band shapes and relative intensities. There are two components, at 2600 and 2530 cm-1, in the weaker band at higher wavenumbers. There is a weak negative band in Figure 2b at 1204 cm-1, the δOD mode of bulk D2O, due to the overcompensation for bulk D2O by the background spectrum. The corresponding loss in the νOD region is masked by the positive bands associated with the TiO2/D2O interface. The 1610 cm-1 band in Figure 1c appears to correspond to the absorption at 1623 cm-1 in Figure 2b. An increase in band wavenumber in this region upon D2O exchange has previously been observed for bidentate carbonate on ZrO2 due to the weaker hydrogen bonding in D2O.25 This observation suggests the association of the 1610 cm-1 band with adsorbed carbonate. A weak band at 1430 cm-1 in Figure 2b is associated with surface carbonate species, and the band lost at 890 cm-1 upon D2O exchange must arise from interfacial H2O absorptions. The remaining absorption bands in the Figure 2b spectrum appear to have been unaffected by D2O exchange. Infrared Spectra of TiO2 Films in NaOH-H2O Solution. Changing the composition of an aqueous solution in contact with the TiO2 film induces a surface perturbation which can be readily monitored using FTIR spectroscopy. This is shown in the Figure 3 spectrum which was obtained from a 5 × 10-3 mol dm-3 aqueous NaOH solution on the film with a background spectrum of water on the film. The Figure 3 spectrum is dominated by a prominent negative band at 3150 cm-1 in the νOH region, which must be partly due to deprotonation of surface OH groups but may also contain some contribution from the surface excess of solvated Na+ ions displacing water in the electrical double layer of the more negatively charged metal oxide. The similarity of the 3150 cm-1 band with that in Figure 1c, and difference from the bulk water absorption in this region, suggest that deprotonation of surface OH groups plays a significant part in this chemical change. The more alkaline solution has also resulted in

TiO2/Aqueous Solution Interface

Figure 3. Internal reflection infrared spectrum of a 5 × 10-3 M aqueous NaOH solution on TiO2 film. The background spectrum was of H2O on TiO2 film.

Figure 4. STIRS spectra from solutions of different pH, but constant 5 × 10-3 mol dm-3 ionic strength, containing TMA+ and ClO4- ions on a TiO2 film. The background spectrum was from the TiO2 film after washing with 5 × 10-3 mol dm-3 aqueous NaOH solution. Spectra are offset from zero absorbance for clarity. The arrow indicates solution sequence.

additional discrete positive absorptions at 3600 and 3480 cm-1 and loss of an absorption at 927 cm-1. Significantly, the change from neutral to alkaline solution has also resulted in absorption losses at 1610, 1385, and 1058 cm-1 due to displacement of surface carbonate species25 and at 1121 cm-1 due to removal of some unknown adsorbate. A similar procedure has been used to clean ZrO2 film surfaces.25 The alkaline wash procedure was therefore employed to clean TiO2 film surfaces in the subsequent work. pH Dependence of Infrared Spectra of the TiO2/ H2O Interface: STIRS Experiment. The initial report of the STIRS technique was based on the TiO2/H2O interface.24 However, only the 1550-1000 cm-1 spectral range was discussed where the principal absorption maxima of TMA+ and ClO4- are observed. Figure 4 shows the previously reported STIRS spectra24 but extended to the 4000-800 cm-1 range. The pH ) 11.7 spectrum is that due to 5 × 10-3 M TMAOH solution on the film with respect to a background of 5 × 10-3 M NaOH solution on the film. The enhanced TMA+ absorptions at 1487, 1419, and 950 cm-1 indicate a net negative surface charge on

Langmuir, Vol. 15, No. 7, 1999 2405

the TiO2 film at pH g 6.8 while the enhancement of the ClO4- absorption at 1103 cm-1 indicates net positive surface charge at pH e 4.3. Zero charge occurs at pH ≈ 5.24 The Figure 4 spectra show additional pH-dependent features. Apart from the enhanced TMA+ positive absorptions there are negative absorptions at 3600, 3450, and 1645 cm-1. These negative absorptions appear to largely arise from loss of interfacial H2O caused by the displacement of hydrated Na+ by the more bulky hydrated TMA+. In the subsequent Figure 4 spectra, at decreasing pH, there is an emergence and progressive increase of a positive absorption at 1623 cm-1 in the δOH water region. There is also a growth of the νOH absorption with a broadening to lower wavenumbers. In the νOH region the growth of an indistinct absorption at about 3400 cm-1 occurs from high to lower pH to reduce the negative 3450 and 3600 cm-1 absorptions arising from the interfacial TMA+. This band passes through an absorbance maximum at pH about 7. For pH e 10.7 a band at 3200 cm-1 becomes evident and appears to parallel the behavior of the 1623 cm-1 band. At pH e 4.3 a broad band at ∼3000 cm-1 emerges and appears to be accompanied by the appearance of an absorption at 927 cm-1. Minor negative absorptions are evident in most spectra at 3600 and at 3500 cm-1. The above STIRS experiment was also carried out on the bare ZnSe prism in the absence of the TiO2 film. The results showed insignificant changes with pH in comparison with those in Figure 4. A STIRS experiment carried out in the reverse solution order to that above, from acidic to alkaline pH, gave identical results and confirmed the stability of the TiO2 film over the experimental pH range. At pH ) 2.3 a sharper absorption emerges at 3580 cm-1. This absorption corresponds with the increase in ClO4absorption at 1103 cm-1 and is due to the influence on the water spectrum of the enhanced perchlorate concentration at the interface. We have observed a similar feature in the infrared spectra of concentrated aqueous perchlorate salt solutions, coupled with an absorbance loss at lower wavenumbers in the νOH region. This νOH absorbance decrease at low pH is masked by other positive bands in Figure 4. Such spectral bands have also been observed at positively charged electrodes in aqueous perchlorate solutions.34 To test the influence of electrolyte ions on the above spectra, we have recorded STIRS spectra where the TMA+ and ClO4- ions have been replaced by the infrared-inactive Na+ and Cl- ions at the same concentrations. Na+ ions are expected to have a smaller effect on water structure34 than the larger TMA+ ions. These STIRS spectra are shown in Figure 5 and have features which are generally similar to those in Figure 4. A 5 × 10-3 M NaOH solution initially on the TiO2 film provides the background spectrum for all the spectra. As in Figure 4, the spectra show a progressive increase in absorbance of a band at 1623 cm-1 with decreasing pH. This band appears to be accompanied by a growing absorption at about 3200 cm-1. A band at about 3400 cm-1 appears and grows at high pH but becomes lower in intensity again at low pH. An absorption at 927 cm-1 which appears at pH e 4.3 appears to correlate with the growth of a broad absorption around 3000 cm-1. The absence of the 1103 cm-1 ClO4- signal in Figure 5, compared with Figure 4, allows the minor peaks at 1121 and 1058 cm-1 to become more evident at low pH. The ∼3600 cm-1 band apparent at low pH in Figure 4, arising from the enhanced interfacial perchlorate concentration, is not observed with the chloride ion. However (34) Kitamura, F.; Ohsaka, T.; Tokuda, K. J. Electroanal. Chem. 1996, 412, 183.

2406 Langmuir, Vol. 15, No. 7, 1999

Figure 5. STIRS spectra from solutions of different pH, but constant 5 × 10-3 mol dm-3 ionic strength, containing Na+ and Cl- ions on a TiO2 film. The background spectrum was from the TiO2 film after washing with 5 × 10-3 mol dm-3 aqueous NaOH solution. Spectra are offset from zero absorbance for clarity. The arrow indicates solution sequence.

Figure 6. Internal reflection infrared spectra of (a) 5 × 10-3 mol dm-3 NaOD/D2O solution on TiO2 film, (b) 5 × 10-3 mol dm-3 NaCl/D2O solution on TiO2 film, and (c) 5 × 10-3 mol dm-3 DCl/D2O solution on TiO2 film. Background spectra were of (a) D2O on TiO2 film and (b, c) 5 × 10-3 mol dm-3 NaOD solution on TiO2 film. Spectra are offset from zero absorbance for clarity. The dashed lines correspond to regions where uncompensated CO2(g) absorptions have been deleted.

a small negative band at 3600 cm-1 is detectable in the Figure 5 spectra collected at pH < 6.7. pD Dependence of Infrared Spectra of TiO2 SolGel Films in D2O. Experiments were carried out using solutions of NaOD and of DCl in D2O. Figure 6a shows the infrared spectrum of 5 × 10-3 M NaOD in D2O solution on a TiO2 film with a background spectrum of 5 × 10-3 M NaCl in D2O on the film. This spectrum is the D2O analogue of the H2O-based spectrum shown in Figure 3 and indicates the changes from neutral to alkaline conditions. In the νOD region there is a broad absorption loss at about 2300 cm-1 but also gain of a weaker 2660, 2580 cm-1 doublet absorption. Comparison with the corresponding νOH data in Figure 3 shows generally analogous features taking into account the isotopic shifts. The Figure 6a spectrum also shows the alkaline conditions have removed carbonates and other surface species by the absorption losses at 1610, 1385, 1121, and 1058 cm-1. Also evident is a negative band at 1193 cm-1, correspond-

Connor et al.

ing to a shift of the 1623 cm-1 band in Figure 5, which suggests a D2O surface species. A further negative band at 920 cm-1 appears to correspond to the 927 cm-1 band in Figure 3. Figure 6b shows the IR spectrum of a 5 × 10-3 M NaCl solution in D2O, using the film under NaOD (Figure 6a spectrum) as the background. This spectrum exhibits the changes occurring on the film with a change from pD ) 11.7 to about pD ) 7 and is thus the analogue of the pH 6.7 spectrum in Figure 5. This spectrum is similar to the Figure 6a spectrum but with most of the bands appearing with opposite polarity. A strong νOD absorption, a sharper band at 1193 cm-1 and a band at 920 cm-1 are present. These three bands are the same as those lost in Figure 6a. The 1600, 1058, and 1385 cm-1 carbonate bands and the 1121 cm-1 band do not reappear, leaving the pD reversible bands. This spectrum has strong similarities to the pH ) 6.7 spectrum of Figure 5 allowing for the isotope shift. The absorption profile in the νOD region is similar to that of the νOH region in Figure 5. The 1623 cm-1 band apparent in Figure 4 has shifted down to 1193 cm-1, which is consistent with the shift of the bending mode of a surface water. The 920 cm-1 band of Figure 6b most likely arises from the same species that forms the 927 cm-1 band at low pH in Figure 5. This shift of 7 cm-1 on deuteration is too small for a δTi-OH or νTi-OH vibration but is still a pH (pD) dependent mode. The Figure 6c spectrum corresponds to the Figure 5, pH ) 2.3, spectrum but with 5 × 10-3 M DCl solution in D2O on the film. The spectrum shows the same bands as the Figure 6b spectrum, with increases in size of the νOD band and the 1193 and 920 cm-1 bands, as would be expected with increasing surface protonation. The significant positive surface charge on the TiO2 at this pD has also lead to the reappearance of the bands at 1610, 1385, 1121, and 1058 cm-1. While the pD decreases from neutral conditions (Figure 6b to Figure 6c), the νOD absorption band increases in intensity and significantly broadens to lower wavenumbers. All the bands in the spectra in Figure 6c can be correlated with bands in the Figure 5, pH ) 2.3, spectrum, and this confirms the association of these absorptions with OH groups. Discussion The internal reflection infrared spectra of metal oxide colloid films immersed in aqueous electrolyte solutions contain contributions from several parts of the system: (i) the bulk metal oxide; (ii) its surface groups, both charged and uncharged; (iii) the interfacial double layer ions, their associated water molecules of solvation, and solvent water; (iv) the bulk solution beyond the double layer. The evanescent wave of the internal reflection decays exponentially with distance from the ZnSe surface into the metal oxide/aqueous phase with a penetration depth35 about the thickness of the colloid film. For the ZnSe/metal oxide film/aqueous solution systems used in this work the contributions to the differential spectra are greatest from ii, significant from iii and usually negligible from i and iv. The simplest spectra are those of the dry film, Figure 1b, and of wet films with H2O or D2O backgrounds, Figures 1c and 2b, respectively. Common to all three spectra are bands at 1610, 1430 or 1385, 1121, 1058, and the large 750 cm-1 TiO2 lattice absorption. The wavenumbers of the other bands are too high to be from TiO2 absorptions. These bands do not significantly shift on deuteration and therefore cannot be associated with a species containing exchangeable protons. A band at 1385 cm-1 corresponds (35) Mirabella, F. M. Appl. Spectrosc. Rev. 1985, 21, 45.

TiO2/Aqueous Solution Interface

for other metal oxides to that of adsorbed carbonate25 which also has an absorption about 1600 cm-1. The 1121 cm-1 band is unassigned and appears to be due to a nonprotonated, anionic species. The profile of the νOH region of Figure 1c is quite distinctive, and a similar profile occurs in the νOD region of Figure 2b.The shift observed for OH absorptions from Figure 1c to Figure 2b is in proportion to the shifts of the νOH and δOH bands of H2O on deuteration. These bands are not expected to be dominated by absorptions of bulk water species because solvent absorptions have been subtracted. The absorptions must be largely due to hydroxide species on the metal oxide surface. Surfacebound water (-OH2) would be expected to have δOH absorptions about 1600 cm-1 which would overlap with the 1610 cm-1 carbonate absorption. However, there is no corresponding absorption in the δOD region of Figure 2b. There do not appear to be significant contributions to the νOH and νOD bands in Figures 1c and 2b from surfacebound water. Changes with pH in the infrared spectra of hydrous metal oxides should reveal the molecular basis of surface charge. A change from neutral to alkaline conditions is expected to remove protons from some surface groups, and this can be seen in the significant absorption loss in the νOH region of Figure 3. The small positive bands seen at 3600 and 3480 cm-1 of Figure 3 may arise from hydroxyls weakly bound to the surface. Alternatively, these bands may be due to interfacial water perturbed by the high negative surface charge.29 Dipolar water molecules may have electric field dependent orientation. A negatively charged surface causes water molecules to adopt a proton down configuration.29 This orientation would prevent H-bonding to other water molecules, giving rise to sharper, higher wavenumber νOH absorptions. The difference spectra resulting from stepwise pH changes show distinctive νOH region features which are largely due to changes in surface hydroxyl species. In the STIRS spectra of Figures 4 and 5 there is the appearance at high pH of a sharp absorption at 1623 cm-1 which grows as the pH is lowered, to reach an absorbance plateau from about pH ) 5, near the TiO2 PZC.24 This is the only pHdependent absorption in the δOH region apart from those of carbonate species. The half-height bandwidth of the bulk water absorption at 1637 cm-1 is 80 cm-1 while that of the 1623 cm-1 peak is 50 cm-1. Thus these absorptions arise from different species. The lower wavenumber and smaller bandwidth of the 1623 cm-1 band compared with bulk water is similar to that observed at 1612 cm-1 (bandwidth of 30 cm-1) by Ataka et al.29 for water at a gold electrode at low potentials. They assigned the 1612 cm-1 band to a water molecule bound to the electrode surface via the O atom, with the protons in unfavorable positions for H-bonding. A weaker H-bonding results in a decrease in δOH wavenumber and is usually accompanied by an increase in the νOH wavenumber.29 However, the 1623 cm-1 band appears to be coupled with the 3200 cm-1 band which is similar to the νOH of bulk water. The decrease in wavenumber of the δOH mode has also been attributed to the changing HOH angle on coordination of the O atom to a surface.30 The 1623 cm-1 band in this case is therefore attributed to a surface coordinated water species i.e., Ti-OH2. The narrow bandwidth of this band indicates that this is more weakly hydrogen bonded than bulk water, as a coordinated water molecule can form fewer H-bonds with adjacent water molecules. The 1623 cm-1 band also shows the expected shift for the deuteration of a water species. This peak is present over a wide pH range and does not appear

Langmuir, Vol. 15, No. 7, 1999 2407

to be closely related to the strong pH dependence of surface charge.24 The narrow bandwidth of the bending mode absorption, and the comparatively narrow corresponding stretching absorption, suggest that the species is not positively charged. This evidence favors assignment of the 1623 cm-1 absorption to a water molecule associated with an uncharged surface site.5,11 The pH dependence of the broad νOH region absorption shown in Figures 4 and 5 appears to indicate the existence of three overlapping absorptions, at about 3400, 3200, and 3000 cm-1. Growth with decreasing pH of the 3200 cm-1 band appears to correlate well with the 1623 cm-1 band and is therefore assigned to the Ti-OH2 group. At pH e 4.3, a band grows in about 3000 cm-1 and its pH dependence appears to correlate with the 927 cm-1 band. The 3000 cm-1 band is much broader than the other bands in the νOH region indicating a greater degree of hydrogen bonding in the species. The νOH absorption in the infrared spectrum of H3O+ is similarly broad and is shifted to lower wavenumbers with respect to that of bulk water.36 This evidence suggests that the 3000 cm-1 band is a positively charged species. The 927 cm-1 band is pH dependent but does not give a significant shift on deuteration. A possible origin for this absorption is the antisymmetric Ti-O stretching mode of the bridging Ti-OH+-Ti species. The wavenumber of this mode is not expected to shift greatly on deuteron exchange. The Ti-O-Ti antisymmetric stretch occurs about 900-800 cm-1 in the infrared spectra of oxo-bridged titanium(IV) compounds.37,38 The proton of the Ti-OH+-Ti species is expected to be weakly bound, and this correlates well with the νOH vibrations of this species being detected at the low wavenumber end of the νOH envelope. Protonation of the bridging oxygen in the Ti-O-Ti unit may affect the dipolar character of this mode, making the absorption more intense. At high to neutral pH there is growth of an absorption at about 3400 cm-1 which reaches a maximum about pH ) 9 and then all but disappears by pH ) 4.3. The pH dependence of this band does not seem to correlate with the pH variation of any other bands in the spectrum. The lack of related bands, and its growth and then diminution as the pH decreases, suggest assignment of the 3400 cm-1 band to the Ti-OH group. This group would be formed by protonation of Ti-O-, with the subsequent absorbance loss due to formation of Ti-OH2+ (see Scheme 1). The Ti-OH2+ species would be expected to be strongly Hbonded with relatively weak O-H bonds. The corresponding νOH band would be expected to occur at the lower end of this region and may therefore overlap with that of the Ti-OH+-Ti species. However, the broadness of the overlapping bands in this region prevents their clear discrimination. No corresponding δOH absorption has been detected but would be expected to be broad and at ∼1650 cm-1. The spectra from experiments with D2O shown in Figure 6b,c exhibit the same pH-dependent band shape as those in Figure 5. Initially, on decrease of pD, the νOD band rises at high wavenumbers with the formation of Ti-OD and Ti-OD2. Whereas at even lower pD the Ti-OD+-Ti and Ti-OD2+ species form with the appearance of a band at ∼ 2240 cm-1. These bands show the expected shift on deuteration from those of Figure 5. No corresponding δOD absorptions were observed. A summary of the above tentative assignments of the infrared absorptions, which are observed to change with (36) Giguere, P. A.; Madec, C. Chem. Phys. Lett. 1979, 37, 569. (37) Clark, R. J. H. The Chemistry of Titanium and Vanadium; Elsevier: Amsterdam, 1968. (38) Okuda, J.; Herdtweck, E. Inorg. Chem. 1990, 30, 1516.

2408 Langmuir, Vol. 15, No. 7, 1999

Connor et al.

Table 1. Assignments of Infrared Absorptions of Hydrous TiO2 Surface Groups wavenumber/cm-1 species Ti-OH

pH range

Ti-OH2

10.7-4.3 (max ∼8) e10.7

Ti-OH2+ Ti-OH+-Ti