J. Phys. Chem. B 2006, 110, 13453-13457
13453
Probe Microscope Observation of Platinum Atoms Deposited on the TiO2(110)-(1 × 1) Surface Akira Sasahara,*,†,‡ Chi Lun Pang,‡,§ and Hiroshi Onishi‡ Japan Science and Technology Agency, Kawaguchi, Saitama 332-0012, Japan, and Department of Chemistry, Faculty of Science, Kobe UniVersity, Nada-ku, Kobe 657-8501, Japan ReceiVed: March 31, 2006; In Final Form: May 14, 2006
Titanium dioxide (TiO2) (110) surfaces with Pt adatoms were examined using a noncontact atomic force microscope (NC-AFM) and a Kelvin probe force microscope (KPFM). Topographic images with NC-AFM identify Pt atoms adsorbed at three different sites. These sites are on the Ti atom rows, on the O atom rows, and in O atom vacancies. Most Pt adatoms were observed on Ti atom rows. Successively recorded images show that the Pt adatoms on Ti atom rows (adatoms A) and O atom rows (adatoms C) are mobile while the adatoms in the O atom vacancies (adatoms B) are not. Adatoms A and adatoms B were identified in KPFM images. However, adatoms C were not visualized in KPFM images because they moved quickly or were swept out by the tip. The KPFM measurements reveal that the work function on adatoms A are lower than that on the surrounding (1 × 1) surface by 0.24 eV whereas adatoms B reduced the work function by 0.26 eV. The work function decrease is interpreted with an electric dipole moment directed toward the vacuum, as a result of electron transfer from the adatoms to the TiO2 substrate. In an O atom vacancy, the adatom B is in contact with two Ti atoms and therefore the electron transfer can be enhanced.
Introduction Platinum-loaded titanium dioxide (TiO2) is an important metal-on-oxide system, finding technological applications as catalysts and gas sensors.1 Recently, the Pt/TiO2 system has attracted further attention due to its photocatalytic activity toward water cleavage.2 To clarify the origin of its chemical performance, model Pt/TiO2 systems consisting of Pt-evaporated TiO2 single-crystal surfaces have been investigated with surfacesensitive analytical methods.3 The (110) face of rutile TiO2 is the most commonly used substrate for such model Pt/TiO2 systems, due to its well-defined structure. Figure 1 shows a ball model of the nonreconstructed (1 × 1) structure of the (110) plane obtained by cleaning in ultrahigh vacuum. Bridging O atoms form rows parallel to the [001] direction and rows of Ti atoms coordinated to five O atoms lie between these O atom rows. The O atom rows inevitably include O atom vacancies formed during vacuum annealing. When Pt is evaporated onto the (1 × 1) surface at room temperature, three-dimensional clusters are formed.4-7 The Pt clusters grow by vacuum annealing, eventually becoming encapsulated by a reduced titanium oxide layer.8,9 Compared to the Pt clusters, little is known about Pt atoms adsorbed on TiO2 surfaces. The adsorption of atoms is the first stage in making an interface between two materials and is therefore a critical step in determining the properties of clusters and/or films. Preferential adsorption of Pt atoms on 5-foldcoordinated Ti atoms has been proposed on the basis of the observation that the intensity of the Ti 2p peak decreases more * To whom correspondence should be addressed. E-mail: sasahara@ kobe-u.ac.jp. † Japan Science and Technology Agency. ‡ Kobe University. § Present address: London Centre for Nanotechnology and Department of Chemistry, University College London, 20 Gordon Street, London, U.K.
Figure 1. Ball model of the TiO2(110)-(1 × 1) surface. Small and large spheres represent Ti and O atoms, respectively. The O atoms are shaded according to their depth.
rapidly than that of the O 1s peak at the initial stage of Pt evaporation.10 Self-consistent field calculations predict that Pt atoms are negatively charged on a (1 × 1) surfacelike Ti4O1616cluster.11 This is consistent with calculations using a twodimensional periodic model which also produce a net negative charge for Pt adatoms.12 On the other hand, an ab initio total energy calculation predicts the most favorable adsorption site for a neutral Pt atom to be a hollow site between one bridge O atom and two in-plane O atoms.13 In this study, Pt adatoms adsorbed on TiO2(110) were examined using a noncontact atomic force microscope (NCAFM)14 and Kelvin probe force microscope (KPFM).15 The Pt adatoms were individually visualized, and electron transfer at the Pt/TiO2 interface is discussed on the basis of the local work function perturbation by the adatoms. The NC-AFM is a probe microscope capable of imaging surfaces with atomic-scale spatial resolution.14 A cantilever with a tip at one end is used
10.1021/jp062000c CCC: $33.50 © 2006 American Chemical Society Published on Web 06/20/2006
13454 J. Phys. Chem. B, Vol. 110, No. 27, 2006
Sasahara et al.
as the probe. The cantilever is approached to the sample surface while oscillating at its resonant frequency (f0). Weak attractive forces such as van der Waals forces, electrostatic forces, and chemical bonding forces between the tip and sample cause a shift of the oscillation frequency (∆f). By regulation of ∆f, the tip-sample distance can also be regulated. The KPFM is a modification of NC-AFM, providing the lateral distribution of the local work function of the sample surface simultaneously with topography. Work function maps are obtained by recording the contact potential difference at each point, using the cantilever like the reference electrode of a macroscopic Kelvin probe.16 Work function maps with atomic-scale contrast have been obtained on Sb/Si(111),17 Si(111)-(7 × 7),18 Au/Si(111),19 and Na/TiO2(110)20 surfaces. Experimental Section Our experiments were carried out by using an ultrahigh vacuum microscope (JSPM-4500A, JEOL) equipped with a KPFM driver (TM-Z50241, JEOL), Ar+ sputtering gun (IG35, OCI), and low-energy electron diffraction optics (BDL600, OCI). The base pressure of the microscope chamber was 2 × 10-8 Pa. Conductive Si cantilevers (NSC12, Mikro Masch) with f0 of ∼300 kHz and spring constants of ∼14 N/m were used. In the NC-AFM mode, the sample bias voltage (Vs) was adjusted so that high contrast images could be obtained with a minimum ∆f. During KPFM measurements, Vs was modulated with an ac voltage (2 V, 2 kHz). The scanning speed for the KPFM measurement was 0.9 or 1.7 s/line, and the image contrast was independent of the scanning speed. A TiO2(110) wafer (7 × 1 × 0.3 mm3, Shinko-sha) was cleaned by cycles of Ar+ sputtering and annealing at 900 K giving the (1 × 1) surface. After being cooled to room temperature, the sputter-annealed surface was exposed to a resistively heated Pt wire (99.98%, Nilaco), which was outgassed beforehand. The distance between the sample wafer and Pt source was about 60 mm, and the increase of pressure during evaporation was less than 8.0 × 10-8 Pa. The amount of Pt deposited was controlled by changing the exposure time. All images were presented without filtering or smoothing. Cross sections were obtained from images filtered with a nine point median operation. Results and Discussions Figure 2 shows the NC-AFM topography of TiO2(110)-(1 × 1) surfaces. The image of the clean surface in Figure 2a consists of atomically flat terraces separated by monatomic steps with the O atom rows observable as bright lines running along the [001] direction. The depressions in the O atom rows are attributed to vacancies formed during annealing.21 Figure 2b-d shows images of the (1 × 1) surfaces following exposure to the heated Pt wire. After exposure for 10 s, bright spots with an atom size appeared as shown in Figure 2b. Four particles with diameters of 2-3 nm can also be seen. When the exposure time was extended to 30 s, the number density of the atomsized spots increased to 0.38 nm-2 as shown in the image of Figure 2c. Five nanometer-sized particles can also be seen in Figure 2c. The image in Figure 2d was obtained after exposure for 30 min. The number density of the atom-sized spots decreased to 0.13 nm-2 while the number of nanometer-sized particles increased to 24. From these results, we assign the atomsized spots and nanometer-sized particles to Pt adatoms and Pt clusters, respectively. The clusters are nucleated on the (110) surface at elevated temperatures.6 The surface was possibly
Figure 2. NC-AFM images of the TiO2(110)-(1 × 1) surface exposed to Pt source for (a) 0, (b) 10, (c) 30, and (d) 1800 s. Some of the O atom vacancies are marked by circles in (a). White and black arrows in (b)-(d) denote Pt adatoms and Pt clusters, respectively. Image size: 20 × 20 nm2. ∆f: (a) -84 Hz; (b) -49 Hz; (c) -84 Hz; (d) -121 Hz. Vs: (a) +0.8 V; (b) +1.0 V; (c) +0.8 V; (d) +0.8 V. Peak-to-peak amplitude of the cantilever oscillation (Ap-p): (a) 6.8 nm; (b) 6.8 nm; (c) 6.8 nm; (d) 6.0 nm.
warmed during 30 min of exposure to the heated wire, and the nucleation was accelerated. Figure 3a shows a narrow-scan NC-AFM image of the surface shown in Figure 2c. Pt adatoms and O atom rows are readily observable. Most Pt atoms are located between the O atom rows and have heights greater than the O atom rows by 0.13 ( 0.01 nm. We will refer to such Pt atoms as adatoms A. Adatoms A are represented as white circles in the schematic model shown in Figure 3b. They lie between two neighboring O atom rows (solid lines) with a symmetric shape in the [11h0] direction, as highlighted in cross section 1 in Figure 3c. We therefore conclude that Pt atoms readily adsorb on the Ti atom rows. Although individual O atoms in the rows are not resolved, we can estimate the position of each bridging O atom from the position of O atom vacancies. The number of removed O atoms was determined from the size of the vacancies along the O atom rows. One-atom vacancy and two-atom vacancies are observed in image a. The intersections of the solid and dotted lines in the schematic in Figure 3b correspond to the center of the bridging O atoms. The centers of adatoms A lie on the dotted lines, which indicated that they sit in a 4-fold hollow site of the in-plane O atoms (or atop the 5-fold-coordinated Ti atoms). This site differs from the preferred adsorption site theoretically predicted for a neutral Pt adatom.13 As mentioned later, our KPFM results indicate an electron transfer from Pt adatoms to TiO2 substrate. The charge of the Pt atoms is probably responsible for the disagreement between our experimental results and the theoretical prediction. The image in Figure 3a shows a minority of adatoms located on O atom rows so we identify two additional types of adatoms. One type has image heights equal to adatoms A, being higher than the O atom rows by 0.13 ( 0.01 nm. They are represented as gray circles in Figure 3b and labeled “B”. The other type of adatom is taller than the O atom rows by 0.16 ( 0.01 nm and labeled “C” in Figure 3b. From the image heights, we deduce that adatoms B and C lie in the O atom vacancies and on the O atom rows, respectively. These three adsorption sites are shown
Pt Atoms Deposited on the TiO2(110)-(1 × 1) Surface
Figure 3. (a) NC-AFM image of the TiO2(110)-(1 × 1) surface exposed to the Pt source for 30 s. Removed bridging O atoms are marked by arrowheads. (b) Model of the image (a). Solid lines along the [001] direction are O atom rows, and dotted lines pass the center of the bridging O atoms and in-plane Ti atoms. Adatoms A-C are presented by white, gray, and black circles, respectively. (c) Cross sections along the broken lines in (b). Key: (1) adatom A; (2) adatom B; (3) adatom C. (d) Ball model of the TiO2(110)-(1 × 1) surface with the Pt adatoms. (e-g) NC-AFM images obtained after the image (a). Image size: 10 × 10 nm2. ∆f: (a) -65 Hz; (e-g) -62 Hz. Vs: (a, e-g) +0.8 V. Ap-p: (a, e-g) 6.6 nm.
in the model in Figure 3d. Four O vacancies and three adatoms B are observable in Figure 3a. Assuming that one adatom B occupies one O vacancy, the number of oxygen vacancies in Figure 3a is 0.07 nm-2, comparable to the number density of O vacancies in Figure 2a, 0.06 nm-2. Hence, our assignment of adatoms B to Pt atoms adsorbed in vacancies is reasonable. The schematic in Figure 3b shows that two of the adatoms B lie on the dotted lines with the other adatom B sitting between two dotted lines. The adatoms B therefore have two possible adsorption sites in the vacancies, either on one Ti atom or between two Ti atoms. The adatoms C do not show any specific position relative to the dotted lines in Figure 3d. Water from the residual vacuum has been shown to dissociate in oxygen vacancies forming pairs of H atoms adsorbed on a bridging O atom.22 Although there is evidence that NC-AFM can image H atoms as bright spots on the bright O atom rows,14 we rule this out as an explanation for adatoms B and C by comparing the density of adatoms B and C with the typical number density of H atoms on the (1 × 1) surface. STM images show that our
J. Phys. Chem. B, Vol. 110, No. 27, 2006 13455 recipe for preparing the (1 × 1) surface gives a number density of ∼0.36 nm-2,20 which is six times higher than the density of adatoms on the O atom rows observed in our images of the Pt-evaporated surfaces. The images in Figure 3e-g were acquired successively following the first image in Figure 3a. The circles in each image show the positions of each Pt adatom in the preceding image. It is clear that some of the adatoms have changed their positions. Some 12% of the adatoms A, 0% of the adatoms B, and 67% of the adatoms C changed their positions between the images in Figure 3a,e. The percentages for migration of adatoms A-C are 16, 0, and 0% between the images in Figure 3e,f and 32, 0, and 67% between the images in Figure 3f,g. Thus, adatoms A and C are both mobile whereas adatoms B are not. Furthermore, our measurements show that adatoms C are more mobile than adatoms A. After adatoms C migrate, O vacancies are not observed in the positions vacated. This supports our assignment of adatoms C to Pt atoms adsorbed on the O atom rows, as opposed to vacancies. Figure 4a,b shows a simultaneously recorded topography image and a work function map of the Pt-evaporated surface, respectively. In the topographic image, both the adatoms and O atom rows are resolved. The scratch noises and incomplete spots are presumably due to adatoms migrating during the scanning. As shown in the histogram in Figure 4e, the height difference between the adatoms A and the O atom rows is 0.09 nm, which is smaller than that taken from the NC-AFM images shown previously. The reduced image height difference is attributed to cancellation of the electrostatic force during KPFM measurement. The electrostatic force cannot be completely canceled during NC-AFM measurement with a constant Vs.23 The long-range electrostatic force might be larger on the Pt adatoms. The tip was more separated from the adatoms than the bare TiO2 surface to keep the ∆f constant by reducing the contribution of the electrostatic force. Some adatoms were observed on the O atom rows with a height greater than the O atom rows by 0.10 nm. We assign the spots on the O atom rows to adatoms B because their heights are comparable to those of adatoms A. Spots assignable to adatoms C cannot be observed in the topographic images recorded during KPFM measurement. One possible reason that adatoms C cannot be visualized during KPFM measurement is their quick migration. The acquisition time for the images in Figure 4 is 463 s compared to only 28 s for the images in Figure 3. It is likely that the mobile adatoms C were not imaged in Figure 4 even though they existed on the surface. Another possible reason for not imaging the adatoms C is because they are swept out by the tip. The adatoms C may be less stable due to a weaker coordination to surface atoms and therefore may be easily removed by interaction with the tip end. The contrast of the work function map in Figure 4b is presented so that areas with larger work functions are shown brighter. The positions of the Pt adatoms are darker than the surrounding TiO2 surface, indicating a work function decrease on the Pt adatoms. As shown in the histogram in Figure 4f for 71 adatoms, the most probable work function decrease lies in the section between 0.24 and 0.26 eV. The distribution for adatoms B, which are represented by the hatched areas, is centered at a larger work function decrease, between 0.26 and 0.28 eV. The cross sections in Figure 4d show that the work function on the Ti atom rows is larger than the O atom rows by 0.03 eV. In our experiments, the work function above the Pt adatoms was always smaller than that of surrounding TiO2 surface, even when the contrast between the O atom rows and
13456 J. Phys. Chem. B, Vol. 110, No. 27, 2006
Sasahara et al. employing models with less restriction on surface atom geometry. For example, atomic relaxation accompanied by electron transfer has been predicted for adsorption of K on the TiO2(100) surface by using ab initio calculations.27 The difference in work function decrease between adatoms A and B is presumably due to the number of Ti atoms which receive the electrons. Electrons from adatoms A are received by one Ti atom which is originally coordinated to five O atoms. When one bridging O atom is removed, two originally 6-foldcoordinated Ti atoms lose one O ligand and are exposed as 5-fold-coordinated Ti atoms. Titanium atoms coordinated to four O atoms appear when two bridging O atoms are removed. The Ti atoms exposed in the O vacancies are less positively charged than the 5-fold-coordinated Ti atoms because the removed O atoms leave electrons. However, it is geometrically possible for an adatom B to have two neighboring Ti atoms, which may enhance the electron transfer. To estimate the extent of the electron transfer, a quantitative treatment of the local work function which takes into account the microscopic shape of the tip end must be undertaken.28 Such an analysis represents an interesting challenge for future work. Summary
Figure 4. Simultaneously obtained (a) topography and (b) work function map of the Pt-evaporated TiO2 surface. (c) Model of the surface (a). Solid lines along the [001] directions are O atom rows. Adatom A and B are presented by white and gray circles, respectively. (d) Cross sections along the broken lines in (c). Distribution of the (e) heights from the O atom rows and (f) work function decrease on the Pt adatoms. White and gray part shows the contribution of adatoms A and B, respectively. Image size: 10 × 10 nm2. ∆f: -78 Hz. Ap-p: 6.6 nm. Scanning speed: 0.9 s/line.
Ti atom rows was sometimes reversed in the work function map. Atom-scale contrast in work function maps include contributions from tip-sample interactions.18,24 Some interaction between the unoccupied 3d orbital of the 5-fold-coordinated Ti atoms and a dangling bond of Si atom at the tip end may be responsible for the contrast between the O atom rows and Ti atom rows in the image in Figure 4b. We assume that the work function is locally perturbed at adatoms by an electric dipole moment. A dipole moment directed from the substrate to the vacuum decreases work function locally. Therefore, our results indicate an electron transfer from the Pt adatoms to the TiO2 substrate. The positively charged 5-fold-coordinated Ti atoms receive electrons from the initially neutral Pt atoms. Evidence for electron transfer from Pt to Ti atoms at the interface between nanometer-sized Pt clusters and a TiO2 support has been provided by electron paramagnetic resonance studies.25,26 Theoretical calculations, however, predict negative charges on the Pt atoms because electron transfer from bridging O to Pt atoms dominates over the transfer from Pt to 5-fold-coordinated Ti atoms.12 Experimentally observed electron transfer would be reproduced by
On the TiO2(110)-(1 × 1) surface, Pt atoms were found in three possible adsorption sites, on the Ti atom rows (adatoms A), in the O vacancies (adatoms B), and on the O atom rows (adatoms C). Adatoms A and B were mobile, while adatoms C did not migrate. The work function was locally decreased on the Pt adatoms, indicating electron transfer from the Pt adatoms to the TiO2 substrate. The adatoms in the O vacancies had work function decreases of 0.28 eV, which was a greater decrease than that on adatoms A by 0.02 eV. This was rationalized by noting that two Ti atoms can receive electrons from an adatom B while only one Ti atom participates in electron transfer from an adatom A. Acknowledgment. This work was supported by Core Research for Evolutional Science and Technology (CREST) from the Japan Science and Technology Agency (JST) and Grant-in-Aid for Scientific Research in Priority Area “Molecular Nano Dynamics from the Ministry of Education, Culture, Sports, Science, and Technology. C.L.P. was supported by the Japan Society for the Promotion of Science (JSPS) Fellowship. References and Notes (1) Henrich, V. E.; Cox, P. A. The Surface Science of Metal Oxides; Cambridge University Press: New York, 1994; p 44. (2) Linsebigler, A. L.; Lu, G.; Yates, J. T., Jr. Chem. ReV. 1995, 95, 735. (3) Diebold, U. Surf. Sci. Rep. 2003, 48, 53. (4) Uetsuka, H.; Pang, C.; Sasahara, A.; Onishi, H. Langmuir 2005, 21, 11802. (5) Gan, S.; Liang, Y.; Baer, D. R.; Grant, A. W. Surf. Sci. 2001, 475, 159. (6) Steinru¨ck, H.-P.; Pesty, F.; Zhang, L.; Madey, T. E. Phys. ReV. B 1995, 51, 2427. (7) Gan, S.; Liang, Y.; Baer, D. R.; Sievers, M. R.; Herman, G. S.; Peden, C. H. F. J. Phys. Chem. B 2001, 105, 2412. (8) Dulub, O.; Hebenstreit, W.; Diebold, U. Phys. ReV. Lett. 2000, 84, 3646. (9) Tauster, S. J.; Fung, S. C.; Baker, R. T. K.; Horsley, J. A. Science 1981, 211, 1981. (10) Schierbaum, K. D.; Fischer, S.; Torquemada, M. C.; de Segovia, J. L.; Roma´n, E.; Martı´n-Gago, J. A. Surf. Sci. 1996, 345, 261. (11) Wei-xing, X.; Schierbaum, K. D.; Goepel, W. J. Solid State Chem. 1995, 119, 237. (12) Thieˆn-Nga, L.; Paxton, A. T. Phys. ReV. B 1998, 58, 13233. (13) Iddir, H.; O ¨ ’’u¨t, S.; Browning, N. D.; Disko, M. M. Phys. ReV. B 2005, 72, 081407(R).
Pt Atoms Deposited on the TiO2(110)-(1 × 1) Surface (14) Noncontact Atomic Force Microscopy; Morita, S., Wiesendanger, R., Meyer E., Eds.; Springer: Berlin, 2002. (15) Nonnenmacher, M.; O′Boyle, M. P.; Wickramasinghe, H. K. Appl. Phys. Lett. 1991, 58, 2921. (16) Yates, J. T., Jr. Experimental InnoVations in Surface Science; Springer: Berlin, 2002; p 424. (17) Okamoto, K.; Yoshimoto, K.; Sugawara, Y.; Morita, S. Appl. Surf. Sci. 2003, 210, 128. (18) Shiota, T.; Nakayama, K. Jpn. J. Appl. Phys. 2002, 41, L1178. (19) Kitamura, S.; Suzuki, K.; Iwatsuki, M.; Mooney, C. B. Appl. Surf. Sci. 2000, 157, 222. (20) Sasahara, A.; Uetsuka, H.; Onishi, H. Jpn. J. Appl. Phys. 2004, 43, 4647.
J. Phys. Chem. B, Vol. 110, No. 27, 2006 13457 (21) Fukui, K.; Onishi, H.; Iwasawa, Y. Phys. ReV. Lett. 1997, 79, 4202. (22) Bikondoa, O.; Pang, C. L.; Ithnin, R.; Muryn, C. A.; Onishi, H.; Thornton, G.; Nat. Mater. 2006, 5, 189. (23) Sasahara, A.; Uetsuka, H.; Onishi, H. Phys. ReV. B 2001, 64, 121406(R). (24) Okamoto, K.; Sugawara, Y.; Morita, S. Jpn. J. Appl. Phys. 2003, 42, 7163. (25) Salama, T. M.; Hattori, H.; Kita, H.; Ebitani, K.; Tanaka, T. J. Chem. Soc., Faraday Trans. 1993, 89, 2067. (26) Huizinga, T.; Prins, R. J. Phys. Chem. 1983, 87, 173. (27) Muscat, J.; Harrison, N. M.; Thornton, G. Phys. ReV. B 1999, 59, 15457. (28) Ono, S.; Takahashi, T. Jpn. J. Appl. Phys. 2004, 43, 4639.
