LETTER pubs.acs.org/Langmuir
Adsorption-Site-Dependent Electronic Structure of Catechol on the Anatase TiO2(101) Surface Shao-Chun Li,† Yaroslav Losovyj,‡ and Ulrike Diebold*,†,§ †
Department of Physics and Engineering Physics, Tulane University, New Orleans, Louisiana 70118, United States Center for Advanced Microstructures and Devices, Louisiana State Univeristy, Baton Rouge, Louisiana 70806, United States § Institute of Applied Physics, Vienna University of Technology, Wiedner Hauptstrasse 8-10/134, 1040 Vienna, Austria ‡
bS Supporting Information ABSTRACT: The adsorption of catechol (1,2-benzendiol) on the anatase TiO2(101) surface was studied with synchrotron-based ultraviolet photoemission spectroscopy (UPS), X-ray photoemission spectroscopy (XPS), and scanning tunneling microscopy (STM). Catechol adsorbs with a unity sticking coefficient and the phenyl ring intact. STM reveals preferred nucleation at step edges and subsurface point defects, followed by 1D growth and the formation of a 2 1 superstructure at full coverage. A gap state of ∼1 eV above the valence band maximum is observed for dosages in excess of ∼0.4 Langmuir, but such a state is absent for lower coverages. The formation of the band gap states thus correlates with the adsorption at regular lattice sites and the onset of self-assembled superstructures.
1. INTRODUCTION TiO2 is a versatile material that finds numerous applications in catalysis, photocatalysis, and solar energy conversion.1 3 In particular, its use in dye-sensitized solar cells (DSSCs) has attracted attention during the past decade.4 The band gap of TiO2 is about ∼3.0 eV for rutile and ∼3.2 eV for anatase, respectively,5 limiting light absorption to the UV range of the solar spectrum. The formation of band gap states leads to a red shift of the absorption edge into the visible. Such a red shift can be induced by “sensitizing” the TiO2 surface with a monolayer of selected organic dye molecules, thus enhancing the efficiency of the DSSCs. The great promise of DSSCs as an inexpensive and effective approach to solar energy conversion has motivated many fundamental studies that are both experimental and theoretical nature.6 12 Catechol is probably the simplest dye as well as a common functional group that links more complex dye molecules to the TiO2 surface. In catechol (1,2-benzendiol), two neighboring H atoms on the benzene ring are substituted by hydroxyls. A recent study of catechol adsorbed on a model surface of single-crystalline rutile (110) has found a strong correlation between the molecule’s adsorption geometry and its electronic structure.13 When catechol adsorbs on a rutile TiO2(110) surface, the H atoms from both hydroxyl groups are split off and donated to bridge-bonded surface O atoms. This results in a catecholate entity bonded to neighboring surface Ti5c sites via two Ti O bonds. Although the molecule in such a fully dissociated (D2) bidentate configuration contributes a band gap state, neither partially dissociated (only one H split off, D1) nor molecularly adsorbed (D0) catechol has its HOMO within the band gap region on rutile (110).13 The facile transfer of H between the O atoms at catechol and at the surface affects the diffusion of the r 2011 American Chemical Society
molecule across the rutile (110) surface.14 This work has motivated us to study the adsorption of catechol on anatase, the (metastable) crystallographic polymorph of TiO2 that is used as the photoactive electrode of DSSCs. We used UPS, XPS, and STM to study the coveragedependent adsorption behavior of catechol on anatase (101), the lowest-energy surface of anatase.15 Not too surprisingly, we find that initial adsorption is preferred at the steps edges and at point defect sites, which, for reduced anatase, are predominantly located at the subsurface.16 After the saturation of these defect sites, catechol grows in self-assembled structures on the terraces.17 Interestingly, it is found that a band gap state originates only from these self-assembled, ordered domains on TiO2 whereas the initial adsorption at defects, in particular, at step edge sites, apparently does not result in the formation of a band gap state. The ramifications for designing optimized DSSCs are discussed.
2. EXPERIMENTAL SECTION UPS measurements were performed at the Center for Advanced Microstructures and Devices (CAMD) in Baton Rouge, Louisiana on the 3 m TGM beamline. The electric field vector of the incident light was directed along the [101] direction of the anatase surface. The emitted photoelectrons were collected with a hemispherical analyzer in a plane subtended by the surface normal and the incident photon beam. STM measurements were carried out with an Omicron room-temperature STM in an ultrahigh vacuum (UHV) chamber with a base pressure of 1 10 10 mbar. Empty-state STM measurements were performed in constant-current mode at room temperature. The STM images were Received: April 27, 2011 Revised: June 13, 2011 Published: June 20, 2011 8600
dx.doi.org/10.1021/la201553k | Langmuir 2011, 27, 8600–8604
Langmuir
Figure 1. Photoemission spectra of clean anatase TiO2(101) (black) and with a saturation coverage of catechol (red). (Photon energy 50 eV, angle of incidence 70°, emission angle 30°, electric field vector along the [101] direction.) The blue arrow marks the defect state in the band gap, and the black arrow marks the photoemission from second-order light. Red arrows mark the states formed upon catechol adsorption. analyzed and processed with free software WSxM.18 XPS was performed with a SPECS Phoibos hemispherical energy analyzer in a separate UHV chamber using a nonmonochromatized Mg KR anode. Cut and polished natural mineral anatase was cleaned by repeated Ar+ sputtering and thermal annealing to 600 °C in UHV. The surface cleanliness and structure were checked by XPS, LEED, and STM before experiments. The anatase sample was reduced through many cycles of sputtering and annealing, leading to subsurface defects as identified in previous work.16 Catechol powder (Alfa Aesar, 99%) was put in a glass vial and thoroughly outgassed by pumping it at a slightly elevated temperature. The purified catechol forms bright-yellow crystallites. Gas-phase catechol was dosed from the vapor sublimed at room temperature through a leak valve. The gas line was pumped by a turbo pump, and the base pressure inside the gas line is ∼10 10 mbar. During dosing, the pressure was controlled to less than 5.0 10 8 mbar. Exposures are quoted in Langmuirs (L), where 1 L corresponds to a dosage of 10 6 Torr 3 s.
3. RESULTS Figure 1 shows the photoemission spectra of the clean anatase TiO2(101) surface (black) and the surface covered with a full layer of catechol (red). The clean surface has a state in the band gap located at ∼1.0 eV (marked by the blue arrow) below the Fermi level; such a state is indicative of defects and derived from Ti 3d.3 For anatase (101), this band gap state originates mainly from the subsurface O vacancies or Ti interstitials in the near-surface region.16 The weak, broad feature at ∼13.5 eV (black arrow) on the clean surface is an experimental artifact (photoemission from the Ti 3s state induced by second-order light) and does not need to be considered further. The UPS spectrum changes markedly upon catechol adsorption; the new states introduced are marked by red arrows in Figure 1. The HOMO is located about ∼3.0 eV below the Fermi level, and this value can be compared with a binding energy of ∼2.4 eV observed for catechol on rutile TiO2(110).13 In polar angle-dependent measurements, an intensity maximum of the HOMO peak was observed for emission angles of 30 40° (Figure S1). Three clearly resolved additional peaks are observed below the valence band region. Similar peaks appear in the spectra of catechol on the rutile (110) surface, but again the binding energies on anatase are slightly higher.13 Figure 2a shows the UPS spectra for incremental dosages of catechol. The photoemission geometry is similar to that in
LETTER
Figure 1 except that the emission angle is now 40°. The bottom black curve represents the spectrum of the clean surface. The following spectra were taken after increasing the dosages in steps of 0.2 L; after a total dosage of ∼1 L was reached, the spectra did not change any more, indicating that saturation had been reached. Figure 2b shows difference spectra that were obtained by subtracting the clean surface spectrum from the catechol spectra in Figure 2a. For dosages higher than ∼0.4 L, the molecular orbitals and the HOMO in the band gap are clearly discerned. Interestingly, however, the HOMO orbital is not apparent for the initial dosages, the valence band is suppressed, and the lower-lying orbitals are visible in the low-coverage regime (i.e., for dosages lower than ∼0.4 L). The peak area of the HOMO as a function of catechol dosage is shown in Figure 2c. The uptake curve shows three distinct stages: a delayed onset, an approximately linear increase above 0.4 L, and a plateau above ∼1.2 L when the surface is already saturated with catechol. To check for consistency in our dosing procedure and/or the possibility of a coverage-dependent sticking coefficient, we performed XPS measurements. Figure 2d shows the XPS C 1s peak as a function of catechol dosage. The peak is composed of two components located at binding energies of 284.8 and 286.2 eV corresponding to C C and C O bonds, respectively. The ratio is about 2:1, consistent with an intact phenyl ring of catechol as also observed in previous studies of catechol on rutile (110).13,19 In contrast to the UPS results, the intensity of the C 1s peak is linearly dependent on the catechol dosage from the very beginning, as can be seen more clearly in the plot of C 1s peak area vs coverage in Figure 2e. The C 1s signal is saturated at about 1.2 L, consistent with the UPS results. Figure 3 shows the STM results for various coverages of catechol on anatase (101). The clean (101) surface is characterized by trapezoidal islands separated by monatomic steps. In a previous study,20 steps oriented parallel to the [010] and [111] /[111] directions (termed B and D, respectively) were found to be dominant for energetic reasons; see the sketch in Figure 3. Figure 3a shows an STM image of a clean anatase TiO2(101) surface. At terraces, rows run along the [010] direction; bright features in atomically resolved STM images extend across the surface Ti5c and O2c atoms.15 The subsurface defects typical of anatase (101)16 are not visible in the STM images presented here. Occasional black spots in Figure 3a are indicative of water molecules adsorbed to these subsurface defect sites.21 Figure 3b shows an enlarged image of step edge B on the clean surface, and a ball-and-stick model of such a step edge after ref 20 is shown in Figure 3g. After a catechol exposure of 0.2 L, STM images show that catechol is located mostly at the step edges and at some surface point defect sites (Figure 3c,d). No preference for a particular step orientation was observed. The higher-magnification image of step B in Figure 3d shows catechol molecules as bright protrusions, ∼4.0 Å higher than the upper terrace and ∼1 nm in apparent width. As the coverage is increased, catechol molecules assemble in elongated islands on top of the terraces.17 (See Figure 3e for a coverage of 0.5 L). At full coverage, catechol forms a 2 1 superstructure with two domains oriented along the (equivalent) [111] and [111] directions, as seen in Figure 3f. The zigzag configuration in Figure 3f is associated with the resulting two antiphase domains.
4. DISCUSSION Uptake curves measured with both UPS and XPS exhibit saturation at dosages of ∼1.2 L, indicating that the sticking 8601
dx.doi.org/10.1021/la201553k |Langmuir 2011, 27, 8600–8604
Langmuir
LETTER
Figure 2. Photoemission for exposure-dependent catechol adsorption. (a) UPS spectra after incremental dosages as indicated. Arrows mark the HOMO (band gap state). (Photon energy 50 eV, angle of incidence 70°, emission angle 40°, electric field vector directed along the [101] direction.) (b) Difference spectra (catechol minus clean anatase) of the spectra in panel a. (c) Intensity maximum of the band gap state as a function of catechol exposure. (d) XPS of the C 1s region of the clean surface (bottom spectrum) in incremental steps up to a total dosage of 1.6 L (top spectrum). (e) Integrated C 1s peak from panel d as a function of catechol exposure.
coefficient of catechol is close to unity. The XPS C 1s peak is a good quantitative measure of the amount of catechol adsorbed on the TiO2 surface because the cross section of the core-level photoemission is largely independent of an atom’s bonding environment. The ratio and the binding energies of the two C 1s components are nearly unchanged over the whole coverage regime, implying that the benzene-like ring on catechol stays intact, independent of whether catechol is adsorbed at step edges or at terraces. The linear dependence of the C area on the dosage shown in Figure 2e suggests the catechol adsorption is a precursor-mediated process. However, the intensity of the HOMO in UPS (Figure 2c) is not directly correlated to the coverage. Different from the XPS results, the gap state in UPS shows a delayed onset up to a total exposure of ∼0.4 L, which might indicate that the electronic structure is rather different during the initial stages of overlayer formation. It is instructive to correlate the evolution of the band gap state with the coverage-dependent STM measurements in Figure 3. Initially, most of the catechol resides at step edges in addition to isolated molecules at terraces (probably located at more reactive subsurface defect sites16). Adsorption at step edges and surface point defects is completed at an exposure of ∼0.4 L, which is when the plateau in the UPS uptake curve ends. During the second stage of HOMO evolution, when the intensity of the HOMO level (gap state) increases with exposure, catechol starts
to adsorb at regular surface sites in self-organized islands with closely spaced molecules. An analysis of the dynamic change in these islands suggests an attractive interaction between neighboring catechol molecules in this higher-coverage regime.17 As mentioned above, the step edges of anatase (101) are either of type B or D (sketch in Figure 3);20 steps that do not run parallel to these directions are composed of small sections of B and D. A statistical evaluation of STM images shows that, on this particular anatase surface, the ratio of step edge lengths of B to D varies between 4:1 and 6:1. Therefore, the initial adsorption occurs mostly on steps of type B. Such step edges essentially resemble small sections of (100)-oriented anatase facets. We have carefully considered that the observed phenomenon is indeed an initial-state effect and not an experimental artifact. The photoemission intensity from molecular orbitals depends on the experimental geometry because of selection rules. It is conceivable that we have fortuitously chosen an experimental geometry that suppresses photoemission from the HOMO state for catechol molecules in the dilute coverage regime. We tested for this possibility. On the basis of the group symmetry of catechol, the HOMO can be assigned as a1, and maximum photoemission should occur along a direction parallel to the phenyl ring. In a previous photoemission study13 of catechol on rutile (110), where the molecule is adsorbed in a tilted fashion, the HOMO-derived gap state indeed exhibits a maximum in the 8602
dx.doi.org/10.1021/la201553k |Langmuir 2011, 27, 8600–8604
Langmuir
Figure 3. STM images (25 nm 12.5 nm, Usample = +1.5 V, Itunnel = 0.2 nA) showing catechol adsorbed on anatase (101) with various coverages. (a, b) Clean (101) surface; (a, c, d) ∼0.2 L; (e) ∼0.5 L; and (f) ∼1.0 L. (g) Structural model of step B; Ti atoms are represented by blue balls, and O atoms are represented by red balls.
range of 15 30°. In this study, we also observe an intensity maximum along an off-normal emission angle in the full-coverage regime (Figure S1), indicating that the molecule in the organized overlayer is adsorbed in a tilted fashion on anatase as well. In these spectra, the intensity of the HOMO is a relatively weak function of the emission angle, however. If one factors in that catechol molecules at steps and defects likely have a more randomized orientation, then it is unlikely that a complete suppression of the gap state should occur in this specific experimental geometry. It is more likely that the stronger coupling between sites at surface steps and the adsorbed molecules changes the position of the HOMO and so could adsorption at the (subsurface) defect sites. The catechol-Ti complex always exhibits strong electronic coupling. It is reasonable to expect that the adsorption geometries of catechol at surface steps are more complex than those at terraces. Although the adsorption of catechol on an anatase (101) terrace has been thoroughly studied recently,17 the geometries of catechol on step edges remain poorly understood. In any event, the resulting hybridization/charge transfer between the HOMO and the anatase defect sites might play a role in modifying the electronic structure of catechol. Intermolecular interactions might also change the alignment of the HOMO. Initially, catechol molecules are isolated at the subsurface defect sites; even when situated along step edges they are ∼1 nm apart (Figure 3c,d). At these large separations, the intermolecular interactions are likely negligible. However, during the following growth, catechol self-organizes into one-dimensional chains and ultimately forms a 2 1 superstructure at full coverage.17 The intermolecular distance in these ordered superstructures is decreased to ∼5 Å, which is much smaller than that of catechol adsorbed on step edges and prone to stronger molecule molecule interactions. Thus, the coupling between
LETTER
neighboring catechols might play an (additional) role in shifting the HOMO into the band gap. Previous optical absorption experiments and theoretical calculations of catechol/TiO2 indicated a direct injection (type II) mechanism: a new broad band appears upon catechol adsorption on TiO2 particles, which is centered at about 430 nm with the tail extending into the visible range.9,12,22 26 In our UPS measurement at full coverage, the HOMO is located within the band gap region and the binding energy of ∼3.0 eV corresponds to a wavelength of ∼413 nm, which is in line with these previous results. However, it is difficult to compare our low-coverage results with previous studies because the coverage calibration is rather different in these studies.27 As shown above, the catechol adsorption configurations are coverage-dependent and thus induce coverage-dependent spectroscopic features. A previous IR study also showed the coverage-dependent configurations.24 It is worthwhile to point out, however, that most of the spectroscopy studies were performed with TiO2 particles in solution, where the pH value and particle size might also play a role in the catechol adsorption. The suppression of the gap state for catechol adsorbed at defects is quite remarkable, especially considering the fact that a large fraction of adsorption sites are located at step edges on small anatase particles. The anatase nanocrystals in a percolating TiO2 network that constitute the photoactive electrode in DSSCs28 have a typical size of ∼15 nm. Thus, less than 10% percent of adsorption sites are located at step edges, assuming ideally shaped bipyramidal crystallites.29 However, once the size of the anatase nanocrystals decreases to ∼2 nm, the number of step edge sites increases to ∼50%. Although smaller particles result in a drastically increased surface area, a relatively large fraction of molecules would be “blind” to optical absorption. Therefore, an optimized particle size distribution should be practically considered in order to maximize the number of effective adsorption sites. In addition, the fundamental insights into catechol adsorption on TiO2 derived from this study might also be applicable to catechol-based bioadhesion studies in water.
5. CONCLUSIONS The coverage-dependent adsorption of catechol on anatase TiO2(101) has been characterized with UPS, XPS, and STM. A band gap state, located ∼1 eV above the valence band edge of TiO2, is found to originate from catechol adsorbed in selforganized islands at terraces. In contrast, catechol adsorbed at surface steps and at isolated point defect sites does not introduce a state in the band gap. ’ ASSOCIATED CONTENT
bS
Supporting Information. Polar-angle-dependent photoemission spectrum. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected].
’ ACKNOWLEDGMENT S.-C.L. and U.D. acknowledge financial support by the DoE-BES under contract DE-FG02-05ER15702. 8603
dx.doi.org/10.1021/la201553k |Langmuir 2011, 27, 8600–8604
Langmuir
LETTER
’ REFERENCES (1) Linsebigler, A. L.; Lu, G.; Yates, J. T. Chem. Rev. 1995, 95, 735–758. (2) Fujishima, A.; Zhang, X.; Tryk, D. A. Surf. Sci. Rep. 2008, 63, 515–582. (3) Diebold, U. Surf. Sci. Rep. 2003, 48, 53–229. (4) Gr€atzel, M. Nature 2001, 414, 338–344. (5) Kavan, L.; Gr€atzel, M.; Gilbert, S. E.; Klemenz, C.; Scheel, H. J. J. Am. Chem. Soc. 1996, 118, 6716–6723. (6) Connor, P. A.; Dobson, K. D.; McQuillan, A. J. Langmuir 1995, 11, 4193–4195. omor, M. I.; Nedeljkovic, J. M. (7) Jankovic, I. A.; Saponjic, Z. V.; C J. Phys. Chem. C 2009, 113, 12645–12652. (8) Araujo, P. Z.; Mendive, C. B.; Rodenas, L. A. G.; Morando, P. J.; Regazzoni, A. E.; Blesa, M. A.; Bahnemann, D. Colloids Surf., A 2005, 265, 73–80. (9) Persson, P.; Bergstrom, R.; Lunell, S. J. Phys. Chem. B 2000, 104, 10348–10351. (10) Redfern, P. C.; Zapol, P.; Curtiss, L. A.; Rajh, T.; Thurnauer, M. C. J. Phys. Chem. B 2003, 107, 11419–11427. (11) Rego, L. G. C.; Batista, V. S. J. Am. Chem. Soc. 2003, 125, 7989–7997. (12) Xu, Y.; Chen, W.-K.; Liu, S.-H.; Cao, M.-J.; Li, J.-Q. Chem. Phys. 2007, 331, 275–282. (13) Li, S.-C.; Wang, J.-G.; Jacobson, P.; Gong, X.-Q.; Selloni, A.; Diebold, U. J. Am. Chem. Soc. 2009, 131, 980–984. (14) Li, S.-C.; Chu, L. N.; Gong, X.-Q.; Diebold, U. Science 2010, 328, 882–884. (15) Hebenstreit, W.; Ruzycki, N.; Herman, G. S.; Gao, Y.; Diebold, U. Phys. Rev. B 2000, 62, R16334–R16336. (16) He, Y.; Dulub, O.; Cheng, H.; Selloni, A.; Diebold, U. Phys. Rev. Lett. 2009, 102, 106105. (17) Liu, L.-M.; Li, S.-C.; Cheng, H.; Diebold, U.; Selloni, A. J. Am. Chem. Soc. 2011, 133, 7816–7823. (18) Horcas, I.; Fernandez, R.; Gomez-Rodriguez, J. M.; Colchero, J.; Gomez-Herrero, J.; Baro, A. M. Rev. Sci. Instrum. 2007, 78, 013705. (19) Jacobson, P.; Li, S.-C.; Wang, C.; Diebold, U. J. Vac. Sci. Technol., B 2008, 26, 2236–2240. (20) Gong, X.-Q.; Selloni, A.; Batzill, M.; Diebold, U. Nat. Mater. 2006, 5, 665–670. (21) He, Y.; Tilocca, A.; Dulub, O.; Selloni, A.; Diebold, U. Nat. Mater. 2009, 8, 585–589. (22) Sanchez-de-Armas, R.; San-Miguel, M. A.; Oviedo, J.; Marquez, A.; Sanz, J. F. Phys. Chem. Chem. Phys. 2011, 13, 1506–1514. (23) Moser, J.; Punchihewa, S.; Infelta, P. P.; Gr€atzel, M. Langmuir 1991, 7, 3012–3018. (24) Lana-Villarreal, T.; Rodes, A.; Perez, J. M.; Gomez, R. J. Am. Chem. Soc. 2005, 127, 12601–12611. (25) Duncan, W. R.; Prezhdo, O. V. J. Phys. Chem. B 2004, 109, 365–373. (26) Wang, Y.; Hang, K.; Anderson, N. A.; Lian, T. J. Phys. Chem. B 2003, 107, 9434–9440. (27) Creutz, C.; Chou, M. H. Inorg. Chem. 2008, 47, 3509–3514. (28) Hagfeldt, A.; Gr€atzel, M. Acc. Chem. Res. 2000, 33, 269–277. (29) Diebold, U.; Ruzycki, N.; Herman, G. S.; Selloni, A. Catal. Today 2003, 85, 93–100.
8604
dx.doi.org/10.1021/la201553k |Langmuir 2011, 27, 8600–8604
