ARTICLE pubs.acs.org/JPCC
Binding of a Benzoate Dye-Molecule Analogue to Rutile Titanium Dioxide Surfaces David C. Grinter,†,‡ Patrick Nickels,†,§ Thomas Woolcot,†,‡ Sulaiman N. Basahel,|| Abdullah Y. Obaid,|| Ahmed A. Al-Ghamdi,^ El-Sayed H. El-Mossalamy,|| Abdulrahman O. Alyoubi,|| and Geoff Thornton*,†,‡ †
London Centre for Nanotechnology and ‡Department of Chemistry, University College London, London WC1H 0AH, U.K. Bio Nano Consulting, London NW1 3BT, U.K. Chemistry Department and ^Physics Department, Faculty of Science, King Abdulaziz University, Jeddah, Kingdom of Saudi Arabia 21589
)
§
ABSTRACT: Scanning tunneling microscopy (STM) has been used to investigate the adsorption of benzoic acid on the rutile TiO2(110)(1 1) and the reconstructed TiO2(110)(1 2) surfaces. Benzoic acid binds to both surfaces dissociatively via a bridging geometry to two Ti5c sites. At a slightly elevated sample temperature during deposition onto the (110) (1 1) surface, a well-ordered (2 1) overlayer was formed at saturated benzoate coverage. On the reconstructed (110) (1 2) surface, benzoate was observed to adsorb between the (1 2) strands leading to a (2 2) superstructure at higher coverage. Elongation along the [110] direction in the STM images indicates a rotation of the benzene ring of 90 relative to the carboxylate group, which is reasonably explained by hydrogen bond interactions between terminating O-atoms on the surface and H-atoms of the ring.
1. INTRODUCTION Ever since the discovery of TiO2-based photocatalysis by Fujishima and Honda in the early 1970s,1 there has been remarkable interest in this field with many publications regarding both the fundamental science and industrial applications.2 There are a number of reviews of TiO2 photocatalysis, in particular the studies from a surface science perspective that aim to correlate the behavior of single-crystal surfaces with the activity of the bulk catalysts.3 6 The properties of TiO2 render it useful in other fields, and thus it has many proven applications including gas sensors, heterogeneous catalysts, pigments, and electrical devices.7 9 It is hoped that the study of single-crystalline TiO2 surfaces will lead to increased understanding of the mechanisms behind titania’s many valuable properties with a view to optimizing efficiency and improving effectiveness. The thermodynamically most stable (110) surface of rutile TiO2, which is easily prepared, has become a model system for metal oxide surface studies with a vast body of experimental and theoretical work reported in the literature.2,3,6,7 The rutile TiO2(110)(1 1) surface has been thoroughly investigated using the full spectrum of surface science techniques, and thus it is rather well understood from the perspective of its defect structure, as well as the adsorption and reactivity of small molecules.7,10 The (1 2) reconstruction, achieved after repeated reduction by annealing in ultrahigh vacuum (UHV), has been less well investigated with controversy still existing over its basic structure. A number of models have been proposed for the (1 2) reconstruction; the two that have had the most attention are (i) a Ti2O3 added-row model that appears consistent with lowenergy electron diffraction (LEED) and scanning tunneling microscopy (STM) data11 14,35 and (ii) a Ti3O6 added-row r 2011 American Chemical Society
model that may explain noncontact atomic force microscopy (NC-AFM) and other STM results.15 17 For the purposes of this paper we will assume the Ti2O3 added-row picture when constructing structural models. However, we note that similar models could also be produced assuming the Ti3O6 added-row structure. The adsorption and behavior of many different small molecules on rutile TiO2(110) have been studied in an attempt to understand the reactivity of the surface and its defect sites.10 One important class of molecules is carboxylates, which have a number of practical applications, including the linking of dye molecules to titania surfaces within Gr€atzel type dye-sensitized solar cells.18 On the TiO2(110)(1 1) surface it is wellestablished that the most common binding geometry of monocarboxylic acids (formic, acetic, and so on) is that of dissociative bidentate bridging to two neighboring Ti5c sites along the [001] direction.10,19 21 Investigations on the reconstructed TiO2(110)(1 2) surface are much more limited, with only a couple of STM studies in the literature. Formic acid was observed by Bennett et al. to adsorb preferentially on the cross-linked sites and even led to the reoxidation of the surface during decomposition at raised temperatures.22 In early studies by Guo et al. 23 the exposure of rutile TiO2(110)(1 1) and -(1 2) surfaces to benzoic acid was investigated with LEED, STM, and electron stimulated desorption ion angular distribution (ESDIAD). It was found to follow behavior similar to the smaller carboxylates,24 where the benzoate Received: October 11, 2011 Revised: December 2, 2011 Published: December 07, 2011 1020
dx.doi.org/10.1021/jp209749m | J. Phys. Chem. C 2012, 116, 1020–1026
The Journal of Physical Chemistry C
ARTICLE
formed a dimer or trimer superstructure. This was explained in terms of interactions between hydrogen atoms on one adsorbate and the π system of an adjacent molecule. Similar adsorbate adsorbate interactions have been identified for pyridine carboxylic acid25 27 and for terephthalic acid.28 31 In the present work, mild heating of the target leads to a more homogeneous ordering on the rutile TiO2 (110)(1 1) surface. Additionally, we provide enhanced quality STM images sufficient to assign the adsorption sites.
2. EXPERIMENTAL PROCEDURE The experiments were conducted in a UHV system with a base pressure of 1 10 11 Torr and separate analysis and preparation chambers. The STM used is an Omicron UHV AFM/STM instrument operated in constant current mode at room temperature using electrochemically etched tungsten tips which were conditioned during scanning with voltage pulses (up to 10 V) and high negative/positive bias scans to obtain stable atomic resolution imaging. All STM imaging was carried out by tunneling into empty states using positive sample bias voltages in the range of 0.7 2 V. The rutile TiO2(110) crystals (Pi-Kem GmbH) were attached to standard sample plates with spotwelded tantalum clips, and the temperature during annealing was monitored using an infrared pyrometer (Minolta Land). To prepare the TiO2(110)(1 1) surface, cycles of Ar+ sputtering (5 min, 1 keV, 10 μA) and annealing in UHV (10 min, 900 K) were performed until any contamination was below the detection limit of AES, and a well-ordered (1 1) LEED pattern with a low background was observed. For the reconstructed TiO2(110)(1 2) surface a separate, heavily reduced dark blue crystal was used, which had undergone a high number of annealing cycles at temperatures up to 1020 K, until a well-ordered (1 2) LEED pattern was observed. The temperatures during annealing and acid exposure were measured with an optical pyrometer or a K-type thermocouple attached to the manipulator arm and positioned near the sample plate. Benzoic acid (99.9%, Sigma-Aldrich) was contained within a glass vial attached to the gas line of the UHV system and admitted into the preparation chamber via a high-precision leak valve. To purify the acid, which is solid at room temperature, repeated melt freeze pump cycles were carried out. During exposure, the gas line and vial were heated to ∼400 K to liquefy the acid, and RGA mass spectrometry was used to monitor the purity during dosing. All exposures are nominal and quoted in langmuirs (1 langmuir = 1 10 6 Torr s) on the basis of the uncompensated pressure within the preparation chamber during dosing. The chamber pressure during benzoic acid exposure was in the range of 5 10 8 to 2 10 7 Torr, with exposure times of 10 20 min. For this work we define a monolayer as one adsorbate per primitive surface unit cell, a density of 5.2 1014 cm 2. 3. EXPERIMENTAL RESULTS 3.1. Benzoic Acid Adsorption on Rutile TiO2(110)(1 1). The rutile TiO2(110)(1 1) surface has been extensively studied with STM over the past decade and is now considered to be well-understood. Subsequently, for our initial investigations of benzoic acid adsorption on titania surfaces we decided to also use TiO2(110)(1 1). The surface is characterized by flat (110)(1 1) terraces separated by monatomic steps of height 0.3 nm. In atomically resolved images of empty states (positive sample bias), the bright rows running along the [001]
Figure 1. STM images of 0.2 monolayer of benzoic acid deposited at 300 K on rutile TiO2(110)(1 1): (a) Large area image (100 100 nm2, Vs = +1.9 V, It = 0.05 nA) with no long-range ordering of the adsorbates visible; (b) zoomed-in image (25 25 nm2, Vs = +1.9 V, It = 0.05 nA) where the individual benzoate molecules can be seen more clearly and are observed to form short strings along the [110] direction, some of which are highlighted by black rectangles. A line profile (blue) across one of these strings is shown as an inset in b.
direction are assigned to Ti5c cations.32 Between the Ti rows are bridging and surface oxygen ions. Hydroxyls sitting at O positions appear as bright features.33 The unit cell has dimensions of 6.49 Å by 2.98 Å. Since the dimensions of the TiO2(110)(1 1) unit cell are well-known, the STM piezos can be calibrated in all directions by comparison with an image of the bare surface and applied to the other experimental results permitting more accurate measurement of distances. Preparation of the clean TiO2(110)(1 1) surface was confirmed by STM and by LEED measurement (80 eV). An Auger electron spectrum was taken of the surface and Ti LMM (385 and 420 eV) and O KLL (512 eV) were identified, with any possible contamination of the surface being under the detection limit of the spectrometer. Exposure of the clean TiO2(110)(1 1) surface to ∼30 L of benzoic acid at room temperature led to a coverage of ∼0.2 monolayer as depicted in the STM images presented in Figure 1. From the large area (see Figure 1a), there does not appear to be any longrange ordering of the acid molecules, which appear as bright round protrusions evenly distributed across the surface. Upon 1021
dx.doi.org/10.1021/jp209749m |J. Phys. Chem. C 2012, 116, 1020–1026
The Journal of Physical Chemistry C zooming into any given region, the adsorbed benzoate molecules appear as bright features of height ∼0.30 nm and width ∼1.2 nm (not compensated for broadening via tip-convolution effects) and form occasional short chains of up to 4 molecules in length, which are oriented along the [110] direction. Some of these are highlighted with black rectangles in Figure 1b. The spacing of the molecules within these chains is measured to be ∼0.62 nm, consistent with the unit cell dimension of the TiO2(110)(1 1) surface in the [110] direction. The benzoate-covered surface was heated to 380 K under UHV in attempt to increase the ordering of the overlayer, although this did not have any appreciable effect. Aizawa et al. have reported that the binding geometry of formate on the TiO2(110)(1 1) surface can be greatly affected by annealing to temperatures around 350 K, including binding to surface oxygen vacancies.34 We do not report any such species, but it is likely that their formation is extremely sensitive to the surface preparation and defect concentration as well as the sample temperature and exposure conditions.10 To further examine the effect of sample temperature on the structure of the adsorbed layer, we exposed the TiO2(110)(1 1) surface to ∼60 L benzoic acid with the sample held at 370 K (see Figure 2). A (2 1) superstructure with a near-saturated coverage measured at 0.45 monolayer is observed (see Figure 2a). Significant tip interaction with the benzoate is observed, characterized by streaking in the fast scan direction (horizontal), which results in noticeable instability during STM imaging. In images obtained at rotated scan angles (45), the appearance of the molecules does not change and streaking is still only observed in the fast scan direction. Figure 2b shows a different area of the same sample, and here, the (2 1) structure is clearly observed. A line profile (blue line) across a number of the benzoate molecules confirms the spacing in the [001] direction to be ∼0.6 nm, twice the lattice spacing of TiO2(110)(1 1) (see Figure 2d). A missing benzoate molecule, which appears as a dark spot, permits measurement of the benzoate height to be 240 pm at the particular scanning parameters used. This height measurement is slightly smaller than that measured on the unordered benzoate (see Figure 1). The discrepancy is likely due to a tip convolution effect. In Figure 2b the unit cell of the (2 1) overlayer is marked with a red rectangle; in the region marked with a black rectangle, there exists the boundary between two domains of the (2 1) layer that are offset by one lattice spacing in the [001] direction, all of which are consistent with a dissociative bidentate binding geometry of the benzoate bridging two neighboring 5-foldcoordinated Ti4+ ions in the [001] direction. A (2 1) ordered overlayer is commonly found for a saturation coverage achieved by exposure of TiO2(110) to carboxylic acids. This is consistent with other experimental data in indicating deprotonation of the acid,10 with maximum exposure in a bidentate geometry to adjacent Ti5c sites along [001] forming a (2 1) unit cell, as shown in Figure 2c. The formation of a (2 1) overlayer here is therefore taken to indicate dissociative adsorption of benzoic acid. Figure 2c shows a model of the binding observed in the highlighted regions of Figure 2b, where the boundary between two (2 1) domains is shifted by a single lattice spacing. This (2 1) superstructure is identical to that observed in previous studies of small carboxylic acids on the rutile TiO2(110)(1 1) surface,10 suggesting that any intermolecular interaction between the aromatic rings of neighboring benzoates adsorbed at the surface is minimal compared with the interaction with the surface under these deposition conditions.
ARTICLE
Figure 2. STM of 0.45 monolayer of benzoic acid deposited at 370 K on the rutile TiO2(110)(1 1) surface: (a) Large area STM image (25 25 nm2, Vs = +1.9 V, It = 0.1 nA) of the saturated surface showing a (2 1) overlayer with an occasional missing adsorbate as highlighted; (b) STM image (15 15 nm2, Vs = +1.6 V, It = 0.1 nA) of the overlayer showing the presence of domains offset by a lattice spacing in the [001] direction (highlighted in black). The (2 1) unit cell is marked in red; (c) model of the (2 1) superstructure formed by the benzoate molecules bridge-binding to two adjacent Ti4+ cations along [001] (the black rectangle is around the junction between two (2 1) domains that are offset by a single lattice spacing in the [001] direction); (d) line profile from the STM image in b across four benzoate molecules and a missing adsorbate confirming their (2 1) spacing and height of ∼240 pm.
3.2. Benzoic Acid Adsorption on Rutile TiO2(110)(1 2). Broad reconstruction of the heavily reduced TiO2 crystal surface was confirmed via STM and LEED. The surface of our sample is almost completely covered in strands running along the [001] direction with a spacing of 1.3 nm that are characteristic of the reconstructed surface.12 On regions close to a step edge, it is observed that the additional (1 2) strands are situated on top of (1 1)-terminated TiO2. The (1 2) strands form characteristic single- and crossed-links that run along the [110] direction. It is observed that the cross-links are spaced every 3.5 nm in the [001] direction, which is 12 times the unit cell spacing of the (1 1) surface consistent with the (12 2) reconstruction observed in LEED. STM images were obtained at positive sample biases (+ 1.2 V), and we are therefore imaging empty states and might expect the Ti4+ cations to appear as bright, as observed in the normal imaging mode on the (1 1) surface.32 The cross-linked TiO2(110)(1 2) surface was exposed to ∼100 langmuirs of benzoic acid, while held at 350 K. STM images after deposition are displayed in Figure 3. From the images of large areas (see Figure 3a,b) it can be observed that there is a clear preference for adsorption along the [001] direction, with an estimated coverage of 0.12 monolayer, with some short-range ordering visible from the STM image shown in Figure 3c. It is observed that the benzoate is adsorbed between the (1 2) strands (one strand marked with red lines for clarity). The crosslinks (highlighted with black rectangles) are completely void of adsorbates of the size of benzoate, which is in contrast to the observations made by Bennett et al.22 on formic acid adsorption on TiO2(110)(1 2), where the acid was observed to favor adsorption on the cross-links with no ordering. The benzoate 1022
dx.doi.org/10.1021/jp209749m |J. Phys. Chem. C 2012, 116, 1020–1026
The Journal of Physical Chemistry C
Figure 3. STM images of 0.12 monolayer benzoic acid adsorbed at 350 K on the rutile TiO2(110)(1 2) surface: (a) Large area image (100 100 nm2, Vs = +1.2 V, It = 0.1 nA); (b) image (60 60 nm2, Vs = +1.2 V, It = 0.1 nA) showing that the benzoate has a preference for adsorption along the [001] direction; (c) zoomed-in region (20 20 nm2, Vs = +1.0 V, It = 0.1 nA) of the adsorbate-covered surface showing that the benzoate is adsorbed between the (1 2) strands, one of which is marked with red lines (the cross-links (highlighted with black rectangles) are clear of any adsorbates, in contrast to the reported results of formic acid adsorption on TiO2(110)(1 2) 21); (d) a line-profile across four of the benzoate molecules in c showing that they have a spacing in the [001] direction that is consistent with an overall (2 2) superstructure.
appears as an oval shape with its long axis running along the [110] direction, and arrangements of molecules in regular distances can be seen in the [001] direction. A line profile across one of these is shown in Figure 3d, and from this it is observed that the intermolecular spacing is ∼0.6 nm, which is consistent with double the lattice constant along [001] and the start of a (2 2) superstructure. This conformation is also suggestive of bidentate bridging between two neighboring Ti5c cations in the [001] direction. to identify the binding of individual benzoate molecules more closely, the cross-linked TiO2(110)(1 2) surface was exposed to ∼10 langmuirs of benzoic acid, again at 350 K (see Figure 4). A much lower coverage of benzoate can be observed, which is estimated to be 0.005 monolayer from a number of images. Figure 4a shows that there are significant areas of the surface (∼20%) that consist of (1 1)-terminated TiO2(110), which are marked with a black boundary. These regions of (1 1) have a large number of small adsorbates that are likely to be hydroxyl groups due to unintentional water dosing during the benzoic acid exposure. Figure 4b shows a region from the center of the image in Figure 4a and displays some features of height ∼140 pm located on top of the (1 2) strands (highlighted with black circles), which, by comparison with the heights of species on the TiO2(110)(1 1) surface, are likely to be either water molecules or hydrogen atoms.32 These appear to be randomly distributed across the surface and, on occasion, are located in close proximity
ARTICLE
Figure 4. STM images of TiO2(110)(1 2) after exposure to 0.005 monolayer of benzoic acid at 350 K: (a) Large area image (60 60 nm2, Vs = +1.2 V, It = 0.1 nA) showing the low coverage of benzoate that appears as bright features between the (1 2) strands (approximately 20% of the surface is (1 1) terminated with a high concentration of hydroxyls due to water that was dosed adventitiously during the benzoic acid exposure); (b) atomically resolved, filled-states STM image (10 10 nm2, Vs = +1.2 V, It = 0.1 nA) showing the presence of additional adsorbates (possibly H) on top of the (1 2) strands (marked with solid circles) as well as the benzoate (dashed ovals); (c) model of two benzoate molecules in the bidentate binding geometry between the (1 2) strands (Ti2O3 added-row model; black, Ti; dark gray, O in (1 1); light gray/red, O in (1 2) strands). The top benzoate molecule has its aromatic ring parallel to [110], and the lower has it parallel to [001]. The origin of the atomic contrast in STM is highlighted in c (white circles correspond to the bright features in the empty states STM images located above Ti3+ positions in the added-rows).
to the (clearly distinguishable) benzoate molecules, which are located between the strands and are highlighted with dashed ovals in Figure 4b. The benzoate molecules here (see Figure 4) have a slightly different appearance when compared with those depicted in Figure 3c. This change in appearance has been attributed to a tip convolution effect in closed packing, which leads to symmetrization. There is a faint atomic contrast visible on the (1 2) strands in Figure 4b, which have an appearance similar to that observed in our studies of the clean surface and other STM work.11 14,35 Here the added rows have lateral separations (∼0.4 nm), and the bright spots are assigned to the Ti3+ cations within these rows. Some of these areas have been marked with white circles and lines in Figure 4b, to demonstrate that the benzoate is indeed in a position consistent with bridge-binding to two adjacent Ti5c sites along the [001] direction between the strands. A model of this binding is presented in Figure 4c, where the white circles above the Ti2O3 rows correspond to the bright spots in the atomically resolved images, and two possible conformations of the aromatic ring are shown: one perpendicular to [001] (upper) and one parallel (lower). Since the benzoate molecules appear in STM as oval shapes with their long axis along [110], it is expected that the upper arrangement in Figure 4c is more likely. This conformation would also allow hydrogen bonding between the ring hydrogen atoms and the oxygen atoms in the (1 2) added 1023
dx.doi.org/10.1021/jp209749m |J. Phys. Chem. C 2012, 116, 1020–1026
The Journal of Physical Chemistry C
ARTICLE
Figure 5. Structural models of an overlayer of benzoate on the rutile TiO2(110)(1 1) (a) and rutile TiO2(110)(1 2) (b) surface where the adsorbates bind to the Ti5c sites between the protruding strands containing bridging oxygen atoms (red). Each model contains two possible conformations of the benzoate: the aromatic ring perpendicular to the carboxylate (along [110]) on the left row and parallel to the carboxylate (along [001]) in the right row. Our STM data indicate that on the (1 2) surface the benzene ring is aligned perpendicularly (along [110]) due to hydrogen bonds between the H atoms (green) and the substrate O atoms (red), whereas on the (1 1) surface this interaction is not strong enough to rotate the ring because of the larger distance: 3.1 Å for (1 1) and 2.6 Å for (1 2).
rows, leading to further stabilization of the binding. In our results (see Figure 4a,b), we do not observe any nucleation on the crosslink. From the STM images we measure the benzoate molecules to have a height of ∼200 pm above the top of the (1 2) strands (∼350 pm in total), which is slightly higher than that recorded on the (1 1)-terminated surface (∼300 pm). This discrepancy has been attributed to the slight difference in bias voltages during scanning and a possible change of tip termination.
4. DISCUSSION Previous work by Guo et al. on benzoic acid adsorption on rutile TiO2(110) surfaces observed dimer and trimer formations of benzoate along the [110] direction, which they attributed to hydrogen bonding between the aromatic ring of one benzoate molecule and a hydrogen on the neighboring benzoate, where its aromatic ring is rotated by 90 (i.e., pointing along [110]).23,24 For illustration, ball models of benzoate adsorbed on the rutile TiO2(110)(1 1) and -(1 2) surfaces are shown in Figure 5a, b respectively. The right-hand row of each model has four molecules with the benzene ring arranged along the [001] direction, and the left-hand row has four molecules with 90 rotated benzene rings with their axes in the [110] direction. Only in the case of the (1 1) surface at low coverage can we observe some ordering along the [110] axis (see Figure 1b), indicating an interaction between the aromatic ring of one benzoate molecule and hydrogen on the neighboring benzoate. In this experiment, dosing was performed at room temperature under conditions similar to those described by Guo and coworkers.23 However, due to the distances involved, it would seem more favorable for these sort of interactions to occur along the [001] direction. This phenomenon is not observed, leading to the conclusion that further stabilization may be provided by hydrogen bonding between the ortho hydrogens on the aromatic ring of the benzoate and the bridging oxygen rows of the
TiO2(110)(1 1) surface, which would only be possible if the aromatic ring is perpendicular to the carboxylate group. Our results from high benzoate coverage on the (1 1) surface do not show any preferences toward ordering into dimer or trimer structures. In contrast, the benzoate overlayer forms a highly periodic (2 1) structure and we conclude that the interaction between single molecules is weak under the experimental conditions used. The resolution of the STM images gives us an almost perfect circular shape for all molecules, and it is therefore not possible to assign the orientation of the benzene ring or establish if hydrogen bonds are formed with the bridging oxygen at the surface. An explanation for the symmetrical appearance of the benzoate in these particular STM images may simply be that, due to the significant tip adsorbate interactions, the tip was terminated by a benzoate molecule transferred from the surface, leading to slightly reduced resolution (although still sufficient for molecular resolution). One reason for the different overlayer structure compared to the results reported by Guo et al.23 could be the elevated temperature of the substrate during dosing, which can overcome the molecule molecule interaction leading to stabilization by hydrogen bonding to the substrate. In experiments conducted with terephthalic acid overlayers on TiO2(110)(1 1) it was observed that at saturation coverage intermolecular interaction leads to inclination of the whole molecule and linkage between neighbors resulting in stable dimer rows along the [001] direction.28,29 STM measurements show three different image types exhibiting real dimer motifs, enhanced contrast between neighboring dimer rows, and (2 1) patterns. All of these patterns appear on the same surface, and they are attributed to changes in the tip mode.30 It is concluded by Zasada et al. in ref 31 that different tip modes only image certain electronic states in the molecular assembly resulting in the difference in appearance. A similar effect, in principle, is 1024
dx.doi.org/10.1021/jp209749m |J. Phys. Chem. C 2012, 116, 1020–1026
The Journal of Physical Chemistry C possible for benzoate on TiO2(110)(1 1). However, despite a large number of scans no dimer structure was ever observed in our measurements. Also, in the case of terephthalic acid, which is substantially larger than benzoic acid, the responsible hydrogen bonds are between the additional apical carboxyl groups, and the interaction between phenyl rings was found to be weak.31 Further evidence for H-bonding of adsorbed molecules with the substrate oxygen was found in the low-density coverage areas on the rutile TiO2(110)(1 2) surface (see Figure 4b). Here, an arrangement along the [110] axis of the benzene ring is shown by the elongated structure of single molecules. If we compare the 90 rotated benzene ring (see Figure 5) on both surfaces, it can be seen that the distance on the (1 2) between H-atoms and surface O-atoms is smaller than for the (1 1) surface. Furthermore, the molecular axis of the benzoate (indicated by the black dotted line in Figure 5) is in line with the O-atoms (assuming Ti2O3 added-rows), whereas on the (1 1) surface the axis lies between two oxygen atoms. Taking the structural parameters reported by Blanco-Rey et al.13 for the (1 2) surface, the standard dimensions of a benzene ring, and the geometry of the carboxylate reported by Sayago et al.,36 we calculate a distance of 2.6 Å between an H atom on an ortho carbon of the benzoate and the nearest surface O-atom compared with a H-bond distance of 3.1 Å for the (1 1) surface. Typical hydrogen bond lengths lie within the range of 2 3 Å,37 possibly explaining the increased interaction between the benzoate and the (1 2) surface leading to the observed rotation of the benzene ring. Interestingly, no benzoic acid was found to bind to the crosslinks on the TiO2(110)(1 2) surface. This is in strong contrast to the previously reported results for formic acid,22 where the cross-links are the favored adsorption sites. We suggest that extra stabilization is a result of hydrogen bonding between the hydrogen atoms in the ortho position on the benzene ring and the terminating surface oxygen in the (1 2) added-rows. We do not observe any direct evidence in our images for the location of the proton (H+) that results from the dissociation of the carboxylic acid group on either the (1 1) or (1 2) surfaces. At low benzoate exposures on the (1 2) reconstructed surface, however (see Figure 4b), there are a number of bright features on top of the added rows, often in very close proximity to the adsorbed benzoate, which may be signatures of the protons resulting from the acid dissociation.
5. CONCLUSIONS We have studied the adsorption of benzoic acid as a model molecule that resembles common dyes for the use in Gr€atzel type solar cells on rutile TiO2(110) surfaces with STM. The rutile TiO2(110)(1 1) prototypical surface and the heavily reduced, (1 2) reconstructed surface have been investigated at low and high benzoate coverage at raised temperatures. On the nearstoichiometric TiO2(110)(1 1) surface, adsorption at slightly raised temperatures (370 K) led to the formation of a saturated (2 1) overlayer. The benzoate binds dissociatively in a bidentate bridging geometry to two adjacent Ti5c cations in the [001] direction, which is consistent with observations of other carboxylic acids on the same surface. Evidence of intermolecular interactions between adsorbed benzoate molecules was also observed when deposited onto the TiO2(110)(1 1) surface at room temperature, with some local ordering of the adsorbates attributed to hydrogen bonding.
ARTICLE
High exposures of benzoic acid on TiO2(110)(1 2) at 350 K resulted in ordered (2 2) superstructures of adsorbed benzoate bound to the Ti5c cations between the (1 2) strands, with no adsorption observed at the cross-links, in contrast to an earlier study of formate adsorption on the TiO2(110)(1 2) surface. Atomically resolved images of the (1 2) reconstruction, with a low coverage of benzoate, confirm this adsorption site. In turn, this suggests that additional interactions between the benzoate and the surface (such as hydrogen bonding between the aromatic hydrogen and the oxygen in the substrate) may play a role in the adsorption and lead to the stable conformation of benzoate, where the aromatic ring and carboxylate π systems are perpendicular to each other. We conclude that the binding geometries of more complex carboxylic acids are likely to differ from smaller molecules, in part because of hydrogen bonding to the substrate.
’ AUTHOR INFORMATION Corresponding Author
*Tel.: +44 (0)20 7679 7979. Fax: +44 (0)20 7679 0595. E-mail:
[email protected].
’ ACKNOWLEDGMENT We thank Quanmin Guo for providing the original figures in ref 23 as well as useful discussions. We also thank Chi-Ming Yim for his assistance in drawing models, as well as Felicity Sartain, Irene Chomiak, and Deena Modeshia for valuable discussions. This work was funded by EPSRC (U.K.) and the Deanship of Scientific Research at King Abdulaziz University (T/80/429). ’ REFERENCES (1) Fujishima, A.; Honda, K. Nature 1972, 238, 37–38. (2) Fujishima, A.; Zhang, X.; Tryk, D. Surf. Sci. Rep. 2008, 63, 515–582. (3) Henderson, M. A. Surf. Sci. Rep. 2011, 66, 185–297. (4) Yates, J. T. Surf. Sci. 2009, 603, 1605–1612. (5) Thompson, T. L.; Yates, J. T. Chem. Rev. 2006, 106, 4428–4453. (6) Linsebigler, A.; Lu, G.; Yates, J. T. Chem. Rev. 1995, 95, 735–758. (7) Diebold, U. Surf. Sci. Rep. 2003, 48, 53–229. (8) Bowker, M. Surf. Sci. 2009, 603, 2359–2362. (9) Chen, M.; Goodman, D. W. Science 2004, 306, 252–255. (10) Pang, C. L.; Lindsay, R.; Thornton, G. Chem. Soc. Rev. 2008, 37, 2328–2353. (11) Onishi, H.; Iwasawa, Y. Surf. Sci. 1994, 313, 783–789. (12) Blanco-Rey, M.; Abad, J.; Rogero, C.; Mendez, J.; Lopez, M.; Martin-Gago, J.; Andres, P. D. Phys. Rev. Lett. 2006, 96, No. 055502. (13) Blanco-Rey, M.; Abad, J.; Rogero, C.; Mendez, J.; Lopez, M.; Roman, E.; Martin-Gago, J.; Andres, P. D. Phys. Rev. B 2007, 75, No. 081402. (14) Pang, C. L.; Haycock, S.; Raza, H.; Murray, P.; Thornton, G. Phys. Rev. B 1998, 58, 1586–1589. (15) Cocks, I.; Guo, Q.; Williams, E. Surf. Sci. 1997, 390, 119–125. (16) Pieper, H. H.; Venkataramani, K.; Torbruegge, S.; Bahr, S.; Lauritsen, J. V.; Besenbacher, F.; Kuehnle, A.; Reichling, M. Phys. Chem. Chem. Phys. 2010, 12, 12436–12441. (17) Bennett, R.; Stone, P.; Price, N.; Bowker, M. Phys. Rev. Lett. 1999, 82, 3831–3834. (18) O’Regan, B.; Graetzel, M. Nature 1991, 353, 737–740. (19) Tao, J.; Luttrell, T.; Bylsma, J.; Batzill, M. J. Phys. Chem. C 2011, 115, 3434–3442. (20) Guo, Q.; Cocks, I.; Williams, E. J. Chem. Phys. 1997, 106, 2924–2931. (21) Gutierrez-Sosa, A.; Martinez-Escolano, P.; Raza, H.; Lindsay, R.; Wincott, P.; Thornton, G. Surf. Sci. 2001, 471, 163–169. 1025
dx.doi.org/10.1021/jp209749m |J. Phys. Chem. C 2012, 116, 1020–1026
The Journal of Physical Chemistry C
ARTICLE
(22) Bennett, R.; Stone, P.; Smith, R.; Bowker, M. Surf. Sci. 2000, 454, 390–395. (23) Guo, Q.; Cocks, I.; Williams, E. Surf. Sci. 1997, 393, 1–11. (24) Guo, Q.; Williams, E. Surf. Sci. 1999, 433, 322–326. (25) Schnadt, J.; Schiessling, J.; O’Shea, J. N.; Gray, S. M.; Patthey, L.; Johansson, M. K.-J.; Shi, M.; Krempasky, J.; Ahlund, J.; Karlsson, P. G.; Persson, P.; Martensson, N.; Bruhwiler, P. A. Surf. Sci. 2003, 540, 39–54. (26) Schnadt, J.; O’Shea, J. N.; Patthey, L.; Kjeldgaard, L.; Ahlund, J.; Nilson, K.; Schiessling, J.; Krempasky, J.; Shi, M.; Karis, O.; Glover, C.; Siegbahn, H.; Martensson, N.; Bruhwiler, P. A. J. Chem. Phys. 2003, 119, 12462–12472. (27) Schnadt, J.; O’Shea, J. N.; Patthey, L.; Krempasky, J.; Martensson, N.; Bruhwiler, P. A. Phys. Rev. B 2003, 67, No. 235420. (28) Tekiel, A.; Prauzner-Bechcicki, J. S.; Godlewski, S.; Budzioch, J.; Szymonski, M. J. Phys. Chem. C 2008, 112, 12606–12609. (29) Rahe, P.; Nimmrich, M.; Nefedov, A.; Naboka, M.; W€oll, Ch.; K€uhnle, A. J. Phys. Chem. C 2009, 113, 17471–17478. (30) Prauzner-Bechcicki, J. S.; Godlewski, S.; Tekiel, A.; Cyganik, P.; Budzioch, J.; Szymonski, M. J. Phys. Chem. C 2009, 113, 9309–9315. (31) Zasada, F.; Piskorz, W.; Godlewski, S.; Prauzner-Bechcicki, J. S.; Tekiel, A.; Budzioch, J.; Cyganik, P.; Szymonski, M.; Sojka, Z. J. Phys. Chem. C 2011, 115, 4134–4144. (32) Sanchez-Sanchez, C.; Gonzalez, C.; Jelinek, P.; Mendez, J.; de Andres, P. L.; Martin-Gago, J. A.; Lopez, M. F. Nanotechnology 2010, 21, No. 405702. (33) Teobaldi, G.; Hofer, W. A.; Bikondoa, O.; Pang, C. L. Chem. Phys. Lett. 2007, 437, 73–78. (34) Aizawa, M.; Morikawa, Y.; Namai, Y.; Morikawa, H.; Iwasawa, Y. J. Phys. Chem. B 2005, 109, 18831–18838. (35) Fukui, K.-I.; Sakai, M. J. Phys. Chem. B 2006, 110, 21118–21123. (36) Sayago, D. I.; Polcik, M.; Lindsay, R.; Toomes, R. L.; Hoeft, J. T.; Kittel, M.; Woodruff, D. P. J. Phys. Chem. B 2004, 108, 14316–14323. (37) Greenwood, N.; Earnshaw, A. Chemistry of the Elements; Butterworth-Heinemann: Oxford, U.K., 1997.
1026
dx.doi.org/10.1021/jp209749m |J. Phys. Chem. C 2012, 116, 1020–1026
