Electronic Structure, Chemical Interactions and Molecular Orientations

Nov 15, 2011 - Hao MaHe WangPeng LiXiaolei WangXiaoxia HanChengyan HeBing Zhao .... Liang Cao , Yu-Zhan Wang , Jian-Qiang Zhong , Yu-Yan Han ...
1 downloads 0 Views 3MB Size
ARTICLE pubs.acs.org/JPCC

Electronic Structure, Chemical Interactions and Molecular Orientations of 3,4,9,10-Perylene-tetracarboxylic-dianhydride on TiO2(110) Liang Cao,†,‡ Yuzhan Wang,‡ Jianqiang Zhong,‡ Yuyan Han,† Wenhua Zhang,† Xiaojiang Yu,§ Faqiang Xu,*,† Dong-Chen Qi,*,‡ and Andrew T. S. Wee‡ †

National Synchrotron Radiation Laboratory, School of Nuclear Science and Technology, University of Science and Technology of China, Hefei, Anhui 230029, People's Republic of China ‡ Department of Physics, National University of Singapore, 2 Science Drive 3, Singapore 117542, Singapore § Singapore Synchrotron Light Source, National University of Singapore, 5 Research Link, Singapore 117603, Singapore ABSTRACT: The electronic structure, molecular orientations, and interfacial energy level alignment of 3,4,9,10-perylenetetracarboxylic-dianhydride (PTCDA) molecules on rutile TiO2(110) 1  1 surface have been investigated using synchrotronbased photoemission spectroscopy (PES) and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy. The evolution of PES spectra as a function of PTCDA coverage is interpreted as possible chemical reactions that strongly couple the TiO2 surface Ti and O atoms with PTCDA molecules. The emergence of an interfacial gap state observed at the lower binding energy side of the highest occupied molecular orbital (HOMO) of PTCDA corroborates the strong electronic coupling between PTCDA molecules and TiO2 substrate. In addition, the molecular orientation of PTCDA is found to vary with coverage due to the strong interfacial interactions. It adopts a slightly tilting, disordered, and nearly lying-down geometry in the submonolayer, monolayer, and multilayer regimes, respectively. The strong electronic coupling and the relative energy alignment at the PTCDA/TiO2 interface could facilitate the interfacial charge delocalization, which is desirable for making PTCDA-based dyesensitized solar cell (DSSC) devices.

’ INTRODUCTION In recent years, organic semiconductors have attracted widespread interest due to their promising applications in the field of organic electronics ranging from organic light emitting diodes (OLEDs), organic field effect transistors (OFETs), organic photovoltaic cells (OPVCs),13 to dye-sensitized solar cells (DSSCs).4,5 It has been widely acknowledged that interfaces between organic semiconductor molecules and their functional substrates largely determine the device performance.611 One of the most intensively studied functional substrates for organic devices is titanium dioxide (TiO2) due to its superior properties for various applications including (photo)catalysis,12 chemical sensors,13 and in particular DSSCs.14 A number of studies on organic molecules such as metal-phthalocyanines,15 metal-porphyrins,16,17 perylene derivatives,18 and several other small molecules19,20 deposited on TiO2 surface have been reported, focusing on the optical and electronic properties at their interfaces. Among the most widely studied organic semiconductor molecules, 3,4,9,10-perylene-tetracarboxylic-dianhydride (PTCDA) is a typical model planar molecule with excellent optoelectronic properties and chemical stability.21,22 The PTCDA molecule, as shown in Figure 1, consists of a perylene core and two anhydride r 2011 American Chemical Society

functional groups. In general, PTCDA molecules form highly ordered structures on various substrates, for example, Au,2 Ag,23 HOPG,24 NaCl,25,26 and other organic semiconductors,27 in a planar geometry with a bulklike herringbone arrangement due to the strong multiple intermolecular CdO 3 3 3 HC hydrogen bonding.2,2731 Several studies on the adsorption geometry of PTCDA on TiO2(011) 2  1 surface have been reported using scanning probe microscopy (SPM).32,33 It was found PTCDA molecules form 1D molecular lines with the long axis of lyingdown molecules oriented along the bridging oxygen rows of the TiO2 substrate. The driving force for the observed self-assembled molecular superstructures was proposed to be the dominant moleculesubstrate interactions, which are very likely to involve chemical reactions between PTCDA and TiO2 although spectroscopic evidence and detailed chemisorption mechanisms are still lacking. Moreover, there are few studies on the electronic structures and energy level alignment at the PTCDA/TiO2(110) interface.34 Received: August 31, 2011 Revised: November 8, 2011 Published: November 15, 2011 24880

dx.doi.org/10.1021/jp2083924 | J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C

ARTICLE

Figure 1. Molecular structure of PTCDA.

In this work, we systematically investigate the electronic structure, chemical interactions, molecular orientations, and energy level alignment, critical factors that influence device performance, of PTCDA on reduced rutile TiO2(110) 1  1 surface using synchrotron-based photoemission spectroscopy (PES) and near-edge X-ray absorption structure (NEXAFS) spectroscopy. The PTCDA molecules were found to grow layer-by-layer on TiO2 surface below 2 monolayers (ML) with a small tilt angle due to strong moleculesubstrate interactions and chemical reactions. At higher coverage, island growth with lying-down geometry dominates due to reduced moleculesubstrate interactions. The intermolecular hydrogen bonding between PTCDA molecules stabilizes the planar geometry, similar to PTCDA on other substrates.35 Finally, we determined the energy level alignment from the interface to the bulk of the organic layers based on thickness dependent PES results.

’ EXPERIMENTAL DETAILS The sample preparation, PES, and NEXAFS measurements were carried out in situ at the SINS beamline of Singapore Synchrotron Light Source equipped with a Scienta R4000 electron energy analyzer.36 All of the measurements were performed at room temperature in an ultrahigh vacuum (UHV) chamber with a base pressure of 1  1010 mbar. The photon energy was calibrated using the Au 4f7/2 core level peak at 84.0 eV of a sputter-cleaned gold foil in electrical contact with the sample. The rutile TiO2(110) single crystal (KMT Corp.) was cleaned by repeated cycles of Ar+ sputtering and annealing at 900 K in UHV. The cleanliness of the sample surface was checked by PES with no carbon contaminations detected. Reduced TiO2(110), which is evidenced by the emergence of gap states arising from oxygen vacancies in the valence band (VB) spectrum, was obtained following this sample preparation procedure. Lowenergy electron diffraction (LEED) showed a clear 1  1 pattern from the clean surface, indicating low concentration of thermally produced point defects. PTCDA molecules (Sigma-Aldrich) were thoroughly degassed for several hours before subliming onto the TiO2(110) substrate using a standard Knudsen-cell. The film thickness was estimated using PES by the attenuation of the substrate Ti 2p3/2 signal intensity.37 The nominal coverage of one monolayer is defined as 3.3 Å assuming flat lying molecular orientation.38 The PES measurements were employed to examine the evolution of electronic structure as a function of PTCDA coverage. The work function was measured using 60 eV photon energy with a 10 V bias applied to the sample to overcome the work function of the analyzer. All PES spectra were collected at normal emission. The least-squares peak fit analysis was performed using Voigt photoemission profiles with constant Lorentzian widths

Figure 2. Ti 2p3/2 core level spectra (a) and intensity ratio of Ti3+/Ti4+ (b) as a function of PTCDA thickness.

(intrinsic lifetime broadening) of 80 meV for C 1s and 100 meV for O 1s respectively, whereas the Gaussian widths (extrinsic broadening) were fitting parameters with constraints.39 All satellite peaks were fitted using pure Gaussian line profiles. The C K-edge NEXAFS spectra were collected in total electron yield (TEY) mode by measuring the sample current with a photon energy resolution of 200 meV. The linear polarization factor of the X-ray beam was measured to be more than 90%. All NEXAFS spectra were first normalized to the incident photon intensity (I0) monitored by refocusing mirror.36 Furthermore, the spectra were subtracted by a I0 corrected background spectrum obtained from the clean TiO2 substrate.40,41

’ RESULTS AND DISCUSSION A. Interfacial Reactions and Electronic Structure. Part a of Figure 2 shows the evolution of Ti 2p3/2 core level state of TiO2(110) substrate upon the deposition of PTCDA molecules. The top spectrum shows the deconvolution of Ti 2p3/2 peak for the clean TiO2(110). The dominant component at ∼459.1 eV is associated with the regular Ti4+ ions in the bulk and 5-fold coordinated Ti (Ti5f) atoms on the surface,42 whereas the shoulder with ∼1.3 eV lower bind energy (BE) is attributed to the surface Ti3+ ions,43 which arises from the presence of surface oxygen vacancies. The BE of both features remain constant with the increasing of PTCDA thickness, whereas their intensities are gradually attenuated by the PTCDA deposition. In general, COOH-containing organic molecules adsorbed on TiO2 surface prefer to bond at the surface O-vacancy sites: the hydroxylic O-atom is deprotonated and covalently bonded to one of the adjacent Ti3+ ions, whereas the carboxylic O-atom is attached to the nearby Ti5f atoms.44 Consequently, this type of reaction leads to the quenching of the Ti3+ components. However, in our case, the adsorbed PTCDA molecules do not suppress the intensity of the Ti3+ component (c.f., part b of Figure 2), suggesting 24881

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C

ARTICLE

Figure 3. O 1s core level spectra (a), intensity ratio of the PTCDA O 1s components O1 and O2 (b), and peak fitting results (c)(f) for PTCDA molecules on TiO2(110) with increasing thickness. Dashed straight lines in (a) indicate the O 1s components originating from PTCDA molecules. The spectrum for clean TiO2 was fitted using a linear background, and all other spectra were fitted using Shirley plus linear background.

Table 1. Peak Fitting Parameters of Energy Positions (EB), Gaussian Full Width Half Maximum (WG) and Relative Intensities (Irel) for O 1s Components of PTCDA on TiO2(110). a Peak Osub

Peak O1

EB (eV)

WG (eV)

Irel %

0

530.46

1.16

100.0

0.2

530.47

1.16

0.5

530.43

1.16

1

530.51

2 4

Peak O2

EB (eV)

WG (eV)

Irel %

EB (eV)

WG (eV)

Irel %

96.9

531.92

1.12

2.4

533.92

1.12

0.7

90.1

531.86

1.31

7.0

533.83

1.12

2.9

1.16

72.7

531.94

1.37

17.3

533.91

1.21

10.0

530.48

1.16

62.5

531.92

1.33

22.4

533.81

1.28

15.1

530.53

1.16

51.6

531.95

1.34

28.7

533.89

1.29

19.7

8

530.48

1.15

35.5

531.97

1.33

37.7

533.93

1.30

26.8

16 28

530.43 530.44

1.15 1.15

21.1 7.5

531.93 531.94

1.28 1.29

45.5 53.4

533.86 533.89

1.27 1.29

33.4 39.1

thickness (ML)

a

Lorentzian full width half maximum (WL) is set to 0.10 eV in all cases. Relative intensities of each peak are evaluated with respect to the total integrated O 1s intensity.

that PTCDA molecules are not likely bonded to O-vacancies on TiO2(110). Part a of Figure 3 shows the O 1s core level spectra measured with 600 eV photon energy at the PTCDA/TiO2(110) interface with increasing PTCDA coverage. The O 1s spectrum of clean TiO2 is characterized by a single bulk component (Osub) at about 530.5 eV with an asymmetric line profile due to the Ti3+ associated tailing states at higher binding energy.45 The substrate Osub signal decreases rapidly in intensity but remains unchanged in BE with PTCDA deposition. These phenomena confirm that the oxygen vacancies are not filled by PTCDA molecules, which otherwise would lead to a band bending on TiO2 surface.44 With the deposition of organic molecules, two new features (labeled O1 and O2) gradually develop at ∼531.9 eV and ∼533.9 eV, respectively. The peak parameters are listed in Table 1. O1 is mainly attributed to the carbonyl (CdO) group of the adsorbate molecules, whereas O2 originates mainly from the bridging (O) oxygen atoms (c.f. Figure 1). Moreover, the O2

component contains contributions from the shakeup of peak O1,39 since the intensity ratio of peak O1 to peak O2 of 1.4 for 28 ML is lower than the stoichiometric ratio of 2(CdO): 1(O). In fact, the intensity ratio between peak O1 and peak O2 decreases rapidly from 3.5 to 1.5 with increasing PTCDA coverage below 2 ML (part b of Figure 3) and becomes relatively stable with further deposition. One possible scenario leading to such thickness dependent behavior is the chemical reaction of the anhydride terminal group (OdCOCdO) with the TiO2 surface through the ring-opening from the COC bonds, which are then covalently bonded to the Ti5f atoms and the adjacent bridging oxygen (Ob) atoms on the TiO2 surface by forming COTi bonds as observed for other perylene dyes with anhydride groups on TiO2 surfaces.35 The BE of oxygen in this bonding arrangement is similar to that of the O1 atoms in isolated PTCDA molecules,17 and hence the reacted molecular species increase the intensity ratio of O1 to O2. However, the O2 component is observed at the beginning of PTCDA deposition, 24882

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C

Figure 4. (a) C 1s core level spectra as a function of PTCDA thickness. All C 1s core level spectra are normalized to the maximum intensity for clarity. (b) Integrated intensities of the C 1s core level spectra, deconvoluted components and intensity ratio as a function of PTCDA thickness on TiO2(110). The intensities of C 1s, peak C2, and peak C1' are set to 1 at the maximum intensity.

indicating that only part of the anhydride groups of PTCDA molecules is involved in the reaction. The evolution of C 1s core level spectra of the PTCDA thin film with step-by-step deposition is shown in part a of Figure 4. The top and bottom spectra present the deconvolution of the peaks for submonolayer (0.2 ML) and multilayer (28 ML), respectively. The peak parameters for different PTCDA thicknesses are listed in Table 2. In multilayer spectra, the typical C 1s core level consists of six components. Peaks C1 and C2 are assigned to the perylene core carbon atoms with CdCC and CH bonds, respectively. Peak C3 is attributed to the anhydride (OCdO) carbon atoms, and peak SC3 is its associated shakeup satellite. The other weak structures (peak SC1 and peak S2C3) are also attributed to shakeup features of peak C1 and C3, respectively.39,46 The energy positions and intensity distribution of the C 1s shakeup satellite structures have been discussed in great details for large π-conjugated molecules.39,47 In general, the intensities of shakeup satellites depends on the overlap integral between the initial frozen core-hole state and the final excited ionic state (with valence excitations). Consequently, for polar molecules or molecules with polar structures the corresponding shakeup satellites tend to be intense, as part of the molecule can act as electron donor and provide electrons to screen the coreionized atomic sites. This is the case for the relatively intense

ARTICLE

satellite peaks associated with the anhydride carbon (C3) and the perylene carbon (C1) due to the high electron donating strength of the perylene core.39 At submonolayer coverage, a new component (Peak C1') at ∼284.3 eV appears at lower BE side, accompanied by spectral weight decrease of peak C1. It suggests that, in addition to the proposed reaction involving the anhydride groups, some C-atoms in the perylene core of PTCDA could also react with the TiO2 substrate, forming a new species giving rise to peak C1'. Part b of Figure 4 shows the relative intensities of C 1s and the corresponding deconvoluted components as a function of PTCDA thickness, which helps us to determine the reaction mechanism between PTCDA and TiO2 surface. First of all, the thickness dependence of the peak C2 intensity follows almost the same trend as that of integrated C 1s peaks. Furthermore, while the intensity of peak C1' is detected at about 1 ML and almost vanish in multilayer, the intensity ratio between peak (C1 + C1') and peak C2 remains almost constant and corresponds well with the stoichiometric ratio of the perylene carbon in CCdC and CH bonding configurations. Both observations suggest that it is the CdCC not CH in the perylene core that takes part in the interfacial reaction, and they are most likely to be covalently attached to the Ob atoms on TiO2. The reaction may also lead to the electron transfer from the TiO2 surface to the perylene core C-atoms, reducing the BE of component C1'. However, the intensity ratio between the peak (C3 + SC3) and C 1s components in the C 1s spectra increases as a function of PTCDA thickness. The lower ratio indicates the reduction of carbonyl (CdO) carbon, which is consistent with the formation of COTi bonds as discussed above. In addition to core level PES, the work function and VB spectra at the PTCDA/TiO2 interface can provide important information on the interfacial interactions. Part a of Figure 5 shows the low kinetic energy region of the PES spectra, which presents the evolution of the secondary electron cutoff with the deposition of PTCDA. Depending on the surface treatment, the work function of pristine TiO2(110) has been reported with values ranging from 3.9 to 6.2 eV.48 In present study, the work function for the clean TiO2 substrate, as indicated by the cutoff energy position, is found to be 5.1 ( 0.1 eV. Upon the deposition of 0.2 ML PTCDA, the work function is slightly decreased to 4.8 ( 0.1 eV, and remains almost constant until 1 ML. Further deposition of molecules leads to gradual decrease of work function to 4.6 ( 0.1 eV for 16 ML PTCDA coverage and more. Part b of Figure 5 shows the VB features for the clean TiO2 (110) surface and with PTCDA deposition. The VB of clean TiO2 is dominated by dispersive O 2p bands hybridized with Ti 3d states lying between 3.19.5 eV.49 The valence band maximum (VBM) occurs at ∼3.1 eV, which puts its Fermi level (EF) close to its conduction band edge. This is consistent with the intrinsic n-type nature of TiO2 due to the native defects. Moreover, a pronounced surface defect state centered at ∼0.9 eV originating from Ti 3d nonbonding state is clearly resolved for the clean surface (part c of Figure 5).42,45,4850 With the adsorption of PTCDA molecules, the intensity of entire VB structures from TiO2 is attenuated. At high molecular coverages, the characteristic molecular frontier orbitals from PTCDA molecules51 are visible with the highest occupied molecular orbital (HOMO) derived state located at about 2.5 eV.52 More information can be found in the close-up of the VB region close to EF as shown in part c of Figure 5. The defect state in the band gap at ∼0.9 eV below EF containing mainly Ti 3d derived character49,50 24883

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C

ARTICLE

Table 2. Peak Fitting Parameters of Energy Positions (EB), Gaussian Full Width Half Maximum (WG) and Relative Intensities (Irel) for Major C 1s Components of PTCDA on TiO2(110)a Peak C10

Peak C1

Peak C2

thickness (ML)

EB (eV)

WG (eV)

Irel %

EB (eV)

WG (eV)

Irel %

EB (eV)

WG (eV)

Irel %

0.2

285.49

0.80

27.3

284.25

0.84

25.7

284.91

0.89

39.2

0.5

285.43

0.89

36.4

284.34

0.85

15.8

284.85

0.94

39.0

1 2

285.46 285.44

0.88 0.88

39.2 43.1

284.30 284.28

0.85 0.78

11.8 7.0

284.88 284.85

0.92 0.92

38.0 37.5

4

285.47

0.86

44.5

284.29

0.72

4.6

284.88

0.91

37.4

8

285.47

0.84

46.1

284.34

0.68

2.6

284.90

0.95

36.8

16

285.42

0.84

49.1

284.84

0.95

36.0

28

285.46

0.83

48.5

284.91

0.95

36.1

Peak SC3

Peak C3 0.2 0.5

EB (eV) 289.01 289.03

WG (eV) 0.83 0.96

Irel % 5.8 6.1

EB (eV) 290.30 290.36

WG (eV) 1.02 1.24

Irel % 2.0 2.7

1

289.01

0.96

7.4

290.36

1.35

3.7

2

288.99

0.96

8.3

290.39

1.26

4.0

4

289.00

0.94

8.8

290.38

1.26

4.5

8

289.01

0.88

9.5

290.36

1.18

5.1

16

288.97

0.86

9.8

290.37

1.16

5.0

28

288.99

0.88

10.2

290.35

1.16

5.2

a

Relative intensities of each peak are evaluated with respect to the total integrated C 1s intensity. Lorentzian full width half maximum (WL) is set to 0.08 eV in all cases.

Figure 5. Low kinetic energy secondary electron cutoff (a), VB spectra (b), and the close-up of VB features close to the EF (c) of PTCDA on TiO2(110) with increasing thickness (hν = 60 eV, normal emission). The arrows in (c) indicate the position of the VBM of TiO2.

is clearly visible. Careful analysis reveals that the intensity of the defect state is slightly enhanced instead of being suppressed with the initial deposition of 0.5 ML molecules. This result confirms that the oxygen vacancies are not quenched by adsorbed PTCDA molecules, which is consistent with the analysis of the core level spectra, because this defect state is associated with the presence of surface Ti3+ ions originating from oxygen vacancies. Its increase in intensity may be caused by the interaction between PTCDA and Ti atoms, which can give rise to a new gap state at about 1.0 eV below EF.53

In addition, the HOMO of PTCDA on TiO2(110) at low coverages (below 1 ML) has an apparent asymmetric peak profile with a shoulder appearing at ∼2.1 eV as indicated by the vertical lines in part c of Figure 5. Its intensity saturates when PTCDA thickness reaches approximately 1 ML, afterward only a tail is visible at the lower BE side of HOMO for PTCDA coverages from 2 to 8 ML. It finally disappears at higher thickness and the HOMO peak is restored to the symmetric shape. Therefore, this new state is closely related to the interfacial reacted PTCDA species. 24884

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C

ARTICLE

Figure 6. Intensity ratio of Ti 2p3/2 as a function of PTCDA thickness (d) on TiO2(110). The core level integrated intensities for Ti 2p3/2 from the TiO2 substrate are normalized to that of clean substrate. The red dashed line represents the theoretical exponential attenuation curve of the substrate Ti 2p3/2 signal for an ideal layer-by-layer growth mode.

It is attributed to newly formed molecular states due to the reaction of the perylene core of PTCDA molecules with the TiO2 substrate. B. Growth Mode and Molecular Orientation. The growth mode of organic overlayers can be deduced from the attenuation of the integrated core level intensities from the substrate as a function of the film thickness.54 As shown in Figure 6, the intensity of Ti 2p3/2 peak decreases exponentially for PTCDA film thickness below 2 ML, which well corresponds to a layer-bylayer growth mode. At higher thicknesses, the intensity decays much slower with the increase of film thickness. The large deviation from the theoretical decay curve (red dashed line) assuming the exponential attenuation dependence for an ideal layer-by-layer growth mode clearly points to an island growth mode. This layer-plus-island growth mode (Stranski-Krastanov mode) is also consistent with the evolution of adsorbate C 1s (c.f., part b of Figure 4) and substrate O 1s core level (not shown here) intensities as a function of PTCDA film thickness. In general, Stranski-Krastanov mode is commonly observed at organic/inorganic interface. The moleculesubstrate interactions dominate for the first one or two monolayers and lead to the layer-by-layer growth. At higher thickness, the molecule substrate interactions are suppressed and the intermolecular interactions facilitate the island growth mode due to much reduced surface diffusion barrier. The molecular orientation of PTCDA molecules on TiO2 is studied by angular dependent NEXAFS. In a molecular system, NEXAFS monitors the resonant electronic transitions from a specific core level to unoccupied molecular orbitals with π* or σ* symmetry. It is well established that the resonant intensity is enhanced if electric field vector E of the incident X-ray is parallel to the molecular orbital vector, and the intensity is suppressed if E is perpendicular to the orbital vector.55 For planar πconjugated molecules like PTCDA, π* and σ* orbitals are oriented essentially in-plane and out-off-plane respectively. Therefore, the molecular orientation of PTCDA in the films on TiO2(110) could be easily derived by the angular dependent intensities of π*/σ* resonances in its NEXAFS spectra. Figure 7 shows the angular dependent C K-edge NEXAFS spectra for submonolayer (0.5 ML), monolayer and multilayer (28 ML) PTCDA on TiO2(110). The four sharp resonant peaks below 291 eV are attributed to the electronic transitions from the C 1s core level into individual π* orbitals. The lowest resonance

Figure 7. Angular dependent C K-edge NEXAFS spectra for PTCDA submonolayer (0.5 ML), monolayer, and multilayer (28 ML) on TiO2 (110) The X-ray incident angles are defined with respect to the substrate surface. Schematic representations of molecular orientation at different coverage are shown on the right.

(peak Cp1) occurring at ∼284.1 eV is attributed to transitions from the aromatic perylene core carbon atoms to the lowest unoccupied molecular orbital (LUMO), whereas the strong resonance (peak Cp2) at ∼285.6 eV corresponds to transitions from the perylene core C-atoms to the next three orbitals of LUMO+1 ∼ LUMO+3. The resonances Ca1 and Ca2 at about 287.8 eV and 288.4 eV originate from transitions from the anhydride C-atoms into LUMO and LUMO+1 ∼ LUMO+3, respectively.56 The broad absorption features at higher photon energy are assigned to the C 1s f σ* transitions. They are in good agreement with previously reported NEXAFS spectra for PTCDA on other substrates.25,28,57 However, a closer look at the first two resonances in the NEXAFS spectra for submonolayer coverage of PTCDA reveals that the intensity ratio between peak Cp1 and peak Cp2 at normal incidence is different from that of monolayer and multilayer. The overall spectral features are also considerably broadened. These spectral variations suggest that the LUMO orbitals are distorted/disrupted because of the reaction between PTCDA molecules and TiO2(110) surface, corroborating the PES results. In general, for all PTCDA coverages, the π* resonances are most intense at grazing incidence (θ = 20°), suggesting the molecules are nearly lying-down on the substrate. A more detailed analysis of the angular dependent NEXAFS reveals that the molecular orientation varies with PTCDA coverage. Assuming a random azimuthal orientation between molecular plane and substrate, the π* resonances intensity (Iπ*) ratio at 90° and 20° incident angles (θ) can be expressed as follows:55 Iπ/ ðα, 90°Þ Pðsin2 α sin2 90° þ 2 cos2 α cos2 90°Þ þ ð1  PÞ sin2 α ¼ Iπ/ ðα, 20°Þ Pðsin2 α sin2 20° þ 2 cos2 α cos2 20°Þ þ ð1  PÞ sin2 α

ð1Þ in which P = 0.90 is the linear polarization factor. And α is the average tilt angle of molecular plane with respect to the substrate surface plane. Using the intensity ratio of the most intense peak Cp2, we have calculated the average tilt angle α to be 40° ( 10°, 24885

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C

ARTICLE

causes for the abrupt work function decrease at the interface,59 such as the Pauli exclusion effect on most metallic surfaces, interfacial charge transfer, intrinsic molecular dipole moment, and so forth, are not applicable to the PTCDA/TiO2 interface, leaving the chemical bond formation at interface as the most possible origin. Further decrease in work function may be caused by the change of molecular orientation with thickness from inclined to flat lying, which is often observed for orientationally ordered molecular assemblies on a substrate surface.60 The LUMO position is estimated by adding the HOMOLUMO gap (Et) of 3.2 eV61 to the measured HOMO peak position, which puts it directly above the CBM of TiO2. This relative energy alignment would benefit the ultrafast electron transfer from the LUMOs of PTCDA dye to TiO2, which will be discussed in a separate report. Figure 8. Schematic energy level alignment diagram at PTCDA/ TiO2(110) interface.

51° ( 10°, and 27° ( 10° for submonolayer, monolayer, and multilayer, respectively. At submonolayer coverage in which all of the molecules are believed to react with TiO2 surface, the molecular plane is found to be significantly tilted away from the substrate instead of the completely planar adsorption geometry usually found on inert substrates.25,26 As discussed in previous sections, the chemisorption of PTCDA on TiO2 is likely to occur through the covalent attachment of one of its anhydride groups to the adjacent Ti5f atoms with its perylene core bounded to the Ob atoms of TiO2. Therefore, the PTCDA molecules adsorb on top of Ob rows. This adsorption geometry, as suggested by a recent DFT calculation of PTCDI molecules, which has the similar molecular structure with PTCDA but the bridging O atoms are substituted by N atoms, naturally leads to a tilted molecular orientation on TiO2(110).18 In addition, the interfacial reaction could also distort or bend the molecular plane, further increasing the molecular tilt angle. For monolayer coverage, the average tilt angle of molecules is further increased, indicating the subsequent deposited PTCDA molecules are tilted further away from the substrate, possibly due to increased steric hindrance of neighboring adsorbed molecules. Consequently, the nominal average α lies close to the magic angle of 54.7°,55 which essentially indicates a disordered molecular orientation. The average tilt angle is considerably reduced when the film gets thicker, suggesting the molecules in multilayer are adopting nearly lying-down geometry and the orientational order is greatly improved. This is readily helped by the strong multiple intermolecular hydrogen bonding which dominates over the moleculesubstrate interactions. In addition, it is known that the tilt angle estimated from standard NEXAFS analysis could be overestimated due to its sensitivity to the nonvanishing intensity at normal incidence.58 Therefore, the actual tilting angle of PTCDA in multilayer may be much smaller or even close to zero. C. Energy Level Alignment. On the basis of the above results, the schematic energy level alignment at PTCDA/TiO2(110) interface is proposed and shown in Figure 8. By extrapolating the VB leading edge, the TiO2 VBM is found at 3.1 eV below EF, thus placing its conduction band minimum (CBM) 0.1 eV above EF by assuming a band gap of 3.2 eV.48 An interfacial dipole (Δ1) of about 0.3 eV is present at the interface, which may originate from the dipolar bond (COTi) formed and associated charge redistribution between PTCDA and TiO2. Other common

’ CONCLUSIONS The chemical interactions, molecular orientations, and energy level alignment of PTCDA on rutile TiO2(110) surface have been systemically investigated using synchrotron-based PES and NEXAFS technologies. From the thickness dependent PES spectra PTCDA molecules are proposed to covalently attach to the TiO2(110) surface through its anhydride group and perylene core. The transformation of thickness dependent molecular orientation from slightly tilting at 0.5 ML, disordered at 1 ML, and lying-down at 28 ML is observed by angular dependent NEXAFS. The tilted orientation of interfacial molecular layers is attributed to the chemical reaction, which becomes less significant as coverage increases and the intermolecular hydrogen bonding redominates for the lying-down geometry in multilayer PTCDA. Finally, the energy level alignment diagram of the interface is plotted, in which the LUMO of PTCDA is directly above the CBM of TiO2. The understanding of the interfacial interactions and electronic structure at the PTCDA/TiO2 interface may have important implications for developing PTCDAbased DSSC devices. In particular, the strong electronic coupling and the relative energy alignment at the interface facilitate the interfacial charge delocalization and are beneficial for the device performance. ’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected], Tel: +86-551-3602127, Fax: +86-5515141078, (F.X.); E-mail: [email protected], Tel: +65-6516-7695, Fax: +65-6773-6734 (D.Q.).

’ ACKNOWLEDGMENT We are grateful for the financial support by National Natural Science Foundation of China (Grant No. 10975138 and 11175172), Singapore ARF (Grant No.R398-000-056-112), and NUS Core Support (Grant No. C-380-003-002-001). ’ REFERENCES (1) Mathine, D. L.; Woo, H. S.; He, W.; Kim, T. W.; Kippelen, B.; Peyghambarian, N. Appl. Phys. Lett. 2000, 76, 3849. (2) Forrest, S. R. Chem. Rev. 1997, 97, 1793–1896. (3) Xue, J.; Rand, B. P.; Uchida, S.; Forrest, S. R. Adv. Mater. 2005, 17, 66–71. (4) Gr€atzel, M. Nature 2001, 414, 338–344. (5) O’Regan, B.; Gr€atzel, M. Nature 1991, 353, 737–740. 24886

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887

The Journal of Physical Chemistry C (6) Zhu, X. Y. Surf. Sci. Rep. 2004, 56, 1–83. (7) Lindstrom, C. D.; Zhu, X. Y. Chem. Rev. 2006, 106, 4281–4300. (8) Qi, D.; Gao, X.; Wang, L.; Chen, S.; Loh, K. P.; Wee, A. T. S. Chem. Mater. 2008, 20, 6871–6879. (9) Qi, D.; Sun, J.; Gao, X.; Wang, L.; Chen, S.; Loh, K. P.; Wee, A. T. S. Langmuir 2009, 26, 165–172. (10) Chen, W.; Qi, D.; Gao, X.; Wee, A. T. S. Prog. Surf. Sci. 2009, 84, 279–321. (11) Qi, D.; Chen, W.; Gao, X.; Wang, L.; Chen, S.; Loh, K. P.; Wee, A. T. S. J. Am. Chem. Soc. 2007, 129, 8084–8085. (12) Hoffmann, M. R.; Martin, S. T.; Choi, W.; Bahnemann, D. W. Chem. Rev. 1995, 95, 69–96. (13) Zuruzi, A. S.; Kolmakov, A.; MacDonald, N. C.; Moskovits, M. Appl. Phys. Lett. 2006, 88, 102904. (14) Schnadt, J.; Br€uhwiler, P. A.; Patthey, L.; O’Shea, J. N.; Sodergren, S.; Odelius, M.; Ahuja, R.; Karis, O.; Bassler, M.; Persson,  P.; Siegbahn, H.; Lunell, S.; Martensson, N. Nature 2002, 418, 620–623. (15) Yu, S.; Ahmadi, S.; Palmgren, P.; Hennies, F.; Zuleta, M.; G€othelid, M. J. Phys. Chem. C 2009, 113, 13765–13771. (16) Rangan, S.; Katalinic, S.; Thorpe, R.; Bartynski, R. A.; Rochford, J.; Galoppini, E. J. Phys. Chem. C 2009, 114, 1139–1147. (17) Rienzo, A.; Mayor, L. C.; Magnano, G.; Satterley, C. J.; Ataman, E.; Schnadt, J.; Schulte, K.; O’Shea, J. N. J. Chem. Phys. 2010, 132, 084703. (18) Sch€utte, J.; Bechstein, R.; Rahe, P.; Rohlfing, M.; K€uhnle, A.; Langhals, H. Phys. Rev. B 2009, 79, 045428. (19) Rahe, P.; Nimmrich, M.; Nefedov, A.; Naboka, M.; W€oll, C.; K€uhnle, A. J. Phys. Chem. C 2009, 113, 17471–17478. (20) Johansson, E. M. J.; Odelius, M.; Karlsson, P. G.; Siegbahn, H.; Sandell, A.; Rensmo, H. J. Chem. Phys. 2008, 128, 184709. (21) Forrest, S. R. J. Phys.: Condens. Matter 2003, 15, S2599. (22) Tautz, F. S. Prog. Surf. Sci. 2007, 82, 479–520. (23) Rohlfing, M.; Temirov, R.; Tautz, F. S. Phys. Rev. B 2007, 76, 115421. (24) Chen, W.; Huang, H.; Chen, S.; Gao, X. Y.; Wee, A. T. S. J. Phys. Chem. C 2008, 112, 5036–5042. (25) Moal, E. L.; M€uller, M.; Bauer, O.; Sokolowski, M. Phys. Rev. B 2010, 82, 045301. (26) Burke, S. A.; Ji, W.; Mativetsky, J. M.; Topple, J. M.; Fostner, S.; Gao, H. J.; Guo, H.; Gr€utter, P. Phys. Rev. Lett. 2008, 100, 186104. (27) Chen, W.; Huang, H.; Chen, S.; Chen, L.; Zhang, H. L.; Gao, X. Y.; Wee, A. T. S. Appl. Phys. Lett. 2007, 91, 114102. (28) Chen, W.; Qi, D. C.; Huang, Y. L.; Huang, H.; Wang, Y. Z.; Chen, S.; Gao, X. Y.; Wee, A. T. S. J. Phys. Chem. C 2009, 113, 12832–12839. (29) Weiss, C.; Wagner, C.; Temirov, R.; Tautz, F. S. J. Am. Chem. Soc. 2010, 132, 11864–11865. (30) Gross, L. Nat. Chem. 2011, 3, 273–278. (31) Mura, M.; Sun, X.; Silly, F.; Jonkman, H. T.; Briggs, G. A. D.; Castell, M. R.; Kantorovich, L. N. Phys. Rev. B 2010, 81, 195412. (32) Tekiel, A.; Godlewski, S.; Budzioch, J.; Szymonski, M. Nanotechnology 2008, 19, 495304. (33) Godlewski, S.; Tekiel, A.; Prauzner-Bechcicki, J. S.; Budzioch, J.; Gourdon, A.; Szymonski, M. J. Chem. Phys. 2011, 134, 224701. (34) Komolov, A. S.; Møller, P. J.; Mortensen, J.; Komolov, S. A.; Lazneva, E. F. Appl. Surf. Sci. 2007, 253, 7376–7380. (35) Edvinsson, T.; Li, C.; Pschirer, N.; Sch€oneboom, J.; Eickemeyer, F.; Sens, R.; Boschloo, G.; Herrmann, A.; M€ullen, K.; Hagfeldt, A. J. Phys. Chem. C 2007, 111, 15137–15140. (36) Yu, X.; Wilhelmi, O.; Moser, H. O.; Vidyaraj, S. V.; Gao, X.; Wee, A. T. S.; Nyunt, T.; Qian, H.; Zheng, H. J. Electron Spectrosc. Relat. Phenom. 2005, 144147, 1031–1034. (37) Graber, T.; Forster, F.; Sch€oll, A.; Reinert, F. Surf. Sci. 2011, 605, 878–882. (38) M€obus, M.; Karl, N.; Kobayashi, T. J. Cryst. Growth 1992, 116, 495–504. (39) Sch€oll, A.; Zou, Y.; Jung, M.; Schmidt, T.; Fink, R.; Umbach, E. J. Chem. Phys. 2004, 121, 10260.

ARTICLE

(40) Outka, D.; St€ohr, J. J. Chem. Phys. 1988, 88, 3539. (41) Sch€oll, A.; Zou, Y.; Schmidt, T.; Fink, R.; Umbach, E. J. Electron Spectrosc. Relat. Phenom. 2003, 129, 1–8. (42) Thomas, A. G.; Flavell, W. R.; Mallick, A. K.; Kumarasinghe, A. R.; Tsoutsou, D.; Khan, N.; Chatwin, C.; Rayner, S.; Smith, G. C.; Stockbauer, R. L.; Warren, S.; Johal, T. K.; Patel, S.; Holland, D.; Taleb, A.; Wiame, F. Phys. Rev. B 2007, 75, 035105. (43) Mayer, J. T.; Diebold, U.; Madey, T. E.; Garfunkel, E. J. Electron Spectrosc. Relat. Phenom. 1995, 73, 1–11. (44) Yu, S.; Ahmadi, S.; Sun, C.; Palmgren, P.; Hennies, F.; Zuleta, M.; G€othelid, M. J. Phys. Chem. C 2010, 114, 2315–2320. (45) G€opel, W.; Anderson, J. A.; Frankel, D.; Jaehnig, M.; Phillips, K.; Sch€afer, J. A.; Rocker, G. Surf. Sci. 1984, 139, 333–346. (46) Gustafsson, J. B.; Moon, E.; Widstrand, S. M.; Gurnett, M.; Johansson, L. S. O. Surf. Sci. 2004, 572, 32–42. (47) Rocco, M. L. M.; Haeming, M.; Batchelor, D. R.; Fink, R.; Sch€oll, A.; Umbach, E. J. Chem. Phys. 2008, 129, 074702. (48) Diebold, U. Surf. Sci. Rep. 2003, 48, 53–229. (49) Zhang, Z.; Jeng, S. P.; Henrich, V. E. Phys. Rev. B 1991, 43, 12004. (50) Vogtenhuber, D.; Podloucky, R.; Neckel, A.; Steinemann, S. G.; Freeman, A. J. Phys. Rev. B 1994, 49, 2099. (51) Azuma, Y.; Hasebe, T.; Miyamae, T.; Okudaira, K. K.; Harada, Y.; Seki, K.; Morikawa, E.; Saile, V.; Ueno, N. J. Synchrotron Radiat. 1998, 5, 1044–1046. (52) Hirose, Y.; Kahn, A.; Aristov, V.; Soukiassian, P.; Bulovic, V.; Forrest, S. R. Phys. Rev. B 1996, 54, 13748. (53) Hirose, Y.; Wu, C. I.; Aristov, V.; Soukiassian, P.; Kahn, A. Appl. Surf. Sci. 1997, 113114, 291–298. (54) L€uth, H. Surfaces and Interfaces of Solid Materials: Springer Study ed.; Springer: Berlin, 1995. (55) St€ohr, J. NEXAFS Spectroscopy; Springer: Berlin, 1992. (56) Taborski, J.; V€aterlein, P.; Dietz, H.; Zimmermann, U.; Umbach, E. J. Electron Spectrosc. Relat. Phenom. 1995, 75, 129–147. (57) Cao, L.; Zhang, W. H.; Chen, T. X.; Han, Y. Y.; Xu, F. Q.; Zhu, J. F.; Yan, W. S.; Xu, Y.; Wang, F. Acta Phys. Sin. 2010, 59, 1681–1688. (58) Cao, L.; Wang, Y. Z.; Chen, T. X.; Zhang, W. H.; Yu, X. J.; Ibrahim, K.; Wang, J. O.; Qian, H. J.; Xu, F. Q.; Qi, D. C.; Wee, A. T. S. J. Chem. Phys. 2011, 135, 174701. (59) Ishii, H.; Sugiyama, K.; Ito, E.; Seki, K. Adv. Mater. 1999, 11, 605–625. (60) Chen, W.; Qi, D. C.; Huang, H.; Gao, X.; Wee, A. T. S. Adv. Funct. Mater. 2011, 21, 410–424. (61) Hill, I. G.; Kahn, A.; Soos, Z. G.; Pascal, J. R. A. Chem. Phys. Lett. 2000, 327, 181–188.

24887

dx.doi.org/10.1021/jp2083924 |J. Phys. Chem. C 2011, 115, 24880–24887