Characterization of the Interactions between Alq3 Thin Films and Al

ARTICLE pubs.acs.org/JPCC

Characterization of the Interactions between Alq3 Thin Films and Al Probed by Two-Color Sum-Frequency Generation Spectroscopy Takayuki Miyamae,*,† Eisuke Ito,‡ Yutaka Noguchi,§ and Hisao Ishii§ †

Nanosystem Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), 1-1-1 Higashi, Tsukuba, Ibaraki 305-8565, Japan ‡ Flucto-order Functions Research Team, RIKEN-HYU Collaboration Research Center, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan § Center for Frontier Science and Graduate School of Advanced Integration Science, Chiba University 1-33 Yayoi-cho, Inage-ku, Chiba 263-8522, Japan

bS Supporting Information ABSTRACT: We present the investigation of the vibrational and electronic states of tris(8-hydroxyquinoline) aluminum (Alq3)/Al (Alq3 on Al) interfaces by using two-color infrared-visible sum frequency generation (SFG) spectroscopy. The visible wavelength dependence of the SFG spectra of the 2 nm thick Alq3/Al consists of the vibrational bands derived from the Alq3 at the Al interfaces. The intensities of the peaks derived from the ring stretching modes of the quinolate ligands were significantly enhanced due to the double resonance effect. In contrast, the SFG electronic spectrum obtained from the output photon energy dependence of the SFG peak amplitudes derived from the CdC bands of the Al on Alq3 interfaces does not show the wavelength dependences, indicating that the electronicresonance associated with the π
π* transitions in the quinolate rings are almost vanished at the Al deposited on the Alq3. The disappearance of the electronic-resonance of the CdC stretching modes must be caused by the perturbation of the HOMO and LUMO of pristine Alq3 by the interaction with the Al. The spectral features of the two-color SFG spectra of the Al/LiF/Alq3 system show quite different behavior from those of Alq3/Al and Al/Alq3. The shift of the CdC stretching modes toward lower frequencies is indicative of the formation of the Alq3 anionic states upon reaction with Li at the interface. Additional broad bands around 1335 and 1450 cm
1, which show the weak excitation wavelength dependence, might be due to the existence of the Li-reacted graphitic carbon-like Alq3.

1. INTRODUCTION Recently, organic semiconductors are attracting attention because of possible applications to electronic devices such as organic light emitting diodes (OLEDs), organic field-effect transistors (OFETs), and organic solar cells. The optical and electronic properties at interfaces of such devices are key factors to understand and improve organic devices, but these are not well examined, especially the properties of buried interfaces have been not well investigated. In organic devices, the charge carriers, both electrons and holes, often have to be injected through organic/electrode interfaces. Especially, an understanding of the interaction between metal electrode and organic molecules is quite important, because the electronic properties of the metal/organic interface directly affect the performance of the OLEDs.1 In order to understand the mechanisms that control the electron energetics of organic/metal interfaces, the determination of the energy barriers between the Fermi level of the metal and the HOMO and LUMO levels of organic materials across the interfaces has been main challenges to surface and interface studies of organic thin films.1 For understanding the occupied states, ultraviolet photoelectron spectroscopy (UPS) is a powerful technique for r 2011 American Chemical Society

studying the valence electronic structure of material. The electron injection barrier at the metal/organic interface is significantly altered by the interfacial dipole layer, by which the vacuum level at the organic layer is shifted relative to that at the metal layer.2 The dipole layer has been studied for organic/metal interfaces by using photoelectron spectroscopic technique. Ishii et al. proposed various origins of the dipole layer at the organic/ metal interface: (1) charge transfer, (2) mirror force, (3) pushback effect due to the surface rearrangement, (4) chemical interaction, (5) interface state, and (6) permanent dipole of the adsorbate.2 In contrast, the band gaps of the organic materials and the energy of LUMO are often determined by the optical absorption measurements of the “bulk”. Because of the surface confinement effect and the interaction between organic molecules and metal, the molecular conformation and the band gap at the buried interface are expected to be different from those in the bulk. Although the optical band gap obtained from the absorption Received: December 2, 2010 Revised: April 11, 2011 Published: April 22, 2011 9551

dx.doi.org/10.1021/jp201051h | J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C spectroscopy is different from the corresponding charge transfer gap due to the exciton absorption to form Frenkel-type exciton, the optical band gaps at buried interfaces are still useful energy parameters to discuss the charge injection and overall efficiency of the OLED. However, it has been a great challenge to measure the buried interfacial electronic states because of a lack of a suitable probing technique. Traditional surface science techniques based on ultrahigh vacuum are not applicable to a buried interface, and absorption and emission spectroscopy do not have the necessary surface sensitivity. Nonlinear optical spectroscopy is a powerful technique for the characterization of these issues due to its high interface sensitivity. Second harmonic generation (SHG) and sum frequency generation (SFG) have been used to investigate the molecular orientation of the materials at the interface. IR-visible SFG spectroscopy has made it possible to study the vibrational spectra of surface or interfacial species.3 SFG is a surface-sensitive tool because its second-order nonlinear optical process is allowed only in noncentrosymmetric media under the electric dipole approximation.4 IR-visible SFG vibrational spectroscopy has been traditionally carried out by using the frequency-fixed visible and tunable IR beams to obtain a surface vibrational spectrum, which identifies the surface chemical species. Recently, a new technique for vibrational SFG spectroscopy by tuning the incident visible and IR frequencies, so-called two-color SFG, has attracted much attention.5
12 When the photon energy of the SFG coincides with electronic transition energies of surface species, the output SFG intensity is drastically enhanced when the IR light is resonant with the vibrational state and the SFG light is resonant with the interfacial electronic transitions. Such an enhanced SFG process is called doubly resonant (DR) SFG.5,6 Our previous two-color SFG study of single-walled carbon nanotube on silver showed remarkable dependence on the visible excitation wavelength, which is ascribed to the resonance of the SFG photon energy with the electronic transition energy of the semiconducting SWCNTs.11 With the capability of tuning both the incident IR and visible frequencies, two-color SFG spectroscopy becomes a powerful multidimensional technique for studying the interface of electronic states coupled to a specific vibrational mode. The SFG electronic excitation profiles, which can be obtained by measuring the visible probe frequency dependence of the vibrational SFG band intensity, allow deduction of coupling electronic transitions and vibrational modes at the interface. In addition, there are several advantages of the two-color SFG technique. The signal enhancement is expected only for species that have an electronic absorption at the photon energy of the SFG. Therefore, two-color SFG offers a kind of molecular selectivity to SFG. Moreover, an electronic excitation spectrum of the interface species for each vibrational band can be obtained. Thus the SFG excitation profiles are useful to investigate mixed interface layers where several chemical species coexist and show complex vibrational spectra, because vibrational bands can be classified with reference to corresponding electronic spectra. In addition, the measurement of SFG excitation profiles may be an effective way to obtain electronic spectra of the molecules, especially at an interface on opaque substrates where electronic absorption spectrum measurement is difficult.12 Furthermore, it can be possible to measure the interface of the OLED materials that show very strong photoluminescence in the visible region, since the output SFG emerges at the anti-Stokes side of the excitation wavelength.12

ARTICLE

Figure 1. Molecular structures for (a) meridional and (b) facial Alq3.

In this study, two-color SFG was used to detect the interfacial vibrational and electronic states of tris(8-hydroxyquinoline) aluminum (Alq3)/Al interfaces. In OLEDs, Alq3 is most widely used as electron transport/light emitting material.13 It is wellknown that Alq3 has two possible geometrical isomers of meridional (C1 symmetry) and facial (C3 symmetry) forms, as shown in Figure 1. In the meridional isomer, the three quinolate ligands around the central Al atom are not equivalent, while they are equivalent in the facial isomer. It has been reported that Rand amorphous Alq3 consist of the meridional isomer,14,15 whereas γ- and δ-Alq3 consist of facial isomer.16 Elucidating the electronic structures of the Alq3/metal interface are required for its applicability to OLED devices. Such necessity should increase, due to the recent report of a significant enhancement of the current injection and OLED output induced by the insertion of an insulating layer such as LiF,17
19 MgO,17 or MgF220 between the Al cathode and the Alq3. Various mechanisms for this enhancement in the device efficiency have been proposed, and investigated using various techniques such as XPS, UPS,21 and high resolution electron energy loss spectroscopy (HREELS).22 One hypothesis is that a thin LiF layer protects the Alq3 from the deleterious reaction with Al. Another hypothesis is that Li atoms produced by the dissociation of LiF by Al deposition lead to formation of the Alq3 radical anion.21,23 On the other hand, it has been reported that a large potential is built in as-deposited Alq3 thick film in the dark observed by the Kelvin probe method and SHG.24
27 According to the Kelvin probe experiments by Ito et al.,24 the large surface potential was observed by the thick Alq3 films (∼100 nm) under dark condition. Furthermore, this behavior is independent of the kind of substrate.24,28 Since the meridional isomers of the Alq3 molecules have permanent dipole moments, a very slight alignment (1
2% of full alignment for the dipole moment per unit volume24) of the molecules is believed to cause the large surface potential of the thick Alq3 films. On the other hand, the large surface potential is reduced by exposure to ambient light.24,25 The reduction of the surface potential was considered with the photoinduced randomization of molecular orientation,29 and the electrostatic screening by photogenerated carrier.30 Recently, the first-order electroabsorption measurements for the Alq3 film confirm that the noncentrosymmetric molecular orientation remains even after light irradiation, indicating that the reduction of the large surface potential is not caused by the orientation randomization.31 It is also suggested that the orientation polarization of the Alq3 film is maintained in the OLED structure even under light illumination after device fabrication.32 Thus the electrostatic screening effect is more plausible for the reduction of the large surface potential. Because the SFG process is allowed 9552

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C

ARTICLE

in noncentrosymmetric media, the SFG spectra is significantly influenced by the noncentrosymmetric dipole orientation in the bulk. In order to avoid the influence of the uniaxial orientation of the thick bulk layer as much as possible, extremely thin Alq3 layers were used in this study. The thickness dependence of the SFG intensity is not our focus in the present study, but the SFG spectra taken from 1, 2, and 5 nm thick Alq3 films deposited on Au substrate are shown in the Supporting Information. Further experiments of the two-color SFG for the thick Alq3 films are underway, and the results of the thickness dependences of the thick Alq3 layers will be published in the near future. The results of this paper are split into four sections. After the Introduction, Theoretical Background, and Experimental Section, we present in section 4.1 the two-color SFG results on the Alq3/Al interface using ultrathin Alq3 layers measured in ambient conditions Here, the notation of A/B indicates a system prepared by depositing A on B. In this section, we present the intensities of the SFG peaks derived from the quinolate ring stretching mode, which are strongly dependent on the visible wavelength due to the doubly resonant effect. In section 4.2, we show the buried Al/Alq3 interfaces investigated by two-color SFG. We discuss the mechanism of the disappearance of the electronic-resonant effect of the CdC stretching modes at the buried Al/Alq3 interface by comparison with the previous theoretical and experimental results. In section 4.3, we present the effect by inserting a thin LiF layer between Alq3 and Al cathode by measuring two-color SFG for the Al/LiF/Alq3 system. We also discuss the different

ð2Þ χijk

chemical interactions between Al/Alq3 and the Al/LiF/Alq3 systems based on the interfacial vibrational and the electronic structural differences. Finally, in section 5, we present our conclusions.

2. THEORETICAL BACKGROUND In the electric dipole approximation, SFG is forbidden in centrosymmetric materials, but not at their interfaces, where the inversion symmetry of the bulk is broken. For an air
metal interface, the SFG intensity reflected from the surface is given by ð2Þ

IðωSF Þ
jχef f : EðωIR ÞEðωvis Þj2

ð1Þ

where χ(2) eff is the effective second-order nonlinear susceptibility tensor and E(ωIR) and E(ωvis) are the input fields. The secondorder nonlinear susceptibility contains nonresonant, singly resonant, and doubly resonant contributions. The latter dominates strongly if the vibrations (ωl) and electronic transitions (ωeg) probed by ω IR and ω SFG are coupled. Both IR
visible (vibrational transition followed by an electronic transition) and visible
IR processes (electronic transition followed by a vibrational transition) contribute to doubly
resonant SFG.6,33 However, the IR
visible sequence is expected to dominate, as a result of the quicker relaxation of the electronic excitation compared to the vibrational one. Assuming harmonic potential surfaces for the electronic states and the Born
Oppenheimer and Condon ap6,33 proximations, the doubly
resonant χ(2) ijk can be described as

* ( )+ pffiffiffiffi
Sl k ¥ X N i j Dμgg Snl 1 1 Sl e ð2Þ ¼
2 μeg μge
þ χNR , ijk Dql ωIR
ωl þ iΓl n ¼ 0 n! ωs
nωl
ωeg þ iΓen, g0 ωs
ðn þ 1Þωl
ωeg þ iΓen þ 1, g0 p ð2Þ μieg

where N is the surface molecular density, represents the i component of electronic transition moment, ql is the normal coordinate, Sl is a dimensionless coupling constant known as the Huang
Rhys factor, n labels the vibrational state, g and e label the ground and excited electronic states, respectively, ωS is the SFG frequency, ωl and ωeg are the resonant vibrational and electronic frequencies, respectively, Γl and Γen,g0 are the damping constants, the angular brackets indicate an average over molecular orientations, and χ(2) NR,ijk describes the nonresonant contributions. Sl is related to the shift dl of the harmonic potential of the vibration in the excited electronic level by Sl ¼

1 ωl dl 2 2p

ð3Þ

DR-SFG occurs thus for ωIR = ωl and for several visible frequencies, when ωS matches an allowed vibronic transition to the excited electronic level. The intensity of each vibronic resonance depends on the Franck
Condon overlap integrals of the vibrational levels involved in the transition. Since the initial and final vibrational states, respectively of the visible and SFG transitions, always differ, the vibration and the electronic transition must therefore be coupled (dl 6¼ 0) to have a nonzero transition probability for the global DR-SFG process. Thus the DR-SFG spectrum allows for the determination of the coupling strength and characteristics. Equation 2 includes all the vibronic transitions series. However, in general, the visible
IR SFG is much weaker and not detected because of the very fast relaxation of the electronic

excitation. Actually, the dephasing times of vibronic transitions are in the femtosecond region for Alq3.34 Therefore, by assuming Γen,g0 . Γe0,g0, the nonzero vibronic transitions can be neglected. It is worth pointing out that a significantly larger Γen,g0 also suppresses the aforementioned visible
IR SFG, which starts with an electronic transition followed by a vibrational transition. To analyze the spectra, we note that with the visible input frequency ωvis fixed, eq 2 can be approximated by the form X Al ð2Þ ð2Þ þ χNR eiξ ð4Þ χijk
ω
ω þ iΓ IR l l l where Al, ωl, and Γl are the peak amplitude describing the electronic resonance, the resonant vibrational frequencies, and the damping constants, respectively. χ(2) NR,ijk and ξ describe the nonresonant contributions and the phase difference between resonant and nonresonant term, respectively. We use eq 4 to fit all of the measured spectra with ωl, Γl, Al, ξ, and χ(2) NR as adjustable parameters.

3. EXPERIMENTAL SECTION 3.1. Sample Preparation and the Characterization. For the preparation of the Alq3/Al samples, the Al substrates were prepared by vacuum evaporation of Al on Si substrates by using a tungsten filament. LiF powder (purity 99.999%) was obtained from Aldrich, and Al Wire (purity 99.999%) was purchased from NILACO Co. Ltd. The sample of Alq3 (Aldrich, sublimated 9553

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C

Figure 2. Schematic arrangement for two-color SFG spectrometer.

grade, the nominal purity was 99.995%) was used without further purification. Deposition of Alq3 was performed using a quartz crucible coiled with a tungsten wire heater. Alq3 were deposited on them under dark condition in a vacuum chamber with base pressure of 1  10
7 Pa. A pumping system was composed of a 2000 L/s turbomolecular pump and a 500 L/min rotary pump for a vacuum deposition chamber and a 50 L/s turbomolecular pump and a 90 L/min rotary pump for a load-lock chamber. To suppress ion-gauge and photoirradiation effects on the polarization property of Alq3 film,35 an ion-gauge in the vacuum chamber was turned off and all viewing ports were covered during the deposition. The thicknesses of Alq3, Al, and LiF films were monitored using a quartz microbalance. The thicknesses of the Al films were about 50 nm. The thickness of the Al film is too thick to transmit the light (O.D. > 4). The deposition rates of the sample materials were about 0.5
1.5 nm/s for Al, 0.01 nm/s for LiF, and 0.02
0.05 nm/s for Alq3, respectively. CaF2 substrates purchased from the Pier Optics Co. were thoroughly cleaned using a procedure with several steps. They were first soaked in toluene for 24 h and then soaked in 5% aqueous detergent (Scat20X-N, Dai-ichi Clean Chemical Inc.) for 24 h. After that, they were rinsed with deionized water and acetone. All of the substrates were then cleaned by a discharge plasma cleaner (PDC-32G, Harrick Plasma) for 4 min immediately before introducing the vacuum chamber. Substrates were tested using SFG, and no signal from contamination was detected. For the observation of the Al/Alq3 interfaces by using SFG, the Alq3 films were directly deposited on 1 mm thick CaF2 substrates, and thick Al layers were then deposited under dark condition. For Al/LiF/ Alq3, a thin LiF film of 1 nm thick was deposited on the Alq3 film from a tungsten basket. Then a 50 nm thick Al layer was deposited on the LiF layer. After the deposition, the two-color

ARTICLE

SFG measurements were subsequently performed in ambient conditions. Transmission infrared measurements were carried out at a 4 cm
1 resolution using a Fourier transform IR spectrometer (Infinity, Mattson). UV
visible absorption spectra were measured on a Shimadzu UV-2500PC spectrometer. 3.2. Two-Color SFG Measurements. The basic SFG system employed in this experiment has been described in detail in the previous publication.11,36 Figure 2 schematically depicts the twocolor SFG experimental setup. Tunable IR and visible laser beams were generated by two optical parametric generators/ amplifiers (OPG/OPA, Ekspla, PG401VIR/DFG) pumped by a mode-locked Nd:YAG laser at 1064 nm (Ekspla, PL-2143D, 25 ps, 10 Hz). The IR beam, tunable from 1000 to 4300 cm
1, was produced by difference frequency mixing of the 1064 nm beam with the output of a LiB3O5 (LBO) crystal mounted in OPG/OPA, which is pumped by the 355 nm beam. The visible beam, tunable from 420 to 640 nm, was generated in a LBO crystal mounted in OPG/OPA pumped by the 355 nm beam. The visible and IR beams were overlapped at sample surface with the incidence angles of 70 and 50, respectively. The spectral resolution of the tunable visible beam was about 8 cm
1, and its frequency was calibrated with the Hg lines. The spectral resolution of the IR beam was 6 cm
1, and the IR frequency was calibrated with the absorption lines of polystyrene standard.37 In order to minimize the irradiation damage, both tunable infrared and visible beams were defocused. The focus size of the infrared and visible beams were >1 and >3 mm, respectively. Further, in order to avoid photoirradiated damage, the fluence of the visible beam was kept below 100 μJ per pulse. The absence of the damage effect was confirmed by repeated SFG measurements. In order to eliminate the scattered visible light and the photoluminescent light from the samples, the sumfrequency output signal in the reflected direction was filtered with short-wave-pass filters (Asahi Spectra Co. Ltd.), prism monochromator (PF-200, Bunkoukeiki Co., Ltd.), and grating monochromator (Oriel MS257). Then the SFG signal was detected by a photomultiplier tube (Hamamatsu R649). The signal was averaged over 300 pulses by a gated integrator for every data point taken at a 3 cm
1 interval and was stored in a personal computer. In the frequency region between 2000 and 1300 cm
1, significant portion of the infrared beam is absorbed by water vapor in the optical path. The effect was minimized by purging the optical path of the IR beam and the sample stage by dry nitrogen gas. Each SFG spectrum was normalized to the visible and IR power to compensate the laser intensity fluctuations. All SFG spectra reported were taken with a PPP polarization combination, in which the IR, visible, and SFG light were polarized in the plane of incidence.

4. RESULTS AND DISCUSSIONS 4.1. Two-Color SFG of Alq3/Al Interface Measured in Ambient Conditions. In Figure 3 we show the two-color SFG

vibrational spectra of the 2 nm thick Alq3 deposited on Al on silicon substrate with various visible wavelength in a PPP polarization combination measured in the ambient condition. We also measured the 1 nm thick Alq3/Al, and the spectral feature shows almost the same to that of the 2 nm thick one (data not shown). Since the film thickness of 1 nm is comparable to the average thickness of the Alq3 monolayers,40 the observed SFG spectra of 2 nm thick Alq3/Al are mainly originated from the 9554

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C

Figure 3. Two-color SFG vibrational spectra of the 2 nm thick Alq3 deposited on Al precoated Si substrate with various visible wavelength in a PPP polarization combination. The solid lines correspond to the calculated fits to the data by using eq 4. Spectra are offset for clarity. Inset shows the experimental geometry.

Alq3/Al interface. The 1 nm thick data is unstable and much worse reproducibility. This must be caused by an instantaneous reaction of Alq3 and/or Al with the atmospheric environment. On the other hand, the SFG spectra of the 5 nm thick Alq3/Al systems show relatively stronger signals than those of the 1 and 2 nm thick Alq3/Al. Such thickness dependent signal enhancements must be due to the effect of the uniaxial orientation of the molecular dipole, as mentioned in the introduction.24
32 In order to minimize the effect of the uniaxial orientation of the Alq3, we used the 2 nm thick Alq3 films for the SFG measurements in the air atmosphere. In the two-color SFG spectra of the 2 nm thick Alq3/Al shows the bands at 1344, 1386, 1426, 1465, 1504, 1589, 1597, and 1612 cm
1. The band at 1386 cm
1 is attributed to mixed modes that contain the contribution from the C
C and C
N stretch of the pyridyl side of the quinolate ligands as well as C
H in-plane bending motions.15,38,39 The bands at 1589, 1597, and 1612 cm
1 are derived from the CdC stretching modes of the quinolate ligands. As shown in Figure 3, remarkable changes in the intensity of these peaks can be clearly observed by changing the visible wavelength. Figure 4a shows the changes in the two representative peak strengths (Al) of the peaks deduced from the fitting of the two-color SFG spectra in Figure 3 using eq 4 as a function of the photon energies of the SFG. In order to observe the enhancement ratios of the peak strength with the output SFG frequencies, the peak strengths are normalized with the strength of the SFG spectrum taken at the visible wavelength of 532 nm. For comparison, we show the optical absorption spectrum of the 20 nm thick Alq3 film on CaF2 in Figure 4. The excitation spectra exhibit a resonance almost coincident with the absorption spectrum for the Alq3 thin film. Especially, a significant increase in intensity is observed for the band at 1586 cm
1. It should be noted that the strengths of the 1586 cm
1 are illustrated by multiplying the original by 0.2. In contrast, the enhancements ratio of the SFG peak strength of the band at 1504 cm
1 is relatively weaker than that of the bands at

ARTICLE

Figure 4. (a) Changes in the strengths (Al) of the 1586 and 1504 cm
1 peaks (open circles) and nonresonant term χNR (filled circles) deduced from the fitting of the two-color SFG spectra in Figure 2 as a function of the photon energies of the SFG. The strengths of the band at 1586 cm
1 are multiplied by 0.2. Optical absorption spectrum of the 20 nm thick Alq3 film on CaF2 is shown in gray line. (b) Evolution of the Fresnel factors of Fzzz (black line) and Fxxz (red line) for the air/Al, Fzzz for the air/Alq3 (green line), and Fzzz for the CaF2/Alq3 (blue line) inter3 2/Alq3 and FCaF are multiplied by 10. faces. The Fair/Alq zzz zzz

1586 cm
1. The vibrational mode around 1586 cm
1 is assigned to the CdC stretching of the quinolate group, whereas the 1504 cm
1 peak is mixed modes that contain the contributions from the C
C and C
H in-plane bending motions.15 They are expected to have different degrees of coupling with the electronic transition. For Alq3, the electronic transition at 390 nm is dominated by the π
π* excitation of the quinolate ligands.41 Thus it is reasonable that the CdC stretching of the quinolate ligands are effectively enhanced due to the resonance with the π
π* transitions. The wavelength dependence of the nonresonant contribution also shows similar behavior. Such wavelength dependence of the nonresonant term is already reported by Raschke et al,6 and the wavelength dependent nonresonant term is due to the contributions from the singly resonant and visible-IR transitions. Recently, Wu et al. reported that the visible
IR transition process contributes the nonzero background in the doubly resonant condition.42 It should be noted that this behavior is not caused by the Fresnel factor at the metal interface, as described below. The wavelength-dependent nonresonant term in Figure 4 is, therefore, due to the singly resonant and visible
IR transition processes. The SFG electronic spectra in Figure 4 are slightly shifted to lower frequency as compared to the optical absorption spectrum of the Alq3. The shift of the electronic transition peak may be suggestive that the electronic excitation gap at the interface becomes smaller than that of the bulk. Since the SFG excitation spectra are not measured in the whole region across the optical transition peak, further experiments with the shorter wavelength excitation are needed to reveal the definitive information of the excitation profiles at the interface. For the analysis of the doubly resonant SFG, it should be important to note that the changes of the Fresnel factors have to be considered since it might change with the variation of the 9555

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C

ARTICLE

Figure 5. Two-color SFG spectra of the thick Al film directly deposited on Alq3 of 5 nm thick film deposited on the CaF2 substrate. The solid lines correspond to the calculated fits to the data by using eq 4. Spectra are offset for clarity. Experimental SFG setup for probing the buried Al/Alq3 interface is shown in the inset.

visible wavelength. The effective the second-order nonlinear susceptibility tensor components of an azimuthally isotropic sample contributes to the PPP SFG signals can be written as Aq, PPP ¼
Lxx ðωSF ÞLxx ðωvis ÞLzz ðωIR Þ cos βSF cos βvis sin βIR χxxz
Lxx ðωSF ÞLzz ðωvis ÞLxx ðωIR Þ cos βSF sin βvis cos βIR χxzx þ Lzz ðωSF ÞLxx ðωvis ÞLxx ðωIR Þ sin βSF cos βvis cos βIR χzxx þ Lzz ðωSF ÞLzz ðωvis ÞLzz ðωIR Þ sin βSF sin βvis sin βIR χzzz

ð5Þ where Lxx(ω) and Lzz(ω) are the Fresnel coefficients at frequency ω; βSF, βvis, and βIR are the reflection angles of the sum frequency, visible, and IR pulses, respectively; and χijks are the nonvanishing elements of the second-order nonlinear susceptibility.43 We found that χxzx and χzxx are much smaller than χxxz and χzxx. Thus the Fresnel factors Fzzz = |LZZ(ωSF)LZZ(ωvis)LZZ(ωIR) sin βSF sin βvis sin βIR| and Fxxz = |LXX(ωSF)LXX(ωvis)LZZ(ωIR) cos βSF cos βvis sin βIR| were calculated using the complex refractive indices of metallic aluminum were taken from the literature.44 In order to simplify, the refractive indices of interfacial layers were assumed to be 1. As shown in Figure 4b, the Fzzz at air/Al interface monotonically decreases with the increase of the photon energy of the SFG, and it does not explain the evolution of the SFG intensities of the Alq3/Al. Thus we conclude that the changes of the Fresnel factors do not much affect on the spectral shape of the SFG excitation profile of the Alq3/Al. The analysis of the Fresnel factor is also important for the investigation of such a thin layered sample. We calculated the Fresnel factor Fzzz at air/Alq3 interface, assuming that the air/ Alq3 interface is azimuthally isotropic. The wavelength dependence of the refractive indices for Alq3 reported by Djurisi_c et al. were used for the evaluation of the Fresnel factors at air/Alq3 and

CaF2/Alq3 interfaces.45 As shown in Figure 4 the Fzzz at air/Alq3 interface are much smaller than the Fzzz at air/Al interface. From this observation, we conclude that the experimentally observed SFG should be mainly from the Alq3/Al interface. In the previous UPS study of the Alq3 deposited on Al systems, an extra occupied state above the HOMO level was detected, suggesting a strong chemical interaction between Alq3 and Al.40 It was suggested that the interaction between Al and Alq3 is somewhat different from the charge transfer as reported for alkaline metal doped Alq3.46,47 On the other hand, a theoretical calculation of Alq3 layer on Al (111) suggested that the interfacial interaction is weak,48 These previous findings are controversial and not easy to discuss consistently, but one possible reason of this discrepancy must be due to the least reactivity of the clean Al(111) flat surface used in the DFT calculations.49 If the strong chemical interaction occurs at the interface, the vibrational SFG spectra and the corresponding electronic excitation profiles should provide different behavior to those of the bulk. On the contrary, our SFG results are almost identical to the IR39 and visible optical absorption spectrum of the pristine Alq3, and we cannot find such spectral changes. The discrepancy between UPS and SFG must be due to the different measurement condition, and our samples of Alq3/Al must be oxidized by exposing to ambient condition. Aluminum is very reactive and oxidizes quickly, even under high vacuum conditions. The 2 nm thick Alq3 layer cannot prevent the oxidation of Al in the air atmosphere. When the sample is exposed to the air, reactive Al surface immediately reacts with atmospheric oxygen and water vapor, and the charge transfer from the Al substrate or chemical interaction between Al and Alq3 must be vanished. We also measured the Alq3 deposited on the air-exposed Al substrate, and the SFG spectra and the corresponding excitation profiles are similar to Figures 3 and 4. This gives another support to the oxidation of the Al layer. Such oxidation of the Al is much reduced in the previous UPS measurements because the sample preparation and the UPS measurements are performed in UHV environments.40 For further investigation of the interfacial reaction between Alq3 and Al by the two-color SFG without the oxidation, measurements under UHV condition are needed. An UHV chamber for two-color SFG spectroscopy will be available in the future. 4.2. Two-Color SFG of Al/Alq3 Systems. The traditional surface analysis techniques such as UPS have been applied to examine the interfaces formed by depositing organic material on metals, which are not much troubled by the factor of chemical reaction.2 Actually, in many cases the UPS spectra of organic-onmetal systems show only rigid shifts on the energy scale, suggesting the absence of strong chemical interaction. In contrast, actual OLEDs are fabricated by organic layers sandwiched by a cathode and anode. The buried interface between metal anode and the organic layer is formed by the deposition of the metal on organic materials. When the metal is deposited on an organic layer by evaporation, the high reactivity of the vaporized hot metal atom often leads to a chemical reaction at the interface,50
52 and diffuse into the organic layer.53,54 Due to the energy transfer from the hot metal atom, deposition of the metal on organic materials may also induce the decomposition of the molecules, polymerization of the molecules, reorientation of the molecules, desorption of the organic materials from the substrate, and so on. Thus the metal-on-organic systems are generally much more complex than the organic-on-metal systems. Due to the high reactivity of the Al atom, reactions between 9556

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C

ARTICLE

Figure 6. Changes in the peak strengths of 1386 and 1586 cm
1 deduced from the fitting of the two-color SFG spectra in Figure 4 as a function of the photon energies of the SFG. Optical absorption spectrum of the Alq3 film is also shown by gray line.

the Alq3 and Al are expected by the deposition of the Al layer. For the characterization of the buried metal-on-organic interface, next, we measure the two-color SFG for the Al deposited on Alq3 film. In contrast to the case of Alq3 on metal, the deposited thick Al layer can act as a superior gas barrier, and the extent of the oxidation of the Al interface is much reduced. In Figure 5, we show the two-color SFG spectra of Al film directly deposited on 5 nm thick Alq3 film on CaF2 substrate. In the case of the Al/Alq3 system, the thickness of Alq3 is set to 5 nm. When we use the Alq3 layer of 2 nm thick sandwiched by CaF2 and Al, the SFG gives a quite weak signal. Low sticking probability of the Alq3 on CaF2 is not plausible judging from the molecular weight of Alq3. Although the exact reason is not clear at present, we tentatively thought that the detectable interface could not be formed by the deposition of the 2 nm thick Alq3. As mentioned above, the metal atoms often diffuse into the organic layer, and this process may prevent the formation of the clear interface for the 2 nm thick Alq3 sandwiched between CaF2 and Al. It should be noted that the SFG signal comes mainly from the Al/Alq3 interface, not from the Alq3/CaF2 interface, because no SFG signals are detected from the thin Alq3 layer deposited on CaF2 (data not shown). This is further supported by the Fresnel factor difference between Al and CaF2 interfaces. We show in Figure 4b, the Fzzz at CaF2/Alq3 interface are negligibly smaller than Fzzz at air/Al interface. One may think that the SFG spectra in Figure 5 comes from the Alq3 bulk, since the thick Alq3 film shows uniaxial orientation. Although the evaluation of the bulk contribution needs transmission experiments,55 if the SFG signals in Figure 5 are mainly originated from the bulk Alq3, the spectral shapes and their behavior should be similar to those of the Alq3 on Al systems. Therefore, we thought that the main contribution of the SFG signals is the Al/Alq3 interface. The IR and visible beams were incident from the CaF2 side, as shown in the inset of Figure 5. The spectral features are different from the SFG spectra of Alq3/Al and the pristine IR spectrum of the Alq3.15,38,39 The new band appears at 1399 cm
1, but the assignment of this band is not clear. Since the thick Alq3 layers showed the remarkable SFG peak at 1400 cm
1 (data not shown), we thought that this peak might be caused by the bulk contribution, as mentioned in the introduction. The relative

Figure 7. Two-color SFG spectra of Al/LiF/Alq3. The solid lines correspond to the calculated fits to the data by using eq 4. Spectra are offset for clarity. Experimental SFG setup for probing the buried Al/LiF/ Alq3 interface is shown in the inset. The arrows indicate the broad features around 1335 and 1450 cm
1, respectively, see text.

intensities of bands derived from the CdC stretching around 1600 cm
1 becomes weak as compared to the case of the Alq3/ Al. Figure 6 shows the changes in the two representative peak strengths (Al) of the vibrational peaks deduced from the fitting of the two-color SFG spectra in Figure 5 as a function of the photon energies of the SFG. The SFG electronic excitation profile obtained from the band at 1386 cm
1 is almost identical with the optical absorption spectrum of the Alq3 film. As shown in Figure 6, the SFG electronic profiles derived from the CdC bands does not agree with the linear optical absorption spectrum of the Alq3, indicating that the electronic-resonant effects associated with the π
π* transitions in the quinolate rings are almost vanished. According to the theoretical calculations for the Al
Alq3 complex by Curioni and Andreoni, the energy diagrams near the gaps are significantly changed by the chemical bonding formation between Al and Alq3.46 In contrast to the case of the alkaline-metal
Alq3 complex, HOMO of pristine Alq3 is destabilized by the interaction with the Al 3s orbital.56 The interaction with the aluminum atom is such that one of the Alq3 HOMOs is repelled to higher energy, and a state with predominant Al character appears in the same energy range. Al
Alq3 interaction also induces the modification of LUMO. Previous NEXAFS study of the Al/Alq3 interface suggests that the Al
Alq3 interaction is not simple electron transfer from Al to Alq3.57 Although the theoretical simulations for the 1:1 Al
Alq3 complex cannot predict the observed NEXAFS results, modification of the HOMO (LUMO) level occurs at the Al/Alq3 interface. Thus we conclude that the disappearance of the doubly resonant effect associated with the π
π* transitions of Alq3 must be caused by the perturbation of the HOMO and LUMO of pristine Alq3 by the interaction of the Al. 4.3. Two-Color SFG of Al/LiF/Alq3 Interface. In this section, we discuss the effects of the insertion of a LiF layer at the Al/Alq3 interface. The two color SFG spectra of the Al/LiF/Alq3 system are shown in Figure 7. It should be noted that the LiF thickness of 9557

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C

ARTICLE

Figure 8. Changes in the peak strengths of the three representative vibrational peaks deduced from the fitting of the two-color SFG spectra in Figure 6 as a function of the photon energies of the SFG.

1 nm is insufficient to cover the Alq3 surface.40 From the optical perspectives, such an insufficient coverage cannot act as the effective “layer” which reflects the lights. Therefore, by adding 1 nm thick LiF does not induce the multiple reflections, because 1 nm is too thin as compared to the visible frequencies. Actually, the phase difference between the resonant and nonresonant term ξ obtained from the fitting does not show a significant difference between Al/Alq3 and Al/LiF/Alq3 systems, as listed in the tables in the Supporting Information. The spectral features are quite different from those of Alq3/Al and Al/Alq3. The new band appears at 1377 cm
1, but the assignment of this band is not clear. Additionally, broad bands, which show the weak excitation wavelength dependence, appear around 1335 and 1450 cm
1. The CdC stretching modes of the quinolate ligands are observed at 1572 and 1607 cm
1. Red-shift of these bands is not observed in the case of the Al/Alq3 system. The frequency shift to lower wavenumber of the CdC stretching mode is also reported in the IR and DFT study of the potassium-doped Alq3.39 The DFT calculations and the IR spectrum for the potassium-doped Alq3 suggested that the CdC stretching frequency of Alq3 anion is lower than that of pristine Alq3 molecule. Consequently, the lower wavenumber shift of the CdC stretching bands in the twocolor SFG spectra is indicative of the formation of the Alq3 anionic states upon reaction with Li at the interface.21,39 This observation is in good agreement with the previous UPS and XPS measurements for the Al/LiF/Alq3 interface,21,58,59 and, to our knowledge, this is the first observation of the anionic state formation at the buried interface in ambient condition. Figure 8 shows the SFG excitation profiles obtained from the two-color SFG spectra in Figure 7 as a function of the photon energies of the SFG. The SFG excitation profiles derived from the 1572 cm
1 band give a maximum around 420 nm; however, it does not show a large shift to the lower photon energy. If the Alq3 at the LiF interface forms the Alq3 anion, the absorption peaks should appear below 600 nm.60 The charges transferred from the Li might smear the SFG excitation profiles of the CdC stretching. Unfortunately, our two-color SFG system cannot generate sufficient power of the light below 640 nm at present. Further experiments with the longer wavelength excitation will reveal the character of the excitation profiles at the charged interface. Next, we mention the origin of the broad peaks at 1335 and 1450 cm
1. Similar broad features are observed by IR and Raman

Figure 9. SFG spectra of the (a) Alq3/Al, (b) Al/Alq3, and (c) Al/LiF/ Alq3 systems excited by the visible light of 450 nm with the PPP polarization combination. Solid lines are derived from the numerical fit using eq 4.

studies of potassium-doped Alq3 by Sakurai et al.39 These bands are also observed in the IR studies of the Al/LiF/Alq3 system.61 It was suggested that these bands are derived from the quinolate ligands reacted with potassium. On the other hand, recent Raman studies performed in UHV environment show similar broad features around 1355, 1405, and 1560 cm
1 for the Alq3/ Al system and 1315, 1435, and 1515 cm
1 for the Alq3/Ca system, respectively.62,63 The assignments of these modes are derived from the G-bands and the D-bands of the graphitic carbon generated by the deposition of Ca or Al onto Alq3 thick film. One may think that the graphite related bands are IR inactive modes; however, these modes become IR active in the nitrogenated amorphous graphite.64 NEXAFS studies also suggest the existence of the chemical interaction different from electron transfer at the Al/LiF/Alq3 system, which differs from Al/Alq3 and Li/Alq3 interfaces.57 Although further studies are necessary to determine the origin of these broad features, the wavelength independent SFG excitation profiles of these features may be suggestive that these bands are originated from the graphite-like bands, not from the pristine Alq3. On the other hands, the red-shift of the SFG vibrational peaks, which shows the remarkable wavelength dependent SFG excitation profiles, may be indicative by the formation of the Alq3 anionic states at the interface, as mentioned above. Because the interface formed by the metal deposition onto the organic materials is much more complicated than the organic-on-metal system, we conclude that the Al/LiF/Alq3 buried interface might be coexistence of the negatively charged Alq3 by the charge transfer from the Li and the Li-reacted graphitic carbon-like Alq3. Finally, we discuss the interfacial vibrational and electronic structural difference between Al/Alq3 and Al/LiF/Alq3 9558

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C interfaces by comparing the two-color SFG spectra. Figure 9 shows the SFG spectra of the Alq3/Al, Al/Alq3, and Al/LiF/Alq3 systems excited by the visible light of 450 nm with the PPP polarization combination. The fitting curves are also shown in Figure 9. From these data, we can confirm three differences by the different interface. First, the red-shift of the CdC stretching modes are observed in the Al/LiF/Alq3 interfaces, while such shift is not observed in the case of the Al/Alq3 system. These spectral behaviors clearly indicate that the chemical interaction at the Al/LiF/Alq3 interface is different from that of the Al/Alq3 system. Red-shift of the SFG vibrational modes is suggestive of the formation of the Alq3 anionic states at the interface. Second, the electronic-resonance effects for the CdC stretching modes are almost vanished for the Al/Alq3 system, while it remains in the case of the Al/LiF/Alq3 system. This observation suggests that the energy diagrams near the band gaps are significantly modified by the chemical reaction between Al and Alq3, rather than the simple charge transfer from Al to Alq3. Finally, the broad features which might be due to the graphite-like carbon formation are observed in the case of the Al/LiF/Alq3 system. Insertion of the LiF thin layer between Al and Alq3 induces significant interaction at the interface, and it emerges the negatively charged Alq3 and the Li-reacted graphitic carbon-like Alq3. Although all of the experimental results obtained here were performed in ambient condition, additional measurements in the lower frequency region65 may be useful for the study of the chemical interaction between metal and the organic materials and the characterization of the isomerization at the interface.39,49 Chemical interaction between Al and Alq3 can be studied directly by measuring lower frequencies from 500 to 1000 cm
1. It is also useful to measure the two-color SFG in UHV condition because of the elimination of the ambient effects.

5. CONCLUSION The interfacial vibrational and electronic states of Alq3/Al, Al/ Alq3, and Al/LiF/Alq3 were studied using two-color SFG spectroscopy. The two-color SFG spectra of the 2 nm thick Alq3/Al show the bands derived from the pristine Alq3. Due to the double resonance effect, remarkable changes in intensities of the SFG peaks derived from the CdC stretching of the quinolate ligands can be clearly observed by changing the visible wavelength. In contrast, significant changes are observed at the Al/ Alq3. The SFG excitation profiles derived from the CdC bands do not show the wavelength dependences, indicating the perturbation of the HOMO and LUMO of pristine Alq3 by the interaction of the Al. The spectral features of the two color SFG spectra of the Al/ LiF/Alq3 system show quite different behavior from those of Alq3/Al and Al/Alq3. The shift of the CdC stretching modes toward lower frequencies is indicative of the formation of the Alq3 anionic states at the buried interface in atmospheric condition. Additional broad bands around 1335 and 1450 cm
1, which show the weak excitation wavelength dependence, might be due to the existence of the Li-reacted graphitic carbon-like Alq3. Coexistence of the negatively charged Alq3 and the Li-reacted graphitic carbon-like Alq3 at the buried interface is proposed. Although the experimental results obtained here were performed in ambient condition, the present study lead to a better understanding of the electronic structure of the Alq3/Al interfaces. In much the same way as multidimensional spectroscopy is suitable for studies of intra- and intermolecular interactions in the

ARTICLE

bulk, the two-color SFG provides a similar opportunity for studies of molecules at interfaces. Further information about the uniaxial orientation in the bulk film, such behavior is often observed in the evaporated organic film, can be studied by measuring the thickness dependence. This two-color SFG technique, which was proved feasible for OLEDs studied in this paper, offers a novel spectroscopy for the characterization of the vibrational and electronic structures at the buried organic interfaces.

’ ASSOCIATED CONTENT

bS

Supporting Information. Sample preparation of Alq3 on Au, the thickness dependent SFG spectra in PPP polarization combination of the Alq3 deposited on gold evaporated films for visible wavelengths of 450 nm, and the fitting parameters for Figures 3, 5, and 7. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*Telephone: 81 29 861 9389. Fax 81 29 861 6236. E-mail: [email protected].

’ ACKNOWLEDGMENT The authors thank Dr. Yoko Sakurai for helpful discussions. This work is supported in part by Grant-in Aid for Scientific Research (Nos. 18550025 and 21245042). ’ REFERENCES (1) Salaneck, W. R., Seki, K., Kahn, A., Pireaux, J. J., Eds.; Conjugated Polymer and Molecular Interfaces—Science and Technology for Photonic and Optoelectronics Applications; Marcel Dekker: New York, 2002. (2) Ishii, H.; Sugiyama, K.; Ito, E.; Seki, K. Adv. Mater. 1999, 11, 605. (3) Shen, Y. R. Nature 1989, 337, 519. (4) Shen, Y. R. The principles of Nonlinear Optics; Wiley & Sons: New York, NY, 1984. (5) Huang, J. H.; Shen, Y. R. Phys. Rev. A 1994, 49, 3973. (6) Raschke, M. B.; Hayashi, M.; Lin, S. H.; Shen, Y. R. Chem. Phys. Lett. 2002, 359, 367. (7) Caudano, Y.; Silien, C.; Humbert, C.; Dressen, L.; Mani, A. A.; Peremans, A.; Thiry, P. A. J. Electron Spectrosc. 2003, 129, 139. (8) Ishibashi, T.; Onishi, H. Appl. Phys. Lett. 2002, 81, 1338. (9) Maeda, T.; Ishibashi, T. Appl. Spectrosc. 2007, 61, 459. (10) Li, Q.; Hua, R.; Chou, K. C. J. Phys. Chem. B 2008, 112, 2315. (11) Miyamae, T.; Miyata, Y.; Kataura, H. J. Phys. Chem. C 2009, 113, 15314. (12) Miyamae, T.; Tsukagoshi, K.; Mizutani, W. Phys. Chem. Chem. Phys. 2010, 12, 14666. (13) Tang, C. W.; VanSlyke, S. A. Appl. Phys. Lett. 1987, 51, 913. (14) Brinkmann, M.; Gadret, G.; Muccini, M.; Taliani, C.; Masciocchi, N.; Sironi, A. J. Am. Chem. Soc. 2000, 122, 5147. (15) Kushto, G. P.; Iizumi, Y.; Kido, J.; Kafafi, Z. H. J. Phys. Chem. A 2000, 104, 3670. (16) Nishiyama, Y.; Fukushima, T.; Takami, K.; Kusaka, Y.; Yamazaki, T.; Kaji, H. Chem. Phys. Lett. 2009, 471, 80. (17) Hung, L. S.; Tang, C. W.; Mason, M. G. Appl. Phys. Lett. 1997, 70, 152. (18) Jabbour, G. E.; Kippelen, B.; Armstrong, N. R.; Peyghambarian, N. Appl. Phys. Lett. 1998, 73, 1185. (19) Shaheen, S. E.; Jabbour, G. E.; Morrell, M. M.; Kawabe, Y.; Kippelen, B.; Peyghambarian, N.; Nabor, M.
F.; Schlaf, R.; Mash, E. A.; Armstrong, N. R. J. Appl. Phys. 1998, 84, 2324. 9559

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560

The Journal of Physical Chemistry C (20) Lee, C. H. Synth. Met. 1997, 91, 125. (21) Mason, M. G.; Tang, C. W.; Hung, L.
S.; Raychaudhuri, P.; Madathi, J.; Giesen, D. J.; Yan, L.; Le, Q. T.; Gao, Y.; Lee, S.
T.; Liao, L. S.; Cheng, L. F.; Salaneck, W. R.; dos Santos, D. A.; Bredas, J. L. J. Appl. Phys. 2001, 89, 2756. (22) He, P.; Au, F. C. K.; Wang, Y. M.; Cheng, L. F.; Lee, C. S.; Lee, S. T. Appl. Phys. Lett. 2000, 76, 1422. (23) Kido, J.; Matsumoto, T. Appl. Phys. Lett. 1998, 73, 2866. (24) Ito, E.; Washizu, Y.; Hayashi, N.; Ishii, H.; Matsuie, N.; Tsuboi, K.; Harima, Y.; Yamashita, K.; Seki, K. J. Appl. Phys. 2002, 92, 7306. (25) Yoshizaki, K.; Manaka, T.; Iwamoto, M. J. Appl. Phys. 2005, 97, 023703. (26) Manaka, T; Yoshizaki, K.; Iwamoto, M. Curr. Appl. Phys. 2006, 6, 877. (27) Seki, K.; Ito, E.; Ishii, H. Synth. Met. 1997, 91, 137. (28) Hayashi, N.; Imai, K.; Suzuki, T.; Kanai, K.; Ouchi, Y.; Seki, K. Proc. Int. Symp. Super-Functionality Org. Devices, IPAP Conf. Ser. 2004, 6, 69. (29) Sugi, K.; Ishii, H.; Kimura, Y.; Niwano, M.; Ito, E.; Washizu, Y.; Hayashi, N.; Ouchi, Y.; Seki, K. Thin Solid Films 2004, 464
465, 412. (30) Ozasa, K.; Nemoto, S.; Isoshima, T.; Ito, E.; Maeda, M.; Hara, M. Appl. Phys. Lett. 2008, 93, 263304. (31) Isoshima, T.; Ito, H.; Ito, E.; Okabayashi, Y.; Hara, M. Mol. Cryst. Liq. Cryst. 2009, 505, 59. (32) Noguchi, Y.; Sato, N.; Tanaka, Y.; Nakayama, Y.; Ishii, H. Appl. Phys. Lett. 2008, 92, 203306. (33) Hayashi, M.; Lin, S. H.; Raschke, M. B.; Shen, Y. R. J. Phys. Chem. A 2002, 106, 2271. (34) Ino, D.; Watanabe, K.; Takagi, N.; Matsumoto, Y. Phys. Rev. B 2005, 71, 115427. (35) Noguchi, Y.; Sato, N.; Miyazaki, Y.; Ishii, H. Appl. Phys. Lett. 2010, 96, 143305. (36) Miyamae, T.; Tsukagoshi, K.; Matsuoka, O.; Yamamoto, S.; Nozoye, H. Langmuir 2001, 17, 8125. (37) International Union of Pure and Applied Chemistry. Commission on Molecular Structure and Spectroscopy Tables of Wavenumbers for the Calibration of Infra-red Spectrometers; Butterworths: London, 1961. (38) Halls, M. D.; Aroca, R. Can. J. Chem. 1998, 76, 1730. (39) Sakurai, Y.; Hosoi, Y.; Ishii, H.; Ouchi, Y.; Salvan, G.; Kobitski, A.; Kampen, T. U.; Zahn, D. R. T.; Seki, K. J. Appl. Phys. 2004, 96, 5534. (40) Yokoyama, T.; Yoshimura, D.; Ito, E.; Ishii, H.; Ouchi, Y.; Seki, K. Jpn. J. Appl. Phys. 2003, 42, 3666. (41) Halls, M. D.; Schlegel, H. B. Chem. Mater. 2001, 13, 2632. (42) Wu, D.; Deng, G. H.; Guo, Y.; Wang, H. F. J. Phys. Chem. A 2009, 113, 6058. (43) Zhuang, X.; Miranda, P. B.; Kim, D.; Shen, Y. R. Phys. Rev. B 1999, 59, 12632. (44) Parik, E. D., Ed.; Handbook of Optical Constants of Solids; Academic Press: London, 1985. (45) Djurisi_c, A. B.; Kwong, C. Y.; Guo, W. L.; Lau, T. W.; Li, E. H.; Liu, Z. T.; Kwok, H. S.; Lam, L. S. M.; Chan, W. K. Thin Solid Films 2002, 416, 233. (46) Curioni, A.; Andreoni, W. J. Am. Chem. Soc. 1999, 121, 8216. (47) Zhang, R. Q.; Lu, W. C.; Lee, C. S.; Hung, L. S.; Lee, S. T. J. Chem. Phys. 2002, 116, 8827. (48) Yanagisawa, S; Morikawa, Y. Jpn. J. Appl. Phys., Part 1 2006, 45, 413. (49) Yanagisawa, S.; Lee, K.; Morikawa, Y. J. Chem. Phys. 2008, 128, 244704. (50) Salaneck, W. R.; Stafstr€om, S.; Bredas, J. L. Conjugated Polymer Surfaces and Interfaces: Electronic and Chemical Structure of Interfaces for Polymer Light Emitting Devices; Cambridge University Press: Cambridge, 1996. (51) L€oglund, M.; Dannetun, P.; Salaneck, W. Handbook of Conducting Polymers; Skotheim, T., Elsenbaumer, R. L., Reynolds, J. R., Ed.; Marcel Dekker: New York, 1997. (52) Miyamae., T.; Aoki, M.; Etori, H.; Muramatsu, Y.; Saito, Y.; Yamamoto, T.; Sakurai, Y.; Seki, K.; Ueno, N. J. Electron Spectrosc. Relat. Phenom. 1998, 88, 905.

ARTICLE

(53) Hirose, Y.; Kahn, A.; Aristov, V.; Soukiassian, P. J. Appl. Phys. 1996, 68, 217. (54) Tarlov, M. J. Langmuir 1992, 8, 80. (55) Wei, X.; Hong, S. C.; Lvovsky, A. I.; Held, H.; Shen, Y. R. J. Phys. Chem. B 2000, 104, 3349. (56) Curioni, A.; Andreoni, W. IBM J. Res. Rev. 2001, 45, 101. (57) Yokoyama, T.; Ishii, H.; Matsuie, N.; Kanai, K.; Ito, E.; Fujimori, A.; Araki, T.; Ouchi, Y.; Seki, K. Synth. Met. 2005, 152, 277. (58) Park, K.; Ding, H.; So, F.; Gao, Y. Proc. SPIE 2008, 7051, 70511V. (59) Ding, H.; Park, K.; Gao, Y.; Kim, D. Y.; So, F. Chem. Phys. Lett. 2009, 473, 92. (60) Ganzorig, C.; Fujihira, M. Appl. Phys. Lett. 2002, 81, 3137. (61) Sakurai, Y., Ph. D thesis, Nagoya University, 2004. (62) Davis, R. J.; Pemberton, J. E. J. Phys. Chem. C 2008, 112, 4364. (63) Davis, R. J.; Pemberton, J. E. J. Phys. Chem. A 2009, 113, 4397. (64) Kaufman, J. H.; Metin, S.; Saperstein, D. D. Phys. Rev. B 1989, 39, 13053. (65) Nomoto, T.; Onishi, H. Phys. Chem. Chem. Phys. 2007, 9, 5515.

9560

dx.doi.org/10.1021/jp201051h |J. Phys. Chem. C 2011, 115, 9551–9560