High-sensitivity ultraviolet photoemission spectroscopy technique for direct detection of gap states in organic thin films

Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

Contents lists available at ScienceDirect

Journal of Electron Spectroscopy and Related Phenomena journal homepage: www.elsevier.com/locate/elspec

High-sensitivity ultraviolet photoemission spectroscopy technique for direct detection of gap states in organic thin films Fabio Bussolotti ∗ Department of Nanomaterial Science, Graduate School of Advanced Integration Science, Chiba University, Chiba 263-8522, Japan

a r t i c l e

i n f o

Article history: Available online 11 April 2015 Keywords: Photoemission spectroscopy Gap states Organic molecules Organic/inorganic interface Organic/organic interface

a b s t r a c t We developed ultrahigh sensitivity, low-background ultraviolet photoemission spectroscopy (UPS) technique which does not introduce detectable radiation damages into organic materials. The UPS allows to detect density of states of the order of ∼1016 states eV−1 cm−3 even for radiation-sensitive organic films, this results being comparable to electrical measurements of charge trapping centers. In this review we introduce the method of ultrahigh sensitivity photoemission measurement and we present some results on the energy distribution of gap states in pentacene (Pn) films deposited on SiO2 and Au(1 1 1) substrate. For Pn/SiO2 thin film the results show that exposure to inert gas (N2 and Ar) atmosphere produces a sharp rise in gap states from 1016 to 1018 states eV−1 cm−3 and pushes the Fermi level closer to the valence band (0.15–0.17 eV), as does exposure to O2 (0.20 eV), while no such gas-induced effects are observed for Pn/Au(1 1 1) system. The results demonstrate that these gap states originate from small imperfections in the Pn packing structure, which are induced by gas penetration into the film through the Pn crystal grain boundaries. Similar results were obtained for CuPc/F16 CuPc thin films, a prototypical example of donor/acceptor interface for photovoltaic application. © 2015 Elsevier B.V. All rights reserved.

1. Introduction: energy level alignment at organic interfaces and role of gap states Organic-based (opto)electronic devices, such as light emitting diodes, solar cells, field effect transistors, generally include multiple interfaces between organic semiconductors and organic or inorganic dielectrics, metals or other types of conducting electrodes. Without exception, the electronic structure and electrical behavior of these interfaces have large impact of the performances of devices. In this context, organic/organic and organic/inorganic interfaces have been the object of extensive experimental and theoretical investigations [1–6]. The transport of charge carriers through a semiconductor interface depends on: (i) the atomic or molecular structure of this interface, i.e. bonding or molecular orientation across the interface, and (ii) the electronic structure of this boundary region, i.e. the relative positions of the relevant electronic transport levels. In a simple metal/organic film/metal system, these levels are: (i) the Fermi level of the metal electrodes, (ii) the lower edge of the lowest

∗ Present address: Institute for Molecular Science, Myodaiji, Okazaki 444-8585, Japan. Tel.: +81 0564557413. E-mail address: [email protected] http://dx.doi.org/10.1016/j.elspec.2015.04.002 0368-2048/© 2015 Elsevier B.V. All rights reserved.

unoccupied molecular orbital (LUMO) band, (iii) the upper edge of the highest occupied molecular orbital (HOMO) band, and (iv) the Fermi level of the organic film. When the system is in the thermal equilibrium, the Fermi levels are aligned at the interface to result in the single Fermi level throughout the system. Once the thermal equilibrium is achieved, the electrical transport through the organic system requires the charge injection from the electrode into the organic film. The hole (electron) injection efficiency, which determines the mobile hole (electron) concentration, depends on the energy barrier defined between the Fermi level of the electrode and the HOMO (LUMO) level of the organic layer. The holes (electrons) transport at the HOMO (LUMO) level through the organic material, then they are finally collected at the other electrode. Without exception, therefore, the relative positions of the electrode Fermi level and the HOMO (for hole) and the LUMO (for electron) levels of the organic layer are the key factor for establishing electrical current between the electrodes, and this important concept is simply called as energy level alignment (ELA). Recently, the ELA mechanism was related to the density of gap states (DOGS) caused by the slight structural imperfections in the molecular packing [1–4]. Starting from the HOMO and LUMO edges, the structure-related DOGSs gradually decrease to have tail states in the energy gap, resulting in a typical exponential- or Gaussian-like tail [1–4].

30

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

Fig. 1. Energy level alignment at a weakly interacting interface of metal and organic film. (Left panel) Before contact. (Right panel) After contact, where the thermal equilibrium is achieved. The work function of the metallic electrode (WFmet ) is assumed to be larger than that of the organic film (WFmet ). The HOMO and LUMO levels (shown by lines) are assumed to have broadening that results in gap states due to imperfection of molecular packing structure. Note that the LUMO and HOMO derived density of states and broadening are, in principle, not symmetric, as the LUMO and HOMO wavefunctions have different spatial extension and degeneracy.

The ELA is achieved after reaching the thermal equilibrium by electron transfer across the interfaces, as schematically illustrated in Fig. 1. The number of transferred electrons depends on the relative position of the two Fermi levels before the contact of the electrode and the organic film. After the contact is established the electrons flow from the lower to the higher work function material to achieve the thermal equilibrium. If the two materials do not interact strongly at the interface, it is possible to assume that no electric dipole exists across the interface. In this case the electron transfer creates electrostatic potential in the organic film relative to the electrode where the potential is constant because it is an electrical conductor. This process is schematically illustrated in Fig. 1, where the organic material is assumed to have the lower work function. Initially, such electrons come from the occupied gap states near the center of the HOMO–LUMO gap of organic material, where a very low density of gap states exists. When the integrated electronic charge over the low density of gap states is not sufficient to achieve the thermodynamic equilibrium, the electrons from the deep gap states near the HOMO edge are also transferred to the substrate until the thermodynamic equilibrium is achieved. As the density of states near the HOMO edge is very large and increases steeply with energy toward the HOMO edge, the Fermi level seems to be located very closely to the HOMO band. With increasing of the film thickness, i.e. distance from the interface, the interfacial charge transfer becomes weaker, hence the integration of over a much smaller energy range is sufficient to ensure the transferring of enough electrons to achieve the thermodynamic equilibrium, thus locating/pinning the Fermi level further away from the HOMO (Fig. 1). As a result, the Fermi level moves further toward the center of the band gap of the organic film depending on the distance from the organic/metal interface. Finally, the Fermi level of the system is indicated with a line in usual energy diagram of the system and the effect of the potential is described in terms of as “curved” HOMO and LUMO levels with respect to the Fermi level (Fig. 1). Thus the Fermi level position within the HOMO-LUMO gap depends: (i) on the distance from the interface i.e. on the thickness of the molecular layer and (ii) on the density of gap states of the organic films as the potential changes with the distance from the interface depends on the excess charge distribution that is related to the energy and spatial distribution of the density of occupied and unoccupied gap states [7–10]. As a result the Fermi level position in the HOMO–LUMO gap is a unique indicator of various interface phenomena and plays a

crucial role for the energy level alignment and bending of the HOMO and the LUMO in the organic film. This means that the gap states control the charge injection barriers and therefore the concentration of the mobile hole and electron in organic layer. Moreover the gap states also act as charge trapping states and reduce the charge mobility significantly [11]. Clarify the true origin of gap states in organic materials is therefore crucial for a deeper understanding of the ELA mechanism at organic/inorganic and organic/organic interfaces, pinning of the Fermi level, and bandbending-like phenomena in organic semiconductor films, all of them having a serious impact on the electrical properties of various organic devices. In this context a direct and quantitative measure of the DOGS in organic semiconductors was highly requested. A possible method for this measurement is the ultraviolet photoelectron spectroscopy (UPS). However, this use of conventional UPS system for was generally hindered by: (i) the very low sensitivity to detect the tiny DOGS in organic materials, and (ii) the radiation damages introduced by the photons and photoelectrons during UPS measurement. In the following we describe the ultrahigh-sensitivity UPS that enables us to observe both of gap states and their energy distribution (Section 2). The technique has been applied for the direct detection of the DOGS of as-deposited and inert gas-exposed Pn thin film deposited on SiO2 and Au(1 1 1) substrate, the corresponding experimental results being presented in Section 3.1 of the present review. The same experimental study was conducted on CuPc/F16 CuPc thin films, a prototypical example of donor/acceptor interface for photovoltaic application (Section 3.2). 2. Methods: UPS system for gap states detection UPS is well known as the most potential tool to investigate the electronic properties of various materials. In UPS study of organic thin film, however, the following drawback may commonly affect the photoemission measurement of organic materials, whose impact on the experimental data must be carefully evaluated: (a) Poor sensitivity: The sensitivity of photoemission spectroscopy is commonly much poorer than that of the electrical measurement of charge trapping states. This makes low-density electronic states in the band gap difficult to measure. (b) Radiation damages of organics: It is not easy to observe organic materials as they are, because organic materials are easily

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

31

Fig. 2. Method of the high sensitivity measurement (left) and a photograph of the UPS instrument (right).

damaged by the incidence ionizing radiation. In some cases physical effects such as disorder or changes in molecular packing structure are introduced in addition to chemical changes of molecules [7]. (c) High surface sensitivity: Very short mean-free-path of photoelectrons [0.1 nm for organics (c.a. with HeI␣ radiation, h = 21.218 eV, as generally used in a large number of photoemission study of organic materials)] makes measurements of the bulk of solids and buried interface impossible. In order to overcome these issues a dedicated experimental apparatus was developed at Chiba University, Japan. Fig. 2 displays the principle of ultrahigh sensitivity measurement with a photograph of the apparatus. The UPS uses a hemispherical electron energy analyzer (MBS A-1) with two-dimensional detector and two monochromators (MBS-1/modified) for the incidence lights. The first monochromator is for photon energy (h) range of HeI␣ (h = 21.218 eV) and HeII␣ (h = 40.814 eV) radiations, and the second one covers lower h range of other gas resonance lines including HeI␣ (h = 8.437 eV/3 P1 emission and 9.570 eV/1 P1 emission: here we conventionally use XeI␣ for the former line) and hydrogen Ly␣ (h = 10.199 eV) and NeI (h = 16.673 eV and 16.850 eV). There are two discharge lamps installed to the UPS system, high-density plasma lamp (MBS L-1) for He discharge and a conventional high intensity discharge lamp (Omicron HIS 13) for the other gases. Two filters, Al thin film or LiF single crystal depending on the desired photon, are alternatively used to reduce the intensity of high/low-h impurity photons from the monochromator. Using a two-dimensional electron detector, moreover, photoelectrons excited by desired photons are selected for lowering secondary electron background produced by inelastic scattering of high kinetic energy photoelectrons excited by the remaining high-h photons, since the compact monochromator is not perfect. As photons reflected by the specimen produces electron background in the measurement chamber when they hit the inner surface of the chamber, we pay attention to reduce these electrons. The electrical wiring of the measurement system is carefully performed to minimize electrical noises by putting the electrical cables at appropriate positions. In particular connectors and coaxial cables for signals are checked frequently to minimize the electrical noise level. The AC power is supplied after passing noise-cut transformer. The system is grounded with higher-class earth line. To be more careful the UPS system is installed on a floor which was mechanically separated from working floor in order to minimize the vibration-induced electrical and magnetic noises.

3. Experimental results and discussion 3.1. Organic/inorganic interface: direct detection of gap states in pristine and gas-exposed 3.1.1. Pn/SiO2 and Pn/Au(1 1 1) thin film Si(1 0 0) wafers (n-type) with a thermally grown SiO2 layer (thickness = 3 nm) were cleaned in an acetone and an isopropanol ultrasonic bath. The SiO2 substrates were then annealed in a UHV preparation chamber (∼4 × 10−8 Pa) at 673 K for 12 h. Pn molecules (C22 H14 , Sigma–Aldrich), purified by three cycles of vacuum sublimation, were vacuum deposited onto the SiO2 substrate at room temperature (293 K, RT). The deposition rate (0.5 nm/min) was monitored by using a quartz microbalance. During the Pn deposition, the pressure remained stably below 6 × 10−8 Pa. The asdeposited Pn thin films were reported to have an upright-standing molecular orientation and herringbone-like intermolecular packing, as judged from the HeI␣ UPS spectral profile [12,13]. The Pn thin films were repeatedly exposed to 1-atm N2 with a purity of 99.99995% (6N5) at RT for a total exposure time of 20 h. Next, the film was exposed to 1-atm O2 (6N5) for an additional 5 h at RT. This gas exposure procedure was repeated using Ar gas (5N5) on different samples. The effect of the N2 exposure on Pn thin film on a clean Au(1 1 1) single crystal surface (thickness = 40 nm, RT deposition) was also evaluated. The gas exposures were performed by

Fig. 3. XeI␣ -full UPS spectra of the as-deposited Pn film (15 nm) on SiO2 before (black curve) and after (red curve) prolonged XeI␣ light irradiation (∼2 h). The typical acquisition time during DOGS detection is ∼40 min. No spectral shift and change in density of gap states (see inset) is detected after irradiation, indicating the absence of a significant effect from irradiation induced damage. The figure is cited from [17] with permission. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

32

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

Fig. 4. (a) XeI␣ -UPS spectra of as-deposited [spectra (1)], N2 -exposed [spectra (2)–(6)], O2 -exposed [spectra (7)] and Ar-exposed [spectra (8)] Pn thin film (15 nm) on SiO2 substrate. Spectra (9) includes the UPS data of as-deposited and N2 -exposed Pn/Au(1 1 1) thin film (40 nm) acquired at RT. The positions of the VL and HOMO derived bands (H1 and H2 ) are indicated by vertical (continuous) bars. The difference between the VL (HOMO) positions of the Pn/SiO2 thin films in the “N2 ” and “Ar” experiments reflects the difference between the initial SiO2 work functions. (b) DOS (log scale intensity plot) of as-deposited (filled (red) square symbols), N2 -exposed (20 h, filled (magenta) circles) and O2 -exposed (5 h, open circles) Pn thin film on SiO2 . Continuous (black) lines are the cumulative fitting curve of the HOMO band (see Section 2 of SI). DOS values were extracted from the UPS data, as described in Ref. [8]. SiO2 data are rescaled to preserve, in the DOS scale, their relative intensity with respect to the as-deposited Pn data. Vertical arrows indicate the threshold energy position (ET ) where the DOS distribution deviates from the cumulative fitting curve and DOGS starts to appear. The dashed line indicates the DOGS of the Pn thin film as determined by transport measurements. Data are adapted from Ref. [19]. (c) XeI␣ -UPS spectra of Pn thin film on Au(1 1 1) (40 nm) before and after N2 exposure. The Fermi edge of the Au(1 1 1) substrate is clearly visible (see also inset). No spectral change was detected upon gas exposure. The figure is cited from [17] with permission. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

using a UHV-compatible gas inlet line connected to the preparation chamber. The UPS measurements were performed at RT using an ultrahigh-sensitivity UPS apparatus with a hemispherical electron energy analyzer (MBS A-1) and a monochromatic XeI␣ (h = 8.437 eV) radiation source with a LiF single crystal filter. The electrical wiring of the measurement system was carefully performed to minimize the background noises by positioning the electrical cables at appropriate positions. The accumulation time of the spectra was typically 40 min. Under these conditions, radiation damage effects [14] to the Pn films are negligible as shown in Fig. 3 [15]. All UPS spectra were measured at normal emission

with an acceptance angle of ±18◦ , and a bias of −5 V was applied to the sample in order to measure the vacuum level (VL). The energy resolution of the UPS system was set to 30 meV. The binding energy scale was relative to the substrate Fermi level. Fig. 4(a) shows the XeI␣ -UPS spectra of the as-deposited and N2 -, Ar- and O2 -exposed Pn film (15 nm) in the cutoff and highest occupied molecular orbital (HOMO) regions [spectra (1)–(7)]. The HOMO band of the as-deposited Pn thin film consists of two main components [labeled H1 and H2 in Fig. 4(a)] with an energy separation of ∼0.45 eV. The H1 (H2 ) position was determined by a least-squares fitting of the HOMO-band UPS by a series of Gaussian functions as illustrated in Fig. 5. The two components correspond

Fig. 5. XeI␣ -UPS data of as-deposited (a), N2 -exposed [(b)–(f)], and O2 -exposed (g) Pn film (15 nm) with deconvoluted H1 (dashed black line), H2 [dash dotted (green) line, green shaded area] and H Gaussian components [dotted (orange) line, orange shaded area]. The H2 and H components are omitted from panels (b)–(g) for clarity. The H1 and H2 Gaussian functions simulate the HOMO band splitting (see the discussion of the main text). The H Gaussian function is used to simulate the contribution of the band dispersion, electron–phonon coupling, and structural disorder to the broadening of the HOMO lineshape. The continuous (red) line is the cumulative fitting curve. Experimental and simulated intensity is plotted on a logarithmic scale in order to highlight the difference between the experimental data and fitting curve in the gap energy region (∼0–0.5 eV). Black arrows indicate the threshold energy (ET ) of the gap state energy distribution which is not reproduced by the fitting curve. The figure is cited from [17] with permission. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

33

Fig. 6. Schematic representation of the gas penetration and mediated imperfections in Pn thin film on SiO2 . During exposure [panel (a)], gas molecules progressively penetrate into the Pn film, where they locally alter the original intermolecular packing geometry [panel (b)]. The gas penetration proceeds through the grain boundaries of the film. Because of the weak N2 –Pn interaction, N2 molecules are easily removed once the film is put back in UHV. The result is a weakly disordered film [(panel (c)]. The figure is cited from [17] with permission.

to the density of states of the HOMO band in the Pn thin film with an upright-standing molecular orientation and herringbone-like intermolecular packing [13,16,17]. Following each N2 exposure, the H1 (H2 ) peak shifts toward low binding energies, while the vacuum level (VL) moves upwards by the same amount [spectra (1)–(6) in Fig. 4(a)]. The N2 -induced “rigid” energy shift almost saturates after a total exposure time of 30 h. The subsequent exposure to O2 (5 h) induces an additional energy shift [spectra (7)]. Interestingly, the ionization potential (IP) of the as-deposited Pn thin film (IP = 4.90 eV) is not affected by the various gas exposure treatments. The IP of an organic thin film was reported to be strongly dependent on the molecular orientation [18] and molecular packing [13]. Therefore, the stability of the IP in the present case indicates that no large-scale structural rearrangement occurs. Similar results are obtained for the Ar-exposed Pn/SiO2 thin film [spectra (8)], while no spectral shift is detected for Pn/Au(1 1 1) thin film after the prolonged exposure (19 h) to 1-atm N2 [spectra (9)]. Fig. 4(b) shows the expanded UPS spectra of the HOMO and gap energy region on log scale for the as-deposited, N2 -exposed (total exposure time = 20 h) and O2 -exposed (total exposure time = 5 h following the 20 h N2 exposure) Pn thin film with the corresponding fitting curves. The UPS spectrum of the SiO2 substrate is shown for comparison purpose. The DOGS of the asdeposited Pn film is very low, but it significantly increases after N2 and O2 -exposure. Ar-exposure also leads to a similar DOGS increase (not shown). In the gap state binding energy region (∼0–0.5 eV), the DOS of the as-deposited Pn film and the SiO2 substrate are comparable [Fig. 4(b)]. Therefore, the SiO2 DOS may overlap with the DOGS of the as-deposited Pn film in a way that depends on the electron mean free path in the Pn overlayer (Pn ) which was estimated to be 9 nm < Pn < 27 nm from the Pn/Au(1 1 1) UPS data [15]. Once the substrate contribution is taken into account, the DOGS of the as-deposited film (∼1016 states eV−1 cm−3 ) turns out to be comparable to that detected by electrical measurements for Pn thin films [11,19]. For the gas-exposed Pn films, the DOGS is ∼10 times larger than the contribution from the substrate, and can therefore be unambiguously related to the Pn film. In particular, for the N2 -exposed sample, an exponential-type DOGS [linear on the log scale of Fig. 4(b)] is clearly visible between the HOMO threshold energy ET = 0.4 eV (where the DOS starts deviating from a Gaussian line shape) and EF . A different energy dependence is observed for the DOGS of the O2 -exposed sample. Post-annealing experiments (50 ◦ C for 18 h) on N2 -exposed Pn films shows a gradual recovery to the original VL and HOMO positions (i.e. before gas exposure). At the same time, a decrease in the DOGS is observed [15]. In principle, a weak structural disorder in organic layers was found to be easily removed by moderate annealing treatments [9].

Consequently, the DOGS of the gas-exposed sample can be ascribed to a slight intermolecular packing disorder resulting from prolonged gas exposure. These gap states may in turn affect the position of EF within the energy gap [8,17]. Note that no residual N2 is detected by X-ray photoemission spectroscopy in the gas-exposed film following re-introduction into UHV [15]. This supports the conclusion that the DOGS is not due to N2 -related states but rather to intermolecular packing disorder. The increase in DOGS and its different energy distribution upon O2 exposure can probably be ascribed to: (i) a more effective penetration of O2 molecules into the Pn film (the penetration of a gas into organic systems strongly depends on its chemical properties and size [20]) and/or (ii) chemical interaction between the O2 molecules and the Pn molecules [21,22]. However, it is worth noting that a prolonged N2 exposure of Pn films prepared at RT on single crystalline Au(1 1 1) does not induce any detectable change in the DOGS or spectral shift [Fig. 4(a) and (c)]. Pn films deposited on Au(1 1 1) was reported to exhibit larger crystallites [size ∼ 200 nm] and significantly fewer grain boundaries than those on SiO2 [23,24], suggesting that the grain boundaries play a crucial role in the penetration of the gas molecules and in the changes of the intermolecular packing geometry as depicted in Fig. 6. According to the UPS data [Fig. 4(b)], the defect density is estimated to be in the range of 1016 to 1018 cm−3 , corresponding to a defect-to-molecule ratio of 10−5 to 10−3 , which is hardly detectable via structural diffraction techniques [25]. The hypothesis that the DOGS is mediated by the structural defects in the Pn film is also supported by theoretical calculations by Kang et al. [26]. They showed that (i) sliding defects along the Pn long axis create shallow gap states and (ii) the HOMO levels at the defect sites are distributed over a range of up to 100 meV from the HOMO of the unperturbed molecules. According to this model, the DOS in the HOMO (gap state) binding energy region is expected to decrease (increase) with the density of defects. To verify this point, the evolution of the UPS intensity upon gas exposure is carefully evaluated. First, the XeI␣ UPS spectra of as-deposited, N2 -exposed and O2 -exposed Pn thin films are aligned to the position of the HOMO band maximum (H2 ) [Fig. 7(a) and (b)]. Next the ratio between the UPS spectra of the gas-exposed and the as-deposited samples is evaluated for each gas-exposure step [Fig. 7(c)]. For long exposure times (≥2 h), the increase in the UPS intensity in the gap state energy region (spectral ratio > 1, E > ET ) corresponds to a reduction in the HOMO band intensity (spectral ratio < 1) within ∼0.2 eV from the threshold position. This result is in qualitative agreement with the theoretical prediction [26]. In the high intensity region of the HOMO band (∼0.5 eV) the slight increase of the electronic states and/or the small broadening in the HOMO lineshape, as due to structural disorder induced

34

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

Fig. 7. (a) XeI␣ -UPS data of as-deposited, N2 -exposed and O2 -exposed Pn/SiO2 thin film in the HOMO band region. The energy scale is referred to the H2 position (dotted line) where the HOMO band reaches the maximum intensity. The vertical, short bars on each plot indicate the position of the Fermi level. The SiO2 substrate signal was not removed from the as-deposited sample spectra. (b) Same as panel (a), but plotted on a log scale. The continuous (black) line indicates the position of the threshold energy (ET ) from which the gap state energy distribution starts to appear. (c) Spectral ratio between the data of gas-exposed film and as-deposited film (see text for details). The figure is cited from [17] with permission.

by gas exposure [9], results in the slight increase of the spectral ratio (observed at ∼0.5 eV). Finally, we comment on the impact of the gas exposure on the observed molecular level shift, i.e. the shift of the HOMO level toward EF (see Fig. 4(a)). We suggest that the lowest unoccupied molecular orbital (LUMO) of Pn also gives rise to a distribution of (unoccupied) gap states resulting from imperfections in the intermolecular packing [26]. The energy distributions of the HOMO- and LUMO-derived DOGS are not symmetric because the corresponding wave functions have different spatial spreads [26]. Such disorderinduced-, non-symmetric DOGS (tailing states) affect the position of EF within the gap of organic semiconductors. Depending on the DOGS and their energy distribution, EF may lie closer to the pristine LUMO or HOMO band [9]. With increasing of the structural disorder (as induced, for example, by gas exposure) EF is expected to lie even closer to the HOMO, as observed in the present work.

described in Section 3.1 for the study of the Pn/SiO2 and Pn/Au(1 1 1) thin films. Fig. 8 shows the UPS spectra of the CuPc(1 nm)/F16 CuPc(6 nm)/ SiO2 and CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin films before [spectra (1) and (3)] and after [spectra (2) and (4)] the exposure to 1-atm N2 in the cutoff, valence band (VB) and CuPc-HOMO binding energy region. The UPS data of the as-deposited and N2 -exposed F16 CuPc(6 nm)/SiO2 thin films are also included. The N2 exposure modifies the positions of the CuPc energy levels. Interestingly, the direction of the N2 -induced energy shift changes with the thickness of the CuPc layer. For low CuPc thickness (1 nm), the N2 exposure leads an increase of the CuPc-HOMO binding energy by 0.07 eV, as evaluated from the HOMO edge positions, while the VL shifts downwards by 0.03 eV. At higher CuPc layer thickness (8 nm) the

3.2. Organic/organic interface: impact of structural DOGS on the ELA A 6 nm thin film of F16 CuPc was deposited on the SiO2 substrate at RT. The organic film (indicated as “F16 CuPc(6 nm)/SiO2 ” in the following) was then exposed to 1-atm N2 for 24 h at RT. A very thin layer (1 nm) of CuPc was further RT-deposited on the F16 CuPc/SiO2 film and the as-obtained organic/organic heterostructure (OOH) [“CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 ”] was exposed to 1-atm N2 for 18 h. Finally, additional 7 nm of CuPc was deposited (“CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 ”] and the N2 exposure was repeated for 15 h. The F16 CuPc films deposited at RT on SiO2 exhibit a nearly upright standing molecular orientation known as ␣-phase [27]. CuPc deposited on top nucleates in the ␣-phase, which is also nearly upright oriented [28,29]. For high deposition temperatures (∼490 K), X-ray diffraction data indicate that an intermixing zone may form at the CuPc/F16 CuPc interface [30]. However the sample preparation for the present experiment was done at RT, where no intermixing between both compounds was detected [31]. The UPS data was acquired in the same experimental condition (sample bias, energy resolution, acquisition time)

Fig. 8. XeI␣ -UPS spectra of the CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 and CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin films before [spectra (1) and (3)] and after [spectra (2) and (4)] the exposure to 1-atm N2 . The UPS spectra of the as-deposited and N2 -exposed F16 CuPc(6 nm)/SiO2 thin film are also shown at the bottom. The VL and HOMO edge positions (dash dotted vertical bars) were evaluated by linear extrapolation of the secondary electron cutoff and the CuPc-HOMO bands, respectively. The position of the HOMO peak of CuPc and F16 CuPc are indicated by vertical solid bar.

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

35

Fig. 9. (a) XeI␣ -UPS (log scale intensity plot) of as-deposited (red circles) and N2 -exposed (yellow triangles) CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 thin film near EF . The UPS spectrum of the N2 -exposed F16 CuPc(6 nm)/SiO2 thin film is also shown (gray shaded area). The F16 CuPc-related feature (L*) significantly contributes to the UPS intensity near EF . (b) and (c) Enlarged log scale intensity plots of the UPS spectra near EF , where the intensity of the F16 CuPc(6 nm)/SiO2 film was normalized to that of as-deposited [panel (b)] and N2 -exposed [panel (c)] CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 at EF . (d) UPS spectra (log scale intensity plot) of as-deposited and N2 -exposed CuPc (1 nm)/F16 CuPc(6 nm)/SiO2 thin film after removal of the F16 CuPc(6 nm)/SiO2 contribution. The UPS data were normalized to the highest intensity point of the CuPc-HOMO band to represent DOGS related changes. DOS values (right scale) were obtained according to the same procedure described in Ref. [9]. The molecular packing density of the bulk ␣-phase of CuPc was used (5.82 × 1021 molecule cm−3 )[31]. Continuous lines are the cumulative fitting curve of the HOMO bands. Vertical arrow indicates the energy position where the UPS data of the N2 -exposed film deviate from the Gaussian like fitting curve and a DOGS-related tail appears (dash dotted line). Inset of panel (d): comparison between the CuPc-HOMO band tail of as-deposited and N2 -exposed CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 thin films. The UPS spectra were aligned to the HOMO edge position of the as-deposited thin film. The figure is cited from [10] with permission. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

CuPc-HOMO binding energy decreases by 0.06 eV after N2 exposure while the VL moves upwards by the same amount. Noteworthy, the energy levels of the F16 CuPc(6 nm)/SiO2 thin film are not affected even by the prolonged exposure to N2 (Fig. 8). Therefore, the N2 induced energy shifts observed for the CuPc/F16 CuPc OOHs can be unequivocally related to the impact of gas exposure on the electronic properties of the CuPc layer. This important finding was confirmed by monitoring carefully the N2 -induced variation in the spectral intensity and lineshape. Fig. 9(a) displays the UPS data of the F16 CuPc(6 nm)/SiO2 film (after N2 exposure) and of the as-deposited and N2 -exposed CuPc(1 nm)/F16 CuPc/SiO2 thin film in the CuPc-HOMO and gap region. All the UPS data are plotted on log intensity scale to highlight the spectral changes induced by the N2 exposure. For both the asdeposited and N2 -exposed films, the CuPc-HOMO band is found to extend up to near EF , where a clear (attenuated) contribution from the F16 CuPc-related state L* [Fig. 9(a)] is also visible. The L* state originates from the partial filling of the F16 CuPc LUMO by electrons from the substrate, to establish thermodynamic equilibrium at the F16 CuPc/SiO2 interface [32]. In particular, the spectral intensity near EF is well reproduced by renormalizing the UPS intensities of F16 CuPc(6 nm)/SiO2 thin film to that of the as-deposited [Fig. 9(b)] and N2 -exposed [Fig. 9(c)] CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 thin film at EF . The difference spectra between the as-deposited and N2 -exposed CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 and (normalized) F16 CuPc(6 nm)/SiO2 are shown on log intensity scale in Fig. 9(d) with the corresponding Gaussian fitting curves. The CuPc-HOMO band of the as-deposited film is very well reproduced by a Gaussian lineshape, even at very low binding energy values. For the N2 -exposed sample, an exponential-like tail [linear on the log scale of Fig. 9(d)] is observed between ∼0.3 eV and EF . At ∼0.3 eV the UPS data starts deviating from a purely Gaussian behavior. The change in CuPc-HOMO band tailing upon N2 exposure are highlighted in the inset of Fig. 9(d), where the UPS data are aligned at the HOMO edge position of the asdeposited film. The impact of N2 -exposure on the HOMO band tailing was confirmed by repeating the gas exposure treatment on a similarly prepared CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 thin film [32].

The UPS detection of an exponential-like HOMO band tail represents the signature of DOGS related to the structural imperfections in the CuPc layer [9,10,17]. These gap states may in turn affect the position of EF within the CuPc HOMO–LUMO energy gap [9,10,17]. In particular, a slight degree of disorder in the molecular packing can be introduced by the N2 penetration into the organic thin film which is favored by the presence of initial structural defects (i.e. grain boundaries, etc.) as already discussed for the Pn thin films [9,17]. Fig. 10(a) shows the UPS spectra (log scale intensity plot) of the as-deposited CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin film in the HOMO and gap energy region. Near EF , the L* state is still visible, due to the large electron mean free path in the XeI␣ -UPS [17] but it is shifted (by ∼0.15 eV) to lower binding energy with respect to the case of the F16 CuPc(6 nm)/SiO2 thin film. This energy shift can be possibly attributed to an increase of the polarization energy of the F16 CuPc layer upon CuPc deposition on top [33,34]. A similar polarization-related energy shift (∼0.1 eV) was reported for the HOMO of a pentacene (Pn) monolayer on Au(1 1 1) after the adsorption of additional Pn layers [34]. As the L* binding energy changes upon CuPc deposition, the spectral intensity of the CuPc (8 nm)/F16 CuPc(6 nm)/SiO2 thin film near EF is not well simulated by the normalized UPS data of the F16 CuPc(6 nm)/SiO2 thin film [Fig. 10(a)]. Therefore, the CuPcHOMO intrinsic lineshape cannot be extracted from the UPS data, thus hindering a direct quantitative evaluation of the impact of N2 exposure on the DOGS of CuPc. However we note that: (i) moderate annealing treatments of the N2 -exposed film (at 50 ◦ C and 70 ◦ C for 17 h and 15 h, respectively) lead to a progressive recovery of the original (i.e. before N2 exposure) CuPc energy level positions [Fig. 10(b) and (c)] and (ii) the intensity of the L* state decreases (increases) following to the N2 exposure (annealing treatments). The energy level shifts following to the N2 exposure and annealing treatments can be ascribed to the weak molecular packing disorder (with related DOGS) introduced by the N2 penetration into the CuPc layer which can be easily removed, in organic thin films, by moderate heating cycles [9]. An imperfect CuPc molecular packing may also result in a shorter mean free path for the photoelectron from the underlying F16 CuPc thin film [9,35]. Therefore, the change

36

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

Fig. 10. (a) XeI␣ -UPS spectra (log scale intensity plot) of as-deposited CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin film (red circles). The UPS data of the N2 exposure F16 CuPc(6 nm)/SiO2 thin film are also displayed. Dashed (black) line shows the UPS spectra of F16 CuPc(6 nm)/SiO2 thin film, as renormalized to the intensity of the CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 spectra at EF . Vertical arrows indicate the position of the F16 CuPc-related L* peak before and after the CuPc deposition. (b)–(d) XeI␣ -UPS spectra of as-deposited [spectra (4)], 15 h N2 -exposed [spectra (5)], 17 h annealed (at 50 ◦ C) [spectra (6)], 15 h annealed (at 70 ◦ C) [spectra (7)] and 18 h N2 -exposed [spectra (8)] CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin film in the VL (b) and CuPc-HOMO region (c). The position of the VL (CuPc-HOMO peak, CuPc-HOMO edge) of the as-deposited (N2 exposed and annealed) CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin film are indicated by dash-dotted red line (solid black line). Annealing treatments induce a gradual recovery of the energy level positions to the pre-exposure values. (d) Enlarged log scale intensity plot of (4)–(8) UPS spectra near EF . Note that the intensity of the F16 CuPc-related feature (L*) decreases (increases) after N2 exposure (annealing). The figure is cited from [10] with permission. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 11. (a) Schematic (qualitative) evolution of the energy level positions at CuPc/F16 CuPc interface as a function of the CuPc thickness. The occupied (unoccupied) energy levels are indicated by black (blue) line. At low CuPc-thickness the EF is located near the CuPc-HOMO. With increasing of the CuPc-thickness EF gradually moves toward the center of the HOMO–LUMO energy gap. Note that the energy scale does not reflect the real binding energy of the molecular levels. (b) and (c) Impact of N2 exposure on the energy level position of CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 [low CuPc thickness, panel (b)] and CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 [high CuPc thickness, panel (c)] thin film. The position of the VL (i.e. the work function ), and CuPc-HOMO edge were determined from the experimental data shown in Fig. 1. The LUMO position was determined according the CuPc electron affinity (EA = 2.92 eV). According to this value a CuPc HOMO–LUMO gap of ∼2.2 eV was estimated. The HOMO and LUMO derived DOGSs (black curves) are represented in log intensity scale (arbitrary scale) in both panels. At high CuPc thickness (8 nm) the EF lies closer to the middle of the CuPc HOMO–LUMO energy gap [panels (a) and (c)], where the CuPc-HOMO-related DOGS and LUMO-DOGS are comparable. As the N2 -induced changes in the CuPc-LUMO-related DOGS are larger than those in the HOMO-related-DOGS [panel (c)] a net electron charge transfer occurs from the F16 CuPc into the CuPc layer. This determines the observed upward movement of the CuPc energy levels. The N2 -induced structural disorder impacts differently on the CuPc-LUMO- and CuPc-HOMO-related DOGS as the CuPc-HOMO and LUMO wavefunctions have different spatial spread and degeneracy (see main text). The figure is cited from [10] with permission. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

in the UPS intensity of F16 CuPc-related L* state after various N2 exposure and thermal annealing treatments [Fig. 10(d)] can be consistently explained by the different degree of structural disorder in the gas-exposed and annealed CuPc layers. The impact of the N2 exposure (and related DOGS) on the CuPc level positions of the as-deposited CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 and CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 thin films is summarized in Fig. 11(b) and (c), as determined from the UPS data shown in Fig. 8. In the as-deposited CuPc(1 nm)/F16 CuPc(6 nm)/SiO2 thin film the EF is located very closely to the steepest part of the CuPcHOMO band tail [i.e. near the HOMO edge, Fig. 11(a) and (b)], as shown by the UPS data in Fig. 8(c). The origin of the EF pinning near the CuPc-HOMO at low CuPc thickness is described in Ref. [4]. Under such conditions, a small increase of the HOMO-derived DOGS [resulting, for example, by the N2 exposure and related disorder, see Fig. 9(c)] would lead to populated gap states above the initial EF . Therefore, a part of these electrons are transferred from the CuPc HOMO-derived DOGS to the F16 CuPc layer and/or SiO2 , in order to establish the thermodynamic equilibrium. This determines the observed downwards shift of the CuPc energy levels [Fig. 11(b)]. In the as-deposited CuPc(8 nm)/F16 CuPc(6 nm)/SiO2 film, the EF is positioned near the center of the CuPc HOMO–LUMO energy gap, where the HOMO- and LUMO-derived DOGS are, in principle, comparable [Fig. 11(a) and (c)]. In this case, therefore, small changes in HOMO and LUMO-derived DOGS can both affect the EF position within the CuPc energy gap and the related energy levels shifts. For pristine organic thin films, the HOMO and LUMO-derived DOGS, both resulting from structural imperfections, are not symmetric as the corresponding wavefunctions have different spatial spread [26] and degeneracy (i.e. CuPc-LUMO wavefunction is doubly degenerate while CuPc-HOMO wavefunction is non-degenerate [36]). Depending on the DOGS and the degree of asymmetry in their energy distributions, EF may be located closer to the pristine LUMO or HOMO band [9,10,17]. For pristine CuPc thin film, EF was reported to lie closer to the HOMO as a result of the gap state distribution induced by structural defects [11]. With increasing structural disorder (as induced, for example, by N2 exposure) EF is therefore expected to move closer to the CuPc-HOMO band, resulting in the observed upwards binding energy shift of the CuPc-molecular level [Fig. 11(c)].

4. Conclusions In this paper we shortly review on the development of experimental UPS system for the direct detection of ultralow density of gap states in organic thin films. The UPS allows to detect density of states of the order of ∼1016 states eV−1 cm−3 even for radiation-sensitive organic films, which is comparable to electrical measurements of charge trapping centers. In particular the technique was successfully applied to the study of the electronic properties of Pn thin film deposited on SiO2 and Au(1 1 1) single crystal, before and after the exposure to gaseous atmosphere. Striking effects due to gas exposure are that: (i) exposure to 1atm of inert gas (N2 or Ar) produces a rise in gap state density, as does exposure to O2 ; (ii) these gap states push the Fermi level closer to the HOMO; and (iii) the gap states originate from small imperfections in the Pn packing structure induced by gas molecule penetration into the film, presumably through the crystal-grain boundaries. Such imperfections remain even after removal of the gas molecules to yield gap states that are responsible to control the EF position within the gap. The present findings demonstrate the significant impact of slight structural disordering on the electronic properties of organic thin films. Furthermore, they reveal that complete control of the organic film structure is a requisite for producing organic devices with the desired properties. Finally, they

37

clearly demonstrate that, contrary to a widely held assumption, processing of organic films such as Pn in inert atmosphere is not without negative impact on the electronic structure of the material. Similar results were obtained for CuPc/F16 CuPc thin films, a prototypical example of donor/acceptor interfaces for photovoltaic applications.

Acknowledgements The ultrahigh sensitivity UPS project is supported by 21st Century COE program (MEXT) and 388 Global COE program (MEXT).

References [1] J. Hwang, E.J. Kim, J. Liu, J.L. Bredas, A. Duggal, A. Kahn, Photoelectron spectroscopic study of the electronic band structure of polyfluorene and fluorene-arylamine copolymers at interfaces, J. Phys. Chem. C 111 (2007) 1378–1384. [2] G. Heimel, S. Duhm, I. Salzmann, A. Gerlach, A. Strozecka, J. Niederhausen, C. Burker, T. Hosokai, I. Fernandez-Torrente, G. Schulze, S. Winkler, A. Wilke, R. Schlesinger, J. Frisch, B. Broker, A. Vollmer, B. Detlefs, J. Pflaum, S. Kera, K.J. Franke, N. Ueno, J.I. Pascual, F. Schreiber, N. Koch, Charged and metallic molecular monolayers through surface-induced aromatic stabilization, Nat. Chem. 5 (2013) 187–194. [3] S. Kera, Y. Yabuuchi, H. Yamane, H. Setoyama, K.K. Okudaira, A. Kahn, N. Ueno, Impact of an interface dipole layer on molecular level alignment at an organic–conductor interface studied by ultraviolet photoemission spectroscopy, Phys. Rev. B 70 (2004) 085304–085309. [4] H.Y. Mao, F. Bussolotti, D.-C. Qi, R. Wang, S. Kera, N. Ueno, A.T.S. Wee, W. Chen, Mechanism of the Fermi level pinning at organic donor–acceptor heterojunction interfaces, Org. Electr. 12 (2011) 534–540. [5] M. Fahlman, A. Crispin, X. Crispin, S.K.M. Henze, M.P. de Jong, W. Osikowicz, C. Tengstedt, W.R. Salaneck, Electronic structure of hybrid interfaces for polymer-based electronics, J. Phys.: Condens. Matter 19 (2007) 183202–183222. [6] H. Fukagawa, S. Kera, T. Kataoka, S. Hosoumi, Y. Watanabe, K. Kudo, N. Ueno, The role of the ionization potential in vacuum-level alignment at organic semiconductor interfaces, Adv. Mater. 19 (2007) 668–676. [7] M. Ono, T. Sueyoshi, Y. Zhang, S. Kera, N. Ueno, VUV induced doping of Cuphthalocyanine thin films: a possibility of n-type doping, Mol. Cryst. Liq. Cryst. 455 (2006) 241–249. [8] T. Sueyoshi, H. Fukagawa, M. Ono, S. Kera, N. Ueno, Low-density band-gap states in pentacene thin films probed with ultrahigh-sensitivity ultraviolet photoelectron spectroscopy, Appl. Phys. Lett. 95 (2009) 183303–183305. [9] T. Sueyoshi, H. Kakuta, M. Ono, K. Sakamoto, S. Kera, N. Ueno, Band gap states of copper phthalocyanine thin films induced by nitrogen exposure, Appl. Phys. Lett. 96 (2010) 093303–093305. [10] F. Bussolotti, J. Yang, A. Hinderhofer, Y. Huang, W. Chen, S. Kera, A.T.S. Wee, N. Ueno, Origin of the energy level alignment at organic/organic interfaces: the role of structural defects, Phys. Rev. B 89 (2014) 115319–115326. [11] W.L. Kalb, S. Haas, C. Krellner, T. Mathis, B. Batlogg, Trap density of states in small-molecule organic semiconductors: a quantitative comparison of thinfilm transistors with single crystals, Phys. Rev. B 81 (2010) 155315–155327. [12] H. Yoshida, N. Sato, Crystallographic and electronic structures of three different polymorphs of pentacene, Phys. Rev. B 77 (2008) 235205–235215. [13] H. Fukagawa, H. Yamane, T. Kataoka, S. Kera, M. Nakamura, K. Kudo, N. Ueno, Origin of the highest occupied band position in pentacene films from ultraviolet photoelectron spectroscopy: hole stabilization versus band dispersion, Phys. Rev. B 73 (2006) 245310–245315. [14] B. Boudaıffa, P. Cloutier, D. Hunting, M.A. Huels, L. Sanche, Resonant formation of DNA strand breaks by low-energy (3 to 20 eV) electrons, Science 287 (2000) 1658–1660. [15] See supplemental material at http://link.aps.org/supplemental/10.1103/ PhysRevLett.110.267602 [16] F. Bussolotti, S. Kera, N. Ueno, Potassium doping of single crystalline Pn thin film, Phys. Rev. B 86 (2012) 155120–155129. [17] F. Bussolotti, S. Kera, K. Kudo, A. Kahn, N. Ueno, Gap states in pentacene thin film induced by inert gas exposure, Phys. Rev. Lett. 110 (2013) 267602–267606. [18] S. Duhm, G. Heimel, I. Salzmann, H. Glowatzki, R.L. Johnson, A. Vollmer, J.P. Rabe, N. Koch, Orientation-dependent ionization energies and interface dipole in ordered molecular assemblies, Nat. Mater. 7 (2008) 326–332. [19] F. De Angelis, S. Cipolloni, L. Mariucci, G. Fortunato, Aging effects in pentacene thin-film transistors: analysis of the density of states modification, Appl. Phys. Lett. 88 (2006) 193508–193511. [20] H. Kajiro, A. Kondo, K. Kaneko, H. Kanoh, Flexible two-dimensional squaregrid coordination polymers: structures and functions, Int. J. Mol. Sci. 11 (2010) 3803–3845. [21] D.V. Lang, X. Chi, T. Siegrist, A.M. Sergent, A.P. Ramirez, Amorphous like density of gap states in single-crystal pentacene, Phys. Rev. Lett. 93 (2004) 086802–094805.

38

F. Bussolotti / Journal of Electron Spectroscopy and Related Phenomena 204 (2015) 29–38

[22] I. Salzmann, S. Duhm, R. Opitz, J.P. Rabe, N. Koch, Impact of low 6,13pentacenequinone concentration on pentacene thin film growth, Appl. Phys. Lett. 91 (2007) 051919–051921. [23] D. Käfer, C. Wöll, G. Witte, Thermally activated dewetting of organic thin films: the case of pentacene on SiO2 and gold, Appl. Phys. A 95 (2009) 273–284. [24] D. Käfer, L. Ruppel, G. Witte, Growth of pentacene on clean and modified gold surfaces, Phys. Rev. B 75 (2007) 1–14. [25] T. Hosokai, et al., In situ structural characterization of picene thin films by X-ray scattering: vacuum versus O2 atmosphere, Chem. Phys. Lett. 544 (2012) 34–38. [26] J.H. Kang, D. Da Silva Filho, J.-L. Bredas, X.Y. Zhu, Shallow trap states in pentacene thin films from molecular sliding, Appl. Phys. Lett. 86 (2005) 152115–152118. [27] T. Hosokai, A. Gerlach, A. Hinderhofer, C. Frank, G. Ligorio, U. Heinemeyer, A. Vorobiev, F. Schreiber, Simultaneous in situ measurements of X-ray reflectivity and optical spectroscopy during organic semiconductor thin film growth, Appl. Phys. Lett. 97 (2010) 063301–063304. [28] A. Hoshino, Y. Takenaka, H. Miyaji, Redetermination of the crystal structure of [alpha]-copper phthalocyanine grown on KCl, Acta Crystallogr. Sect. B 59 (2003) 393–403.

[29] W. Chen, S. Chen, H. Huang, D.C. Qi, X.Y. Gao, A.T.S. Wee, Molecular orientation dependent interfacial dipole at the F16 CuPc/CuPc organic heterojunction interface, Appl. Phys. Lett. 92 (2008) 063308. [30] R. Ye, M. Baba, K. Suzuki, K. Mori, Structure and morphology of CuPc and F16 CuPc Pn heterojunction, Appl. Surf. Sci. 254 (2008) 7885–7888. [31] A. Hinderhofer, Growth studies of organic heterostructures for photovoltaic applications (PhD dissertation), University of Tübingen, Germany, 2011. [32] See Supplemental Material at http://link.aps.org/supplemental/10.1103/ PhysRevB.89.115319 [33] F. Amy, C. Chan, A. Kahn, Polarization at the gold/pentacene interface, Org. Electr. 6 (2005) 85–91. [34] N. Koch, A. Vollmer, S. Duhm, Y. Sakamoto, T. Suzuki, The effect of fluorination on pentacene/gold interface energetics and charge reorganization energy, Adv. Mat. 19 (2007) 112–116. [35] M. Vos, S.G. Anderson, J.H. Weaver, Inelastic mean free paths for electrons at disordered interfaces, Phys. Rev. B 39 (1989) 3274–3279. [36] T. Schwieger, H. Peisert, M.S. Golden, M. Knupfer, J. Fink, Electronic structure of the organic semiconductor copper phthalocyanine and K-CuPc studied using photoemission spectroscopy, Phys. Rev. B 66 (2002) 155207.