Article pubs.acs.org/JPCC
Orientation of Nonplanar Molecules in Polycrystalline Layers from Infrared Spectra: Core-Chlorinated Naphthalene Tetracarboxylic Diimides Robert Lovrinčić,†,∥ Jens Trollmann,†,‡ Carl Pölking,† Jan Schöneboom,‡,§ Christian Lennartz,‡,§ and Annemarie Pucci*,†,‡ †
Kirchhoff-Institute for Physics, Heidelberg University, Germany InnovationLab GmbH, Heidelberg, Germany § BASF SE, GVC/E B009, 67056 Ludwigshafen, Germany ‡
S Supporting Information *
ABSTRACT: By means of infrared spectroscopic ellipsometry in the range of 350−5000 cm−1, the anisotropic dielectric functions of dichlorinated naphthalene diimide with fluoroalkyl chains (NDI-F), a novel high-performance n-type organic semiconductor, thin films were determined with high precision. Films of varying thicknesses (between 5 and 60 nm) evaporated in vacuum or spin coated from solution on different silicon wafers can all be described by the same anisotropic vibrational susceptibility, indicating weak dependence of the molecular orientation on processing procedure and film thickness for this material. It is shown how to determine the predominant molecular orientation by thorough comparison of ellipsometric measurements with calculations based on density functional theory. The method used additionally reveals subtle discrepancies between experiment and single molecule calculation which certainly arise from intermolecular interaction.
■
INTRODUCTION Huge research efforts have been devoted in recent years to organic semiconductors and electronic structures based thereon.1−3 Among the most persuasive advantages organic electronic devices have to offer are mechanical flexibility and solution processability.4,5 In most instances, the device performance depends on molecular orientation due to the highly anisotropic nature of π-conjugated molecules. In case of small molecules, the processing techniques (vacuum deposition or solution processing) can also strongly affect performance,6 usually yielding worse results for the potentially cheaper (and therefore desirable) solution processing. The intention of this work is 2-fold: first, in more general perspectives, we want to show how the orientation of novel nonplanar molecules in anisotropic layers (polycrystalline or amorphous with substrate-induced preferred orientation) can be derived reliably by combining results from infrared spectroscopic measurements and density functional theory (DFT) calculations. Second, we will use this method to study vibrational and structural properties of a novel high-performance n-type organic semiconductor, a core-chlorinated naphthalene diimide with fluoroalkyl chains 7 (NDI-F, C22H6Cl2F14N2O4, see Figure 1; the molecule is referred to as 3b in ref 7), for samples prepared by both vacuum deposition and spin coating. This material exhibits high stability in air, good processability, due to close π−plane distances large πstack overlap and high crystal density, and excellent field-effect © 2012 American Chemical Society
Figure 1. (a) Spherical angles θ,ϕ and τ,σ to characterize the relative orientation of the substrate surface, crystal basis, and E⃗ -field amplitude vector. θ,τ are polar angles (0 ≤ θ,τ ≤ π) and ϕ,σ the associated azimuthal angles (0 ≤ ϕ,σ ≤ 2π). τ is controlled during experiments. (b) Illustration how the trihedron from (a) represents the dimer basis.
mobilities (0.75 cm2 V−1 s−1 in air), making it a very promising material for applications in organic field effect transistors.7 It crystallizes in the monoclinic structure with a dimolecular basis. Received: November 17, 2011 Revised: January 11, 2012 Published: January 31, 2012 5757
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763
The Journal of Physical Chemistry C
■
For films prepared by vacuum evaporation, an edge-on conformation of the molecules in the film was found.7
■
(1)
where rp and rs are the reflection coefficients for light polarized parallel and perpendicular to the plane of incidence and Ψ and Δ are the standard ellipsometric parameters.12 The measured IRSE spectra have to be modeled to obtain a best-fit parametrized description of the dielectric function (DF) ε(ω) = ε1 + iε2
(2)
of the sample. ε1 and ε2 are the real and imaginary parts of the DF, respectively. The susceptibility of n vibrational modes contributing to the DF can be described in most cases by a sum of Lorentzian oscillators n
χ=
∑ i
Si ωi 2 ωi − ω2 − iωγi
EXPERIMENTAL DETAILS AND DATA ANALYSIS
NDI-F thin films were produced by either vacuum evaporation of the pure material or spin coating from tetrahydrofuran (THF) solution. The silicon substrates were highly resistive (low doping) Si wafers covered with native oxide. The evaporated films were prepared on 1 mm thick double-sided polished wafers, while spin coating was done on single-sided polished wafers with 525 μm thickness. The IRSE measurements at different angles of incidence Φ and a resolution of 2 or 4 cm−1 were performed with a Woollam IR-VASE rotatingcompensator ellipsometer, covering the range of 350−5000 cm−1. Modeling of the ellipsometric spectra was done using the WVASE-32 software package, which appropriately considers the layered structure of the samples. To obtain the orientation of the molecules with regard to the substrate surface, the experimentally observed vibrational modes were assigned to absorption peaks calculated for a single molecule by means of DFT. DFT results deliver not only peak positions of vibrational modes but also the directions of the corresponding dipole moments relative to the molecule. We developed a calculation tool that reproduces from DFT results the anisotropic DF obtained experimentally by varying the orientation of a basis molecule set relative to the substrate. The molecular orientation for which the agreement between experiment and DFT is best is then considered to be the preferred orientation of the molecules in the film. The optimization is performed on the basis of a freely adjustable number of DFT calculated oscillators that have to be allocated to experimental absorption peaks. The approach that we applied for the allocation was for it to find matches with as little shift in wavenumber required as possible, keeping in mind that DFT tends to be more reliable in oscillator energy than oscillator strength. The tool is built on the principle that how strongly an incident linearly polarized electromagnetic (EM) wave couples to an infrared-active vibrational mode can be determined from the overlap between the induced dipole moment associated with that mode of vibration and the electric-field vector of the EM wave. In a transmittance measurement with p-polarized light, this overlap can experimentally be controlled by the angle of incidence of the IR beam. In ellipsometry, one angle of incidence is already sufficient as the polarization state is changed during measurement (see Supporting Information for a comparison between transmittance and ellipsometric measurements). The uniaxial anisotropy of polycrystalline films effects that as the angle of incidence of the EM wave measured against the substrate normal increases, the absorption caused by the IR-active modes is either increased, decreased, or remains constant. This leads to characteristic peak behaviors that specify how the peak intensity observed in experiments changes with the orientation of the incoming electrical field. On the basis of a selection of oscillators, the aforementioned software tool computes the absorption peaks for various orientations of the molecular basis relative to the substrate. The molecular orientations that need to be inspected follow from the structure of the film (crystalline/polycrystalline), the symmetry and structure of the molecule, the crystalline basis (especially the mutual orientation of the molecules in the primitive unit cell), and the constraint that the induced IR dipole moments are only quantifiable as absolute value. In a polycrystalline layer, a given preferred orientation of the crystallites still allows for arbitrary rotations around the
INFRARED SPECTROSCOPIC ELLIPSOMETRY Infrared spectroscopic ellipsometry (IRSE) is a very sensitive technique to determine the dielectric function of thin films.8 The optical response of a material in the IR region provides information on free charge carriers,9 optical phonon modes,10 and crystal orientation and structure.11 Ellipsometry measures the complex reflectance ratio ρ = rp/rs = tan(Ψ)exp( −iΔ)
Article
(3)
with Si, ωi, and γi being the oscillator strength, resonance frequency, and damping of the i-th IR-active mode, respectively. For highly unordered systems, small fluctuations of the resonance frequencies caused by the amorphous structure have to be accounted for by additional Gaussian broadening.13 IRSE is most suitable to investigate the IR optical response of anisotropic single crystals.14 In the case of polycrystalline films, the random rotation of the crystallites around the surface normal results in an effectively uniaxial anisotropy with the optical axis parallel to the surface normal, even if the single molecules are orientated with an angle off the surface normal.11,15,16 IR excited molecular vibrations with dipole moment parallel (in-plane, ip) and perpendicular (out-of-plane, oop) to the substrate surface can be easily distinguished from IRSE spectra, as they influence rp and rs differently. For angles of incidence smaller than the Brewster angle, ip modes cause dip-up peaks in Ψ and oop modes dip-down peaks.16 The oop modes appear due to the Berreman effect.17 Their relative strengths compared to ip modes also depend on film thickness18 and refractive index of the substrate.
■
DFT CALCULATION OF VIBRATIONAL DATA Molecular geometries were optimized using the BP86 density functional,19,20 in combination with a triple-ζ basis set (TZVP) including polarization functions on all heavy atoms.21 The convergence criteria used were 10−8 au for the scf energy, whereas 10−7 au and 10−4 au were involved regarding energy changes and the Cartesian gradient norm during geometry optimization. The same level of theory was applied for the analytic calculation of vibrational frequencies. Resolution of identity techniques22,23 were used throughout. All calculations were carried out with the TURBOMOLE software package.24 Scaling factors for the vibrational frequencies were not used. 5758
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763
The Journal of Physical Chemistry C
Article
τ-behavior of all oscillators under investigation and will depend on σ,
substrate surface normal. To account for this in-plane isotropy, the tool averages over a large set of crystallites that are twisted against each other by small angles around this normal. To derive a figure of merit for each of these configurations, the peak intensities for a given oscillator are normalized with regard to the largest peak intensity measured and calculated for the same oscillator, respectively. The figure of merit, henceforth referred to as a rating, is based on the deviations between the normalized peak intensities from experiment and from spectral simulation. For comparison with ellipsometric measurements, this implies matching the normalized in-plane and out-of-plane intensities of spectral simulation with their respective counterparts obtained from modeling the DF with the DFT-based oscillator data. An additional scaling factor that accounts for the number of oscillators included in the analysis ensures that ratings will always lie within the range from zero to one, with a rating of zero indicating perfect agreement between experiment and simulation. For mathematical details, it is assumed that the behavior of Np oscillators with intensities Ai0, 1 ≤ i ≤ Np, are investigated at angles τj, 1 ≤ j ≤ Nτ, between the E-field and substrate surface. The peak intensity observed at angle τj, Ai(τj), will generally differ from Ai0 and is proportional to the square of the dipoletransition matrix element. The latter involves calculating the overlap between electric-field amplitude vector E⃗ (τj) inside the thin film and the induced dipole moment p⃗i of the ith vibration. Symbolically, Ai(τj) ∝ Ai0|p⃗i·E⃗ (τj)|2. Seeing that the induced dipole moments p⃗i are predicted by DFT, one can extract numerical values for the peak intensities both from the experimental spectra and from simulation, yielding two sets of intensities Aiexp(τj) and Aith(τj). The latter cannot be calculated directly since the orientation of the molecules within the thin film on the substrate is not previously known. To fully specify the relative orientations in the threecomponent system substrate−crystal basis−electric-field vector, four independent coordinates are needed. Assuming the substrate surface normal as the reference direction, any orientation of the molecular basis and the amplitude vector can be indexed by spherical angles θ,ϕ and τ,σ, respectively (see Figure 1 for definitions). A trihedron inscribed into the crystallite basis allows us to uniquely define molecular orientation. As a result, the configuration of the whole system is divided into the subconfigurations {θ,ϕ} and {τ,σ} of the crystal basis and field vector, respectively. The dependence of the peak intensities Aith on the system configuration is written as A ith ({θ, ϕ}, {τj , σ}) ∝ A i 0|pi⃗ ({θ, ϕ}) ·E ⃗({τj , σ})|2
Q ({θ, ϕ}, {σ}) =
×
Np
Nτ
i=1
j=1
1 Nτ ∑k ρk
∑ ρi ∑ [Ã ith ({θ, ϕ}, {τj , σ}) − Ã iexp (τj)]2 (5)
à i···
The symbol denotes that the peak intensities have been normalized as explained above. The statistical peak weight factor ρi can be used to emphasize certain oscillators in the analysis or just can be chosen constant. The prefactor 1/Nτ ensures that ratings Q({θ,ϕ},{σ}) will always adopt values between zero and one. Further simplification is achieved for polycrystalline layers, where ϕ differs from crystallite to crystallite. This is the same as stating that there is no unique value for σ that fits a particular system configuration. Replacing 1 Ã ith ({θ, ϕ}, {τj , σ}) → 2π
∫0
2π
à ith ({θ, ϕ}, {τj , σ})dσ (6)
in eq 5, i.e., using averages over all σ for the one obtains a rating Q({θ,ϕ}) that entirely depends on molecular orientation. Due to symmetry properties of the basis, it sometimes suffices to investigate Q({θ,ϕ}) for only a limited region in the θ−ϕ plane. For the NDI-F layer, the dimer basis of the crystallites features a mirror symmetry in addition to the inversion symmetry of the monomers. This implies that all physically distinguishable configurations {θ,ϕ} are located within the range 0 ≤ θ ≤ π/2 and 0 ≤ ϕ ≤ π. Aith,
■
RESULTS AND DISCUSSION Ellipsometry and Dielectric Function. Figure 2 shows measured and best-fit calculated Ψ and Δ spectra for an NDI-F film evaporated on SiO2/Si. The thickness of the film as determined by UV/vis ellipsometry (not shown) was 56 nm. Only the fingerprint range of the measured spectrum is shown (no further peaks were observed, except for weak C−H stretching vibrations). Both ip (dip-up in Ψ) and oop (dipdown) peaks are visible, indicating anisotropic optical properties. Measurements at different sample orientations (rotated around surface normal) always yielded the same results, indicating that the samples are optically uniaxial with optical axis normal to the surface. The DF was modeled with a component parallel to the substrate (εip(ω)) and one perpendicular (εoop(ω)). All measured molecular vibrations could be described by Lorentzian oscillators, adding a Gaussian broadening did not significantly improve the accordance between experiment and model. This result proves that short-range order exists in the layer. As expected for a material with Eg = 2.9 eV,7 no free charge carrier contribution to the DF was measured. Both real and imaginary parts of the modeled εoop(ω) and ip ε (ω) are shown in Figure 3. Twenty-two ip and 15 oop vibrational modes were found. The high-frequency dielectric constant was determined to be ε∞ = 2.4. The table in the Supporting Information summarizes the obtained best-fit parameters for all Lorentzian oscillators including their orientation and the DFT predicted resonance frequencies that they were assigned to. The deviations between measured
(4)
where the polar angle τ has been identified with the angle τj determined by the experimental setup. Equation 4 includes DFT results for Ai0 which are afflicted with a certain error. To prevent this error from affecting the comparison between simulation and experiment, the Aith({θ,ϕ}, {τj,σ}) and Aiexp(τj) are normalized independently from each other with respect to the largest peak intensity observed for any τj. For polycrystalline materials, this will either be the smallest or the largest of the τj. The “quality” (rating) Q of a molecular configuration {θ,ϕ} is estimated by evaluating the difference in 5759
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763
The Journal of Physical Chemistry C
Article
and calculated vibrational energies are in the range of 5−40 cm−1. Most measured peaks could be allocated to DFT predicted vibrations. We do not allocate peaks in the range of 870−980 cm−1 because all these oop modes are energetically close, making a correct allocation difficult. The deviations between experiment and theory are in the typical range of anharmonicity corrections, which are not explicitly included in the present DFT calculations. However, there is a partial error cancellation in the BP86-frequency calculations: harmonic frequencies are generally too small with respect to true harmonic frequencies, but the shift has the same sign and similar size as an anharmonic correction so an unscaled direct comparison with the experiment is generally meaningful.26 To investigate the influence of wet-chemical processing, NDI-F films spin coated from THF solution were measured as a comparison to the films evaporated in vacuum. Figure 4
Figure 2. Measured ellipsometric parameters Ψ and Δ (open symbols) for a 56 nm thick NDI-F film evaporated on silicon and best-fit modeled data (continuous lines). Corrections due to reflections from the substrate backside were taken into account. The film was modeled as optically uniaxial (optical axis normal to surface). All vibrational modes of the film can be adequately described by Lorentzian oscillators. The broad peaks in Ψ at 600 and 1100 cm−1 originate from two-phonon absorption25 and Si−O vibrations in the Si substrate, respectively.
Figure 4. Ellipsometric parameters measured at an incidence angle of 70° for NDI films spin-coated on silicon with two different thicknesses (as indicated) and calculated spectra using the DF obtained from the film evaporated on Si in vacuum. The accordance is obviously very good, though only the film thickness was adjusted for the model calculation. Note that the ψ-spectrum for the 12 nm film was shifted by +1° for clarity.
shows data measured on spin-coated films of 5 and 12 nm thickness and the corresponding modeled spectra. Note that the peaks are clearly above the signal-to-noise ratio even for the 5 nm film. For the modeled spectra, the DF obtained from the evaporated films was used, and only the film thickness was adjusted. We investigated several spin-coated films in the thickness range from 5 to 20 nm, and all could be described by the same DF as the evaporated films. This finding allows for
Figure 3. Real (ε1) and imaginary (ε2) part of the DF of NDI-F for polarization parallel and perpendicular to the substrate surface. The DF was obtained from a 56 nm thick NDI-F film evaporated on a Si substrate with native oxide (see Figure 2). 5760
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763
The Journal of Physical Chemistry C
Article
for a (1 × 1) μm2 area. For the vacuum-deposited film, seemingly crystalline facets with lateral dimensions in the micrometer range are visible, whereas the spin-coated film consists of smaller features (∼100 nm) that are less welldefined. While this difference was to be expected for the two different deposition techniques and different thickness, it seems somewhat surprising that both types of films can be described by the same anisotropic dielectric function, as stronger peak broadening and less pronounced optical anisotropy would have been expected for a more disordered film. This finding indicates that the short-range order is very similar for both films. We therefore suggest that the reduction in charge carrier mobility by 2 orders of magnitude observed for spin-coated films7 can be mainly attributed to scattering at grain boundaries. Orientation of Molecules from Comparison with DFT Results. To obtain the orientation of the molecules, we optimized the agreement between anisotropic DF and DFT results by rotating a basis set of molecules around two solid angles ϕ and θ with regard to the substrate plane. The optimization was performed based on six pronounced peaks for which the allocation to DFT-calculated modes could be done unambiguously. For every orientation, a figure of merit Q for the accordance was calculated. Figure 6 shows the resulting Q(ϕ,θ) for all six peaks individually. The ratings in plots (a)(b) and (d)-(f) are all consistent with ϕ ≈ 0° and θ ≈ 90° ((d) and (e) have limited discriminatory power). In contrast, the peak at about 1436 cm−1 cannot be brought in line with the other ratings, as displayed in Figure 6(c). Note that allocating this peak to other (neighboring) DFT-calculated modes could not solve this problem. For a possible explanation of this behavior, we first want to recall that the DFT calculation was performed for a single molecule only. The vibrations near 1440 cm−1 involve vibrations of the atoms in the molecular plane (see table in Supporting Information) with the resulting dipole perpendicular to the plane, whereas the other five peaks for our analysis belong to vibrations of only a few atoms on the molecule. As the π-plane distance is very small for this material (around 0.3 nm7), intermolecular forces could have a strong effect on the vibrational modes near 1440 cm−1. The problem
several conclusions: first, no residuals of the used solvent (THF) can be detected in the spin-coated films. Second, the orientation of the molecules in the film does not depend on the processing method. Moreover, in the investigated range of 5− 20 nm, the orientation is not changing with film thickness either. The orientation of the molecules in the first few layers (5 nm corresponding to roughly three layers) is of special interest for applications in bottom gate OFETs, as charge transport occurs mainly in the vicinity of the gate dielectric. Atomic-Force Microscopic Results. Additional to the ellipsometry study, atomic force microscopy (AFM) measurements were performed on the vacuum-deposited and spincoated films. Figure 5 shows typical height and phase profiles
Figure 5. Height (left column) and phase (right column) plots from tapping mode AFM measurements of evaporated (upper line, average thickness 56 nm) and spin-coated (lower line, average thickness 15 nm) NDI-F films on Si. The scan area was (1 × 1) μm2 in both cases. Note the different height scales.
Figure 6. Rating contour plots for the orientation (ϕ,θ) of the crystal basis relative to the substrate for six pronounced absorption peaks. Smaller rating means better agreement between experimentally derived and DFT predicted absorption at given molecular orientation. Note the different rating scales in (a) to (f). 5761
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763
The Journal of Physical Chemistry C
Article
methods.27 In such cases, our method can prove crucial to the understanding of the effect of molecular orientation on device performance, as it was recently shown in ref 27 for charge separation in organic solar cells. In such cases, the ellipsometric analysis yields more accurate and significant results than achievable by only comparing ordinary and extraordinary refractive indexes in the visible range.28,29 In addition, we showed how a thorough comparison between DFT and IR spectroscopic data can help identify effects of intermolecular interactions. For the specific system studied in this work, a predominant head-on orientation was found. NDI-F films prepared by vacuum deposition and spin coating on oxidized silicon wafers can be described by the same DF over a broad thickness range, indicating that no thickness-dependent change of molecular orientation occurs in spite of differences in grain size.
might hence be due to intermolecular interactions that are not accounted for in the single-molecule DFT calculations and needs further theoretical studies. Figure 7 shows the ip and oop imaginary parts of the dielectric functions and the DFT predicted positions of
■
ASSOCIATED CONTENT
S Supporting Information *
The obtained fit parameters and a comparison to results from transmittance measurements. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Present Address ∥
Department of Materials and Interfaces, Weizmann Institute of Science, Rehovot 76100, Israel. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work is part of the leading edge cluster Forum Organic Electronics, funded by the Bundesministerium für Bildung und Forschung. We thank Jochen Brill from BASF SE and U. Zschieschang and H. Klauk from the Max Planck Institute for Solid State Research for sample preparation.
■
Figure 7. ip and oop modes of ε2 from ellipsometry measurements (left ordinate scale) together with the absorption peaks as calculated by using DFT results (right ordinates scale in units of km/mol) and optimizing the orientation of a basis set of molecules on the substrate. The concluded orientation of the basis set of molecules on the substrate is schematically shown as the inset.
(1) Koch, N. ChemPhysChem 2007, 8, 1438−55. (2) Coropceanu, V.; Cornil, J.; da Silva Filho, D. A.; Olivier, Y.; Silbey, R.; Brédas, J.-L. Chem. Rev. 2007, 107, 926−952. (3) Koch, N.; Vollmer, A.; Salzmann, I.; Nickel, B.; Weiss, H.; Rabe, J. Phys. Rev. Lett. 2006, 96, 156803. (4) Gelinck, G. H.; et al. Nat. Mater. 2004, 3, 106−110. (5) Chua, L.-L.; Zaumseil, J.; Chang, J.-F.; Ou, E. C. W.; Ho, P. K. H.; Sirringhaus, H.; Friend, R. H. Nature 2005, 434, 194−199. (6) Kronenberg, N. M.; Steinmann, V.; Bürckstümmer, H.; Hwang, J.; Hertel, D.; Würthner, F.; Meerholz, K. Adv. Mater. 2010, 22, 4193− 4197. (7) Oh, J. H.; Suraru, S.-L.; Lee, W.-Y.; Könemann, M.; Höffken, H. W.; Röger, C.; Schmidt, R.; Chung, Y.; Chen, W.-C.; Würthner, F.; Bao, Z. Adv. Funct. Mater. 2010, 20, 2148. (8) Röseler, A. Infraredred Spectroscopic Ellipsometry; Akademie Verlag: Berlin, 1990. (9) Lovrinc̆ić, R.; Pucci, A. Phys. Rev. B 2009, 80, 205404. (10) Klevenz, M.; Wetzel, S.; Trieloff, M.; Gail, H.-P.; Pucci, A. Phys. Status Solidi B 2010, 247, 2179. (11) Schubert, M.; Bundesmann, C.; Jacopic, G.; Maresch, H.; Arwin, H. Appl. Phys. Lett. 2004, 84, 200−202. (12) Schubert, M. Infrared Ellipsometry on semiconductor layer structures; Springer-Verlag: Berlin, Heidelberg, 2005; Vol. 209.
absorption peaks for the corresponding optimized orientation of the molecules. For better comparability, a constant Lorentzian broadening was superimposed on the DFT peaks. The intensities are plotted as calculated, and no normalization was applied. The inset displays a schematic drawing of the concluded orientation: the basis molecules are standing upright on the substrate.
■
REFERENCES
SUMMARY
In summary, we have shown how to reliably conclude the orientation of complex organic molecules in a thin film from a combination of IR spectroscopic measurements and DFT calculations. This approach is especially advantageous if applied to molecules with a nonplanar structure forming X-ray amorphous layers with a preferred orientation but an insufficient degree of long-range order for diffraction-based 5762
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763
The Journal of Physical Chemistry C
Article
(13) Klevenz, M.; Wetzel, S.; Möller, M.; Pucci, A. Appl. Spectrosc. 2010, 64, 298−303. (14) Schubert, M.; Tiwald, T. E.; Herzinger, C. M. Phys. Rev. B 2000, 61, 8187−8201. (15) Schubert, M.; Bundesmann, C.; Jakopic, G.; Maresch, H.; Arwin, H.; Persson, N. C.; Zhang, F.; Inganäs, O. Thin Solid Films 2004, 455− 456, 295−300. (16) Hinrichs, K.; Silaghi, S. D.; Cobet, C.; Esser, N.; Zahn, D. R. T. Phys. Status Solidi B 2005, 242, 2681−2687. (17) Berreman, D. W. Phys. Rev. 1963, 130, 2193−2198. (18) Harbecke, B.; Heinz, B.; Grosse, P. Appl. Phys. A: Mater. Sci. Process. 1985, 38, 263−267. (19) Perdew, J. P. Phys. Rev. B 1986, 33, 8822−8824. (20) Becke, A. D. Phys. Rev. A 1988, 38, 3098−3100. (21) Schäfer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829−5835. (22) Eichkorn, K.; Treutler, O.; Oshm, H.; Hasser, M.; Ahlrichs, R. Chem. Phys. Lett. 1995, 240, 283−289. (23) Eichkorn, K.; Weigend, F.; Treutler, O.; Ahlrichs, R. Theor. Chim. Acta 1997, 97, 119. (24) Ahlrichs, R.; Bär, M.; Häser, M.; Horn, H.; Kölmel, C. Chem. Phys. Lett. 1989, 162, 165−169. (25) Shirley, E.; Lawler, H. Phys. Rev. B 2007, 76, 054116. (26) Reiher, M.; Liégeois, V.; Ruud, K. J. Phys. Chem. A 2005, 109, 7567−7574. (27) Ojala, A.; Petersen, A.; Fuchs, A.; Lovrincic, R.; Pölking, C.; Trollmann, J.; Hwang, J.; Lennartz, C.; Reichelt, H.; Höffken, H. W.; Pucci, A.; Erk, P.; Kirchartz, T.; Würthner, F. Adv. Funct. Mater. 2012, 22, 86−96. (28) Lin, H.-W.; Lin, C.-L.; Chang, H.-H.; Lin, Y.-T.; Wu, C.-C.; Chen, Y.-M.; Chen, R.-T.; Chien, Y.-Y.; Wong, K.-T. J. Appl. Phys. 2004, 95, 881−886. (29) Yokoyama, D.; Sakaguchi, A.; Suzuki, M.; Adachi, C. Appl. Phys. Lett. 2008, 93, 173302.
5763
dx.doi.org/10.1021/jp211077t | J. Phys. Chem. C 2012, 116, 5757−5763