Conference-ICSFS-16-Molecular Orientation of Copper Phthalocyanine Molecules on Crystalline and Amorphous Silicon Substrates

The orientation of semiconducting organic molecules, like in copper phthalocyanine (CuPc) thin films, is of importance for the electrical properties of organic/inorganic hybrid devices. Amorphous silicon films (a-Si), deposited onto crystalline Si-wafers by DC-pulsed magnetron sputtering, were utilised as substrates for the investigation. Crystalline silicon (111), with and without native oxide, was used as reference. Thin films of CuPc were fabricated using Organic Molecular Beam Deposition (OMBD) under high vacuum conditions (3 × 10−7 mbar). For one group of the substrates the native SiO2 layer was removed by hydrofluoric acid, and the surface was Hydrogen passivated. The average molecular tilting angle of CuPc molecules on both substrates was calculated from the anisotropic absorption coefficients as determined by ex situ Variable Angle Spectroscopic Ellipsometry (VASE). Significant differences in the molecular orientation depending on the substrate and its preparation are observed. For H-passivated silicon substrates it is found that the tilting angle changes from (66± 5)◦ to (83± 9)◦ when going from crystalline to amorphous substrates. Furthermore, CuPc molecules grew on silicon with native oxide in a “standing up” configuration at an angle of 90 ◦ relative to the substrate. Isotropic layers of CuPc were produced at substrate temperatures of around −155 ◦C. [DOI: 10.1380/ejssnt.2012.553]


I. INTRODUCTION
The electrical performance of copper phthalocyanine layers strongly depends on the orientation of the molecules on the substrate used [1].Especially for photovoltaic applications it is desired to deposit organic molecules in a "lying down" orientation to achieve the highest values of mobility and electrical conductivity perpendicular as well as of the optical absorption parallel to the substrate plane.For other devices, like Organic Field Effect Transistors (OFETs), a "standing up" configuration is preferred since a charge transport parallel to the interface is then wanted.Amorphous silicon is an interesting material for solar cells since it can be produced quite inexpensively.Its optical and electrical properties can be markedly influenced by varying the deposition conditions, for example the substrate temperature and the hydrogen flow rate during the deposition process.The optical absorption range of amorphous silicon can be extended by adding the advantageous absorption properties of organic molecules, like CuPc in the present case.
Optical techniques like Variable Angle Spectroscopic Ellipsometry (VASE) can be used to determine the complex refraction index ñ = n+ik.The absorption spectrum of CuPc shows two distinct absorption peaks in the Q band at energies around 2 eV [2].To calculate the average molecular tilting angle θ of flat molecules (approximated by a disc in the following evaluation) it is necessary to determine the integral of the extinction coefficient in the Q-band peak region (1.5 eV to 2.5 eV) within the molecular plane (A xy ) as well as out of the molecular plane (A z ) FIG. 1: Scheme of the hybrid silicon/CuPc structures [3]: The molecule is assumed to have only two degenerate transition dipole moments within the molecular plane.
With an additional comparison to isotropic films, error bars can be determined accurately.

II. EXPERIMENTAL
Figure 1 schematically shows the structure of the hybrid organic-inorganic systems used in this work.Crystalline (111)-oriented as well as amorphous silicon substrates were used.
The amorphous silicon layers were deposited by DC magnetron sputtering onto crystalline silicon.The sputtering technique has the advantage that the Hydrogen flow rate can be controlled, and therefore also the Hydrogen content within the a-Si layers.Introducing Hydrogen into amorphous silicon can saturate dangling bonds which results in a decreasing number of localized states in the middle of its band gap.Here two different kind of amorphous silicon substrates were used: either they were prepared in a pure Argon atmosphere on a nonheated substrate holder, or they were sputtered in a mixed Argon-Hydrogen-atmosphere onto a heated substrate (T s = 150 • C).For the latter ones the Hydrogen flow rate was kept at F H2 = 10 sccm, which resulted in a Hydrogen content of C H ≈ 12 at.%within the a-Si:H layers.
All substrates were cleaned in an ultrasonic bath of acetone, ethanol, and deionized water, for 10 minutes each.Furthermore, some of the samples (later marked as "Hpassivated") were dipped in 5 % hydrofluoric acid for 2 minutes.To remove the remaining F − ions from the surface of the samples were rinsed in dionized water.Afterwards they were dried by a N 2 gas flow.
The CuPc layers were deposited in vacuum at a gas pressure of 3×10 −7 mbar and a deposition rate of 0.3 nm min .The substrate was kept at room temperature for most of the samples.In order to deposit isotropic CuPc layers, the substrate holder was cooled to T s = −155 • C by liquid nitrogen.For this experiment only silicon substrates with natural surface oxides were used.
Directly after deposition the samples were measured ex situ with VASE in the energy range between 0.73 eV and 5.03 eV at incidence angles of 65 • , 70 • , and 75 • using a Woollam Inc. T-Solar M2000 ellipsometer.The spot size of the ellipsometer was measured to be 1.2 mm.Therefore, the ellipsometric parameters Ψ and ∆ are a measure for the average optical response of the sample.

III. RESULTS AND DISCUSSION
The measured ellipsometric parameters Ψ and ∆ were fitted with the software WVASE32 from Woollam Inc.The amorphous silicon layers were first measured without a CuPc layer on top in order to obtain their optical parameters using a Tauc-Lorentz-Oscillator model.Directly after depositing CuPc onto the substrates samples with three different thicknesses of the CuPc layer on each substrate type were measured.In that manner a good accuracy was achieved.In the ellipsometric modeling these three data sets were coupled.The data were first fitted in the transparent energy region from 0.73 eV to 1.1 eV with a Cauchy model layer representing the CuPc in order to determine the layer thickness.Afterwards an anisotropic (uniaxial) model of Gaussian oscillators was applied to obtain the complex refraction index of the CuPc layer.Employing Eq. ( 1) the average molecular tilting angle was calculated.

A. CuPc on different H-passivated silicon substrates
The comparison of the influence of the substrate on the average molecular tilting angle of CuPc was carried out for Hydrogen passivated silicon substrates.Figures 2  and 3 show the anisotropic extinction coefficients, k xy and k z , as well as the corresponding indices of refraction, n xy and n z .
As it can be seen from the figures, the out-of-plane component of the extinction coefficient is always bigger than the related in-plane component.This shows that the CuPc molecules adopt a standing up orientation relative to the substrate.Table I presents the calculated average molecular tilting angles.For the Hydrogen passivated crystalline silicon (111) substrate this value is comparable to earlier works of the group [4].The error bars were

Axy
Az by 10 %.It has to be taken into consideration that ellipsometry is less sensitive to the out-of-plane-component compared to the in-plane component of the complex refraction index.Therefore, the error bars are overestimating the error bars which are solely given by the fit.
The molecular tilting angle was slightly smaller for CuPc on Hydrogen passivated crystalline silicon when compared to amorphous substrates.The lower order of the amorphous silicon surfaces, especially in the long range, might result in decreased interactions between the CuPc molecules and the substrate.Non-hydrogenated amorphous silicon, deposited onto an unheated substrate, is known to have a high density of localised states within the band gap [5].These are unsaturated, broken silicon-silicon bonds which might interact with the CuPc molecules.The number of localised states can be reduced by an increasing Hydrogen content within the a-Si layers.In our case less dangling bonds might result in a decreased interaction between the CuPc molecules and the substrate.
The analysis of atomic force microscope (AFM) measurements showed that the rms-roughness was higher for HF etched silicon substrates compared to the precleaned ones (see Table II

B. CuPc on H-passivated silicon substrates compared to Si-substrates with natural oxides
The average molecular tilting angle θ of copper phthalocyanine was always smaller on Hydrogen passivated silicon substrates when compared to substrates with native oxide, regardless of the substrates used (c-Si(111), a-Si, a-Si:H) (see Table I).It seems that the natural oxide and remaining particles on the silicon surface reduce the interacting forces between the CuPc molecules and the substrate.Stronger molecule-molecule interaction compared to the molecule-substrate interaction then lead to a "standing" molecular orientation.Peisert et al. [6] also observed a standing up orientation of CuPc on oxidized silicon substrates.
For the substrates with natural surface oxide the ratio between the integral absorption in the z-direction of the molecule (A z ) was always bigger than twice the integral absorption in the xy-plane (A xy ).Then Eq. ( 1) does not hold anymore.The model is not taking into account the interaction between the molecules in a molecular crystal since it is based on single molecules.Furthermore, it was assumed that the transition dipoles within the molecular plane are equal [3], and that there is no coupling between the dipole elements.The more general form of Eq. ( 1) is given by [3]: where a and b are the diagonal elements of the 3D dipole matrix, θ and γ are the tilting angle with respect to the substrate plane and the angle of rotation of the disklike molecule around the central axis perpendicular to the molecular plane, respectively.Furthermore: Volume 10 (2012) Breyer, et al.The equation only holds for disk-like molecules with no dipoles out of the molecular plane.In case of an isotropic distribution of rotation angles γ Eq. ( 3) is only real for Az Axy ≤ 2. That is the same limitation as in the case assumed for this work (Eq.( 1)).In order to solve the equation for the ratios Az Axy determined from the measured VASE spectra for CuPc on oxidized silicon substrates a preferred angle γ has to be taken into consideration.The case γ = 0 • , but 0 < a 2 b 2 ≤ 1 (upper graph in Fig. 4), gives a lower limit of the calculated molecular tilting angle for a given ratio a 2 b 2 , as it can also been seen exemplary in Fig. 4 in the lower graph.The ratio a 2 b 2 = 0.7 was chosen for a better visualisation.
For the experimentally determined ratios of the integral absorption, Az Axy , the ratio a 2 b 2 has to be smaller than 1, otherwise Eq. ( 3) is not defined.To calculate the molecular tilting angle θ with the general formula it is necessary to have more information about the ratio a 2 b 2 .

C. Isotropic CuPc layers
Amorphous copper phthalocyanine layers were deposited by cooling the substrate holder to −155 • C by liquid nitrogen.The ellipsometric parameters Ψ and ∆ were fitted by an isotropic model of Gauss oscillators.The resulting isotropic extinction coefficients can be seen in Fig. 5. Furthermore, a uniaxial fit (not shown here) was used to have a comparison to the isotropic model.Within the uniaxial model the in-and out-of-plane component of the extinction coefficient were the same within the error bars.That leads to the conclusion that the CuPc molecules are not arranged, but in an isotropic order on both substrates, cooled c-Si(111) as well as cooled a-Si.The amorphous phase of organic molecules was observed for films deposited at temperatures below room temperature by other groups [7] (for CuPc), [8] (for pentacene).

IV. CONCLUSION
Hybrid organic-inorganic thin film heterostructures were prepared by Organic Molecular Beam Deposition.Ex situ VASE measurements were performed for CuPc layers on different substrates: c-Si (111), a-Si and a-Si:H.Smaller average molecular tilting angles were observed for CuPc on Hydrogen passivated silicon substrates when compared to silicon substrates with natural surface oxides.Furthermore, the tilting angle tends to be higher for CuPc molecules on Hydrogen passivated a-Si substrates when compared to a Hydrogen passivated c-Si (111) substrate.The deposition of CuPc layers at T s = −155 • C onto silicon substrates with natural surface oxides resulted in an isotropic CuPc layer, most probably an amorphous phase.Using AFM measurements it was observed that the average molecular tilting angle did not show a distinct trend with respect to the root mean square value of the roughness.

FIG. 3 :
FIG. 3: Anisotropic extinction coefficients and refraction indices for CuPc on Hydrogen passivated a-Si and a-Si:H.The a-Si:H substrate was deposited at Ts = 150 • and contains CH ≈ 12 at.%Hydrogen.Errors bars are not included in the figure for better visualisation.The highest error bar in magnitude was given by 5 %, though usually it was a factor of 10 smaller, i.e. around ±0.5 %.

FIG. 4 :
FIG.4: Variation of the average molecular tilting with the ratio a 2 b 2 of the diagonal elements of the 3D dipole matrix and γ for the three absorption ratios observed for CuPc layers on silicon substrates with natural surface oxides.For the plot θ vs. γ a ratio of a 2 b 2 = 0.7 has been chosen as an example.

Figure 4 2 b 2
Figure 4 presents the variation of the theoretically calculated average molecular tilting angle with varying ratio a 2 b 2 and γ for the Az Axy ratios.

FIG. 5 :
FIG.5: Isotropic extinction coefficient of CuPc on c-Si (111) substrates (light green dots) and on a-Si substrates (dark green dots).Both substrates had a natural surface oxide layer and were cooled to -155 • C during CuPc deposition.

TABLE I :
The table summarizes the average molecular tilting angles for CuPc layers on Hydrogen passivated silicon substrates and on silicon substrates with natural surface oxides.Substrates with natural surface oxide.The average molecular tilting angle was assumed to be 90 • for Az Axy > 2.

TABLE II :
The table summarizes the root mean square roughness (rms-roughness) of the substrates used for an area of (10x10) µm 2 .Topography images of the given area, which were used to analyze the roughness of the sample, were taken by atomic force microscopy.