Influence of film thickness and air exposure on the transport gap of manganese phthalocyanine

Phthalocyanines (Pcs) and porphyrins are chemically and thermally stable versatile organic molecules with tunable physical and chemical properties by e.g. changing the central metal ion. Having long spin lifetimes organic molecules are also promising candidates for future applications in molecular spintronics.¹ Among the transition metal phthalocyanines (TMPcs) series, manganese phthalocyanine (MnPc) is one of the most interesting due to its magnetic properties in bulk^2,3 and in films⁴ at low temperatures. MnPc has a high spin state of S = 3/2,^2,5 due to 3 unpaired 3d electrons which form a singly occupied highest molecular orbital (SOMO) originating from the 3d states of the central metal ion with 1e_g symmetry above the highest occupied molecular orbital (HOMO) distributed on the phthalocyanine ligand.^6,7 The magnetism of a single MnPc molecule was addressed by scanning tunneling microscopy demonstrating a practical way to modulate the Kondo resonance.⁸ Further theoretical studies on a single MnPc molecule sandwiched between two single walled carbon nanotubes reveal its nearly perfect spin filtering properties.⁶ It was already shown that the high spin ground state of MnPc S = 3/2 can be changed to S = 1 by adsorption on Bi(110) surface,⁹ or S = 1/2 by reaction with CO,⁹ or O₂.¹⁰ A higher spin state of S = 5/2 can, on the other hand, be obtained by doping with alkali metals.⁵ 

In the view of its application in spintronic devices, the knowledge of the electronic properties of MnPc at the interface with a ferromagnetic electrode with high spin polarization, such as cobalt,¹ is vitally important. In particular the evolution of the energy level alignment between the occupied and unoccupied molecular states and the substrate Fermi level as well as the evolution of the molecular transport gap with the films thickness are of importance for the device performance.

While for inorganic semiconductors the transport gap can be considered to be equal to the optical gap and can thus be determined from optical absorption or spectroscopic ellipsometry measurements, the strong excitonic effects present in organic molecules are responsible for the large mismatch between the optical and transport gaps of organic semiconductors.^11,12 The optical band gap can thus be smaller by up to 1.4 eV than the transport band gap.¹¹ Therefore, for a correct transport band gap determination, the combination of direct and inverse photoemission measurements should be employed, rather than optical absorption based methods (see e.g. Ref. 12).

The occupied electronic states of MnPc were intensively studied by ultraviolet photoelectron spectroscopy (UPS).^13,14,7 The unoccupied electronic states of MnPc were determined by inverse photoemission spectroscopy (IPS) by Yoshida et al.¹⁵ The latter study concentrated on the 3d levels of the central metal ion for different TMPcs and a detailed comparison with unoccupied electronic levels determined by X-ray absorption near edge structure was presented.¹⁶ To date, however, no data regarding the transport gap of MnPc were reported.

Regarding the optical gap of MnPc, contradictory values can be found in literature. Whereas the absorption for most Pcs is negligible in the energy range below typically 1.6–1.7 eV, MnPc exhibits absorption features also for lower energies.¹⁴ Optical absorption spectra presented by Kraus et al.¹⁷ correspond well to data derived from electron energy loss spectroscopy (EELS) and show a well resolved absorption peak centered at ∼0.5 eV. This would be an indication that the optical gap amounts to 0.5 eV. Other studies based on absorption measurements on thin films,¹⁸ or on sandwich structures,^19–21 suggest an optical band gap above 2.5 eV,^19,20 or even above 3 eV.^18,21 Even though spectroscopic ellipsometry is widely accepted as a reliable method for the determination of dielectric functions and optical gaps, ellipsometry measurements of MnPc films are still rare in literature with the exception of a study concentrating on film morphology without presenting optical properties of MnPc.²² 

Recent spin polarized UPS and IPS measurements on MnPc deposited on Co(001) reveal a high interface spin polarization at room temperature.²³ It is a purely ligand related interface phenomena concentrated at the first monolayer of organic molecules.²⁴ Focusing on possible applications for electronic and spintronic devices our study concentrates on electronic and optical properties of MnPc films with thicknesses in the nanometer range.

In this work, we report the determination of the energy level alignment between MnPc and a polycrystalline Co substrate and the evolution of the transport gap of MnPc films as a function of film thickness and subsequent air exposure by the combination of UPS and IPS. These results are complemented by in situ optical investigations by means of spectroscopic ellipsometry. Co was chosen as a substrate due to its attractiveness for organic spintronics applications provided its high spin polarization properties.^1,23

The UPS measurements were recorded at normal emission using a hemispherical analyzer, the excitation source being a He I discharge lamp (21.22 eV). The IPS spectrometer is a home built system operating in the isochromatic mode. A low energy electron gun having a mono-energetic, collimated beam was used as electron source. The current density (in the range of 10⁻⁶ A/cm²) is low enough not to damage the organic molecules. A Geiger-Müller tube filled with an Ar and Ethanol gas mixture having an MgF₂ entrance window is used as a fixed energy photon detector (10.5 eV). The resolution of the UPS and IPS setups is 0.20 eV and 0.50 eV, respectively, determined from the Fermi edge of clean, Ar sputtered Au and Co foils. Both setups are attached to the same UHV chamber with a base pressure in the low 10⁻¹⁰ mbar range. The samples were prepared in an additional chamber by organic molecular beam deposition (OMBD) thermally subliming MnPc (Strem Chemicals, Inc., 97% purity) from a Knudsen cell onto pure Co foil (Alfa Aesar®, 99.997% Puratronic®). Film thicknesses were monitored by a quartz crystal microbalance placed in the vicinity of the sample. The deposition rate was in the range of 0.2 nm/min. The base pressure in the preparation chamber was 8 × 10⁻¹⁰ mbar. The substrates were cleaned by Ar⁺ ion bombardment. The lack of contamination is proven firstly by the work function of the substrate (5.0 ± 0.1) eV as reported in literature,²⁵ and secondly by the clearly resolved Fermi edges both in the UP and IP spectra. The transfer system between the preparation and measurements chambers with a base pressure of 1 × 10⁻⁸ mbar ensures the transfer without breaking the vacuum. The final nominal film thickness of MnPc was kept below 20 nm in order to avoid charging. In addition, it was shown earlier that higher thicknesses do not yield further information in thickness dependence studies.²⁶ Data evaluation was performed by Unifit2010© software.²⁷ 

In situ SE measurements in the spectral range of (0.7 - 5.0) eV were performed using a T-Solar™ ellipsometer (based on the M-2000® Rotating Compensator SE) from J.A. Woollam Co., Inc. It is attached at a fixed angle of incidence 69.5° to a vacuum chamber having the base pressure of 5 × 10⁻⁷ mbar. The substrate was p-type, (111) oriented silicon with native oxide. Si is an ideal substrate due to its well-known optical properties. The final nominal film thickness was as large as 90 nm in order to obtain a precise optical characterization. All substrates were kept at room temperature during the experiments. The raw data were processed in the CompleteEASE© program using a layer model for data fitting including the Si substrate as a semi-infinite medium with a native oxide layer onto which the MnPc layer is modeled by an isotropic B-spline layer.²⁸ Even though Pc films usually have anisotropic optical properties,^29,22 and the surface of the films can be rough, these properties were neglected due to the measurements at fixed angle. To lower the correlation between the film thickness and the dielectric function, the data recorded at ten different thicknesses were coupled together during the fitting procedure.

Figure 1 shows the evolution of the UP and IP spectra with increasing film thickness. The width (W) of the UP spectrum is taken from the secondary electron cutoff (Fig. 1(a)) up to the Fermi edge for the Co substrate or to the highest occupied molecular orbital (HOMO) onset (Fig. 1(c)) for the MnPc film. The work function (Φ) of the substrate was derived by subtracting W_Co from the photon energy. Changes in the energy position of the secondary electron cutoff reveal the occurrence of an interface dipole (Δ). The ionization energy (IE) of MnPc and concomitant the position of the vacuum level (E_vacuum) was determined by subtracting W_MnPc from the photon energy. From the IP spectra (Fig. 1(d)) the lowest unoccupied molecular orbital (LUMO) onset position is determined. The difference between energy position of E_vacuum and the LUMO onset determines the electron affinity (EA). Schematic representations and detailed explanations on interface formation between organic and inorganic materials can be found e.g. in Ref. 12,30.

For exact determination of IE and EA both UP and IP spectra were fitted with Gaussian shaped peaks and cubic polynomial backgrounds.²⁶ The HOMO region is defined by two peaks: the large one having a pure ligand nature, and the smaller one closer to the Fermi level having Mn 3d character.⁷ The LUMO was fitted by a single peak. It is broad due to the convolution of the original LUMO with the Gaussian type experimental resolution of approximately 0.5 eV and was thus deconvoluted in order to reduce the influence of the instrumental broadening.¹² The inset of Fig. 1(b) exemplifies the onset determination for the spectra of a 20 nm thick film with backgrounds subtracted. The onset positions determined are marked in Fig. 1(c) and 1(d) by vertical bars. The difference between the HOMO and LUMO onsets determines the transport gap (E_t).^11,12 In the case of very thin films of 1 nm and 2 nm an extra Gaussian was used in the fit to compensate for the Co substrate signal at the Fermi level.

Due to image potential contributions at the metal-organic contact in the nanometer range a reduction of the barrier height is expected.^31–34,1 We observe the resulting opening of the transport band gap within the first 4 nm film thickness by a value of 0.15 eV to 0.20 eV for both HOMO and LUMO regions. At the MnPc/Co interface a constant interface dipole Δ = (−0.7 ± 0.1) eV is observed indicating charge transfer between the MnPc molecules and the metallic substrate, as already observed in the case of MnPc/Au.³⁵ This behavior is different from what was observed for e.g. Copper Phthalocyanine (CuPc) deposited on a semiconducting substrate, namely hydrogen passivated Si (H-Si). There a band bending type behavior for HOMO and LUMO positions was found at the semiconductor-organic interface with Δ increasing slowly with film thickness.²⁶ 

The electronic properties of the thickest MnPc film investigated here (20 nm) are summarized in Fig. 2(b) as an energy band diagram, and tabulated in Table I. These values should represent MnPc not affected by the substrate any more. For comparison, the values reported for one of the most investigate Pc molecule, CuPc, for which the 3d states of the central metal ion are well separate from the π states of the organic ligand,¹² are also given in the table. The much smaller transport gap of MnPc (1.2 ± 0.3) eV compared to CuPc (2.2 ± 0.3) eV is the result of additional electronic levels close to the Fermi level,⁷ both in the HOMO and the LUMO region. The presence of states with metallic 3d contribution having e_g symmetry is supported by DFT calculations.^7,10

The imaginary part of the dielectric function determined for a MnPc film, and plotted as green line in Fig. 3 resembles very well the absorption spectrum derived from EELS measurements in the energy range from 1 eV to 3 eV.^17,14 We therefore consider the value reported in Ref. 17 and 14 for the lowest energy peak as the optical gap E_opt = 0.5 eV. Due to the limited spectral range (0.7 – 5.0) eV in our SE measurements, we can only confirm by ellipsometry the presence of an absorption feature below 1 eV, but we cannot determine with good confidence its energetic position (Fig. 3). The difference between E_t and E_opt provides a value for the exciton binding energy of (0.7 ± 0.3) eV. This value is close to that reported for CuPc, i.e. (0.5 ± 0.2) eV,¹² or (0.6 ± 0.4) eV.¹¹ 

We attribute the optical absorption peaks at 1.38 eV and 1.74 eV in Fig. 3 as the direct transitions between the first two occupied electronic bands, i.e. the HOMO onset (0.45 eV) and HOMO-1 onset (0.95 eV), and the LUMO onset (−0.75 eV) as seen in our combined PES-IPS study (Fig. 2(a)). While the HOMO has significant Mn 3d contribution, the HOMO-1 has a pure ligand π character.⁷ 

It is well known that MnPc is reactive in air.¹⁰ To date, however, only the alteration of optical properties was monitored experimentally. Here the impact of air exposure (1 h) on the MnPc electronic structure was investigated by performing UPS-IPS measurements afterwards. In order to test the reversibility of the oxidation process the sample was re-measured after annealing it in vacuum up to 200 °C for 10 minutes. The corresponding spectra are plotted in Fig. 1 in full range and in Fig. 2(a) in a smaller energy range. As a general trend, all peaks decrease in intensity and become broader upon air exposure. In particular, the low energy peak at 0.9 eV in the UP spectrum disappears almost completely and shifts towards higher energies. As discussed above, this 1e_g state stems from the 3d-orbitals of the Mn central ion.⁷ The peak at 1.5 eV having π nature⁷ does not shift. Even though the peaks related to unoccupied states also weaken upon air exposure, they are still well visible and have an onset very close to the one prior to air exposure. These observations provide clear evidence for the hypothesis that the oxygen molecule attaches to the metal center of the MnPc molecule and thus the uppermost occupied 3d states of the Mn ion are affected by this oxygen attachment. This is in agreement with theoretical results of DFT calculations, which predict for a MnPc-O₂ complex formation a small binding energy of only 0.5 eV.¹⁰ The broadening of the peaks might be related to the appearance of additional occupied and unoccupied states related to the MnPc-O₂ complex,¹⁰ as well as by a distortion of the molecular geometry during the oxygen diffusion into the MnPc film.

In addition, we observe a lowering of the interface dipole and an increase of the HOMO onset position by 0.2 eV upon air exposure.

The UP spectrum of the sample annealed at 200°C for 10 min shows a recovery of the low lying peak denoting desorption of the oxidizing agent. It does, however, not have the same shape as the one of the pristine material probably due to a morphology change upon annealing.³⁶ In a second annealing step at 270°C the molecules completely desorbed from the substrate.

SE measurements performed on MnPc films after air exposure reveal marked changes in the dielectric function (Fig. 3, gray line). The changes due to air exposure were attributed mainly to molecular oxygen.¹⁰ The optical response changes fast due to air exposure, however, it requires significant time until it reaches saturation. The air exposure time for the results in Fig. 3 was one day.

The evolution of the optical spectrum (Fig. 3) is in concordance with the UPS-IPS measurements during air exposure. The disappearance of the HOMO peak at 0.9 eV is mostly related to the reduction of the peak at 1.38 eV in the optical spectrum. The second peak at 1.74 eV (Q band for phthalocyanines)²⁹ probably experiences an increase in oscillator strength due to a charge redistribution and/or deformation of the ligand upon the O₂ “adsorption”.

In summary, we report the combined UPS-IPS study to determine the transport gap of MnPc thin films on a ferromagnetic substrate. We observe the formation of an interface dipole Δ = (−0.7 ± 0.1) eV having a constant value for all film thicknesses as a result of charge transfer at the MnPc/metal interface. The transport gap increases by 0.3 eV from 1 nm to 20 nm MnPc film. The bulk transport gap, determined for the thickest films, amounts to (1.2 ± 0.3) eV, in agreement with DFT calculations, and is thus smaller compared to e.g. CuPc ((2.2 ± 0.3) eV)¹² due to presence of the metallic 3d SOMO states in the vicinity of HOMO and LUMO states.

We also determined the dielectric function of MnPc films in situ from spectroscopic ellipsometry measurements. This is in good agreement with previous absorption measurements by Kraus et al.,¹⁷ suggesting an optical band gap value of 0.5 eV. From E_t - E_opt an exciton binding energy of (0.7 ± 0.3) eV was determined. This value is in the same range as for other TMPcs.^11,12

Both types of techniques, i.e. combined UPS-IPS and SE measurements, confirmed previous suggestion that upon air exposure, mainly the metallic 3d electronic states are affected.¹⁰ 

We gratefully acknowledge financial support by the Deutsche Forschungsgemeinschaft (DFG) Research Unit FOR 1154 “Towards Molecular Spintronics”.